基于改进BP神经网络的机床基础部件可再制造性评价模型

为了利用样本数据准确完成机床基础部件可再制造性评价，提高机床基础部件可再制造性评价预测精度，提出一种采用模拟退火遗传算法优化BP神经网络的机床基础部件可再制造性评价模型。该评价模型以机床基础部件可再制造性经典评价模型评价结果为样本数据，建立机床基础部件可再制造性评价BP神经网络预测模型，采用模拟退火遗传算法优化BP神经网络模型，寻找更优初始网络权值、阈值，以提高收敛速度和避免局部收敛。以一台机床基础部件可再制造性评价为例，验证了基于模拟退火遗传算法优化的BP神经网络评价模型具有更好的预测精度。

Evaluation Model for Machine Tool Basic Parts Remanufacturability Based on Optimized BP Neural Network

To utilize sample data to accomplish the remanufacturability evaluation of the machine tool basic parts, and to improve the prediction accuracy of remanufacturability evaluation of the machine tool basic parts, a BP neural network remanufacturability evaluation model optimized by the simulated annealing algorithm and genetic algorithm was proposed. A BP neural network remanufacturability evaluation prediction model of the machine tool basic parts was built according to the evaluation results of typical remanufacturability evaluation model. The BP neural network evaluation model optimized by the simulated annealing algorithm and genetic algorithm has better initial weights and thresholds to increase the convergence rate and avoid the local convergence. Remanufacturability evaluation of a machine tool basic parts was taken as an example to demonstrate that the remanufacturability evaluation model optimized by simulated annealing algorithm and genetic algorithm has higher prediction accuracy.