Title :
New methods to train a BP network and their application
Author :
Chen, Jun Qing ; Jiang, Jing Ping
Author_Institution :
Dept. of Electr. Eng., Zhejiang Univ., Hangzhou, China
Abstract :
To estimate reaction consistency of the continuous stirred tank reactor (CSTR) system a kind of hybrid genetic algorithms (HGA) is presented. It combines the merits of BP algorithm and canonical genetic algorithms (CGA). The BP algorithm is inserted between the two reproduction parts-selection and reproduction, the results demonstrate the proposed HGAs can get quite good effect. We also replace CGA with an evolution strategy, which the simulation results show gives more accurate results
Keywords :
backpropagation; chemical technology; genetic algorithms; neural nets; state estimation; BP network training; CGA; CSTR; HGA; backpropagation; canonical genetic algorithms; continuous stirred tank reactor; evolution strategy; hybrid genetic algorithms; neural net; reaction consistency estimation; Biological cells; Continuous-stirred tank reactor; Decoding; Encoding; Genetic algorithms; Genetic mutations; Instruments; Neural networks; Nonlinear systems; State estimation;
Conference_Titel :
Neural Networks, 1999. IJCNN '99. International Joint Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-5529-6
DOI :
10.1109/IJCNN.1999.832631
