Virtual load dominant parameter identification method based on incremental learning

Abstract

The invention discloses a virtual load dominant parameter identification method based on incremental learning. The virtual load dominant parameter identification method comprises the steps that (1) enabling virtual load model dominant parameters to be subjected to random value simulation; (2) establishing a deep learning neural network; (3) carrying out deep neural network incremental learning; and (4) carrying out online rapid identification and cyclic training. The feasibility of incremental learning applied to power system analysis is mainly introduced and is combined with load parameter identification, the training efficiency is improved while the identification precision is ensured, the storage overhead is maintained while catastrophic forgetting is prevented, a new thought is provided for processing training samples in parameter identification, and a technical support is provided for online identification of dominant parameters of the virtual load model; by means of the thought of continuous training online rapid identification, the convolutional neural network is applied to parameter identification of the load model, and online identification, continuous circulation and continuous training of dominant parameters of the virtual load model are achieved on a power grid big data platform.

Description

technical field [0001] The invention relates to the field of electric power system load identification, in particular to a virtual load dominant parameter identification method based on incremental learning. Background technique [0002] In actual power system operation monitoring, an accurate load model plays a very important role in the safe and stable operation of the power system. Different load models lead to very different results of stability calculations, even completely opposite results. Therefore, how to establish an accurate load model and obtain accurate model parameters has always been a hot topic for scholars, and has received extensive attention for a long time. There are two main methods of load modeling, which are statistical synthesis method and overall measurement and identification method. The statistical synthesis method first classifies the loads, and then counts the characteristics of each type of loads to obtain the overall characteristics of the lo...

AI Technical Summary

Problems solved by technology

However, in practical applications, it is usually impossible to obtain all the training samples at once, but to obtain them gradually over time, and the information reflected by the samples may also change over time.

If new samples need to relearn all the data after they arrive, it will consume a lot of time and space, so batch learning algorithms cannot meet this demand.

Images