A Dynamic Resource Allocation Strategy with Reinforcement Learning for Multimodal Multi-objective Optimization

Dang, Qianlong; Xu, Wei; Yuan, Yangfei

doi:10.1007/s11633-022-1314-7

Cited by 21 publications

(5 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When transforming the encoding matrix into the positions of the particles, every priority within the path between every node is the coordinate of one axis of the particle, i.e., the dimension of the particle is 4a. Also, when initializing a population particle, its priority takes a value in the range of (0, 1], and if a transport mode does not exist within the path, the priority of that transport mode is set as 0. where 𝑣 (𝑡 + 1) and 𝑥 (𝑡 + 1) are the velocity and position of particle 𝑖 after one iteration, 𝑣 (𝑡) and 𝑥 (𝑡) are the velocity and position of particle 𝑖 before the iteration, 𝜛 is the inertia weight of the particle, 𝑐 and 𝑐 are the learning factors, 𝑟 and 𝑟 are random numbers between 0 and 1 (Dang et al, 2022), 𝑃 (𝑡) is the optimal position experienced by particle 𝑖 (excluding particles that exceed the limit), and 𝐺 (𝑡) is the best position experienced by the particle population after excluding particles that exceed the limit.…”

Section: Multimodal Transport Path Optimization Model Under Transport...mentioning

confidence: 99%

Rail Consignment Path Planning Based on Multimodal Transport: Considering the Time Uncertainty Condition

2023

EPPM-Journal

View full text Add to dashboard Cite

Multimodal transport has emerged and applied to logistics work as single-mode transport was unable to meet the growing demand for logistics. Due to the difference in cost and time consumption between different modes of transportation, the path planning of multimodal transport is different from that of single-mode transport. This paper firstly established a multimodal railroad consignment path optimization model under the condition of transportation time uncertainty and then optimized the path and transportation mode with the particle swarm optimization (PSO) algorithm according to the path optimization model. In addition, the PSO algorithm was optimized by a genetic algorithm to avoid the optimization process of the PSO algorithm from falling into the locally optimal solution. Finally, simulation experiments were carried out on the improved PSO algorithm, and it was compared with the traditional PSO and genetic algorithms. The results showed that the improved PSO algorithm converged to stability after about 200 iterations, the traditional PSO algorithm converged to stability after about 240 iterations, and the genetic algorithm converged to stability after about 400 iterations; the total transport cost and time of railway transportation were USD 114,300 and 52.5 h; the total transport cost and time of the path obtained by the genetic algorithm were USD 60,400 and 35.6 h; the total transport cost and time of the path obtained by the traditional PSO algorithm were USD 37700 and 24.3 h; the total transport cost and time of the path obtained by the improved PSO algorithm were USD 3,1900 and 23.4 h.

show abstract

Section: Multimodal Transport Path Optimization Model Under Transport...mentioning

confidence: 99%

Rail Consignment Path Planning Based on Multimodal Transport: Considering the Time Uncertainty Condition

2023

EPPM-Journal

View full text Add to dashboard Cite

show abstract

“…In Ref. [35], in order to distinguish the potentials of each subspace, the reinforcement learning method is used to dynamically allocate computing resources to each subspace. From the results, it is an effective strategy to assist the ZS in solving the MMOPs.…”

Section: Related Workmentioning

confidence: 99%

A Zoning Search-Based Multimodal Multi-Objective Brain Storm Optimization Algorithm for Multimodal Multi-Objective Optimization

et al. 2023

View full text Add to dashboard Cite

For multimodal multi-objective optimization problems (MMOPs), there are multiple equivalent Pareto optimal solutions in the decision space that are corresponding to the same objective value. Therefore, the main tasks of multimodal multi-objective optimization (MMO) are to find a high-quality PF approximation in the objective space and maintain the population diversity in the decision space. To achieve the above objectives, this article proposes a zoning search-based multimodal multi-objective brain storm optimization algorithm (ZS-MMBSO). At first, the search space segmentation method is employed to divide the search space into some sub-regions. Moreover, a novel individual generation strategy is incorporated into the multimodal multi-objective brain storm optimization algorithm, which can improve the search performance of the search engineering. The proposed algorithm is compared with five famous multimodal multi-objective evolutionary algorithms (MMOEAs) on IEEE CEC2019 MMOPs benchmark test suite. Experimental results indicate that the overall performance of the ZS-MMBSO is the best among all competitors.

show abstract

“…Intelligent algorithms improve the problem solving capability and optimize the resource allocation in solving multi-objective problems. Researchers combine multimodal multi-objective optimization with deep learning techniques to rationalize the space allocation to improve this model's area search capability and find optimal solution better [17]. In the field information analysis of drilling operations, some scholars have used support vector machines combined with simulated annealing algorithms for multi-objective optimization model building.…”

Section: Introductionmentioning

confidence: 99%

The optimization method of CNC lathe performance based on Morris sensitivity analysis and improved GA algorithm

2024

J. vibroeng.

View full text Add to dashboard Cite

This study achieved the goal of guiding bed design and optimization by conducting multi-objective optimization research on the performance of CNC lathe beds. In this study, Morris analysis was first performed on the sensitivity of the parameters, and then out to optimize the parameters using a combination of neural network and genetic algorithm. The loss function value, RMSE error accumulation, recall, sensitivity and specificity of the ASSGA-BP optimization model were better. The maximum error between the predicted and true values of the ASSGA-BP model was 0.28 mm. In the performance study of the multi-objective optimization method based on the Morris sensitivity analysis and the improved GA algorithm, the average MAE value is 0.91 %. The average RMSE value is 0.59 %. Also, the new model is significantly better than the NSGA-II, EGA, and FGA algorithms in terms of both the number of final non-dominated solutions and the speed of reaching convergence. The above results demonstrate that the model proposed in this study has high performance, can achieve faster convergence and has the best stability of the convergence state. The innovation of this article lies in the use of the Morris method to screen and evaluate numerous parameters in order to improve the accuracy of the calculation results and ensure the effectiveness of the optimization results. The improved algorithm overcomes the problems of BP neural network and can effectively improve the generalization performance of the neural network, thereby improving the prediction accuracy of the model.

show abstract

A Dynamic Resource Allocation Strategy with Reinforcement Learning for Multimodal Multi-objective Optimization

Cited by 21 publications

References 57 publications

Rail Consignment Path Planning Based on Multimodal Transport: Considering the Time Uncertainty Condition

Rail Consignment Path Planning Based on Multimodal Transport: Considering the Time Uncertainty Condition

A Zoning Search-Based Multimodal Multi-Objective Brain Storm Optimization Algorithm for Multimodal Multi-Objective Optimization

The optimization method of CNC lathe performance based on Morris sensitivity analysis and improved GA algorithm

Contact Info

Product

Resources

About