Power system measurement missing value reconstruction method based on depth learning and application thereof

Abstract

The invention relates to a power system measurement missing value reconstruction method based on depth learning. The reconstruction method comprises the following steps: cleaning and selecting the measured historical data from SCADA to obtain a training set of a neural network; an improved generative countermeasure network based on Wasserstein distance being constructed and trained; Adam being chosen as the optimizer to optimize the latent variable of WGAN network, and the improved GAN based on Wasserstein distance being obtained; the measurement data with missing values and corresponding binary mask matrix being inputted into the WGAN network model, and the reconstructed measurement data being finally obtained. The method can be applied to data cleaning and correction or to reconstructionof lost measurements when the system is attacked by communication. As the neural network of the invention automatically learn the complex spatio-temporal relationship which is difficult to be explicitly modeled such as the correlation between measurement and the load fluctuation rule, the feasibility of taking the reconstructed data as a pseudo measurement is ensured.

Description

technical field [0001] The invention belongs to the technical field of power system data cleaning and restoration, and in particular relates to a method for reconstructing missing values ​​of power system measurements based on deep learning and its application. Background technique [0002] Massive measurement devices in the power system constitute a complex data acquisition and monitoring control system SCADA. The measurement data collected by SCADA system is of great significance to power system state estimation, power equipment evaluation, and system operation optimization. Especially in recent years, with the vigorous development of big data technology, the transmission, storage and analysis of power grid data has become an important research direction. [0003] However, only the research conclusions based on real and reliable collected data have practical application value and can correctly reflect the operating characteristics and objective laws of the power system. ...

AI Technical Summary

Problems solved by technology

However, when the number of missing measurements is large and the observability conditions are not met, the state estimation itself cannot be calculated, and this processing method is no longer applicable

There are also literatures that repair missing data through mathematical methods such as mean value filling method, hot and cold card filling method, regression filling method, and closest distance filling algorithm. However, this processing method only analyzes from the perspective of data distribution and ignores the timing characteristics of measurements in power systems. , the correlation between different measurement points, the correlation between measurement variables and the historical load change law, the reconstruction effect of missing measurement data in the power system is not ideal

[0004] The timing characteristics of measurements in the power system, the correlation between measurements, and the law of load changes can all be used as important basis for missing data reconstruction, but the difficulty lies in the complex time-space relationship between these factors, which makes it difficult to use a clear mathematical model. Modeling description

[0005] Therefore, based on these problems, a method is provided to automatically learn complex distribution laws between data in an unsupervised form through neural network training, and then generate new data that satisfies objective laws, thereby solving the problem of data repair. The value reconstruction method can solve the problem of high data distribution dimension and complex modeling

Images