A fault detection method based on automatic encoder and Bayesian network

Abstract

The invention provides a fault detection method based on an automatic encoder and a Bayesian network, belonging to the field of chemical process fault diagnosis. The off-line phase of the method is toselect variable data from any chemical continuous production process to construct chemical process data set and sample data set. The automatic encoder model is trained with sample data set and the statistic T2 and SPE detection threshold are calculated. Bayesian Networks are constructed with Chemical Process Data Set and the Conditional Probability is estimated; at the present stage, the real-time data are acquired and input into the automatic encoder model to obtain the correspond estimated value, and the corresponding T2 and SPE values of the input data are calculated and compared with thedetection threshold value: if the condition is satisfied, the chemical production process is normal; if not, the contribution of each variable is calculated, and the root cause of the fault is found through the Bayesian network. The invention automatically extracts features from process data of chemical production, effectively applies to a non-linear dynamic chemical process, and realizes the detection and rapid diagnosis of chemical process faults.

Description

technical field [0001] The invention relates to the field of chemical process fault diagnosis, in particular to a fault detection method based on an automatic encoder and a Bayesian network. Background technique [0002] The chemical production process is a complex process. In order to maximize the economic benefits, the chemical production process often requires "safe, stable, long, full and excellent", and the production equipment is required to be able to run smoothly for a long period of time. With the continuous development of automatic control technology, the control of the device is tightly coupled with the state variables. When a major variable in the chemical production process is abnormal, these abnormalities will be propagated through the device through mass transfer, heat transfer, etc., thereby This causes the entire device to fluctuate, causing the device to flood with alarms, making it difficult for operators to diagnose the fault correctly. If the operator ...

AI Technical Summary

Problems solved by technology

The encoder and decoder of traditional autoencoders are generally composed of multi-layer feedforward neural networks, which do not consider the expansion of dynamic time

Images