自适应变系数粒子群和径向基神经网络在短期电价预测中的应用(英文)

Short-Term Electricity Price Forecast Model Based on Adaptive Variable Coefficients Particle Swarm Optimizer and Radial Basis Function Neural Network Hybrid Algorithm

分析了传统的粒子群优化(particle swarm optimization,PSO)算法和径向基(radial basis function,RBF)神经网络的优缺点,提出一种自适应变系数粒子群优化算法(adaptive variable coefficients particle swarm optimizer,AVCPSO)。该算法与RBF神经网络结合形成自适应变系数粒子群-径向基(AVCPSO-RBF)神经网络混合优化算法。基于此优化算法,建立了短期电价预测模型,并利用贵州电网历史数据进行短期电价预测。仿真计算结果表明,AVCPSO-RBF混合优化算法在短期电价预测中优于传统RBF神经网络法和PSO-RBF神经网络方法,克服了上述2种方法的缺点,改善了RBF神经网络的泛化能力,具有输出稳定性好、预测精度高、收敛速度快等特点,使用该方法得到的各日预测电价的平均百分比误差可控制在2%以内,平均绝对误差最大值为1.652RMB/MW·h。

The traditional particle swarm optimization （PSO） algorithm, radial basis function （RBF） neural network and their respective advantages and shortcomings are analyzed, an adaptive variable coefficients particle swarm optimizer （AVCPSO） is put forward. This algorithm is combined with RBF neural networks to form an AVCPSO-RBF neural network hybrid optimization algorithm. Based on AVCPSO-RBF neural network hybrid optimization algorithm, a short-term electricity price forecast model is set up, and is used in price forecast simulation of Guizhou power grid with historical data. The simulation results show that AVCPSO-RBF neural network algorithm is better than RBF neural network and PSO-RBF neural network algorithms during short-term price forecast, and overcomes the shortcomings of traditional algorithms. It improves the generalization capability of RBF neural network, and has good output stability, high forecast precision and fast convergence speed. It also has higher computational precision, the average percentage error is not more than 2%, and the maximum of average absolute error is 1.652 RMB/MW·h.