Research On Maintenance Resource Scheduling Optimization For Military Vehicles Based On Hybrid Evolutionary Algorithm

The scheduling optimization for the special vehicles maintenance support resources isthe core in the equipment maintenance and support system. Unreasonable allocation ofresources often leads to part of resources being critical shortage or excessive concentration,directly influencing the equipment in good combat readiness and maneuver capability. For along time, when the equipment maintenance and service institutions face the detailedservice support task configuration or schedule the service support resources, they always goby historic records and experimental data to optimize with the direct computing method, theanalog calculation method and traditional network planning technique. In this case, theexperience is more than the theory supporting, so the accuracy is bad. In recent years, someresearchers have studied the related intelligent optimization algorithm, and have made someachiEvements. HowEver, the efficiency and the performance of the algorithm need to beimproved, due to the particularity and the complexity of the special vehicle maintenancesupport resurce scheduling, as well as make the algorithm structure design conforms to theactual maintenance support resource scheduling.The optimized scheduling for the special vehicles service support resources belongs tothe resource-constrained project scheduling problem (RCPSP). But the former has toomuch constraint condition to use the RCPSP algorithm. When the conditions of the modelchange a little, the original algorithm may no longer apply. This paper studies on themaintenance support resource scheduling oprimizaation problems from four typicalproblems and designs a bybrid particle swarm-genetic algorithm. The improved algorithmhas been validated with the actual maintenance support resource scheduling optimizationproblem.（1）For the “Duration fixed-Resource leveling” scheduling optimization problem.This paper studies on the maintenance support resources levelling scheduling optimizationproblem, its process of operation time is “certain” and optimization goal is “resourcelevelling”. It studies the existing algorithms, and designs the hybrid particle swarm-geneticalgorithm based on the dynamic float. It reconstructs the particle swarm optimizationalgorithm using the genetic algorithm by adopting the elite reserved strategy and mutationoperation, and improves its Evolution strategy. And it makes the improved algorithmoptimize human resources configuration and process follow of the repair and engineeringvehicle’s deployed operation. The rescource intensity of the optimal solution is reduced by89.6%than the initial scheme, with the demand of human resources is reduced by12.5%.Finally, the results, compares with genetic algorithm and branch and bound approximationalgorithm, confirmed the feasibility and effectiveness of the algorithm in this paper.(2) For he “Resource constrained-minimum duration” scheduling optimizationproblem. This paper studies on the maintenance support resources constrained schedulingoptimization problem, its process of operation time is “incomplete certain” andoptimization goal is “the minimum duration”. It presents a kind of dual real number codingscheme based on processes and human resources, takeing one type of special vehicle’s three-level maintenance process for example, and gives a new parallel scheduling schemeused to decode. It introduces logistic mapping technology to the algorithm, which can helpthe algorithm to skip the local optimum by the ergodicity of chaotic systems, and gives aHybrid Chaos Particle Swarm Optimization. And, for the characteristic of encoding, itimproves the update strategy by the elite reserved strategy and using crossover operatorinstead of standard particle swarm update mechanism on the basis of the general population.The application of the hybrid chaos particle swarm algorithm optimizes the specialvehicle’s three-level maintance process, the working hours per person being reduced from13h to8.8h and the efficiency being increased to32.3%. It analyzes the optimal schedulingscheme by comparing with different quanlity and different level mechanic configuration,and, compares with generation algorithm and the algorithm of priority rules based onmaintenance experience. The result shows that the algorithm is superior to the other twoalgorithms. The core algorithm organized by this method has been applied into the ServiceReform Project—“The Vehicle Service Technological Process Program AssistantDecision-Making System”.(3) For the multi-objective optimization scheduling problem. By three Evaluation indexin the resource scheduling process concerned by decision makers, such as the maximumcompletion time, the total load of human resource and the key human resources load, itestablishes the multi-objective repair and maintenance allocation of resource schedulingmodel based on the research of “Duration fixed-Resource leveling” and “Resourceconstrained-minimum duration”. Based on the Pareto optimal solutions, it improves thehybrid chaos particle swarm algorithm on parallel scheduling scheme, fitness assignment,the selection of particle individual extremum, the selection of particle species extremum,and particle swarm update strategy, according to the characteristics of the multi-objectiveoptimization problem. The algorithm is designed to optimize the special vehicle’sthree-level maintance process and compared with multi-objective generation algorithm,which prove the feasibility and availability of the algorithm. Finally, it assesses the Paretooptimal solution by the werghting method and chooses the best proposal met the decisionmaker’s requirement.(4) For the multi-objective fuzzy period scheduling optimization problem. This paperproposed fuzzy period maintenance support resources scheduling optimization problembased on the certain resource-constrained scheduling optimization problem in the formerresearches. It introduces the fuzzy period concept into the maintenance support resourcesscheduling model, and gives three Evaluation index, such as fuzzy completion time, the keyhuman resources load and the agreement index. As to the scheduling problem with bothfuzzy processing time and fuzzy period, it adopts the common triangular fuzzy number torepresent the fuzzy processing time and fuzzy completion time of the process and thetrapezoidal fuzzy number to represent the expected period for the project, and designs thefuzzy parallel scheduling scheme. With the improved hybrid particle swarm-geneticalgorithm for the special vehicle’s three-level maintance process fuzzy scheduling, itanalyzes the results of the Pareto optimal solution and the influence under different indexeson the scheduling scheme. And the result shows that the solution matches the practicalsituation, its operability is much more than the accuracy scheduling. It provides the theory support for making the reasonable scheduling scheme by the scheduling department.Finally, it sums up all works done and innovation points in this paper, prospects thefuture researching direction.