Particle swarm optimization-based multi-resolution wavelet neural network power consumption prediction method

Abstract

The invention relates to a particle swarm optimization-based multi-resolution wavelet neural network power consumption prediction method, and belongs to the field of neural network prediction research. According to the technical scheme, the method involves contents such as a wavelet neural network, a particle swarm optimization-based wavelet neural network and a particle swarm optimization-based multi-resolution wavelet neural network; the wavelet neural network, a particle swarm algorithm and multi-resolution analysis are combined; and high efficiency of the prediction method is verified by predicting power consumption. The feasibility and high efficiency of the prediction method are verified specifically through a simulation experiment: a wavelet neural network prediction method, a particle swarm optimization-based wavelet neural network prediction method and a particle swarm optimization-based multi-resolution wavelet neural network prediction method are compared. A target function and a regression analysis graph obtained by analysis can clearly show that by adopting the last prediction method, a target function value is converged more quickly, the prediction precision is higher, the influence caused by crossed overlapping of neurons in a hidden layer is effectively avoided, falling into local minimum is avoided, and the prediction effect is better.

Description

technical field [0001] The invention relates to a multi-resolution wavelet neural network electricity consumption prediction method based on particle swarm optimization, which belongs to the research field of neural network prediction. Background technique [0002] Under the environment of rapid development of the smart grid, the power grid is moving towards the general direction of networking, digitalization, integration, and standardization, so the requirements for power load forecasting are gradually deepening. The forecast of electric load includes the forecast of electricity sales and consumption. Electricity sales refer to the electricity sold by power enterprises to users and the electricity used by the non-electric production infrastructure, overhaul and non-production departments of the enterprise; electricity consumption refers to the electricity sold by the power grid and the electricity generated and used by self-provided power plants. The sum of electricity and...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a multi-resolution wavelet neural network electricity consumption prediction method based on particle swarm optimization, aiming at the slow convergence speed, low prediction accuracy and overlapping of hidden layer neurons, which are easy to fall into local For extremely small problems, the idea of ​​particle swarm algorithm and multi-resolution analysis is added to the framework of wavelet neural network, and the advantages of both are combined to establish a model based on particle swarm optimization and multi-resolution wavelet neural network to predict power consumption

Images