基于曲轴扭振信息识别的柴油机失火气缸判别方法

为解决柴油发动机在高转速、低负荷工况下难以准确判断出发生失火故障的气缸的问题,提出了通过检测曲轴转速信号中的扭振信号起始点来判定失火气缸的故障诊断方法。该方法首先通过分析单缸失火时的转速信号以确定曲轴扭振的自由频率,然后基于该频率构造正弦检测信号,将该检测信号与各缸做功转角范围内的转速信号做内积运算,并得到的内积值作为失火气缸的指示特征。在6缸柴油机上的失火试验证明该方法能够在高转速低负荷情况下准确识别发动机的单缸失火和两缸失火,弥补了传统失火诊断方法工况覆盖率低的不足。     

基于SSD和MPE的滚动轴承故障诊断方法

为了解决滚动轴承故障特征难以提取的问题,提出了一种奇异谱分解(SSD)和多尺度排列熵(MPE)的故障特征提取方法,结合K近邻(KNN)算法识别滚动轴承故障类型。首先对滚动轴承振动信号用SSD进行分解,得到3个奇异谱分量(SSC),根据峭度最大原则选择主分析分量;然后用MPE计算主分析分量的熵值,实现滚动轴承的故障特征进行提取;最后将熵值作为特征向量输入KNN分类器中,完成滚动轴承的状态识别。将该方法应用于实验数据分析,并与VMD和MPE相结合的故障诊断方法做比较,结果证明,该方法能够有效地提取故障特征,实现故障诊断。     

基于双谱特征的行星齿轮箱故障诊断方法

为了研究行星齿轮箱齿面磨损全生命周期实验的退化过程,使用了可以抑制高斯噪声和对信号中产生的频率耦合进行解耦的双谱方法去进行故障特征提取,提出了两个基于双谱的特征指标,双谱熵以及非高斯强度,并通过特征指标评价方法去检验其指标性能。结果表明基于双谱熵具有良好的指标性能,而且对于行星齿轮箱齿面磨损的早期微弱故障十分敏感,适用于行星齿轮箱齿面磨损的故障诊断。     

A process fault diagnosis method using multi-time scale dynamic feature extraction based on convolutional neural network

Unlike many other techniques used in process control, which are widely applied in practice and play significant roles, abnormal situation management (ASM) still relies heavily on human experience, not least because the problem of fault detection and diagnosis (FDD) has not been well addressed. In this paper, a process fault diagnosis method using multi-time scale dynamic feature extraction based on convolutional neural network (CNN) consisting of similarity measurement, variable ranking, and multi-time scale dynamic feature extraction is proposed. The CNN-based model containing the fixed multiple sampling (FMS) layer can extract dynamic characteristics of process data at different time scales. The benchmark Tennessee Eastman (TE) process is used to verify the performance of the proposed method.     

Robust fault diagnosis and fault tolerant control for PEMFC system based on an augmented LPV observer

In order to improve the safety and reliability of proton exchange membrane fuel cell system, this paper proposes a novel robust fault observer for the fault diagnosis and reconstruction of the PEMFC air management system. First, considering the complexity and large computation of the nonlinear PEMFC system, a linear parameter-varying (LPV) model is introduced to describe the system behavior and reduce the computation cost. Then, an augmented state observer based on the LPV model is proposed for simultaneously estimating the internal states and component faults. The robustness is guaranteed by taking the system disturbances and measurement noises into consideration when designing the observer gain. The observer design is transformed into a process of solving a set of linear inequality matrices. According to the results, the augmented robust observer can accurately estimate the system states and faults under different conditions. Moreover, to realize the fault tolerant control of the air supply, the oxygen stoichiometry estimator is designed taking consideration of system fault information and a corresponding controller is employed for air compressor voltage following the net power maximization strategy.     

基于混合时间序列卷积神经网络的轴承故障诊断

在传统滚动轴承故障诊断中,绝大多数方法采用了从振动信号提取特征的诊断模式,但是这种模式必然会使原始信号降维进而导致故障信息的丢失。卷积神经网络(CNN)通过权重共享和稀疏连接直接对原始信号进行操作,实现自适应特征提取,最大化保留故障信息。受CNN原理启发,开发出了一种基于工业振动信号特征的新型诊断框架,称之为混合时间序列CNN(HTS-CNN)。首先,利用估计总体比例的方法自适应确定模型训练样本数目;其次,通过对时间序列片段进行随机组合的方式,使模型能够提取非相邻信号特征;最后,利用Softmax激活函数在模型输出端执行多分类任务。通过对凯斯西储大学及CUT-2平台轴承数据进行分析,实验结果表明:该方法能够准确、有效的对滚动轴承故障进行分类。     

基于VMD瞬时能量与GA- RBF的滚动轴承故障诊断

针对滚动轴承早期故障的有效识别,提出了一种基于VMD瞬时能量与GA优化的RBF神经网络的滚动轴承故障诊断方法,可以有效对滚动故障做出诊断。首先,VMD将滚动轴承振动信号进行分解成合适数目的本证模态函数;其次,计算本证模态函数分量的瞬时能量并组成特征向量;最后,将特征向量输入到GA优化的RBF神经网络实现轴承故障识别。通过滚动轴承故障诊断实验对该方法进行验证。结果表明,该方法识别滚动轴承故障的准确率为96.43%,较默认参数的RBF神经网络和EEMD瞬时能量与GA-RBF神经网络有明显的提高,证明了所提方法的可行性。     

基于 MPDE-EEMD 及自适应共振解调的轴承故障特征提取方法

针对滚动轴承振动信号具有非线性、非平稳性和非高斯性,并且故障特征往往淹没于系统噪声之中而难于识别的问题, 提出了以多种群差分进化(multiple population differential evolution, MPDE) 算法来改进集合经验模式分解( ensemble empirical mode decomposition, EEMD) 的 MPDE-EEMD 消噪方法,并与自适应共振解调技术( adaptive resonance demodulation technique, ARDT)相结合实现故障特征提取。 首先,为了解决 EEMD 中加入参数依靠人工选择且难以准确获取的问题,建立极值点分布 特性评价函数,利用 MPDE 来寻优获取最佳白噪声幅值,实现 EEMD 自适应分解。 然后,采用峭度与相关性相结合的准则对分 解后的 IMF 分量进行自动筛选,将满足条件的有效信号进行重构,实现对原始振动信号的降噪处理。 最后,采用 ARDT 自动确 定对消噪信号进行带通滤波的带宽和中心频率,再通过包络解调提取出滤波信号的特征频率。 将轴承仿真故障信号与实际故 障信号用于算法的验证,结果表明 MPDE-EEMD+ARDT 能有效提取出轴承故障特征。     

Fault detection and diagnostic method of diesel engine by combining rule-based algorithm and BNs/BPNNs

The stable operation of diesel engine is critical to the normal production of the industry, and the prevention, monitoring, and identification of faults are of great significance. At present, the fault research on diesel engines still has some defects, such as only few types of faults diagnosis are identified, the accuracy of fault diagnosis is still low, and fault identification is located at a fixed speed. A novel fault detection and diagnostic method of diesel engine by combining rule-based algorithm and Bayesian networks (BNs) or Back Propagation neural networks (BPNNs) is proposed. The signals are processed by wavelet threshold denoising and ensemble empirical mode decomposition. The signal-sensitive feature values are extracted from the decomposed intrinsic mode function. Seven faults are roughly identified using rule-based algorithm and finely identified using BNs or BPNNs. Results show the proposed fault diagnosis method has a good diagnostic performance for a wide range of rotation speeds when the training data for BNs and BPNNs are from fixed speeds. In addition, the influences of the layers of decomposed signals, sensor noise and external excitation interference on the fault diagnostic performance are also researched.