【摘要】 运用BP类神经网络建立化学生物絮凝工艺模型,考察了单一的多输入多输出(MIMO)和多输入单输出(MISO)模型的适应性。研究表明,MISO模型的预测相对误差均在9.0%以下,低于MIMO模型。采用多输入单输出的MISO模型组合建立化学生物絮凝工艺模型是合理的,能够用于实际工艺控制的需要。该项研究为实现化学生物絮凝工艺的优化设计与控制提供了可行的途径。更多还原

【Abstract】 Based on BP neural network,the models of chemical-biological flocculation process were set up,and the adaptabilities of the single Multi-Input/Multi-Output(MIMO)model and Multi-Input/Single-Output(MISO)model were studied.Results showed that the relative errors in forecasting could be controlled below 9.0% in the application of MISO models,which was lower than in the use of the MIMO model.It was reasonable in the application of the MISO model to set up the modeling of the chemical-biological flocculation process,which could satisfy with the needs of the real-time control.The study could provide the feasible way for the optimal design and real-time control of the chemical-biological flocculation process.更多还原