【摘要】 将模拟退火思想和加速遗传特性相结合,改进选择策略和遗传算子,建立加速遗传模拟退火算法(AGSA);基于支持向量机(SVM)的非线性回归和改进混合遗传算法的因子筛选,构造了支持向量机模型参数的自适应优化算法,提出连续微滤系统(CMF)产水预测模型;通过实测中试规模连续微滤系统产水量变化对模型进行验证,结果表明:该模型较好地揭示了CMF系统产水变化规律,模拟与实测结果间的误差小、相关性强(R2=0.91、MAE=0.0132、SSE=0.0055、RMSE=0.0155),说明模型具有较强的预测能力;采用留一法对训练样本进行交叉验证(R2=0.89、MAE=0.0164、SSE=0.0073、RMSE=0.0178),表明该模型同时具有良好的稳健性。此外,将基于AGSA-SVM的模型与神经网络BP算法进行了比较,结果显示:应用AGSA-SVM建立的模型在稳健性和预测能力都优于BP算法,因此该算法更适合于对CMF系统进行产水预测研究。更多还原

【Abstract】 Combined the accelerating genetic algorithm(AGA) and simulated annealing algorithm(SA),and through improved select tactics and genetic operators to form a new accelerating genetic and simulated annealing algorithm(AGSA).Based on improved hybrid genetic algorithm and support vector machine(SVM),it formed a new self-adapting optimized algorithm applied in the SVM parameters.A modified method to develop the flow rate prediction model of continuous micro-filtration(CMF)system was proposed.The prediction models were verified by flow rate experiments in pilot-scale continuous micro-filtration system,results showed that this model could reveal the rule of flow rate variation in CMF.It had small error and strong correlation(R2=0.91,MAE=0.0132,SSE=0.0055,RMSE=0.0155)between predicted values and measured values which explained the model had a strong predictability.Based on leave-one-out cross validation of training samples,the model also showed good robustness(R2=0.89、MAE=0.0164、SSE=0.0073、RMSE=0.0178).Moreover,the model developed by AGSA-SVM was compared with the model constructed by BP neural network.The former algorithm showed the optimal predictive capability and robustness in the comparison,indicating more suitable for the flow rate prediction of CMF.更多还原