Lei Wang

Distributed manufacturing introduces new challenges to traditional production shop scheduling, and the combination of machine learning and metaheuristic algorithms offers new approaches to solve these problems. To address the distributed hybrid flow shop scheduling problem with missing operations (MDHFSP), a shuffled frog leaping algorithm with Q-learning (QSFLA) is proposed to minimize the maximum completion time. A dual-string encoding method is proposed to represent factory assignment and job sequencing, with heuristic methods utilized during decoding to determine machine assignments. The state set is constructed based on changes in the minimum and average objective values of solutions in the population, while the action set is built from different optimized solutions and learned solutions during the memeplex search process. Symmetry-driven Q-learning is employed to dynamically adjust the optimization objects based on the state of the population. Testing on 140 benchmarks and a real-life example shows that symmetry-driven Q-learning plays a positive role within QSFLA, and QSFLA effectively solves the MDHFSP.

In recent years, the distributed assembly permutation flow-shop scheduling problem (DAPFSP) has been studied frequently. Due to the high complexity of DAPFSP, the results of the benchmark instances are not particularly desirable. In order to solve the disadvantages of meta-heuristics in DAPFSP, such as being easy to fall into local optima and poor stability, a genetic biogeography-based optimization (GBBO) algorithm is proposed to address DAPFSP with minimize makespan. A DAPFSP mathematical model is established. The crossover operator of genetic algorithm is integrated into the migration operator to improve the quality of the emigration solution. The mobility model is modified to improve the searching ability of the algorithm, and the method of product sequence fusion (PSF) is designed for deep searching of partial solutions. Through simulation experiments, GBBO refreshed 155 benchmark instances in both small-sized and large-sized instances.

The traditional A-Star algorithm has the problem of performing much invalid traversal work of nodes near concave obstacles when dealing with path planning tasks in the presence of concave obstacles. To solve this problem, a fusion algorithm (called IAPF-A-Star here) based on improved artificial potential field (IAPF) and A-Star is proposed. In IAPF-A-Star, to better measure the distances between the nodes and the target node, the potential energy of the artificial potential field is used to represent the node’s proximity to the target node, and an iterative correction method is used to improve the potential field in order to eliminate the local low-potential energy that may exist in the artificial potential field. The improved potential field effectively avoids the occurrence of local low-potential fields and makes it possible to measure the position of nodes in relation to their target node by means of the potential field. Finally, the obtained path is converted to a curve path by using a cubic spline curve. In addition, to solve the problem that the curve path generated by the cubic spline curve may collide with obstacles, a path optimization strategy based on an adaptive cubic spline curve is proposed to make the final path smooth and collision avoidance, shorter and closer to the actual situation. Simulation experiments show that the proposed IAPF-A-Star algorithm in this paper can effectively avoid the problem of additional local energy consumption caused by concave obstacles when processing path planning tasks, and the traversal of invalid nodes facing different complex concave obstacles is effectively avoided.

Energy saving has always been a concern in production scheduling, especially in distributed hybrid flow shop scheduling problems. This study proposes a shuffled frog leaping algorithm with Q-learning (QSFLA) to solve distributed hybrid flow shop scheduling problems with energy-saving(DEHFSP) for minimizing the maximum completion time and total energy consumption simultaneously. The mathematical model is provided, and the lower bounds of two optimization objectives are given and proved. A Q-learning process is embedded in the memeplex search of QSFLA. The state of the population is calculated based on the lower bound. Sixteen search strategy combinations are designed according to the four kinds of global search and four kinds of neighborhood structure. One combination is selected to be used in the memeplex search according to the population state. An energy-saving operator is presented to reduce total energy consumption without increasing the processing time. One hundred forty instances with different scales are tested, and the computational results show that QSFLA is a very competitive algorithm for solving DEHFSP.

Cloud computing is emerging as an advanced stage of enterprise information technology in a highly competitive society, and is now in a phase of rapid development. Enterprises using cloud computing can reduce the cost of building infrastructure and reap huge benefits. Due to the convenience of cloud computing, more and more enterprises are inclined to use cloud services to build their business systems. However, there are many different cloud service providers in the market, and enterprises need scientific decision-making tools to determine which cloud service provider to choose. The cloud service provider selection is viewed as the multi-attribute decision-making (MADM). In this paper, the cross-entropy method under triangular fuzzy neutrosophic sets (TFNSs) is built based on the traditional cross-entropy method. Firstly, the TFNSs is introduced. Then, combine the traditional fuzzy cross-entropy method with TFNSs information, the triangular fuzzy neutrosophic number cross-entropy (TFNN-CE) method is established for MADM under TFNSs. Finally, a numerical example for cloud service provider selection has been given and some comparisons is used to illustrate advantages of cross-entropy method with TFNSs.

Aiming at the shortcomings of single ant colony optimization such as many redundant nodes, slow convergence and low efficiency, based on the idea of “selection-crossover” of genetic algorithm, an improved fusion algorithm of ant colony optimization and genetic algorithm is proposed. In this paper, the fusion algorithm includes “optimal strategy” and “genetic region strategy”. The optimal strategy is that high-quality parents are selected by roulette in the first [ ] paths of each generation; genetic region strategy is that according to the path information of the parents, the grid map is divided into genetic area and nongenetic area. Genetic area refers to the area where the offspring ants can pass, and nongenetic area refers to the area where the offspring ants can’t pass; finally, the offspring ant searches the path in the genetic region to reduce the search range of the offspring ant and improve the convergence speed. Simulation results show that the fusion algorithm has faster searching speed and more stable convergence than the basic ant colony optimization and other improved ant colony optimization.

The basic genetic algorithm usually has the following problems when solving robot path planning problems. First, the chromosome (path) of the first generation is randomly generated, so its fitness is often too high, resulting in too slow convergence speed and even unable to obtain the global optimal solution. Second, in the crossover operator, the selection of crossover points is often random, which cannot ensure that the fitness of cross generated offspring is better than that of their parents. Finally, in the mutation stage, because the traditional mutation strategy is to replace the selected mutation node in the individual chromosome with a random node, the offspring path generated by the mutation operator is likely to be discontinuous, so it cannot reflect the effectiveness of the mutation operator. Based on the mentioned contents, some improved strategies are proposed for the discussed three shortcomings of the basic genetic algorithm. First, a new generation mechanism of the first generation is constructed by simulating the path finding rules of ants in ant colony optimization, which greatly improves the quality of the first generation. Second, during crossover operator, the chromosomes of the two parents are disconnected at the same node, and the optimal chromosome fragments in each segment are combined based on greed, so as to gather all the excellent gene fragments of the parents in an individual as much as possible, improve the quality of the offspring after crossover operator, and speed up the convergence speed. Finally, a new mutation strategy is proposed to eliminate redundant sections and the number of corners, so that the individuals always can be mutated in a good direction. A large number of simulation results show that the improved genetic algorithm is effective in solving the robot path planning problem, and the overall performance is better than the basic genetic algorithm and some other improved genetic algorithms. Finally, by modeling and simulating the real enterprise factory environment based on the ROS platform, the experimental results also verify the effectiveness and feasibility of the improved genetic algorithm.