【摘要】 飞灰含碳量是实现锅炉效率在线测量的重要参数之一，然而目前的飞灰含碳量测量装置存在测量周期长和故障率高等缺点。为此，通过改进模型更新方法提出一种新的改进自适应最小二乘支持向量机(IALSSVM)算法，并且将其用于建立某660 MW燃煤锅炉飞灰含碳量的动态软测量模型，其中采用皮尔逊相关性分析筛选出重要变量，利用核主成分分析(KPCA)法融合重要变量信息。仿真测试结果表明：该软测量模型在测试集上的平均绝对预测误差(MAE)、平均绝对百分比误差(MAPE)分别为0.171%和19.814%,拟合优度(R~2)为0.843,具有较高的精度和稳定性。另外，新的模型更新方法在计算速度上相比于传统方法提升30%左右，对促进该模型的在线应用和实现锅炉闭环燃烧优化具有重要作用。更多还原

【Abstract】 The carbon content of fly ash is one of the important parameters to realize the online measurement of boiler efficiency, but the current measurement device for fly ash carbon content has the disadvantages of a long measurement period and a high failure rate. To solve this problem, a novel improved adaptive least squares support vector machine(IALSSVM) algorithm with improved model update method was proposed, and this algorithm was used to establish a dynamic soft measurement model of fly ash carbon content for a 660 MW coal-fired boiler, in which important variables were selected by Pearson correlation analysis, while the information of important variables was fused by kernel principal component analysis(KPCA) method. Simulation results show that, the average absolute prediction error(MAE) and mean absolute percentage error(MAPE) of the soft measurement model on the test set are 0.171% and 19.814%, respectively, and the goodness of fit(R~2) is 0.843, which has a high level of accuracy and stability. In addition, compared with the traditional method, the calculation speed of the novel model update method can be improved by about 30%, which plays an important role in promoting the online application of the model and realizing boiler closed-loop combustion optimization.更多还原