A Method of Electrical Equipment Fault Prediction Based on Multidimensional Time Series

Abstract

A method for predicting equipment faults based on multi-dimension time series presents a data mining method of multi-dimension time series on the basis of high intensive sampled online operating electrical measurement data in a power system; a time series decomposition algorithm, a feature event generation algorithm and an association rule-based fault association relation mining algorithm are set up through historical time series training data; the variation features of other equipment associated with the fault, namely 'forewarning events', are mined out; an equipment fault predication model is formed based on the relationship, and in combination with online monitoring data, powerful support is provided for fault prediction and judgment of complex non-linear electrical equipment. Massive high-density operating monitored historical data of the equipment can be made an effective use by above method to predict beforehand faults or impacts that probably occur in the core equipment of a power enterprise, so that prevention measures can be taken in time to avoid such faults or impacts.

Description

technical field [0001] The invention belongs to the interdisciplinary research field of computer technology and electric power, and specifically proposes a multi-dimensional time series-based equipment failure prediction method for electric power systems. Background technique [0002] In the power industry, some equipment is large-scale equipment that maintains the operation of the power grid, such as transformers in substations, steam turbines, generators, and excitation systems in power stations. The normal progress will also cause huge losses. Serious accidents of large steam turbines at home and abroad are typical examples. Therefore, in order to take preventive measures in time and avoid unnecessary losses, it is very important to predict the failure of these core devices. [0003] The traditional time series forecasting is to use a linear model to fit the data series, which has a good result for the linear system, but it is not suitable for the forecasting of the non...

AI Technical Summary

Problems solved by technology

In the current research, the transient data generated when the fault occurs, such as wave recording files, alarms, etc., are often used for relatively independent single analysis, and it is difficult to realize the prediction of these faults and impacts

Images