Power load probability prediction system and method for multi-core intelligent meter

Abstract

The invention discloses a power load probability prediction system and method for a multi-core intelligent meter, and the method comprises the steps: building a sample set through a CNN (convolutionalneural network) and a GRU (recurrent neural network); training a mixed density network through a training set, so as to optimize network parameters; verifying the trained mixed density network through a verification set; carrying out power consumption load probability prediction by adopting the hybrid density network passing the verification; wherein the hybrid density network comprises a full connection layer, a hybrid parameter output layer and a hybrid probability density function output layer which are connected step by step; and using a mixed probability density function output layer foracquiring the mixed weight, the probability density function variance and the probability density function mean value output by the mixed parameter output layer to construct a mixed density function,and taking the mixed density function as power load probability density distribution. Training efficiency is high, the uncertainty of the sample can be well learned, a highly fluctuating time sequence can be dealt with, and the prediction accuracy is improved.

Description

technical field [0001] The invention relates to the field of electric load prediction, in particular to a method for predicting the probability density function of electric load. Background technique [0002] In the context of the construction of the electric power Internet of Things, the power system can collect, record and store massive load data through the extensive deployment of smart meters. Electricity meters, load forecasting methods based on data processing and analysis benefit from this, and have more data foundation support. The accuracy of load forecasting is affected by many factors such as seasonal dynamics and regional power generation, and has certain uncertainties. How to use the massive data obtained by multi-core smart meters to build an efficient and accurate load forecasting model, and how to describe the uncertainty in load forecasting Issues such as sexuality have become research hotspots in recent years. [0003] Load forecasting methods can be main...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the deficiencies of the above-mentioned technologies, the present invention provides a probabilistic prediction method for power loads for multi-core smart meters, which solves the need to rely on the prediction results of multiple models to construct the final Technical Issues Predicting Outcomes

Images