Online Monitoring Data Cleaning of Transformer Considering Time Series Correlation

A data cleaning strategy based on association rule analysis and neural network is proposed to solve the problem of data missing and abnormal data in the process of state evaluation of transformer. Firstly, a mathematical model using association rules is established to measure the degree of correlation between state monitoring quantities. Then we use the k-means clustering algorithm to detect the missing values and outliers in the sequence, and propose cleaning procedures and rules considering the sequence correlation which can distinguish the abnormal sensor data and the abnormal equipment state. We use the improved wavelet neural network model for missing data prediction and error data correction for the washable data points. The cleaning efficiency and accuracy of the network are improved.