CAJ | 학술논문

针对污水处理过程中关键水质参数如生化需氧量(BOD)在线测量难的问题，利用Libsvm工具箱建立BOD的支持向量回归机(SVR)的软测量模型，并与文献[2]中所用神经网络的方法进行了比较，仿真结果表明：相对于神经网络算法结构性不足，SVR预测模型泛化性更好，预测精度更高，表现出更好的泛化性和预测精度。
침대오수처리과정중관건수질삼수여생화수양량(BOD)재선측량난적문제，이용Libsvm공구상건립BOD적지지향량회귀궤(SVR)적연측량모형，병여문헌[2]중소용신경망락적방법진행료비교，방진결과표명：상대우신경망락산법결구성불족，SVR예측모형범화성경호，예측정도경고，표현출경호적범화성화예측정도。
Aiming at sewage treatment process in the key water quality parameters such as biochemical oxygen demand (BOD), the difficulty of on-line measurement, using Libsvm toolbox establish BOD support vector regression machine (SVR) of soft measurement model, and compared with the literature [2] the method of neural network used in the comparison, the simulation results show that the structural deifciencies compared with the neural network algorithm, the SVR forecasting model generalization is better, higher prediction precision, better generalization and precision.