【摘要】 针对铣刀磨损量预测精度低的问题，提出一种高精度铣刀磨损量预测方法。该方法通过遗传算法(GA)寻出长短期记忆网络(LSTM)的最优参数，并将参数输入LSTM实现改进模型GA-LSTM。采用时域、频域及时频域方法提取特征，应用皮尔逊相关系数法筛选出与铣刀磨损量高度相似的特征向量，输入GA-LSTM模型进行训练，并对测试数据进行预测。实验结果表明：与传统的机器学习方法BPNN或深度学习方法FE-LSTM、CNN相比，GA-LSTM的均方根误差分别下降了41.3%、39.0%、51.5%,平均相对误差分别下降了48.3%、40.8%、56.7%,模型的预测识别精度有较大提高，实现了铣刀磨损量的有效预测。更多还原

【Abstract】 Aiming at the problem of low prediction accuracy of milling tool wear volume, a high-precision milling tool wear volume prediction method was proposed.The optimal parameters of the long short-term memory network(LSTM) was found through the genetic algorithm(GA),and the parameters were input into the LSTM to realize the improved model GA-LSTM.The time domain, frequency domain and frequency domain methods were used to extract features, and the Pearson correlation coefficient method was used to screen out the feature vectors that are highly similar to the milling tool wear volume, the GA-LSTM model was input for training, and the test data were predicted.The experimental results show that compared with the traditional machine learning methods BPNN or deep learning methods FE-LSTM,CNN,the root mean square error of GA-LSTM decreases by 41.3%,39.0% and 51.5%,and the mean absolute percentage error decreases by 48.3%,40.8% and 56.7%,respectively, the prediction and recognition accuracy of the model is greatly improved, and the effective prediction of the milling tool wear volume is realized.更多还原