Prediction of order completion time based on the BP neural network optimized by GASA

Accurate prediction of order completion time (OCT) is an important guarantee for the workshop to dynamically adjust production plans and ensure timely delivery of products. In order to improve the arracy of predicting the OCT based on the BP neural network, this paper proposes a prediction method of the optimized BP neural network based on the genetic algorithm and the simulated annealing algorithm (GASA-BP). Because the simulated annealing algorithm (SA) has strong local searching ability and can avoid falling into the local optimum in searching process, the Metropolis acceptance criteria in SA is introduced to GA, which compares the new fitness of GA with the fitness of the last iteration to find the optimal solution. The improved GA continuously optimizes the weights and thresholds of the BP neural network and we use the trained BP neural network to obtain the optimal prediction value. Finally, the performance of the proposed GASA-BP in predicting the OCT is compared with the optimized BP neural network based on the genetic algorithm(GA-BP). And simulation experiments demonstrate the accuracy and feasibility of GASA-BP.