Power transformer state monitoring data cleaning method

Abstract

The invention discloses a power transformer state monitoring data cleaning method, comprising the following steps: establishing a stack noise reduction auto-encoder for monitoring data of gas in oil of the oil-immersed power transformer; performing stacking and fine adjustment processing after training each auto-encoder in the stack noise reduction auto-encoders one by one, so as to obtain a finalstack noise reduction auto-encoder data cleaning model; and finally inputting monitoring data of gas in oil of the oil-immersed power transformer into the stack noise reduction auto-encoder data cleaning model for processing to achieve the purpose of reducing noise of original data. According to the power transformer state monitoring data cleaning method, a data cleaning model is established on the theoretical basis of a stack noise reduction auto-encoder; compared with a traditional data cleaning mode based on a statistical method, the power transformer state monitoring data cleaning methodgets rid of constraints of data indexed evaluation standards from data essential characteristics, can well eliminate data noise, and provides reliable monitoring data for equipment state evaluation, equipment service life prediction and other work.

Description

technical field [0001] The invention relates to a method for cleaning state monitoring data of a power transformer. The method can eliminate data noise and provide reliable monitoring data for work such as equipment state evaluation and equipment life prediction, and belongs to the technical field of transformers. Background technique [0002] The power transformer is one of the core components in the power system. In order to ensure its efficient, durable and stable operation, many power transformer state assessment methods and fault diagnosis methods have emerged as the times require. At the same time, with the popularization of the application of various sensor technologies today, the data volume and data dimensions of the monitoring data that can be collected for power transformers have shown explosive growth, gradually showing the characteristics of multi-source and heterogeneous data. [0003] Due to the different operating conditions of power transformers and the loss...

AI Technical Summary

Problems solved by technology

[0008] At present, there are many research results related to data cleaning at home and abroad. Lou Jianlou et al. proposed an optimal intra-group variance cleaning algorithm for the wind speed and power data of wind turbines. The selection of the threshold is critical, and it is difficult to deal with high data dimensions

In the process of cleaning the abnormal data of equipment fault information, Yan Yingjie and others adopted a data cleaning method based on time series analysis, and used a double-cycle iterative inspection method to analyze the wire temperature data and gas CH in oil. 4 The monitoring can achieve the purpose of correcting data noise points and filling data gaps, and achieved good experimental results, but there is a large error between the time series cleaned by this method and the original time series near the time when the outlier occurs

Similarly, Yan Yingjie and others processed the equipment state quantity by using the time series autoregressive model of single state quantity data, and quantified the time series by using the self-organizing neural network, mining the multi-state quantity data of the equipment, and based on the equipment state quantity Compared with the method of judging whether the equipment state is abnormal by threshold value, it simplifies the complex correlation relationship of multi-dimensional parameters of equipment state quantity, but its processing results are easily affected by the abnormality of the external environment.

However, at this stage, there are few studies on the data cleaning model for the daily monitoring data of power equipment. Research on the Cleaning and Correction Method Specially Aiming at the Condition Monitoring Data of Power Equipment

Images