Real-time valve fault diagnosis system and method based on time sequence analysis

Abstract

The invention relates to a real-time valve fault diagnosis system and method based on time sequence analysis, and belongs to the technical field of valve fault detection and diagnosis. The method concretely comprises that a vibration accelerometer and a sound pressure sensor monitor a valve in real time; a signal collection chip collects monitored vibration acceleration and sound pressure signalsto a diagnosis processor; five-point three-times smooth denoising is carried out on the data; diagnosis data training sets of different fault diagnosis types are generated; a fault diagnosis sequential model is established; a forward algorithm is used to calculate the likelihood probabilities of different HMM models successively, and a fault diagnosis result is obtained; and a Bluetooth model transmits the fault diagnosis result to an industrial display screen for display. Fault of the valve is diagnosed by time sequence analysis, fault information can be identified and the corresponding faulttype can be obtained at initial stage of the fault, the limit that fault can only be observed in single time is changed, the false detecting rate of faults is reduced, and the using performance of the valve is substantially improved.

Description

technical field [0001] The invention belongs to the technical field of valve fault detection and diagnosis, and in particular relates to a valve fault real-time diagnosis system and a diagnosis method based on time series analysis. Background technique [0002] In most industrial fields, the valve is one of the essential components. If the valve fails seriously, the entire operating system will not work normally, which may easily lead to serious work accidents. Time series is a sequence formed by sorting the values ​​of some statistical indicators in chronological order. The identification and prediction of time series is to analyze the time series, and according to the development process, direction and trend reflected in the time series, carry out analogy or extension, identify or predict the level that may be reached in the next period of time or in the next few years. The anomaly detection of time series is to analyze the historical data to check whether the current dat...

AI Technical Summary

Problems solved by technology

At the same time, the existing methods can only detect whether the valve is faulty, but cannot obtain the type of fault, which has great limitations and is inconvenient for later maintenance.

Images