中文摘要：

提出在火电厂关键参数建模中采用反向建模方法，以规避传统建模方法在实际应用中的建模难题．以超临界直流锅炉中间点温度为例，利用某600MW超临界机组的实际运行数据，采用反向建模方法建立了该参数的数学模型．建模算法选用最小二乘支持向量机（LS—SVM），应用粒子群算法（PSO）解决了LS—SVM参数寻优问题，并将PSO—LS—SVM所得模型与LS-SVM、偏最小二乘（PLS）以及BP神经网络所得模型进行了对比，结果表明：基于PSO—LS—SVM的中间点温度数学模型计算速度快、精度高，验证了反向建模思想的有效性和可行性．

英文摘要：

Reversed modeling method was proposed to avoid the difficulty of the traditional modeling method in power plant critical parameters modeling. Based on the actual operation data of a 600 MW supercrifical once-through boiler, mathematic model of the intermediate point temperature was constructed by reversed modeling method. The modeling algorithm was least square support vector machines （LS-SVM）, and particle swarm optimization （PSO） was used to solve the LS-SVM optimal parameters question. A comparison has been made among models respectively obtained by PSO-LS-SVM, LS-SVM, partial least square（PLS）, BP artificial neural networks（BP-ANN）. Results show that with the intermediate point temperature model based on PSO-LS-SVM, faster and accurate calculation can be achieved, proving the reversed modeling method to be effective and feasible.