Spare Parts Requirements Predictive Model Study Based On Group Intelligence Algorithm

As an important part of manufacturing service value chain coordination,spare parts demand prediction plays a key role in promoting the transformation from production-oriented manufacturing to service-oriented manufacturing.At present,the traditional prediction model of spare parts demand has some problems,such as weak pertinence of prediction model and poor generalization ability of prediction method,which leads to low accuracy of prediction.Therefore,this thesis takes the optimization and improvement of swarm intelligence algorithm as the technical means,takes the construction of spare parts demand prediction model as the core,and takes the spare parts demand prediction of CRH train as the base point to carry out research.The research content of this thesis is as follows:(1)A elimination mechanism and chaotic search particle swarm optimization algorithm(ECPSO)is proposed.Aiming at the problem of premature convergence of particle swarm optimization,an inertia weight calculation method based on iteration times and fitness value is proposed to improve the global search ability and local search ability of the algorithm.In order to enrich the diversity of the algorithm population,an elimination mechanism is introduced to update the particles in the population.In order to improve the convergence accuracy of the algorithm,combine with chaos theory,chaotic search is carried out for global optimal solution.The experimental results show that the convergence performance and the stability of the ECPSO algorithm are improved.(2)A differential search and opposition-based learning squirrel search algorithm(DOSSA)is proposed.In order to improve the optimization efficiency of the squirrel search algorithm,a position updating method with a quadratic reverse search strategy is proposed,and an adaptive predator probability based on the number of iterations is adopted to balance the exploration capability and exploitation capability of the algorithm.Aiming at the problem that the algorithm is easy to fall into the local optimal solution,a differential search strategy for the optimal solution is proposed based on the difference idea to improve the ability of the algorithm to jump out of the local optimal solution.The experimental results show that the convergence accuracy and convergence speed of DOSSA on the test function are better than those of the comparison algorithm.(3)The two improved algorithms are applied to the parameter optimization of spare parts demand prediction model.Aiming at the continuous spare parts demand forecasting problem,a spare parts demand forecasting model is constructed based on BP neural network optimized by ECPSO algorithm(ECPSO-BP).Aiming at the demand prediction problem of intermittent spare parts,a spare parts demand prediction model is constructed based on support vector regression optimized by DOSSA algorithm(DOSSA-SVR).The experiments on the CRH train spare parts demand dataset show that the goodness of fit of the two spare parts demand prediction models constructed in this thesis are 0.802 and 0.815 respectively,which are better than the selected comparison models.In summary,the thesis focuses on the performance improvement of the spares demand prediction model,firstly proposes two improved swarm intelligence algorithms,and then uses the improved algorithm to optimize the core parameters of the spares demand prediction model.Finally,experiments verify that the improved algorithm can effectively improve the prediction accuracy and stability of the spares demand prediction model.