基于PSO-LSSVM算法的造纸过程短期电力负荷预测模型

Short-term Power Load Forecasting Model for Papermaking Process Based on PSO-LSSVM Algorithm

中文关键词:  数学建模  短期预测  电力负荷  最小二乘支持向量机  粒子群优化

Key Words:mathematical modeling  short-term forecasting  power load  LSSVM algorithm  PSO algorithm

中文摘要:

      对造纸厂的用电负荷进行预测有利于对生产调度进行合理安排，从而降低能耗。本课题提出了一种粒子群优化算法（PSO）和最小二乘支持向量机（LSSVM）相结合（PSO-LSSVM）的短期电力负荷预测方法，该方法可对造纸厂未来每30 min的电力负荷进行预测。结果表明，采用PSO-LSSVM算法对短期电力负荷进行预测时，预测结果的相对百分误差绝对值的平均值约为0.75%，精度高于其他行业的电力负荷预测值，模型具有良好的可行性和有效性。

Abstract:

      Papermaking process consumes large amount of electricity for production.The forecast of the power load for the paper mill is conducive to the production scheduling and energy consumption reduction.A short-term power load forecasting method based on least-squares support vector machine (LSSVM) and particle swarm optimization (PSO) algorithms was proposed, which was used to forecast the power load for the next half hour in the paper mills.Compared with the industrial data collected from a paper mill, the forecasting performance showed that the mean relative error of the proposed PSO-LSSVM model was around 0.75%, which demonstrated good feasibility for the papermaking process.