基于机制转换非线性模型的短期负荷预测

Short term load forecasting based on regime-switching nonlinear models

研究了负荷时间序列波动性,提出了一种基于机制转换非线性模型的短期负荷预测方法。在二阶矩层面建立了标准机制转换非线性模型(LSTAR),有效地解决了TAR模型的间断点问题。提出了基于厚尾假设的机制转换模型。借助模型的不对称参数,分析了不同性质冲击下的不同机制。用实际算例验证了该方法的可行性和有效性,并比较了模型的预测能力,得到厚尾LSTAR模型效果最优。

This paper analyzes the volatility of load time series and proposes a feasible approach for short-term load forecasting based on regime-switching nonlinear models. The standard regimeswitching models-LSTAR are specified to solve the break point puzzle of TAR effectively. Furthermore, generalized models with fattail distribution assumption are proposed. With the help of asymmetric parameter, the mechanism of regime against different shocks is analyzed. Load forecasting results on a practical example are presented to clearly demonstrate and verify the proposed algorithm. The forecast performance comparison of all the regime-switching models shows that fat-tail LSTAR model outoerforms others.