基于Bi-LSTM的质量控制图模式识别

为提高制造过程质量智能控制的控制效果,提出了一种基于双向长短时间记忆网络(BidirectionalLSTM,Bi-LSTM)的控制图失控模式识别方法。文中分析了其分类的基本原理,构建了控制图模式识别模型,并通过蒙特卡洛仿真方法生成仿真数据集,进行仿真实验验证。仿真实验结果表明,Bi-LSTM用于控制图模式识别,准确率相对多层感知机(MLP)、贝叶斯分类器有了显著提升,相比支持向量机(SVM)具有效率上的明显优势,且在大样本下识别准确率更高。

Quality Control Chart Pattern Recognition Based on Bidirectional LSTM

In order to improve the control effect of intelligent control of manufacturing process, a method of recognition for out-of-control patterns in control chart based on Bidirectional LSTM (Bi-LSTM) is proposed in this paper.The basic principle of LSTM for classification is analyzed first.Followed by that, the pattern recognition model of control charts is constructed and a simulation data set is generated by the Monte Carlo simulation method for training and testing.The simulation experiments results show that the recognition accuracy of Bi-LSTM on control chart patterns is significantly superior than those of multi-layer perceptron (MLP) and bayesian classifier.In comparison with support vector machine (SVM), it also shows obvious advantages in efficiency and gets higher recognition accuracy under the case of large samples.