Classifying Nature-Inspired Swarm Algorithms for Sustainable Autonomous Mining
Joven Tan1*, Noune Melkoumian1, David Harvey2 and Rini Akmeliawati2
1Discipline of Mining and Petroleum Engineering, School of Chemical Engineering, The University of Adelaide, Adelaide, Australia
2School of Electrical and Mechanical Engineering, The University of Adelaide, Adelaide, Australia
Submission:June 03, 2024; Published:June 19, 2024
*Corresponding author: Joven Tan, Discipline of Mining and Petroleum Engineering, School of Chemical Engineering, the University of Adelaide, Adelaide, SA 5005, Australia
How to cite this article: Joven T, Noune M, David H, Rini A. Classifying Nature-Inspired Swarm Algorithms for Sustainable Autonomous Mining. Insights Min Sci technol.2024; 4(3):555636. DOI: 10.19080/IMST.2024.06.555636
Abstract
Over the resent decade, swarm-based algorithms have been utilized for automation in the mining industry. However, there is lack of understanding of their specific contributions at different stages of the mining process, in the broader sense. This paper classifies the optimization of mining lifecycle and swarm robotic systems based on reviewing nine nature-inspired algorithms for sustainable mining. Namely, the following swarm-based algorithms have been considered: Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), Firefly Algorithm (FA), Bat Algorithm (BA), Krill Herd Algorithm (KHA), Grey Wolf Optimizer (GWO), Salp Swarm Algorithm (SSA) and Grasshopper Optimization Algorithm (GOA). In this study, we conduct a systematic review of their impact on spatial organization, navigation, and collective decision-making, which in their turn can help to improve exploration accuracy, mine planning precision, and transportation efficiency. This research highlights the utility of nature-inspired algorithms that can contribute to specific mining phases and operations and should allow to achieve a more efficient and targeted mine optimization, greater environmental sustainability and improved mine safety.
Keywords: Mining Lifecycle; Mining Optimization; Swarm Robotics; Nature-inspired Algorithm; Sustainable Mining
Introduction
Mining can be regarded as a global resource supply, supporting the gross domestic product (GDP) of various countries, and having a key role in their stable industrial and sustainable economic development [1]. As the human population increases and urbanization accelerates, the demand for resources is also growing rapidly. The mining industry needs to change to meet this demand by seeking innovation and advancement in mining technologies to increase efficiency in all stages of mining, and energy efficiency [2]. This goes beyond providing environmental sustainability and a safe work environment for miners to operate. Mining innovations are becoming more complex involving smart sensors, remote operations, advanced power systems, reliability and resilience, robots operating in harsh environments such as hot and humid underground tunnels, high topology, and desert climates [3]. The prime example of the application of the current revolution in the mining technologies is the Gudai-Darri iron ore mine in Pilbara, Western Australia, operated by Rio Tinto and known as the SMART Mine. Rio Tinto has implemented autonomous mining transport trucks, trains and drill rigs, there is no need for manual labour at this mine site, and all engineers and operators work remotely. Each operator can control up to eight trucks one of which is equipped with a combination joystick, at the Mining Control Station located at approximately 110 kilometres northwest of the Gudai-Darri, at Newman, Western Australia [4]. These innovative technologies have transformed traditional mining to automated robotic mining, integrating machine learning, robotics, and remote operations to maximize mine operational efficiency, reduce human safety risks and mitigate environmental sustainability. Innovation and transformation in the mining industry have increased iron ore production at the mine from 159 tons in 2000 to 836 tons in 2020 [5]. The Rio Tinto Financial report states that the integration of automation technology can not only increase productivity but also reduce operating and maintenance costs [5]. The success of this Rio Tinto operation has inspired further research into perfecting autonomous mining operations through fully automated remote control without the need for humans, while swarm robots can leverage nature-inspired swarm algorithms for decision-making and consensus building.
In this paper, nine nature-inspired swarm algorithms have been systematically analyzed through theoretical and mathematical models based on swarm behaviours and collaborations found in nature. The results from these analyses can contribute to the wider implementation of swarm robotics in mining and hence, achieving improved mining optimisation and operations. This research offers a structured and comprehensive framework of swarm robotics integrated mining that can be helpful for the further advancement of future mining technologies.
Analysis of Swarm-Based Bioinspired Algorithms
This section analyses theoretical and mathematical models of nine nature-inspired swarm algorithms, focusing on understanding how corresponding animal and insect models from nature collaborate, cooperate, and survive in large groups.
Ant Colony Optimization (ACO) Algorithm
The Ant Colony Optimization (ACO) algorithm is a metaheuristic algorithm inspired by the foraging behavior of ants, and was proposed by Dorigo in 1992 [6,7]. The ACO algorithm illustrates the concept of stigmergy, utilizing indirect communication through ant pheromone experiments to imitate ants’ pathfinding techniques, and uses a population-based approach in pheromone search experiments to solve optimization problems [8]. ACO schematic diagram [9] and finite state flow chart [10] includes initializing algorithm parameters and agents, building solutions through state transition rules, and refining these solutions through fitness-based evaluation, as shown in (Figure 1).
Equation (1) on the probabilistic decision-making equation (Pkij) ,integrates the pheromone trace (τijα) and the heuristic desirability (ηijβ) between nodes for the search and harvest process. The uniqueness of the ACO algorithm lies in the dynamic feedback mechanism (τij) in Equation (2). Using the pheromone evaporation (ρ) and ant deposition (τijk) to update the pheromone trial can avoid local optimality and converge to the optimal solution. Through a complex interplay of exploration, exploitation, and adaptive feedback, ACO leverages biomimetic approaches to efficiently solve computational problems.
![Click here to view Large Figure 1](images/IMST.MS.ID.555636.G001.png)
Particle Swarm Optimization (PSO) Algorithm
The particle swarm optimization (PSO) algorithm is a metaheuristic algorithm inspired by the social behavior of birds and fish and was proposed by Eberhart and Kennedy in 1995 [11,12]. The PSO algorithm illustrates the interactions and collaborations of fish schools and birds to optimize problems through social learning. PSO schematic diagram [13] and finite state flow chart [14] involve initializing the agents in the swarm and iteratively updating their speed and position using the individual experience and consensus in the swarm, to update to the final optimal position through the global position, thereby effectively navigating to the optimal solution, as shown in (Figure 2).
Equation (3) on the updated speed mechanism (Vit+1) integrates the inertia of particles with cognitive and social components through personal memory and global consensus. The acceleration coefficients (c1 & c2) and random variables (Ut1) and (Ut2) are used to balance global and local searches. (Pit+1) in the position updates Equation (4) incorporates the updated speed, and the optimal position is refined through the fitness value. Through complex collective intelligence and social learning, PSO can adapt to involve complex optimization strategies and search for the global optimum.
![Click here to view Large Figure 2](images/IMST.MS.ID.555636.G002.png)
Artificial Bee Colony (ABC) Algorithm
The artificial bee colony (ABC) algorithm is a swarm-based metaheuristic algorithm inspired by the foraging behavior of bees and was proposed by Karaboga in 2005 [15,16]. The ABC algorithm illustrates the social role allocation of bees, such as onlooker bees, employed bees, and scout bees. The mission of the scout bees is to explore the quality and location of nectar and inform the onlooker bees, who will decide on the nectar collection location through decision-making, and the recruits or employed bees will do the harvesting [17]. The ABC schematic diagram [18,19] and finite state flow chart [20] involve initializing the agent, searching for suitable nectar sources, and refining the solution through greedy selection, as shown in (Figure 3).
Equation (5) on the selection probability (Pi) combines the selection of observation bees with the adaptability of food sources, and selects sources with higher quality. The new exploration position (X*ij)in the Equation (6), implements the perturbation vector (∅ij) to modify the position of the bee to find the nearby food sources. The updated positions (Xkt+1,Ykt+1) in Equation (7), illustrates that the hired bees use new information and old locations to refine the optimal location. Through sophisticated position updating and forging strategies, ABC can contribute to complex optimization tasks and optimization problems.
![Click here to view Large Figure 3](images/IMST.MS.ID.555636.G003.png)
![Click here to view Large Figure 4](images/IMST.MS.ID.555636.G004.png)
Firefly Algorithm (FA)
The Firefly Algorithm (FA) is a meta-heuristic algorithm inspired by the attraction of fireflies’ bioluminescence and was proposed by Yang in 2007 [21, 22]. The FA algorithm illustrates the attraction of fireflies to different bioluminescence intensities (solution qualities) to find the best solution in the search space. FA schematics diagram [23] and finite state flow chart [24] involve initializing a population of agents in a search space and then updating their positions by light intensity to evaluate fitness, as shown in (Figure 4).
Equation (10), the updated position equation (Xjt+1) , is integrating optimal position updates based on light intensity (I) and attraction (β). Light intensity and attraction decrease with distance [25]. With complex position updates through optical attraction, FA can help to solve optimization problems to search for local and global optimal solutions.
Bat Algorithm (BA)
The Bat Algorithm (BA) is a meta-heuristic algorithm inspired by the echolocation behavior of microbats and was proposed by Yang in 2010 [26]. The BA algorithm illustrates the foraging behavior of bats and uses sound pulses to detect food or prey. This can also be used to avoid obstacles caused by bats’ poor vision and find the optimal solution in the search space [27]. BA schematics diagram [28] and finite state flow chart [29] involve initializing the agent and then evaluating position and velocity updates based on local search and pulse return feedback (e.g. pulse rate and loudness), as shown in (Figure 5).
Equation (11), the velocity update (Vit), combines the frequency of bat echolocation with the global optimal position of the colony. The new velocity update is applied to the position update (Xit) in Equation (12). By adjusting the frequency and intensity of the pulse and reflection, and the bat’s sensory modulation of prey detection through amplitude modulation (A), the echolocation accuracy and the optimal solution can be improved. The new optimal position (Xnew) is obtained by Equation (13). By performing complex position updates via pulse frequency and intensity adjustments, BA can help to solve optimization problems on navigating and searching in multidimensional space problems.
![Click here to view Large Figure 5](images/IMST.MS.ID.555636.G005.png)
Krill Herd (KH) Algorithm
The krill herding (KH) algorithm is a swarm-based metaheuristic algorithm inspired by the krill swarm behavior and proposed by Gandomi and Alavi in 2012 [30]. The KA algorithm illustrates the induced motion, foraging movement, and physical diffusion for navigating and searching in multidimensional spaces [30, 31]. The KH schematic diagram [32] and finite state flow chart [33] involve initializing the agent and fitness, followed by three motion evaluations to search for optimal solutions and maintain cohesion within the group, as shown in (Figure 6).
Equation (14), the perceived distance between krill (ds,i) is calculated by combining the three movements with the global movement speed i (dXi/dt) in Equation (15), and to calculate the best new position Xi(s + Δs) Equation (16) is used. This movement mechanism demonstrates the social dynamics of krill movement of searching and navigating in multi-dimensional space to effectively converge, avoid falling into local optima, and maintain group cohesion [33].
![Click here to view Large Figure 6](images/IMST.MS.ID.555636.G006.png)
Grey Wolf Optimization (GWO) Algorithm
The Grey Wolf Optimization (GWO) algorithm is a metaheuristic algorithm inspired by the social hunting behavior and hierarchical structure of grey wolves and was proposed by Mirjalili and Lewis in 2014 [34]. The GWO algorithm illustrates the social structural roles of Alpha, Beta, Delta, and Omega used in solving optimisation problems in wolf siege and hunting strategies [35]. GWO schematic diagram [34] and finite state flow chart [36] involve initializing each agent in the search space, then updating the position according to the movement of Alpha, Beta and Delta, and further recalculating the distance according to the hierarchical role to adjust the strategy, as shown in (Figure 7).
Equation (17), the position of the leader wolf ) (D →uses the coefficient vectors ) (A →and ) (C →to integrate hunting movements and to update the positions of other pack members X→(t+1) in Equation (18). These hunting mechanisms illustrate the social dynamics of grey wolves and can be used in complex multidimensional optimization problems with efficient convergence capabilities.
![Click here to view Large Figure 7](images/IMST.MS.ID.555636.G007.png)
Salp Swarm Algorithm (SSA)
The salp swarm algorithm (SSA) is a meta-heuristic algorithm inspired by the leader-follower formation of salp chains and was proposed by Mirjalili in 2017 [37]. The SSA algorithm illustrates the formation of a leader-follower, in which the salp leader guides the group to search for plankton, and the followers update their positions based on the salps ahead to search in multidimensional space. The SSA schematic diagram [37] and finite state flow diagram [38] involve initializing the agent and fitness, followed by position updates based on the leading salp to maintain the formation, as shown in (Figure 8).
Equations (19) and (20), show the updated positions of the leader and follower, and the salp motion is determined by integrating the Newtonian motion principle. The leader salps update food location, search range and randomness (yi), (ubi) and (lbi) respectively, to search in multi-dimensional space. These chain-forming mechanisms allow SSA algorithm counters to optimize tasks and avoid local maxima.
Grasshopper Optimization (GOA) Algorithm
The grasshopper optimization (GOA) algorithm is a metaheuristic algorithm inspired by grasshopper swarming and foraging behavior and was proposed by Saremi, Mirjalili, and Lewis in 2017 [39,40]. The GOA algorithm illustrates the three-motion account for social interaction, gravity, and advection in navigating and moving in large groups [41]. The GOA schematic diagram [42] and the finite state flow chart [28] involve initialising the agent and fitness, and then conducting three motion evaluations to search for the optimal solution and maintain cohesion within the group, as shown in (Figure 9).
The movement of the grasshopper (X_i) in Equation (22) is determined by integrating the social interaction, gravity, and wind direction, 〖(S〗_i),〖(G〗_i) and (A_i) respectively, and Euclidean distance between the grasshoppers. These grasshopper swarm dynamics mechanism allows the GOA algorithm to search in multidimensional space.
![Click here to view Large Figure 8](images/IMST.MS.ID.555636.G008.png)
![Click here to view Large Figure 9](images/IMST.MS.ID.555636.G009.png)
Swarm Algorithms in Robotics
Currently, Swarm robotics is being successfully applied to solve problems in various fields, such as in agriculture, where SAGA uses bee foraging models for field mapping and weeding [43], or in construction, where TERMES uses the termite colony concept to build autonomous building structures [44]. The applications of swarm robots are impressive, but the knowledge behind how swarm robots collaborate, reach consensus, communicate, and share information, and take inspiration from nature is an important subject to understand.
According to Brambilla’s 2013 study on the classification of swarm robots, the behavior of swarm robots can be divided into four categories: spatial organization, navigation, decisionmaking, and miscellaneous [46,47], as shown in (Figure 10). This classification facilitates designing swarm robots to meet specific operational needs. In this study, a comprehensive framework will be established to further analyse the research on nine swarmbased algorithms considered in this paper, that can be applied to swarm robot behavior classification.
![Click here to view Large Figure 10](images/IMST.MS.ID.555636.G0010.png)
Spatial organization
Spatial organization refers to the collective intelligence of swarm robots that can interact and organize within allocated areas to aggregate, execute patterns, assemble, or collect objects [46,47]. The PSO algorithm illustrates the aggregation and pattern formation of fish schools that swim to avoid predators [48], similar to the KH algorithm where large groups of krill gather and form large swarms to avoid predators and stay cohesive [49]. The GOA algorithm illustrates the aggregation of group dynamics of grasshopper movements in large groups [40]. The ABC algorithm illustrates object clustering based on how bees cluster themselves and the nectar in their hives [50]. The FA algorithm illustrates the clustering toward high light intensity [51]. The SSA algorithm illustrates the pattern formation of how salps form chains [52].
Navigation
Navigation refers to the collective intelligence of a swarm of robots that can determine a known location and guide themselves or other robots to a specific location. Namely, it includes swarm exploration, movement under set coordinates, swarm transportation and localization [46,47]. The foraging behavior of the ACO, ABC, KHA, SSA, BA, GOA and GWO algorithms reflects collective exploration and localization, and they tend to explore and locate food in large groups to achieve continuous harvesting. The FA algorithm’s bioluminescent light attraction illustrates collective localization, where all fireflies follow basic rules to attract towards higher light intensities [53]. The synchronized movement of the PSO [54], KHA [55] and GOA [40] algorithms illustrate coordinated movement, with the entire colony having local and global positions to adjust to with the aim to avoid predators, similar to SSA, where salps always adjust their position as the position of the leading salp is updated [37]. ACO’s unique harvesting behavior exemplifies collective transportation, with ants tending to pick up heavier or larger objects and place them into the hive.
Decision making
Decision-making refers to the collective intelligence of a swarm of robots that is capable of reaching consensus, allocating tasks amongst the swarm, detecting failures, sensing and adapting to the surrounding environment, performing synchronized tasks, and controlling swarm size [46,47]. The collective swarm behavior of the ACO, PSO, ABC, FA, BA, KHA, GWO, SSA and GOA algorithms illustrates the tendency of consensus mechanisms and collective perception to perceive and make consistent decisions within large groups to reach consensus, such as ants based on pheromone trails or bees based on waggle dance to harvest food; birds, fish, and krill use global motion to update their positions; fireflies and bats receive brightness and echoes to determine direction, and wolves update their positions based on social structure.
The specific role assignments for ACO, ABC and GWO illustrate the distribution of specific tasks within the group. Leafcutter ants separate tasks such as cutting and transporting leaves and cultivating fungi [56]; bees separate tasks such as exploring nectar sites, selecting high-quality nectar, and collecting nectar [15]; wolves separate tasks such as exploring, surrounding prey, attacking, and hunting. The self-diagnostic capabilities of SSA and KHA illustrate collective fault detection, with old or damaged salps in the chain detaching from the chain or changing positions with more fit salps [57], similar to krill who sense and moving toward healthy krill position [30], and both mechanisms are able to sense neighbouring agents to maintain formation. The FA algorithm illustrates synchronisation by the blink technique, whereby male fireflies tend to respond to female fireflies with synchronized flashes [58].
Miscellaneous
Under miscellaneous category is referred to the collective intelligence of a swarm of robots that can heal themselves, replicate their members, and interact with humans. The ACO’s rerouting strategy illustrates self-healing, and when pheromone trial is hindered, the ants execute a self-healing rerouting strategy [59, 60].
Swarm formation control
Swarm robot formation control refers to a swarm of robots collaborating while maintaining a predetermined formation [61]. There are two types of formation control for swarm robots: centralized control and decentralized control. Centralized control relies on a single command or master switch to control all robots, which is more efficient, but has limitations in scalability and adaptability [62]; on the other hand, decentralized control allows a group of robots to use behaviours inspired by nature, according to interactive and local information from nearby neighbouring agents or environments automatically make decisions, making them more scalable, resilient, and adaptable in unknown environments [63].
Virtual structure formation control was proposed by Lewis and Tan in 1997 [64]. This is a type of centralized control that connects all robots with virtual structures to maintain geometry and provide coordinated motion. However, it has limitations due to its susceptibility to single points of failure and the difficulty of formation adjustment. The SSA algorithm illustrates virtual structure formation control, in which a chain of salps guides the entire formation through a virtual structure connecting the leading and following salps in the chain formation [65].
Behaviour-based formation control was proposed by Balch and Arkin in 1998 [66]. This is a type of decentralized control inspired by nature’s forming behavior. This structured network of interactive behaviours receives information from other robots and derives decisions from a behavioural coordinator, requiring less group communication load than that for the centralized control. However, its limitation lies in formation convergence. Swarm algorithms such as ACO, PSO, ABC, FA, BA, GWO, KHA, SSA and GOA illustrate behaviour-based formation control [63]. Each algorithm has its own unique way of communicating and exchanging information with neighbouring agents, such as pheromone trial in ants, position adjustments of neighbouring agents in fish, fireflies, bats, wolves, krill, salps and grasshopper, and waggle dances in honeybees.
Leader-follower formation control was proposed by Desai, Ostrowski, and Kumar in 2001 [67]. This is a type of centralized control where all robots rely on or are controlled by a leader, with followers adjusting and obeying commands accordingly to provide superior, simplified, and complete control. However, its limitation is that the leader’s failure can lead to the failure of the entire formation or system. The GWO [68] and SSA [69] algorithms illustrate leader-follower relationships, where the leader wolf and the leader salp will be the main controllers and all followers will obey and follow the leader’s decision.
Graph-based formation control was proposed by Desai, Ostrowski, and Kumar in 1998 [70]. This is a type of decentralized control where all robots are modelled in a mathematical graph and each robot is treated as a vertex with edge connections representing the flow of information from one agent to another. The ACO algorithm illustrates graph-based formation control, where ants use graph-based functions to detect pheromone intensity to find the shortest path [71-73].
Artificial potential formation control was proposed by Khatib in 1986 [74]. This is a type of decentralized control where all robots interactively control the distance and spacing between adjacent agents using attractive and repulsive forces. The KHA algorithm illustrates an artificial potential formation control, where each krill has a sensed distance to adjust its position relative to the adjacent krill and uses attraction to move toward the krill with a higher fitness, and uses repulsion to maintain the distance between krill to maintain in a large group [75]. Similar to the GOA algorithm, each grasshopper is attracted and repelled by the group movement when it is too close to another grasshopper to maintain its comfort zone [39, 40].
Swarm behaviour and control
Although swarm robotics has been successfully applied to various fields and industries, such as construction, agriculture, entertainment, medical care, etc., , there is a very limited research on the application of swarm robotics for the machinery in mining operations. While some mines are already applying remote operating systems to control autonomous haul trucks, and other processes at a mine site, the decentralized control where robots can collaborate and communicate with each other to perform mining tasks , thus achieving a fully automated mining systems is still in development. This research examines further integration of swarm algorithms into specific mining operations with the aim to improve overall mining productivity, safety, and environmental sustainability. Table 1 summarises the comprehensive study of the nine nature-inspired swarm algorithms considered in this paper, each grouped into specific classifications of swarm behavior and swarm robot formation control.
![Click here to view Large Table 1](table/IMST.MS.ID.555636.T001.png)
Applications of Swarm Algorithms in Mining
Mining is the process of extracting natural resources or minerals from the earth. The mining life cycle [76] begins with the exploration phase, where mineral deposits are explored, mining sites are identified based on core log data from drill samples, and the presence of valuable minerals is assessed. The next stage is the planning stage, where pre-feasibility studies and feasibility studies are conducted to determine the economic feasibility of the project, considering market demand, mining methods, market prices, ore quality, environmental impacts, and regulatory requirements. If the project is feasible, the construction phase follows that includes the construction of mining facilities, roads, mine trucks, tailings dams and more. Mining operations involve extraction and secondary processing of minerals for sale. The final phase is the mine closure that involves reclamation and restoration, as well as continuing to address environmental impacts, as shown in (Figure 11).
Mine exploration and assessment
Mine exploration phase includes core drilling, geological analysis, and identification of mineral sites and deposits. A variety of swarm algorithms have been integrated into the mine exploration stage to improve exploration accuracy and efficiency. Nhleko and Musingwini’s 2019 PSO study demonstrated how PSO algorithms can be used in conjunction with surveying techniques to delineate underground stopes, further improve resource extraction efficiency and operational safety [77]. Optimisation of mine mapping processes demonstrated PSO’s ability to improve the efficiency and safety of underground mining exploration. The study by Jafrasteh and Fathianpour in 2017 details the fuzzy artificial bee colony (FABC) algorithm for evaluating three dimensional ore characteristics to optimise ore body size, such as spatial positioning, azimuth, and inclination of exploration boreholes [78]. Compared with traditional optimisation techniques, the FABC algorithm reduces the kriging variance by adjusting parameters, greatly improving the accuracy of mineral resource estimation, and further improving the quality of mineral resource assessment and exploration. The precise positioning of the borehole improves the accuracy of discovering ore bodies and reflects ABC’s ability to improve mine exploration efficiency through effective core drilling. A 2023 study demonstrated the implementation of the Bat Algorithm (BA) in a hybrid support vector machine (SVM) for improving the accuracy of coppergold mineralisation mapping, showing a 10% improvement in accuracy compared to traditional methods, with an average lower square value error of 6.6% and the accuracy being 94.3% [79], thus further improving the accuracy of mineral exploration. The improved accuracy of mineralisation maps demonstrates BA’s ability to leverage accurate geological data to improve mine exploration and analysis efficiency. Research by Saremi, Mirjalili, and Lewis in 2017 details the implementation of the GOA algorithm in mine exploration [39], using grasshopper swarming behavior to identify and pinpoint valuable mining areas by balancing global and local search mechanisms, and iterating over time to make improvements. Effective mine mapping demonstrates GOA’s ability to enhance mineral exploration. The comprehensive analysis on swarm algorithm into mine exploration and assessment stages has been reviewed and classified in Table 2.
![Click here to view Large Table 2](table/IMST.MS.ID.555636.T002.png)
Mine planning and design
Mine planning and design phase includes mine layout, mining method selection, costs, operational assessment, mine safety and environmental sustainability. Studies show that the implementation of the PSO algorithm can be successfully used to determine efficient mine operations by improving open pit mine layout [80]. The PSO algorithm integrates variants that transform traditional block-level scheduling problems by iteratively improving these solutions using greedy heuristics to account for constraints and uncertainties. The refinement of the open pit mine layout demonstrates PSO’s ability to enhance mine planning and scheduling problems. Korzeń and Kruszyna elaborated on the implementation of the ACO algorithm in the Wrocław underground railway project in their 2023 study [81], using the foraging behavior of ants to search for the best route, by considering the dense population, heavy traffic nearby, and calculations of public transport routes. The selection of the optimal route demonstrated the ACO’s ability to enhance decision-making and optimise mine route planning. Khan’s 2018 study demonstrated the application of the BA algorithm in mine planning to address long-term scheduling challenges under grade uncertainty [82]. The Bat algorithm incorporates uncertainty and shows higher efficiency than traditional commercial software. The generation of effective solutions illustrates BA’s ability to enhance mine design and scheduling. Research by Tolouei and Moosavi in 2021 demonstrated the implementation of the GWO algorithm using the augmented Lagrangian relaxation (ALR) method to carry out long-term production scheduling (LTPS) in open pit mine design [83]. The hybrid ALR-GWO model showed more advancement than the traditional method, achieving a net present value increment of 13.39%. The improvement in net present value demonstrates GWO’s ability to enhance mine planning through improved economic outcomes. Research in 2023 details the application of the SSA algorithm in an extreme learning machine (ELM) model to improve predictions of ground vibration intensities caused by explosions in the Coc Sau coal mine [84]. The hybrid SalSOELM model recorded 216 blasting performances and surpassed the traditional model with an accuracy of 90.5%. Enhanced peak particle velocity predictions demonstrate SSA’s ability to enhance mine blast planning and improve mine safety. The comprehensive analyses on swarm algorithm applications in mine planning and design stages have been reviewed and classified in Table 3.
Mine operation and construction
Mine operation and construction phases include mine extraction for primary production and mine processing for secondary production. Research conducted at the Shenbao open pit mine in 2020 demonstrated the use of the PSO algorithm to optimize the mine equipment mismatch problem [85]. The PSO algorithm combined statistical analysis and uses triangular fuzzy numbers to achieve scheduling randomness, reducing the number of mining trucks, truck scheduling, transportation costs and queuing time. The optimization of mining equipment reflects PSO’s ability to enhance mine operations and construction in terms of efficient mine operation scheduling. The study by Yan and Feng in 2013 detailed the implementation of the ACO algorithm in the Unified Tunnel Weight Calculation Model to analyse the search for escape routes during mine construction [86]. The Max-Min Ant System (MMAS) method was applied to establish a tunnel network partitioning strategy to search for the best route, and further testing was conducted in large domestic coal mines. The optimal escape route obtained was highly reliable and suitable for real-life situations. Enhancements to search escape route planning demonstrate ACO’s ability to enhance the utility of mine construction with greater reliability. The 2021 study demonstrated the implementation of FA algorithm at Sungun Copper Mine to optimise mining fleet management [87]. The FA algorithm transformed fixed scheduling into flexible scheduling dispatch method, thereby improving mine performance, increasing productivity by 20%, and reducing idle time by 20%. Research in 2020 demonstrated the implementation of the GWO algorithm in a support vector machine (SVM) to optimise parameters to solve the problem of fault diagnosis of belt conveyor transportation systems in underground mines [88]. The integrated model was tested using the hybrid wolf optimizer, and the fault detection accuracy was as high as 97.22%, which is suitable for practical applications and avoids impact on safety and mine production. The improvements in belt conveyor reliability and safety reflect GWO’s ability to enhance the construction of conveyor belt mining operations. The comprehensive analysis on the applications of the swarm algorithm to mine operation and construction stages has been reviewed and classified in Table 4.
![Click here to view Large Table 3](table/IMST.MS.ID.555636.T003.png)
![Click here to view Large Table 4](table/IMST.MS.ID.555636.T004.png)
Mine closure and rehabilitation
Mine closure and restoration phase includes restoring the landscape, waste management, soil detoxification, planting vegetation, creating wildlife habitat, mitigating environmental impacts, and ensuring public safety. The 2023 study illustrated the implementation of PSO in the support vector regression algorithm (PSO-SVR) to monitor the environment of the Hongshaquan mining area [89]. The integrated hybrid model has been applied to UAV measurements of vegetation index, surface temperature, salinity index and soil respiration (R²=0.959, RMSE=0.497, AIC=- 0.561). Advanced remote sensing models demonstrated the PSO’s ability to continue mitigating and monitoring environmental sustainability during mine closures. The comprehensive analysis on swarm algorithm application in the mine closure and rehabilitation stages has been reviewed and classified in Table 5.
![Click here to view Large Table 5](table/IMST.MS.ID.555636.T005.png)
Overview of mine integration
The implementation of nine swarm algorithms in various mining stages has been comprehensively analysed and summarised in pervious sections of this paper. The contributions of swarm algorithms are comprehensively reviewed and classified in (Figure 12) to demonstrate that the application of swarm algorithms in the mining field can improve mine efficiency, mine safety, and environmental sustainability.
![Click here to view Large Figure 11](images/IMST.MS.ID.555636.G0011.png)
![Click here to view Large Figure 12](images/IMST.MS.ID.555636.G0012.png)
Conclusion
This paper provides a comprehensive study of nine swarm algorithms, their nature, mathematical and theoretical models of swarm behavior, and presents their implementation and classification in swarm robotics and mining operations. Swarm algorithms show good results in the applications to all stages of the mining life cycle (mine exploration and evaluation, mine planning and design, mine operation and construction, mine closure and rehabilitation). They can successfully be applied to improve aspects such as mapping accuracy and efficiency, drilling hole accuracy, mine layout and scheduling, production scheduling, blasting efficiency and safety, mining equipment scheduling and matching, escape route planning, conveyor belt fault diagnosis and environmental monitoring. Besides, integration of NIA also can increase the net present value of profits by reducing costs, enhancing benefits and safety, and improving environment impact, achieve increased productivity by having more reliable systems, more accurate feasibility studies and more environmental considerations. Current findings on the applications of swarm algorithms to mining are promising and further research and exploration will allow for more widespread applications of bioinspired swarm algorithms in the mining industry. The classification of swarm behavior (spatial organization, navigation, decisionmaking and miscellaneous) and mining optimisation show that the applications of swarm algorithms and swarm robotics can contribute to creating more autonomous and robust smart mines without human intervention. Such innovation and development in global mining operations will help towards achieving more environmentally sustainable mining operations. This study explored the application of swarm intelligence algorithms to the mining industry, highlighting how nature-inspired algorithms can skillfully cope with the complexities of the mining life cycle. It has provided an in-depth review and evaluation on the integration of these algorithms with existing mining practices, demonstrating their superior performance compared to traditional methods, and has highlighted the potential of swarm algorithms to revolutionise the mining process and industry by providing more efficient, accurate and sustainable solutions.
References
- Dorin I, Diaconescu C, Topor DI (2014) The Role of mining in National Economies. Int J Acad Res Account Financ Manage Sci 4(3): 155-160.
- Wilburn DR, Goonan TG, Bleiwas DI (2001) Technological Advancement, a Factor in Increasing Resource Use. US Geological Survey, US Department of the Interior.
- Marshall JA, Bonchis A, Nebot E, Scheding S (2016) Robotics in mining. Springer handbook of Robotics pp. 1549-1576.
- Fisher BS, Schnittger S (2012) Autonomous and remote operation technologies in the mining industry. BAeconomics Pty Ltd, February.
- (2023) Annual Report 2023 year in review, Rio Tinto.
- Dorigo M, Maniezzo V, Colorni A (1996) Ant system: optimization by a colony of cooperating agents. IEEE transactions on systems, man, and cybernetics, part b (cybernetics) 26(1): 29-41.
- Dorigo M, Di Caro G (1999) Ant colony optimization: a new meta-heuristic. In Proceedings of the 1999 congress on evolutionary computation-CEC99 (Cat. No. 99TH8406) IEEE 2: 1470-1477.
- Kumar V, Yadav SM (2022) A state-of-the-Art review of heuristic and metaheuristic optimization techniques for the management of water resources. Water Supply 22(4): 3702-3728.
- Phan HD, Ellis K, Barca JC, Dorin A (2020) A survey of dynamic parameter setting methods for nature-inspired swarm intelligence algorithms. Neural Comput Applic 32(2): 567-588.
- Okonta CI, Kemp AH, Edopkia RO, Monyei GC, Okelue ED (2016) A heuristic based ant colony optimization algorithm for energy efficient smart homes. In Proc. 5th Int Conf Exhib Clean Energy p. 1-12.
- Eberhart R, Kennedy J (1995) A new optimizer using particle swarm theory. In MHS'95. Proceedings of the Sixth International Symposium on Micro Machine and Human Science. IEEE p. 39-43.
- Marini F, Walczak B (2015) Particle swarm optimization (PSO) A tutorial. Chemometrics and Intelligent Laboratory Systems 149: 153-165.
- Van Quan T, Giang NH, Tan NN (2023) A Data-Driven Approach for Investigating Shear Strength of Slender Steel Fiber Reinforced Concrete Beams. J Sci Technol Civil Eng (JSTCE)-HUCE 17(2): 133-144.
- Yousefi M, Omid M, Rafiee S, Ghaderi SF (2013) Strategic planning for minimizing CO2 emissions using LP model based on forecasted energy demand by PSO Algorithm and ANN. Int J Energy and Environ 4.
- Karaboga D (2005) An idea based on honeybee swarm for numerical optimization. Technical report-tr06, Erciyes University, Engineering Faculty, Computer Engineering Department 200: 1-10.
- Chalotra S, Sehra SK, Sehra SS (2016) A systematic review of applications of bee colony optimization. In 2016 International Conference on Innovation and Challenges in Cyber Security (ICICCS-INBUSH). IEEE pp. 257-260.
- Khader AT, Al-betar MA, Mohammed AA (2013) Artificial bee colony algorithm, its variants and applications: a survey.
- Chao KH, Li JY (2022) Global maximum power point tracking of photovoltaic module arrays based on improved artificial bee colony algorithm. Electronics 11(10): 1572.
- Sharma TK, Pant M, Singh VP (2012) Improved local search in artificial bee colony using golden section search. arXiv preprint arXiv:1210.6128.
- Yang J, Peng Z (2018) Improved ABC algorithm optimizing the bridge sensor placement. Sensors (Basel) 18(7): 2240.
- Yang XS (2010) Firefly algorithm, stochastic test functions and design optimization. Int J Bio-Inspired Computation 2(2): 78-84.
- Johari NF, Zain AM, Noorfa MH, Udin A (2013) Firefly algorithm for optimization problem. Applied Mechanics and Materials 421: pp. 512-517.
- Sharma S, Jain P, Saxena A (2020) Adaptive inertia-weighted firefly algorithm. In Intelligent Computing Techniques for Smart Energy Systems: Proceedings of ICTSES 2018. Springer Singapore pp. 495-503.
- Nordin N, Sulaiman SI, Omar AM (2018) Hybrid artificial neural network with meta-heuristics for grid-connected photovoltaic system output prediction. Indonesian J Electrical Eng Comput Sci 11(1): 121.
- Tighzert L, Fonlupt C, Mendil B (2018) A set of new compact firefly algorithms. Swarm and evolutionary computation 40: 92-115.
- Yang XS (2010) A new metaheuristic bat-inspired algorithm. In Nature inspired cooperative strategies for optimization (NICSO 2010) Springer, Berlin, Heidelberg pp. 65-74.
- Iglesias A, Gálvez A, Suárez P (2020) Swarm robotics-a case study: bat robotics. In Nature-Inspired Computation and Swarm Intelligence. Academic Press pp. 273-302.
- Topal AO, Altun O (2016) A novel meta-heuristic algorithm: dynamic virtual bats algorithm. Info Sci 354: 222-235.
- Templos-Santos JL, Aguilar-Mejia O, Peralta-Sanchez E, Sosa-Cortez R (2019) Parameter tuning of PI control for speed regulation of a PMSM using bio-inspired algorithms. Algorithms 12(3): 54.
- Gandomi AH, Alavi AH (2012) Krill herd: a new bio-inspired optimization algorithm. Commun Nonlinear Sci Numerical Simulation 17(12): 4831-4845.
- Wang GG, Gandomi AH, Alavi AH, Gong D (2019) A comprehensive review of krill herd algorithm: variants, hybrids and applications. Artificial Intelligence Rev 51: 119-148.
- Ren YT, Qi H, Huang X, Wang W, Ruan LM, et al. (2016) Application of improved krill herd algorithms to inverse radiation problems. Int J Thermal Sci 103: 24-34.
- Regad M, Helaimi MH, Taleb R, Othman AM, Gabbar HA (2020) Frequency control of microgrid with renewable generation using PID controller-based krill herd. Indonesian Journal of Electrical Engineering and Informatics (IJEEI) 8(1): 21-32.
- Mirjalili S, Mirjalili SM, Lewis A (2014) Grey Wolf Optimizer Adv Eng Softw 69: 46-61.
- Faris H, Aljarah I, Al-Betar MA, Mirjalili S (2018) Grey wolf optimizer: a review of recent variants and applications. Neural Comput Appl 30: 413-435.
- Guha D, Roy PK, Banerjee S (2016) Load frequency control of large scale power system using quasi-oppositional grey wolf optimization algorithm. Eng Sci Technol Int J 19(4): 1693-1713.
- Mirjalili S, Gandomi AH, Mirjalili SZ, Saremi S, Faris H, et al. (2017) Salp Swarm Algorithm: A bio-inspired optimizer for engineering design problems. Adv Eng Software 114: 163-191.
- Zhang J, Wang Z, Luo X (2018) Parameter estimation for soil water retention curve using the salp swarm algorithm. Water 10(6): 815.
- Saremi S, Mirjalili S, Lewis A (2017) Grasshopper optimisation algorithm: theory and application. Adv Eng Software 105: 30-47.
- Meraihi Y, Gabis AB, Mirjalili S, Ramdane-Cherif A (2021) Grasshopper optimization algorithm: theory, variants, and applications. IEEE Access 9: 50001-50024.
- Abualigah L, Diabat A (2020) A comprehensive survey of the Grasshopper optimization algorithm: results, variants, and applications. Neural Comput Appl 32(19): 15533-15556.
- Nabavi S, Gholampour S, Haji MS (2022) Damage detection in frame elements using Grasshopper Optimization Algorithm (GOA) and time-domain responses of the structure. Evolving Syst 13(2): 307-318.
- Trianni V, IJsselmuiden J, Haken R (2016) The saga concept: swarm robotics for agricultural applications. Technical Report.
- Petersen K, Nagpal R, Werfel J (2011) TERMES: An Autonomous Robotic System for Three-Dimensional Collective Construction. Robotics: Sci Syst 7: 257-264.
- Kshetri N, Rojas-Torres D (2018) The 2018 winter olympics: A showcase of technological advancement. IT 20(2): 19-25.
- Brambilla M, Ferrante E, Birattari M, Dorigo M (2013) Swarm robotics: a review from the swarm engineering perspective. Swarm Intelligence 7(1): 1-41.
- Schranz M, Umlauft M, Sende M, Elmenreich W (2020) Swarm robotic behaviors and current applications. Front Robotics AI 7: 36.
- Benuwa BB, Ghansah B, Wornyo DK, Adabunu SA (2016) A comprehensive review of Particle swarm optimization. International J Eng Res Afr 23: 141-161.
- Watkins J (2000) Aggregation and vertical migration. Krill: biology, ecology and fisheries pp. 80-102.
- Eyer M, Greco MK, Lang J, Neumann P, Dietemann V (2016) No spatial patterns for early nectar storage in honey bee colonies. Insectes Sociaux 63: 51-59.
- Senthilnath J, Omkar SN, Mani V (2011) Clustering using firefly algorithm: performance study. Swarm and Evolutionary Computation 1(3): 164-171.
- Sutherland KR, Weihs D (2017) Hydrodynamic advantages of swimming by salp chains. J Royal Society Interface 14(133): 20170298.
- Aliwi M, Aslan S, Demirci S (2020) Solving uav localization problem with firefly algorithm. In 2020 28th Signal Processing and Communications Applications Conference (SIU) IEEE p. 1-4.
- Tongur V, Ülker E (2019) PSO-based improved multi-flocks migrating birds optimization (IMFMBO) algorithm for solution of discrete problems. Soft Comput 23(14): 5469-5484.
- Tarling GA, Thorpe SE (2017) Oceanic swarms of Antarctic krill perform satiation sinking. Proceedings of the Royal Society B: Biol Sci 284(1869): 20172015.
- Hölldobler B, Wilson EO (2010) The leafcutter ants: civilization by instinct. WW Norton & Company.
- Zhang J, Wang JS (2020) Improved salp swarm algorithm based on levy flight and sine cosine operator. Ieee Access 8: 99740-99771.
- Buck J, Buck E (1976) Synchronous fireflies. Scientific Am 234(5): 74-85.
- Liu J, Weng H, Ge Y, Li S, Cui X (2022) A self-healing routing strategy based on ant colony optimization for vehicular Ad Hoc networks. IEEE Internet of Things J 9(22): 22695-22708.
- Lakshmi CB, Rao SM (2014) Bio-inspired self-healing routing to improve lifetime of wireless sensor networks. In 2014 International Conference on Communication and Network Technologies. IEEE pp. 134-138.
- De La Cruz C, Carelli R (2006) Dynamic modeling and centralized formation control of mobile robots. In IECON 2006-32nd annual conference on IEEE industrial electronics. IEEE pp. 3880-3885.
- Mehrjerdi H, Saad M, Ghommam J (2010) Hierarchical fuzzy cooperative control and path following for a team of mobile robots. IEEE/ASME Transactions on Mechatronics 16(5): 907-917.
- Kamel MA, Yu X, Zhang Y (2020) Formation control and coordination of multiple unmanned ground vehicles in normal and faulty situations: A review. Annual Rev Control 49: 128-144.
- Lewis MA, Tan KH (1997) High precision formation control of mobile robots using virtual structures. Autonomous Robots 4(4): 387-403.
- Anderson PAV, Bone Q (1980) Communication between individuals in salp chains. II. Physiology. Proceedings of the Royal Society of London. Series B. Biol Sci 210(1181): 559-574.
- Balch T, Arkin RC (1998) Behavior-based formation control for multirobot teams. IEEE transactions on robotics and automation 14(6): 926-939.
- Desai Jaydev P, Ostrowski James P (2001) Kumar Vijay. Modeling and control of formations of nonholonomic mobile robots, Robotics and Automation, IEEE Transactions 17(6): 905-908.
- Xie Y, Han L, Dong X, Li Q, Ren Z (2021) Bio-inspired adaptive formation tracking control for swarm systems with application to UAV swarm systems. Neurocomputing 453: 272-285.
- Duan Q, Wang L, Kang H, Shen Y, Sun X, Chen Q (2021) Improved salp swarm algorithm with simulated annealing for solving engineering optimization problems. Symmetry 13(6): 1092.
- Desai JP, Ostrowski J, Kumar V (1998) Controlling formations of multiple mobile robots. In Proceedings. 1998 IEEE International Conference on Robotics and Automation (Cat. No. 98CH36146). IEEE. 4: 2864-2869.
- Wang J, Fan X, Zhang C, Wan S (2014) A graph-based ant colony optimization approach for integrated process planning and scheduling. Chinese J Chem Eng 22(7): 748-753.
- Lei T, Luo C, Ball JE, Rahimi S (2020) A graph-based ant-like approach to optimal path planning. In 2020 IEEE congress on evolutionary computation (CEC). IEEE p. 1-6.
- Torkzadeh S, Soltanizadeh H, Orouji AA (2021) Energy-aware routing considering load balancing for SDN: a minimum graph-based Ant Colony Optimization. Cluster Comput 24(3): 2293-2312.
- Khatib O (1986) Real-time obstacle avoidance for manipulators and mobile robots. In Autonomous robot vehicles. Springer, New York, NY pp. 396-404.
- Gandomi AH, Alavi AH (2016) An introduction of krill herd algorithm for engineering optimization. J Civil Eng Manage 22(3): 302-310.
- Sprott D (2022) Mine closure overview, Mine closure overview Mine Closure Hub.
- Nhleko AS, Musingwini C (2019) Analysis of the particle swarm optimization (PSO) algorithm for application in stope layout optimisation for underground mines. In Proceedings of the Mine Planner’s Colloquium.
- Jafrasteh B, Fathianpour N (2017) Optimal location of additional exploratory drillholes using afuzzy-artificial bee colony algorithm. Arabian J Geosci 10: 1-16.
- Mohammadzadeh M, Mahboubiaghdam M, Nasseri A, Jahangiri M (2023) A New Frontier in Mineral Exploration: Hybrid Machine Learning and Bat Metaheuristic Algorithm for Cu-Au Mineral Prospecting in Sonajil area, E-Azerbaijan.
- Khan A, Niemann-Delius C (2014) Production scheduling of open pit mines using particle swarm optimization algorithm. Adv Operat Res.
- Korzeń M, Kruszyna M (2023) Modified Ant Colony Optimization as a Means for Evaluating the Variants of the City Railway Underground Section. Int J Environ Res Public Health 20(6): 4960.
- Kha A (2018) Long-term production scheduling of open pit mines using particle swarm and bat algorithms under grade uncertainty. J South Afr Institute Mining Metallurgy 118(4): 361-368.
- Tolouei K, Moosavi E (2020) Production scheduling problem and solver improvement via integration of the grey wolf optimizer into the augmented Lagrangian relaxation method. SN Appl Sci 2: 1-12.
- Nguyen H, Bui XN, Topal E (2023) Reliability and availability artificial intelligence models for predicting blast-induced ground vibration intensity in open-pit mines to ensure the safety of the surroundings. Reliabil Eng Syst Safety 231: 109032.
- Bao H, Zhang R (2020) Study on optimization of coal truck flow in open-pit mine. Adv Civil Eng 2020: 1-13.
- Yan G, Feng D (2013) Escape-route planning of underground coal mine based on Improved Ant Algorithm. Math Problem in Eng.
- Ghaziania HH, Monjezi M, Mousavi A, Dehghani H, Bakhtavar E (2021) Design of loading and transportation fleet in open-pit mines using simulation approach and metaheuristic algorithms. J Mining Environ 12(4): 1177-1188.
- Li, X, Li Y, Zhang Y, Liu F, Fang Y (2020) Fault diagnosis of belt conveyor based on support vector machine and grey wolf optimization. Mathe Proble Eng 2020: 1-10.
- Liu Y, Lin J, Yue H (2023) Soil respiration estimation in desertified mining areas based on UAV remote sensing and machine learning. Earth Sci Informatics 16(4): 3433-3448.