Adaptive fault monitoring method based on +/-x sigma boundary interval model

Abstract

The invention discloses an adaptive fault monitoring method based on a + / -x sigma boundary interval model. The method comprises a preparation stage, a learning stage and a monitoring stage. Discrete boundary model interval points are formed by calculating a sample mean value and a standard deviation in real time, the discrete boundary model interval points are sequentially connected to form upperand lower boundaries of a learning stage, a boundary model finally used for monitoring is generated according to a preset boundary fusion algorithm, and finally fault monitoring is performed accordingto a pre-designed criterion strategy for faults. According to the technical scheme, the adaptability is good; the invention is simple, easy to understand, suitable for engineering popularization, small in calculated amount and suitable for online monitoring; the fault monitoring accuracy and reliability can be greatly improved, and the data utilization rate is improved; use cost is low.

Description

technical field [0001] The invention belongs to the technical field of fault monitoring, in particular to an adaptive fault monitoring method based on a ±xσ boundary interval model. Background technique [0002] As we all know, there is a common conclusion in mathematical statistics that if a random variable (such as a certain signal and a certain feature) obeys a normal distribution, then the probability of the variable being in the (μ-3σ, μ+3σ) interval is 99.73%, which belongs to the large Probability event, if a small probability event that is considered impossible to occur beyond this interval occurs, there is a high degree of confidence that some abnormal event affecting the random variable has occurred. [0003] Fault monitoring is involved in all walks of life in various fields, such as industrial production, aerospace, transportation, medical treatment, family and so on. [0004] In many fields of fault monitoring, in the face of the urgent need of online monitorin...

AI Technical Summary

Problems solved by technology

[0004] In many fields of fault monitoring, in the face of the urgent shortage of online monitoring methods, some fields do not have online monitoring methods, or the algorithm implementation is complex and the amount of calculation is large, resulting in low real-time performance, poor algorithm adaptability, and high cost. It is difficult to promote, and has been put into use Various online methods, low monitoring accuracy, poor reliability, or confidentiality is difficult to use

[0005] For example, in the field of tool status monitoring for machine tool processing in industrial production and manufacturing, many theoretical research results in this field have poor real-time performance due to the complex calculation of the algorithm (specifically due to various model algorithms), and the algorithm is affected by The experimental environment and parameters have a large impact. The adaptability is not high, and the utilization rate of high-frequency data is mostly ignored. Moreover, the high cost of algorithm implementation affects the promotion and use of projects. Although the reliability of many results (in the experimental environment) is good, it has to be sacrificed. cost or (and) real-time

Images