基于自我联想神经网络的谷氨酸发酵在线故障诊断 - 中国酿造.PDF

2009 No．2 · · Serial No．203 China Brewing 96 Innovation and Knowledge Transfer 基于自我联想神经网络的谷氨酸发酵在线故障诊断 1 2 1 董传亮，潘 丰，史仲平 （1.江南大学生物工程学院工业生物技术教育部重点实验室，江苏 无锡 214122 ；2.江南大学信控学院，江苏无锡 214122） 摘 要：探讨了利用自我联想神经网络（AANN）对谷氨酸发酵进行在线故障诊断的有效性和实用性。自我联想神经网络是一种带瓶 颈层的、5层结构的特殊神经网络。通过对发酵过程的表观变量进行适当筛选，将某些变量输入到AANN 网络中进行训练，发酵过程 的关键特征特性可以在瓶颈层得到提取，不同性能的发酵过程可以得到聚类。优化得到的AANN 网络可以对谷氨酸发酵进行及时、 准确的故障预判，从而为排除故障、尽可能快地恢复发酵性能提供信息。 关 键 词：谷氨酸发酵；故障诊断；自我联想神经网络 中图分类号：TQ922.1 文献标识码：A 文章编号：0254－5071 （2009）02-0096-04 On-line fault diagnosis of glutamate fermentation based on an autoassociative neural network 1 2 1 DONG Chuanliang , PAN Feng , SHI Zhongping （1. Key Laboratory of Industrial Biotechnology, Ministry of Education, School of Biotechnology, Wuxi 2 14 122 China; 2. School of Communication Control; Jiangnan University, Wuxi 2 14 122, China） Abstract: In this study, an autoassociative neural network （AANN）model was used for on-line fault diagnosis in glutamate fermentation processes. The AANN is a special feed-forward neural network with 5 layers including a bottleneck layer. By selecting state variables data and training the AANN with those data, the major characteristics of glutamate fermentation process could be extracted from the bottleneck layer of the AANN, and the fermentations with different performance could be clustered. The optimized AANN could provide on-line and accurate fault alarm for glutamate fermentations, and was potentially helpful in supplying useful information for removing fault and recovering abnormal fermentation. Key words: glutama