【摘要】 小波神经网络是一种新兴的电力负荷预测方法。研究了小波分析与BP神经网络相结合的小波神经网络,利用小波变换对负荷样本做序列分解,得到不同尺度下的小波系数,然后对小波系数进行阈值选择,由BP神经网络对作用阈值后的小波系数进行预测。同时总结历史负荷数据长期的发展变化规律,汲取专业人员的经验知识,形成一系列的相关规则,模拟人类专家的推理和判断过程,从而形成专家系统。最后使用专家系统对小波神经网络预测数据进行修正,得到预测结果。通过陕西汉中电网负荷数据,很好地实现了在小波神经网络中嵌入专家系统的方法,同时提高了预测精度。更多还原

【Abstract】 Wavelet neural network is a new power load forecasting method.In this paper,wavelet analysis and BP neural networks are combined.A sample decomposition of load sequence is done by wavelet transform and the different scales of wavelet coefficients are got.Then the wavelet coefficients to be threshold are chosen and the wavelet coefficients are predicted by BP neural network.At the same time the development of long-term change of historical load data is derived,combined with learning from the experience of professional knowledge form a series of relevant rules and simulating human expert’s reasoning & judging process,a expert system is formed.Finally the expert system is used to amend prediction data of the wavelet neural network.In this paper,based on the load data of Shaanxi Hanzhong Grid,the method to embed a expert system in the wavelet neural network is implemented,and prediction accuracy is improved.更多还原