Multiple Fault Diagnosis - To solve the problems of low fault diagnosis rate and poor efficiency of AC-DC drive traction converter, a fault diagnosis method based on improved multiscale permutation entropy and wavelet analysis is proposed based on the multiple fault characteristics of input current curve in frequency domain. Firstly, the curve of the traction converter is decomposed by wavelet transform, and the modal components of different time scales are obtained. Then the fault characteristic parameters of different components are calculated by improved multi-scale permutation entropy. Finally, the multivariable support vector machine algorithm based on decision tree is used to obtain the tree-like optimal fault interval surface through small sample training, so as to achieve the fault classification of traction converters. The experimental results show that this method can effectively distinguish the fault types of traction converters, and improve the accuracy and efficiency of fault diagnosis, which has good adaptability and practical significance.

Multiple Fault Diagnosis - Aiming at the difficulty of extracting fault features on the aircraft landing gear hydraulic system, traditional feature extraction methods rely heavily on expert knowledge, and the accuracy of fault diagnosis is difficult to guarantee. This paper combined convolutional neural network (CNN) and support vector machine classification algorithm (SVM) to propose a fault diagnosis model suitable for aircraft landing gear hydraulic system. The diagnosis model adopted the onedimensional multi-channel CNN network structure, took the original pressure signal of multiple nodes as input, adaptively extracts the feature value of the pressure signal through CNN, and built a multi-feature fusion layer to realize the feature fusion of the pressure signal of each node. Finally, input the fused features into the SVM classifier to complete the fault classification. In order to verify the proposed fault diagnosis model, a typical aircraft landing gear hydraulic system simulation model was built based on AMESim, and several typical fault types such as hydraulic pump leakage, actuator leakage, selector valve clogging and accumulator failure were simulated, and corresponding Fault type data set, and use overlapping sample segmentation for data enhancement. Experiments show that the diagnosis accuracy of the proposed fault diagnosis algorithm can reach 99.25\%, which can realize the adaptive extraction of the fault features of the aircraft landing gear hydraulic system, and the features after multidimensional fusion have better discrimination, compared with traditional feature extraction methods more effective and more accurate.

Multiple Fault Diagnosis - Traditional mechanical and electrical fault diagnosis models for high-voltage circuit breakers (HVCBs) encounter the following problems: the recognition accuracy is low, and the overfitting phenomenon of the model is serious, making its generalization ability poor. To overcome above problems, this paper proposed a new diagnosis model of HVCBs based on the multi-sensor information fusion and the multi-depth neural networks (MultiDNN). This approach used fifteen typical time-domain features extracted from signals of exciting coil current and angular displacement to indicate the operational state of HVCBs, and combined the multiple deep neural networks (DNN) to improve the accuracy and standard deviation. Six operational states were simulated based on the experimental platform, including normal state, two typical mechanical faults and four typical electrical faults, and the coil current and angular displacement signals are collected in each state to verify the effectiveness of the proposed model. The experimental results showed that, compared with the traditional fault diagnosis model, the Multi-DNN based on multi-sensor information fusion can be applied to finding a better equilibrium between underfitting and overfitting phenomenon of the model.

Multiple Fault Diagnosis - Multiple fault diagnosis is a challenging problem, especially for complex high-risk systems such as nuclear power plants. Multilevel Flow Models (MFM) is a powerful tool for identifying functional failures of complex process systems composing of mass, energy and information flows. The method of fault diagnosis based on MFM is generally based on the assumption that only a single fault occurs, and based on this, the Depth First Search (DFS) is adopted to identify the abnormal functions at the lower level of an MFM. This paper presents a method based on Multilevel Flow Models (MFM) for diagnosing multiple functionally related and coupled faults. An MFM model is firstly transformed into a reasoning Causal Dependency Graph (CDG) model according to a group of alarm events. The CDG model is further decoupled to generate causal trees by a DFS algorithm, each of which represents an overall explanation of a cause of alarm events. The paper presents a comparative analysis of cases. It proves that the method proposed in the paper can give more comprehensive diagnostic results than the existing method.

Multiple Fault Diagnosis - In this article, fault detection (FD) method for multiple device open-circuit faults (OCFs) in modified neutral-point- clamped (NPC) inverters has been introduced using Average Current Park Vector (ACPV) algorithm. The proposed FD design circuit is loadindependent and requires only the converter 3- phase output current. The validity of the results has been demonstrated for OCF diagnostics using a 3-level inverter with one faulty switch. This article examines ACPV techniques for diagnosing multiple fault switches on the single-phase leg of 3-step NPC inverter. This article discusses fault tolerance for a single battery or inverter switch during a standard, active level 3 NPC inverter with connected neutral points. The primary goal here is to detect and locate open circuits in inverter switches. As a result, simulations and experiments are used to investigate and validate a FD algorithm based on a current estimator and two fault localization algorithms based on online adaptation of the space vector modulation (S VM) and the pulse pattern injection principle. This technique was efficiently investigated and provides three-stage modified NPC signature table that accounts for all possible instances of fault. The Matlab / S imulink software is used to validate the introduced signature table for the convergence of permanent magnet motors.

Multiple Fault Diagnosis - In order to solve the problem of real-time fault diagnosis of UAV flight control system, a fault diagnosis method based on hybrid diagnosis engine is proposed. Aiming at the multiple fault modes and cross-linking relationships of each node in the flight control system, the system reference model is established by qualitative and quantitative methods, and then a corresponding domain model is established according to the flight control system of a specific model. Finally, the fault diagnosis reasoning engine based on the model and the hybrid diagnosis engine realizes the diagnosis of the current fault of the system. The results show that this method can determine the time and location of the fault in real time and accurately, which provides an effective guarantee for improving the efficiency of UAV fault diagnosis and improving the flight safety of UAV.

Multiple Fault Diagnosis - Bearings are key transmission parts that are extensively used in rolling mechanical and equipment. Bearing failures can affect the regular running of machines, in serious cases, can cause enormous losses in economy and personnel casualties. Therefore, it is important to implement the research of diagnosing bearing faults. In this paper, a bearing faults diagnosis method was developed based on multiple image inputs and deep convolutional neural network. Firstly, the 1Dvibration signal is transformed into three different types of two-dimensional images: time-frequency image, vibration grayscale image and symmetry dot pattern image, respectively. Enter them into multiple DCNNs separately. Finally, Finally, the nonlinear features of multiple DCNN outputs are fused and classified to achieve bearing fault diagnostics. The experimental results indicate that the diagnosis accuracy of this proposed method is 98.8\%, it can extract the fault features of vibration samples well, and it is an effective bearing fault diagnosis methodology.

Multiple Fault Diagnosis - Diagnosis of faults in logic circuit is essential to improve the yield of semiconductor circuit production. However, accurate diagnosis of adjacent multiple faults is difficult. In this paper, an idea for diagnosis of logic circuit faults using deep learning is proposed. In the proposed diagnosis idea, two adjacent faults can be accurately diagnosed using three deep learning modules. Once the modules are trained with data processed from fault simulation, the number of faults and the location of the faults are predicted by the modules from test responses of logic circuit. Experimental results of the proposed fault diagnosis idea show more than 96.4\% diagnostic accuracy.

Software based scan diagnosis is the de facto method for debugging logic scan failures. Physical analysis success rate is high on dies diagnosed with maximum score, one symptom, one suspect and shorter net. This poses a limitation on maximum utilization of scan diagnosis data for PFA. There have been several attempts to combine dynamic fault isolation techniques with scan diagnosis results to enhance the utilization and success rate. However, it is not a feasible approach for foundry due to limited product design and test knowledge and hardware requirements such as probe card and tester. Suitable for a foundry, an enhanced diagnosis-driven analysis scheme was proposed in [1] that classifies the failures as frontend-of-line (FEOL) and backend-of-line (BEOL) improving the die selection process for PFA. In this paper, static NIR PEM and defect prediction approach are applied on dies that are already classified as FEOL and BEOL failures yet considered unsuitable for PFA due to low score, multiple symptoms, and suspects. Successful case studies are highlighted to showcase the effectiveness of using static NIR PEM as the next level screening process to further maximize the scan diagnosis data utilization.

In the operation of information technology (IT) services, operators monitor the equipment-issued alarms, to locate the cause of a failure and take action. Alarms generate simultaneously from multiple devices with physical/logical connections. Therefore, if the time and location of the alarms are close to each other, it can be judged that the alarms are likely to be caused by the same event. In this paper, we propose a method that takes a novel approach by correlating alarms considering event units using a Bayesian network based on alarm generation time, generation place, and alarm type. The topology information becomes a critical decision element when doing the alarm correlation. However, errors may occur when topology information updates manually during failures or construction. Therefore, we show that event-by-event correlation with 100% accuracy is possible even if the topology information is 25% wrong by taking into location information other than topology information.