基于一维卷积神经网络深度学习的工业过程故障检测

针对化工过程的非线性和动态性，以TE过程为背景，应用深度学习中的一维卷积神经网络算法对TE过程进行故障检测，解决了BP神经网络算法用于故障检测时测试识别率低的问题。用训练数据集分别对BP神经网络模型和一维卷积神经网络模型进行训练，将测试数据集输入已经训练好的神经网络，最后统计出了BP神经网络模型和卷积神经网络模型对故障的识别率。仿真结果表明BP神经网络和卷积神经网络对故障的检测具有较好的效果，但BP神经网络算法收敛速度慢，很容易就陷入局部最小值，从而会导致整体的检测性能下降，而卷积神经网络构建出的一维卷积模型能很好地解决存在的问题，通过比较充分体现了卷积神经网络在故障检测方面的优越性。

Fault Detection of Industrial Process Based on Deep Learning of One-Dimensional Convolution Neural Network

Aiming at the non-linearity and dynamics of chemical process,the TE process is used as the background and the one-demension convolutional neural network algorithm in the field of deep learning is used to detect faults in TE process,which solves the problem of low detection rate of traditional fault detection algorithm.The BP neural network model and the one-dimensional convolutional neural network model were trained by the training data set respectively and then the test data sets were input into the trained neural network.Finally,the fault recognition rate of the BP neural network model and the convolutional neural network model is counted.The simulation results show that the BP neural network and the convolutional neural network have good fault detection effects,but BP neural network algorithm with slow convergence speed is easy to fall into local minimum,which leads to the decline of the overall detection performance.However,the one-dimensional convolution model constructed by convolutional neural network will not have these problems.The comparison between the two fully demonstrates the superiority of convolutional neural network in fault detection.