Fault diagnosis method of multi-line sensor

Abstract

The invention relates to a fault diagnosis method of a multi-line sensor. The method comprises the following steps of: (1) sampling in two modes including a uniform sampling mode and a random sampling mode; (2) recording the state change values on various sub-lines; (3) finding the maximum bit alpha with state change; (4) normalizing data; (5) judging and locating faults: presenting that the line is normal when the formula is satisfied and presenting the line fails if the value is far larger than or smaller than 1; and (6) rechecking faults. The invention analyzes the fault characteristics under different sampling conditions, combines a fault diagnosis method based on statistical results and a fault diagnosis method based on related changes, successfully solves the fault diagnosis problem of the multi-line sensor in the practical use process, can accurately find and locate various defaults of the multi-line sensor and have a certain real-time property and low requirements for software and hardware.

Description

technical field [0001] The invention relates to a fault diagnosis method, in particular to a multi-line sensor fault diagnosis method, which belongs to the field of fault diagnosis. Background technique [0002] Aiming at the problem of sensor fault diagnosis, there are literatures based on the model of the sensor application object, using Kalman filtering and its improved form, and diagnosing the fault of the sensor by estimating the parameters reflecting the state of the sensor in the model, and some using intelligent methods and information fusion. However, in the actual application process, many objects are difficult to establish models, and the model-based method cannot be applied. In order to get rid of the limitation of the object model, some literature uses methods such as wavelet transform to analyze the output signal of the sensor, but this method Only faults can be found, but it is difficult to locate faults. Some systems need to be learned, and the real-time perf...

AI Technical Summary

Problems solved by technology

[0002] Aiming at the problem of sensor fault diagnosis, there are literatures based on the model of the sensor application object, using Kalman filtering and its improved form, and diagnosing the fault of the sensor by estimating the parameters reflecting the state of the sensor in the model, and some using intelligent methods and information fusion. However, in the actual application process, many objects are difficult to establish models, and the model-based method cannot be applied. In order to get rid of the limitation of the object model, some literature uses methods such as wavelet transform to analyze the output signal of the sensor, but this method Only faults can be found, and it is difficult to locate faults. Some systems need to be learned, and the real-time performance is not strong

Real-time diagnosis is another difficult problem in sensor fault diagnosis, especially when the output of the sensor is random, in the absence of a reference output, it is impossible to know the real value of its current point, and it is impossible to effectively locate and isolate the fault

Images