Fault diagnosis method based on Stacked LSTM

Abstract

The invention discloses a fault diagnosis method based on Stacked LSTM (Long Short Term Memory). The method comprises the following steps: acquiring measurement data of each measurement device of a chemical system in a normal state and K fault states and acquiring training samples by construction , then, creating a fault diagnosis model based on Stacked LSTM; wherein the model comprises a StackedLSTM network, a full connection layer and a softmax layer, wherein the Stacked LSTM network is obtained by superposing D layers of LSTM networks; and training the fault diagnosis model by adopting thetraining sample, collecting actual operation data in the operation process of the chemical engineering system, constructing an input sequence, and inputting the input sequence into the fault diagnosis model based on Stacked LSTM to obtain a fault identification result. The Stacked LSTM network is formed by stacking multiple layers of LSTM networks, the dynamic time sequence information of the original data can be automatically extracted under different time scales, and the method has relatively high expression capability for complex nonlinear data, so that the accuracy and robustness of faultdiagnosis are improved.

Description

technical field [0001] The invention belongs to the technical field of chemical process fault diagnosis, and more specifically relates to a fault diagnosis method based on StackedLSTM. Background technique [0002] At present, with the increasing size and complexity of modern chemical processes, the operating conditions and operating environments are becoming more and more changeable, which not only improves the production efficiency of enterprises, but also increases the probability of process failures. If a fault occurs during the production process, it may cause huge economic losses or even personal injury. Therefore, to ensure the reliable and efficient operation of the production process, accurate and timely fault diagnosis technology is required. [0003] In recent years, with the development of machine learning, pattern recognition, artificial intelligence, etc., data-driven fault diagnosis methods have become a research hotspot in the field of fault technology resear...

AI Technical Summary

Problems solved by technology

However, the existing traditional fault diagnosis technology does not consider the timing relationship between observed signals, and assumes that the signals collected at each moment are independent of each other, making it difficult to accurately extract timing information

Images