论文部分内容阅读
提出了一种用于多工况对象系统故障检测与分离的模糊方向神经网络．神经网络用模糊集表示故障模式．模糊集是由模糊超体聚集形成的集合体，模糊超体是由单位方向、夹角和两个半径确定．模糊方向神经网络能在一次循环学习中形成非线性方向边界．并不断融合新样本信息和精炼已存在的故障模式．发动机故障检测与分离的仿真研究验证了模糊方向神经网络分类器的优越性能。
A fuzzy direction neural network for fault detection and separation in multi-working object systems is proposed. Neural networks use fuzzy sets to represent failure modes. Fuzzy sets are aggregates formed by fuzzy super-body aggregation. Fuzzy super-bodies are determined by unit direction, angle and two radii. Fuzzy direction neural network can form a non-linear direction boundary in a learning cycle. And continue to integrate new sample information and refine existing failure modes. The simulation research of engine fault detection and separation verifies the superior performance of fuzzy directional neural network classifier.