【摘要】 由各时点负荷分量组成的负荷时间序列中,各数据点间具有一定的相关性和差异性,在进行短期负荷预测时模型一般无法兼顾数据的共性和差异性。作者采用一种改进的主成分分析法,在不损失负荷原始数据主要信息的前提下提取负荷数据的主成分,有效地减少了预测模型的输入量。同时,针对电力系统短期负荷受温度影响较大的特点,将温度因素引入BP神经网络进行短期负荷预测,实例分析验证了该方法的有效性。更多还原

【Abstract】 In the load-time series composed by the load components at different points of time a certain correlation and discrepancy exist among different data points. In general, during the load forecasting it is unable for mathematical model to consider generality and difference among the data at the same time.An improved PCA is utilized to extract the principal component of the load data under the prerequisite that the main information of original load data is not lost, therefore the input of forecasting model is effectively reduced. Meanwhile, according to the property of short-term load is easily influenced by ambient temperature, as an influencing factor the temperature is led into the BP neural network to conduct the short-term load forecasting. The effectiveness of this method is verified by case results and analysis.更多还原