关键词: 裂解炉, 热值, 模糊神经网络, 小脑模型关节控制器

Abstract: A new neural algorithm based on fuzzy Cerebella Model Articulation Controller (CMAC) is researched for the application of industry process. By the computer simulation, it is proved that the generalization and smoothing ability of CMAC can be improved compared with the traditional CMAC. Based on the algorithm, the enthalpy soft sensing system is built, which has high veracity confirmed by long time applications in industry plant. The system can demonstrate the change of cracking fuel gas enthalpy adequately, and take effect in the control system of average coils output temperature of cracking furnaces.

Key words: cracking furnace, enthalpy, fuzzy-neural network, CMAC