论文部分内容阅读
从近十多年的应用进展情况来看,常用的催化剂定量构效关系(QSAR)建模方法包括以多元线性回归(MLR)、主成分回归(PCR)和偏最小二乘法(PLS)为主的线性方法和以基于反向传播算法(BP)和径向基函数神经网络(RBFNN)为主的人工神经网络(ANN)非线性方法两种。最有效的线性方法是PLS法,其优点是模型机理明确,缺点是有时不如RBFNN模型的预测能力强;最有效的非线性方法是RBFNN法,其优点是模型的预测能力往往比PLS模型强,但缺点是机理不够明确。最成功的也是最具应用前景的方法是综合采用PLS法和RBFNN法同时建立某一具体催化剂的PLS线性模型和RBFNN非线性模型。用PLS模型指导新型高效催化剂的结构设计,而用RBFNN模型来预测所设计催化剂的性能,反过来修正所设计催化剂的结构,从而减少催化剂合成实验的工作量。
From the application progress of the past ten years, the commonly used QSAR modeling methods include multiple linear regression (MLR), principal component regression (PCR) and partial least squares (PLS) And two artificial neural network (ANN) nonlinear methods based on Back Propagation (BP) and Radial Basis Function Neural Networks (RBFNN). The most effective linear method is the PLS method. The advantage of this method is that the mechanism of the model is clear and the disadvantage is that the RBFNN model is sometimes not as good as the prediction capability of the RBFNN model. The most effective nonlinear method is the RBFNN method, which has the advantage that the prediction capability of the model is often stronger than the PLS model. But the disadvantage is that the mechanism is not clear enough. The most successful and promising method is to combine the PLS and RBFNN methods to establish a PLS linear model and a RBFNN nonlinear model simultaneously for a specific catalyst. The PLS model is used to guide the structural design of a novel and efficient catalyst. The RBFNN model is used to predict the performance of the designed catalyst. In turn, the structure of the designed catalyst is modified to reduce the workload of the catalyst synthesis experiment.