【摘要】 为了解决已有的基于小波-卡尔曼滤波的短期负荷预测方法由于未考虑温度积累效应而在温度变化较大时预测误差偏大的问题,提出了一种改进方法:将日负荷表示为日平均负荷与波动部分的乘积,对日平均负荷和波动部分分别进行预测。提出了一种利用人工神经网络预测平均负荷的新方法:将日平均负荷表示为温度敏感分量与平稳的温度不敏感分量之和。温度不敏感分量根据温度不敏感季节同时期的若干负荷数据统计得出。根据前若干天的温度敏感分量值、温度信息以及预测日的温度信息,采用BP网络构成的负荷预测器,得出预测日的温度敏感分量的预测值。对于波动部分沿用基本的小波-卡尔曼滤波的方法,在对波动部分进行多分辨分析的基础上,将小波系数作为状态变量,利用卡尔曼滤波算法得出波动部分的预测值。实例分析表明,提出的改进方法显著提高了预测准确性。更多还原

【Abstract】 Without considering the temperature accumulation effects, the errors of the existing wavelet-Kalman filter based load forecasting method are rather large when temperature changes rapidly. To solve such problem, a load was expressed as the product of an average component and a varying component. Such two components were forecasted separately. A new method based on artificial neural networks was proposed and applied in forecasting the average load. The average load was expressed as the summation of a temperature-insensitive part being stationary and a temperature-sensitive part. The former was obtained according to the data of temperature insensitive seasons. The latter was obtained by a BP network whose inputs were history information of load, temperature (including the averaged temperature, the highest temperature and the lowest temperature) and the temperature of the predict day. The varying component was forecasted by basic wavelet-Kalman filter based method. The wavelet coefficients of the varying load component were obtained by the decomposition scheme of multi-resolution analysis. The wavelet coefficients were modeled as the state variables of the Kalman filter. The predicted varying load component was obtained by the recursive Kalman filter algorithm. Load forecasting results show that the improved method increases the forecast precision.更多还原