Locomotive and rolling stock enterprises are typical order-type manufacturing enterprises. Facing the economic integration of global economy, individualized customer demand and high-speed delivery of products, the rolling stock industry is facing rapid response to customer needs, shortening product delivery time and improving products. Quality and pressure to improve product service. In order to survive and develop in the fierce market competition, enterprises need to optimize from the production scheduling. Production scheduling is a research hotspot of manufacturing system. The task of scheduling is to determine the specific processing path, time, machine and operation for each object according to the production target and constraints. The excellent scheduling strategy is to shorten the production time and reduce the cost of the product. Has a great effect.
The order production scheduling problem belongs to a class of NP-hard combinatorial problems, so it is almost impossible to find an optimal algorithm with polynomial complexity. The traditional heuristic algorithm, intelligent simulated annealing algorithm, taboo algorithm, neural network method and other algorithms are common methods to find the approximate solution or satisfactory solution to meet the actual needs of the production line optimization problem, but they are all local optimization methods. Various statistical optimization methods, such as simulated annealing and genetic algorithms, provide a new way to solve the scheduling optimization problem, but there are also a certain degree of enumeration. Generally speaking, the convergence to the optimal solution is slow. And it is difficult to judge the optimality of the solution. Combining the genetic algorithm with the optimization algorithm has higher performance for solving the order production scheduling problem.
1 hybrid genetic algorithm
1.1 Genetic algorithm
Genetic algorithms are widely used in many fields due to their simplicity of operation and high efficiency of solving problems. It has been proved theoretically that the genetic algorithm can find the optimal solution to the problem in a random way in the sense of probability. But on the other hand, the application practice shows that there are some unsatisfactory problems in the application of genetic algorithms. These problems are mainly caused by the premature phenomenon and the poor local search ability. And in general, for many problems, the solution of the basic genetic algorithm is often not the most effective way to solve this problem, it is less efficient than the knowledge-based heuristic algorithm specifically for this problem, although this knowledge-based inspiration The algorithm does not guarantee that the global optimal solution of the problem will be found. In addition, the genetic algorithm can not avoid the situation of multiple search for the same feasible solution, which is also an important factor affecting the efficiency of the genetic algorithm.
1.2 Hybrid genetic algorithm
Some optimization algorithms such as gradient method, hill climbing method, simulated annealing algorithm and list optimization method have strong local search ability, while genetic algorithm runs because of its simplicity and other heuristic algorithms with problems and related knowledge. The efficiency is also relatively high. Therefore, it can be expected that the idea of ​​combining these optimization methods in the search process of genetic algorithm constitutes a hybrid genetic algorithm is an effective means to improve the efficiency and quality of genetic algorithm.
Hybrid genetic algorithm is the idea of ​​combining local search algorithm in standard genetic algorithm. Its characteristics are mainly reflected in the following two aspects:
(1) A local search process was introduced. Based on the phenotypes corresponding to each individual in the group, a local search is performed to find the local optimal solution corresponding to each individual in the current environment, so as to achieve the purpose of improving the overall performance of the group.
(2) The coding conversion operation process is added. The local optimal solutions obtained by the local search process are transformed into new individuals through the encoding process, so that the next generation genetic evolution operation can be performed based on a new group with better performance.
2 Order-oriented production scheduling model for the rolling stock industry
There are many optimization goals or comprehensive optimization goals in the order production scheduling problem. The optimization goal of the scheduling problem is usually considered from the aspects of production cost and production time. In terms of production cost, the optimization objectives are: minimum product, minimum inventory, highest equipment utilization, etc. From the perspective of production time, the optimization objectives are: minimum completion time, maximum delivery time, minimum flow time And minimum waiting time, etc. The optimization goals of these two directions are not isolated from each other, and many of them have very close links, some promote each other, some conflict with each other, and some have no connection. The algorithm adopts the minimum total processing time as the target, which is specifically expressed as: 
Where: Rk represents the total number of processes that need to be machined on device k; Aijk represents the time required for the jth process of the i-th workpiece to complete processing on device k. This time includes the waiting time Wijk before starting the processing of the process and the required time from the start of processing to the completion of the process Nijk; Cj indicates that the production time of the order j is the sum of the time of completion of all the components included in the order j; Cmax In the case of parallel production for all orders, the maximum completion time, which is the completion time of the latest order for each order completion time; the purpose of the objective function F(x) is to make the longest required time as short as possible. In this way, after several iterations of the algorithm, the approximate optimal solution of the scheduling can be generated, and a satisfactory scheduling result is obtained.
This article refers to the address: http://
3 Hybrid genetic algorithm production scheduling design
According to the above ideas, the basic steps of the hybrid genetic algorithm designed are as follows:
(1) Enter the data and initialize the population.
(2) Calculate the sample variance FSD of the adapted value of the population.
(3) Determine whether the stopping criterion of the genetic algorithm satisfies the requirements. If the stop algorithm is met, output the result, otherwise continue with the following steps.
(4) Selecting a crossover operation based on the current population.
(5) Calculate the sample variance fsd of the population, and determine whether fsd is smaller than FSD/n. If it is smaller, proceed to step (6). If it is greater, perform the mutation operation.
(6) Initialize the tabu search algorithm, the current chromosome is the initial solution.
(7) Determine whether the iteration criterion of the tabu search algorithm satisfies the requirement. If the end tabu search is satisfied, proceed to step (2); otherwise continue the following operation.
(8) Generate a candidate solution set.
(9) Select a solution based on the set level of craving and the taboo table, and update the taboo form.
(10) Go to step (7).
(11) Go to step (2).
4 Simulation experiment
In the simulation test of this paper, the scheduling problem of 10 parts and 10 machines of a locomotive and vehicle industry is taken as an example. The optimal workpiece sorting matrix obtained by processing time array, machine sequential array and GATS algorithm calculation 10 times and its corresponding optimal processing Gantt graph are given below. As shown in Fig. 1, the optimal solution is 87 unit time. At the same time, the Gantt chart of the scheduling problem also clearly reflects the complexity of solving this problem. Through the Gantt chart, you can clearly understand the start time and end time of the workpieces processed on each machine and the processing of each process. If the optimal solution of the scheduling is not unique, the scheduling algorithm can also search all of them, and the optimal scheduling result of the example is not unique. FIG. 1 is only one of the best scheduled Gantt graphs. 
The following is a representation of each of the matrices, where the machine sequence is Jm; the processing time array is T; and the workpiece array is MJ.
(1) The machine sequential array Jm, Jm(i, j) represents the machine number of the jth operation of the machining i workpiece.
(2) Processing time array T, T(i, j) is the processing time of j workpiece on the machine.
(3) The workpiece array MJ, MJ(i, j) is the workpiece number of the jth machining on the i machine. 
The experiment was calculated 10 times using GATS and GA in the simulation environment, and the minimum value of the objective function was recorded each time, as shown in Table 1. The results of Table 1 show that the tabu search algorithm reaches the minimum value twice in 10 operations, and the deviation between the worst value and the optimal value is only 3, and the fluctuation is not large. 
In order to verify the validity of the algorithm, in the case where the crossover algorithm and the mutation algorithm and some parameter settings are the same, the optimal solution obtained by repeating the operation 10 times by the conventional genetic algorithm is 95, and the deviation from the worst value is 8. At the same time, it can be compared that the tabu search hybrid genetic algorithm has better stability than the traditional genetic algorithm.
On the basis of fully analyzing the characteristics of production scheduling in the rolling stock industry, the hybrid genetic algorithm and genetic algorithm combined with the tabu search algorithm are applied in production scheduling. The experimental data show that the improved hybrid genetic algorithm not only preserves the respective advantages of the genetic algorithm and the tabu search algorithm, but also improves the respective deficiencies, so that the hybrid genetic algorithm overcomes the dependence of the tabu search algorithm on the initial solution and overcomes the problem. The genetic algorithm has the disadvantages of poor mountain climbing ability and easy precocity. Experiments show that the hybrid genetic algorithm proposed in this paper can improve the original algorithm and is feasible.
Rainproof Cummins Diesel Generator
Waterproof Diesel Generator,Rainproof Cummins Diesel Generator,600Kw Cummins Rainproof Generator,600Kw Rainproof Cummins Diesel Generator
Jiangsu Lingyu Generator CO.,LTD , https://www.lygenset.com