基于电力消耗的行业总产值预测模型及算法研究

Prediction Model and Algorithm of Industrial Gross Output Value Based on Electric Power Consumption

在"大数据"技术背景下,获取广东省规模以上工业企业电力消耗及总产值月度数据,基于人工神经网络结构建立行业总产值预测模型,并提出一种新的带抱团行为的粒子群优化算法完成对神经网络预测模型的参数优化,进而实现各行业基于电力消耗的总产值有效预测。仿真分析表明,新的改进型带抱团行为的粒子群优化算法具有更快的收敛速度和更高的寻优精度,能够有效地优化神经网络模型参数,实现基于电力消耗的行业总产值的有效、可靠预测。

In the context of big data technology,a prediction model for industrial gross output value based on artificial neural network was built by achieving monthly data of electric power consumption and output value of industrial enterprises above Guangdong provincial designated size. Meanwhile,a kind of new particle swarm optimization with gathering behavior was proposed to finish optimization on parameters of ANN prediction model and realize effective prediction on gross value based on electric consumption of all industries. Simulation analysis indicated that the new improved PSO was provided with faster convergence speed and higher optimizing precision which was able effectively optimize parameters of ANN model and realize effective and reliable prediction on gross output value of industries based on electric power consumption.