最大相关系数预测模型在热电厂短期热负荷预测中的应用

Application of Maximum Correlation Coefficient Prediction Model in Short-term Thermal Load Forecasting of Power Plants

热电厂短期负荷预测一般采用神经网络预测法以及模糊预测等预测方法,但这些预测方法都必须进行大量的样本训练,预测所需要的时间比较长,收敛速度慢,而且预测模型外推性较差[1-3],无法判定预测过程中的问题所在,针对这一问题提出最大相关系数预测法。根据短期负荷预测不同类型的数据之间的相关系数的稳定性,计算出不同类型数据前n项和待预测数据的前n项之间的相关系数[4-5],通过比较不同类型数据与待预测组之间相关系数的绝对值,取其中绝对值最大的一组作为待预测数据的预测依据进行短期预测。实际计算结果证明该预测模型有较高的精准度。利用最大相关系数法对热电厂回水温度和供水流量进行预测,具有建模简单,预测过程快速,精准度较高的优势。预测结果表明,该方法适合数据样本不大,且外推性要求高的短期负荷预测,在实际的负荷预测中有一定的实用价值。

Short-term load forecasting in thermal power plants generally uses neural network prediction methods and fuzzy prediction methods.However,these prediction methods must perform a large number of sample training.The prediction takes a long time,the convergence speed is slow,and the extrapolation of the prediction model is poor[1-3],which can not determine the problem in the prediction process.This research proposes the maximum correlation coefficient prediction method for this problem.According to the short-term load forecasting the stability of the correlation coefficient between different types of data,calculate the correlation coefficient between the first nitems of different types of data and the first nitems of the data to be predicted[4-5].By comparing the absolute value of the correlation coefficient between the different types of data and the group to be predicted,and take the group with the largest absolute value as the prediction basis with the data to be predicted for short-term prediction.The actual calculation results show that the prediction model has higher accuracy.Using the maximum correlation coefficient method to predict the return water temperature and water supply flow of thermal power plants,which has the advantages of simple modeling,fast prediction process and high accuracy.The prediction results show that the method is suitable for short-term load forecasting with small data samples and high extrapolation requirements,and it has certain practical value in actual load forecasting.