Abstract: Abstract A fault detection system for a liquid propellant rocket engine was proposed for the realtime conditions based on the fuzzy hypersphere neural network.It extracted the template class representing for the normal conditions using the pattern recognition method,and then it compared the current point with the template class to judge how the engine was getting on.Test data analyses showed that the fuzzy hypersphere neural network is very sensitive to the input patterns.

摘要： 基于模糊超球神经网络，提出了一种液体火箭发动机故障的实时检测系统。它采用模式识别技术，在建立正常工作状态的样板模式之后，把当前样本与样板模式进行比较，进而判断发动机工作状态。发动机试车数据分析表明：模糊超球神经网络对输入样本非常敏感

关键词: 液体推进剂火箭发动机;人工神经元网络;故障检测;模式识别