模拟退火优化的支持向量机在氨法脱硫效率预测中的应用

Application of Support Vector Machine Optimized with Simulated Annealing Algorithm in Predicting Ammonia Desulfurization Efficiency

运用支持向量机对氨法烟气脱硫效率问题进行建模,利用模拟退火算法对所建模型参数进行优化,得到最优参数组合。以脱硫过程中的烟气量、氨浓度、吸收液浓度、液气比、进口烟气温度、耗氨量和喷淋塔浆液pH值为输入变量,脱硫效率为输出量建立预测模型。采用氨法脱硫系统的运行数据对模型进行校验和寻优,并利用寻优后的模型对20组脱硫数据进行预测评判,最大相对误差小于4%,表明此模型具有较高的预测精度。

The support vector machine was used to model ammonia desulfurization efficiency before having it optimized with simulated annealing algorithm so as to obtain an optimal combination of parameters;through taking flue gas volume,ammonia concentration,absorption solution concentration,liquid-gas ratio,gas inlet temperature,ammonia consumption and spray tower pH as input variables and taking desulfurization efficiency as output,a prediction model was established and calibrated and optimized with operation data from the desulfurization system.Applying this optimized model to predicting 20-group desulfurization data proves its higher prediction accuracy because of its maximum relative error less than 4%.