采用正交信号修正法与偏最小二乘回归的中长期负荷预测

介绍正交信号修正法的基本思想并详细推导该算法的实现步骤，将一种改进后的正交信号修正法(orthogonal signal correction，OSC)与偏最小二乘法(partial least square method，PLS)相结合，对原始数据通过OSC消除正交分量，利用PLS建立中长期负荷预测模型。该方法能有效地去除自变量系统中与因变量无关的正交数据信息，增强自变量、因变量之间的相关性，在有限的成分中提高成分解释能力。通过算例将PLS与OSC-PLS进行比较分析，结果表明，运用OSC-PLS进行中长期负荷预测，尽管预测模型提取的成分个数变少了，但模型成分的解释性却大幅度增强， 预测精度明显提高，具有较强的实用性。

Medium and Long Term Load Forecasting Based on Orthogonal Signal Correction and Partial Least-squares Regression

The fundamental ideas and detailed calculating steps of orthogonal signal correction (OSC) were introduced and presented an integrated method, which combined partial least square (PLS) method with the improved OSC was proposed. The algorithm eliminated orthogonal component and built the model of load forecasting using the method of PLS. This method can remove the useless orthogonal information between the independent variable X and dependent variable Y effectively. The correlation between X and Y is strengthened and the explanatory ability of model’s component under the condition of limited components is improved. Compared with PLS method, the results of example show that the model of OSC-PLS is feasible in medium and long term load forecasting, with fewer components extracted from the forecasting model but better explanatory ability and forecasting accuracy,