联合灰色模型和神经网络的短期电力负荷预测

Short Term Power Load Forecasting Based on Combination of GM and BP Neural Network

电力负荷受温度、湿度等因素影响,因而呈现出典型的非线性和随机性特征,传统单一模型存在预测精度低,泛化能力弱等问题,为此提出一种联合灰色模型(Grey Model,GM)和BP神经网络的组合模型实现对短期电力负荷的高精度预测。GM在小样本、贫信息条件下具有较强的预测能力,但是面对非线性问题时表现不佳,而BP神经网络具有强大的非线性映射能力,理论上能够以任意精度逼近于任意非线性函数,所提组合模型通过对2种方法取长补短实现优势互补,达到提升预测性能的目的。同时针对BP神经网络初值选取困难,易陷入局部极值问题,提出一种改进粒子群优化算法(Improved Particle Swarm Optimization,ImPSO)自动进行全局寻优。基于某地区实际电力负荷数据开展试验,结果表明,所提方法相对于单一模型能够获得更高的预测精度,并且对小样本、低信噪比条件具有更强的泛化能力。

The power load is affected by temperature,humidity and other factors,and presents typical nonlinear and random characteristics,which leads to problems such as low prediction accuracy and weak generalization ability of the traditional single model.A model by combining Grey Model(GM)with BP neural network is proposed to achieve high-precision prediction of short-term power load.GM has strong predictive ability under the conditions of small samples and poor information,but it does not perform well in the face of nonlinear problems,while BP neural network has strong nonlinear mapping ability,which can theoretically approximate any nonlinearity with arbitrary precision.The proposed combined model achieves complementary advantages by complementing the two methods and improves the prediction performance.At the same time,in order to solve the problem of selecting the initial value of BP neural network,an improved particle swarm optimization algorithm(ImPSO)is proposed to automatically perform global optimization.Based on the actual power load data in a certain area,the results show that the proposed method can obtain higher prediction accuracy than a single model,and has stronger generalization ability for small samples and low signal-to-noise ratio conditions.