中文摘要：

提出了1种混合的fuzzy ARTMAP（FAM）智能诊断方法,对泵车主油缸早期泄漏及异常泄漏进行了诊断分析。由于FAM性能受训练样本输入顺序及油缸泄漏故障样本少的影响,运用信号处理方法和改进的距离区分技术抽取了反映油缸泄漏状态的敏感特征参数,然后将不同排序的学习样本分别输入到多个FAM神经网络进行诊断分析,并结合贝叶斯置信法,获取最终的诊断结果。试验结果表明,该方法不仅能对油缸早期泄漏和异常泄漏故障进行有效诊断,诊断精度分别高于单个FAM及混合BP和RBF网络6%、38%和42%,而且具有较好的鲁棒性。

英文摘要：

Based on the fuzzy ARTMAP,a hybrid intelligent diagnosis method is proposed to diagnose the slight state and abnormal state of cylinder leakage in this paper.Since the performance of fuzzy ARTMAP is affected by the input sequence of training samples and less of leakage samples,some salient feature parameters,which are used to depict the fault-related information about the cylinder,are extracted and selected by the signal processing method and the improved distance discriminant technique.Then,the training samples arranged in different order are input into multiple fuzzy ARTMAP models to diagnose the faults respectively,and the final diagnosis result is obtained by the Bayesian belief method.Through the diagnosis of the leakage of the hydrocylinder,the diagnosis results imply that the proposed hybrid method can not only identify the slight leakage and abnormal state effectively,and its diagnosis accuracy is higher 6%,38% and 42% than that of single fuzzy ARTMAP,hybrid BP and hybrid RBF neural network,but also has good robustness.