Prediction of Optimal Rescheduling Mode of Flexible Job Shop Under the Arrival of a New Job

Abstract

Aiming at the arrival of a new job in the flexible job shop, how to quickly and accurately select the optimal rescheduling method was investigated. And improved probabilistic neural network (PNN) model was used to intelligently decide the method of rescheduling. Simulation methods is used to solve the difficulty of obtaining all on-site samples. The problem is caused by the complexity of the on-site working environment. In simulation experiments, a large number of labeled samples were obtained based on performance indicators and stability indicators. Part of the samples were used to train the PNN model to improve the accuracy of the prediction results. The genetic algorithm was used to improve the smoothing factor of the PNN, and the experimental results indicated that the accuracy of the rescheduling method was significantly improved by using the proposed model.