基于组合方法的短期燃气负荷预测技术探究

Research on Short-term Gas Load Forecasting Technology Based on Combination Method

燃气负荷预测能够为燃气公司制定科学的供气方案、提高燃气使用效率、保证自身经济效益提供必要的支持。通过构建负荷预测模型,以历史负荷数据为基础,寻找燃气使用变化规律,可以对未来一定时间段内的燃气负荷值进行准确预测。本文首先介绍了常用的几种短期负荷预测技术,如SVM支持向量机预测、BP神经网络预测、小波分析法等。随后提出了一种将神经网络与小波分析相结合的负荷预测理论,并对该组合方法的负荷预测方式、算法流程进行了简要概述。通过设计仿真实验,证明了小波神经网络在短期燃气负荷预测方面,平均绝对误差、标准误差更小,预测结果精度更高。

Gas load forecasting can provide necessary support for gas companies to formulate scientific gas supply plans,improve gas use efficiency,and ensure their own economic benefits.By constructing a load forecasting model,based on the historical load data,to find the change law of gas usage,it is possible to accurately predict the gas load value within a certain period of time in the future.This paper firstly introduces several commonly used short-term load forecasting techniques,such as SVM support vector machine forecasting,BP neural network forecasting,wavelet analysis and so on.Then,a load forecasting theory combining neural network and wavelet analysis is proposed,and the load forecasting method and algorithm flow of the combined method are briefly summarized.Through the design of simulation experiments,it is proved that the wavelet neural network has smaller mean absolute error and standard error in short-term gas load prediction,and the prediction result is more accurate.