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 Based on fuzzy hypersphere neural network, a real-time detection system of liquid rocket engine fault was proposed. It uses pattern recognition technology, after establishing the model mode of normal working state, compares the current sample with the model mode, and then judges the working state of the engine. Engine test data analysis shows that the fuzzy hypersphere neural network is very sensitive to input samples