Fault Diagnosis of PMSM Using Artificial Neural Network

Brushless DC (BLDC) motors are used extensively in the industry for a wide variety of applications. In the Aerospace industry, they are mainly used in the control actuation system in launch vehicles. Different types of faults can develop in a BLDC motor. It is essential to detect these faults in the early stage itself before they develop into system failures. In this paper, fault detection in BLDC motors using parameter estimation technique is presented. Detailed fault diagnosis is subsequently carried out by a fuzzy inference system. Discrete Square Root Filtering in Covariance form (DSFC) algorithm is used initially for estimating the physical parameters of the BLDC motor. These parameters are then passed on to the fuzzy inference system which carries out the detailed diagnosis. Simulation studies carried out indicates that the developed fault detection and isolation scheme is capable of isolating different types of faults like open circuit, short circuit faults, performance degradation of magnets etc. in BLDC motors.

Permanent magnet synchronous motor (PMSM) is currently the most attractive application electric machine for several industrial applications. It has obtained widespread application in motor drives in recent time. However, different types of faults are unavoidable in such motors. This paper focuses on stator winding faults diagnosis. This paper proposes the ratio of third harmonic to fundamental FFT magnitude component of the three-phase stator line current and supply voltage as a parameter for detecting stator winding turn faults under different load conditions and using artificial neural network (ANN). Discrimination among unbalancing of supply voltage conditions and stator turn short circuit poses a challenge that is addressed in this paper. The presented approach yields a high degree of accuracy in fault detection and diagnosis between the effects of stator winding turn fault and those due to unbalanced supply voltages using artificial neural network. All simulations in this paper are conducted using finite element analysis software.

Current signature analysis of induction motor becomes very much popular for its Condition monitoring purpose. Most of these studies in recent years are focused on hardware. Though, to build an experimental set-up for fault emulating machines is not an easy task. It involves too much cost and risk of damage. Moreover, it is not possible to create different types of faults in the same machine. In this paper an accurate but simple model of 3-phase Induction motor is being proposed which describes the machine at low frequency as well as at high frequency for different types of fault situations. The model, based on DM, CM characteristics of the machine, is well suited for power frequency (50 Hz) and high frequency applications like adjustable speed drives (ASDs). A mechanical model of a three-phase induction machine is incorporated to this electrical model to make a universal model of dynamic squirrel cage induction motor for overall simulation. The proposed model has the feature to incorporate different types of electrical fault like, stator inter-turn fault, phase-to-phase and phase-to ground insulation fault etc. The premature conditions of the above faults can also be emulated.

This paper presents a stator winding faults detection in induction generator based wind turbines by using artificial neural network (ANN). Stator winding faults of induction generators are the most common fault found in wind turbines. This fault may lead to wind turbine failure. Therefore, fault detection in induction generator based wind turbines is vital to increase the reliability of wind turbines. In this project, the mathematical model of induction generator based wind turbine was developed in MATLAB Simulink. The value of impedance in the induction generators was changed to simulate the inter-turn short circuit and open circuit faults. The simulated responses of the induction generators were used as inputs in the ANN model for fault detection procedures. A set of data was taken under different conditions, i.e. normal condition, inter-turn short circuit and open circuit faults as inputs for the ANN model. The target outputs of the ANN model were set as ‘0’ or ‘1’, based on the fault conditions. Results obtained showed that the ANN model can detect different types of faults based on the output values of the ANN model. In conclusion, the stator winding faults detection procedure for induction generator based wind turbines by using ANN was successfully developed.

Fault analysis is an important consideration in power system planning, protection and overall system reliability assessment. When a fault occurs at some point in the network, normal operating conditions are upset; if the fault is persistent severe loss of load, property damage and steep economic losses can arise as undesirable consequences. Relay, circuit breakers and other protective elements are used to prevent such damages. Different types of faults in busbar are classified using the bus voltages and line fault current. In this paper, we have proposed an effective way of fault detection and classification in busbars using Artificial Neural Network (ANN). This can make the power system protection more effective. We have considered IEEE 9-bus system and a dataset has been generated using PSAF CYME software. This dataset is used to train and test our network in MATLAB software. The algorithm can be used for any bus system given the voltage magnitude and angles, which will be helpful for the authorities to get notified and solve the problem as soon as possible, since repair mechanism of each type of fault is different from others.

The proposed technique consists of a pre-processing unit based on Continuous Wavelet Transform (CWT) in combination with an Artificial Neural Networks (ANN). CWT acts as an extractor of distinctive features in the transient current signals at the sending and receiving end of the transmission line. This information is then fed to the ANN for detecting the fault type and location of fault. The results presented clearly indicate that the present technique is very fast, computationally efficient and intelligent enough to accurately identify three different types of fault (LG & LLG) and their locations.

Unlike the more prevalent alternating current transmission systems, the high voltage direct current (HVDC) electric power transmission system transmits electric power using direct current. In order to investigate the precise remedy for fault detection of HVDC, this research proposes a method for the HVDC fault diagnostic methodologies with their limits and feature selection-based probabilistic generative model. The main contribution of this study is using the wavelet transform based on ant colony optimization and ANN to detect the different types of faults in HVDC transmission lines. In the proposed method, ANN uses optimum features obtained from the voltage, current, and their derivative signals. These features cannot be accurate to use in ANN because they cannot give reliable accuracy results. For this reason, first, the wavelet transform applies to the fault and non-fault signals to remove the noise. Then the ACO reduces unimportant features from the feature vector. Finally, the optimum features are used in the training of ANN as faulty and non-faulty signals. The multi-layer perceptron used in the suggested method consists of many layers, enabling the creation of a probability reconstruction over the inputs by the model. A supervised learning method is used to train each layer based on the selected features obtained from the ant colony optimization-discrete wavelet transform metaheuristic method. The artificial neural network technique is used to fine-tune the model to reduce the difference between true and anticipated classes’ error. The input signal and sampling frequencies are changed to examine the suggested strategy’s effectiveness. The obtained results demonstrate that the suggested fault detection and classification model can accurately diagnose HVDC faults. A comparison of the Support vector machine, Decision Tree, K-nearest neighbor algorithm (K-NN), and Ensemble classifier Machine techniques is made to verify the suggested method’s unquestionably higher performance.

A combined image processing and Nearest Neighbor Algorithm tool for classification of incipient faults in induction motor drives

A general framework and a method for model based diagnosis are suggested. In the framework, faults and modeling of faults are formally described. Then the method is used to construct a diagnosis system based on a structure of hypothesis tests. Using these principles, it is possible to diagnose a large variety of different types of faults. The framework and the method are applied to the design of a diagnosis system for an automotive engine. This application clearly shows the strength of these principles because several different types of sensor faults and leakage need to be diagnosed. The resulting diagnosis system is demonstrated using experiments on a real automotive engine.

Objectives: The manufactured Integrated circuits are usable only when they are free from any types of faults or error. This is even more essential in complex circuits like analog and mixed signal circuits. This paper focuses on the fault classification in the analog portion of mixed signal circuits using artificial neural networks. Methods/Statistical Analysis: All possible catastrophic possible faults in mixed signal circuits are introduced and simulated in an exhaustive manner using Cadence Simulation package. Faults are introduced one at a time and parameters were recorded. The parametric variations obtained through simulation were normalized and are used to suitably train artificial neural networks by creating them as database. The artificial neural network is trained such that it can identify correct functioning of the circuit from its faulty operation and also to further distinguish it into the component at fault, along with the type of catastrophic or hard fault. Findings: Comparative results of Feed forward neural networks trained with Levenberg-Marquardt algorithm and Radial basis function networks in the classification of faults are provided. Application/Improvements: This can be extended for the other mixed signal circuits by creating the training data using the software simulations. The hardware implementation is possible using the embedded controllers and other types of ANN can also be selected if necessary.

The artificial neural network is a powerful tool for the detection of the transmission line faults due to its ability to differentiate between various patterns. In this paper, simulation of power system under normal and faulty conditions are carried out using electromagnetic transient program. The voltage and current waveforms at the relay location for normal and fault conditions are extracted. The waveforms obtained are preprocessed in order to improve the performance of the neural network used. Three neural networks are built one for fault detection, one for fault type and a third for fault location. The proposed technique is tested with different types of faults and successive decision was reached.

Consumption of electrical energy has increased many folds in the recent time due to its use for industrial, commercial, agricultural, domestic and social purposes. HVDC technology is increasingly used for bulk power transmission due to advantages like low loss, ease of control, etc. However, the protection system for the HVDC system is not as mature as AC system. HVDC system requires fast and accurate fault detection. Recently, various time- and frequency-domain algorithms are proposed. In this paper, Stockwell (S-) transform is utilized for real-time fault detection. Different type of faults in HVDC transmission line are simulated using PSCAD. The generated fault signals are analyzed using MATLAB. S-transform can detect the transient signal effectively in time-frequency domain. It can be used to robustly differentiate the different type of faults and the load variation.

Occurrence of unwanted resistance at terminal of IGBTs is a type of Incipient faults which occurs in 3-phase voltage source Inverter. This type of faults does not trigger the protective circuit and in due course of time leads to severe faults. The present work had made an attempt to identify this type of faults related to two IGBTs in the VSI scheme. Rigorous simulation is done with different cases of faults in a PSIM software platform. The current profiles obtained from the simulation are processed by data processing tools to obtain Mean Current vector (MCV). A two phase reference frame shows the distribution of MCV for different types of faults. The distribution is found to follow a particular pattern. This has encouraged the authors to use SVM classifier for analyzing the distributed pattern and classify different types of faults.

Electric motor drives with a wide variety of applications are usually derived with inverters, where the inverter switches are always prone to different types of faults. Short circuit faults can rapidly shut down systems, and open-circuit ones can lead to secondary damage if they are not detected and tolerated in time. Due to this fact, in this paper, a novel data-driven fault detection and diagnosis (FDD) method has been proposed to detect and locate all types of inverter switch faults. Three deep learning algorithms, including fully connected neural networks (FCNs), convolutional neural networks (CNNs), and bidirectional long short-term memory (BiLSTM), have been implemented and compared. The BiLSTM network with 98.45% accuracy outperforms the others and can detect all types of faults in less than half a fundamental period under different and variable speeds with the existence of noise. The results show that the proposed method is highly effective and is a great candidate for real-time applications.

This paper discuss the application of artificial neural network-based algorithms to identify different types of faults in a power transformer, particularly using DGA (Dissolved Gas Analysis) test. The analysis of Random Neural Network (RNN) using Levenberg-Marquardt (LM) and Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithms has been done using the data of dissolved gases of power transformers collected from Punjab State Transmission Corporation Ltd.(PSTCL), Ludhiana, India. Sorting of the preprocessed data have been done using dimensionality reduction technique, i.e., principal component analysis. The sorted data is used as inputs to the Random Neural Networks (RNN) classifier. It has been seen from the results obtained that BFGS has better performance for the diagnosis of fault in transformer as compared to LM.