Research On Condition Evaluation Of Power Equipment Based On Big Data Analysis Technology

The traditional method of power transmission and transformation equipment condition assessment are generally based on the mechanism and causality model established by theoretical analysis,simulation and experimental testing,etc,and the construction and development of smart grid promote the state evaluation to the direction of information integration and comprehensive analysis based on panoramic state.This paper intends to use a large number of equipment status information,power network operation information and environmental meteorological information collected by the increasingly perfect power information platform to study the relationship and trend of equipment status evolution from the perspective of inherent law of data itself.Big data analysis method for equipment status assessment and anomaly detection provides a new solution ideas and technical means.In this paper,multi-statistic analysis,correlation analysis and other big data analysis techniques for equipment panoramic state information(standing book,operating records,online monitoring,testing,power grid operation,environmental weather,etc.)are proposed to establish a static,quasi-real-time,real-time parameters of the power equipment key parameters system.According to the real-time parameters,we use the multivariate time series association pattern mining method to extract the parameters that the sequence itself has a strong correlation and the rate of change of the sequence is related.For the static and quasi-real-time parameters,we use principal component analysis to achieve the reduction of parameters.Finally,the validity of the key parameter system and the accuracy of the evaluation are verified by an example,which can provide data support for advanced applications such as data cleaning,anomaly detection and status evaluation.Big data technology,including time series analysis,clustering analysis and neural network method,provides a new idea for data cleaning and anomaly detection of power transmission and transformation equipment.In the aspect of data cleaning,this paper studies the double-iterative cycle cleaning method of state monitoring data based on time series analysis.The principle is to use time series model to identify time series of each state quantity,and to identify and classify outliers with association rules.To achieve the purpose of correcting the noise data and filling the missing values,to complete the state of the amount of data cleaning.In the aspect of anomaly detection,the continuous monitoring data is discretized by unsupervised learning methods such as time series and SOM neural network model.The iterative learning of the state monitoring parameters is analyzed by repeated iterative learning.The method of anomaly detection of state monitoring data is put forward to make it suitable for the state monitoring data flow of power transmission and transformation equipment,which can quickly detect the abnormal data flow and solve the problem that the traditional threshold judgment method is difficult to accurately detect the abnormal problems of power equipment.As a new large data analysis method,high dimensional stochastic matrix theory can integrate various kinds of state monitoring data into high dimensional matrix,and study the characteristics and data distribution of matrix from the perspective of probability and statistic.Based on the construction method of high-dimensional stochastic matrix theory,the high-dimensional matrix of each key performance is constructed.According to the distribution of the eigenvalues,the correlation of the matrix elements and the maximum eigenvalue of the maximum eigenvalues are obtained by using the correlation analysis to extract the state quantity of the key performance of the equipment.The distribution of the characteristic function of the device is realized by using the statistic of the eigenvalue in the ring.The distribution of the characteristic vector elements is used to detect the rows and columns of the anomalies in the matrix,including abnormal state and abnormal moments.The big data analysis algorithms proposed in this paper are applied in the "big data state evaluation system of power transmission and transformation equipment".The data collected in the field are used to test and apply the data cleaning,anomaly detection and state evaluation proposed in this paper.The results show that the method proposed in this paper improves the accuracy and real-time performance evaluation of equipment,and can prepare equipment for identifying abnormal operation and provide effective decision support for condition maintenance.