Fault diagnosis method and system combining multi-point statistics and health trend prediction

Abstract

The invention provides a fault diagnosis method and system combining multi-point statistics and health trend prediction, and the method comprises the steps: determining a fault detection method according to the fault features of a to-be-detected device; according to the fault detection method, determining a time delay post-judgment interval length; initializing the preset parameters; acquiring sensor output data of the to-be-detected device, wherein the sensor output data comprises measurement data of the to-be-detected device at a target moment; determining a corresponding device health change trend prediction method according to the self characteristics of the to-be-detected device and the measurement data, and predicting a health change trend according to the device health change trend prediction method to obtain a failure index sigma of the to-be-detected unit; performing single-point fault detection on the to-be-detected unit at the current moment; updating the preset parameters, and judging whether the later judgment interval is ended or not; and if so, analyzing and diagnosing the fault of the to-be-detected device to obtain a fault diagnosis result. Therefore, the accuracy of fault diagnosis can be improved.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a fault diagnosis method and system combining multi-point statistics and health trend prediction. Background technique [0002] More and more attention has been paid to the safety and reliability of complex systems, leading to the rapid development of fault detection and diagnosis technology. [0003] At present, detection methods for various types of measurement data and various types of fault conditions are becoming more and more mature, and some single-point detection methods have been widely used in engineering applications. However, due to the influence of the actual working environment and sensor measurement errors, the fault detection algorithm for a single point of time usually cannot avoid a certain degree of false alarm or missed detection. The requirements for fault diagnosis methods are not limited to the ability to detect faults under ideal conditions, but a...

AI Technical Summary

Problems solved by technology

Multi-point continuous detection avoids the misjudgment of fault diagnosis caused by single-point false alarm caused by sensor measurement error, and has a low false alarm rate, but the single-point missed detection caused by random error also increases the leakage of continuous multi-point diagnosis method. The detection rate is low, and the detection sensitivity for slowly changing faults of the sensor is low

[0004] To sum up, the multi-point continuous diagnosis method widely used at present has a high failure detection rate.

Images