基于学习向量量化的传感器免疫网络诊断模型

Immune Network of Sensor Fault Diagnosis Based on Learning Vector Quantization

为克服传感器免疫网络连接权值获取困难,提出了一种基于学习向量量化的传感器免疫网络诊断模型。该模型将学习向量量化(LVQ)的概念引入到传感器免疫网络模型中,在训练模式下,LVQ用于提取传感器正常工作下的相关性;在诊断模式下,根据LVQ获取的知识即可以确定传感器之间的检测结果,同时给出了诊断模型的性能优化算法。航空发动机传感器的仿真结果表明,所提出的方法能够准确获得网络节点之间的连接权值,保证免疫网络具有较高的检测灵敏度。

It is difficult to determine the weights of immune network of sensor fault diagnosis, thus we propose an immune network for sensor fault diagnosis based on learning vector quantization. The model has two execution modes. In training mode, the LVQ extracts a correlation between each two sensors from their outputs when they work properly. In diagnosis mode, the LVQ contributes to testing each two sensors using the extracted correlation, and the immune network contributes to determining faulty sensors by integrating the local testing results obtained from the LVQ. The algorithms of model are given. Results of simulation on aircraft engine show that this method can effectively detect the sensor failures, it is sensitive to fault and is robust to noise interference.