Fault detection in PV systems is a key factor in maintaining the integrity of any PV system. Faults in photovoltaic systems can cause irrevocable damages to the stability of the PV system and substantially decrease the power output generated from the array of PV modules. Among'st the various AC and DC faults in a PV system, the clearance of the AC side faults is achieved by conventional AC protection schemes,the DC side, however , there still exists certain faults which are difficult to detect and clear. This paper deals with the modeling, detection and classification of these types of DC faults. It is essential to be able to simulate the PV characteristics and faults through software. In this thesis a comprehensive literature survey of fault detection methods for DC side of a PV system is presented. The disparities in the techniques employed for fault detection are studied . A new method for modeling the PV systems information only from manufacturers datasheet using both the Normal Operating Cell temperature conditions (NOCT) and Standard Operating Test Conditions (STC) conditions is then proposed.The input parameters for modeling the system are Isc,Voc,Impp,Vmpp and the temperature coefficients of Isc and Voc for both STC and NOCT conditions. The model is able to analyze the variations of PV parameters such as ideality factor, Series resistance, thermal voltage and Band gap energy of the PV module with temperature. Finally a novel intelligent method based on Probabilistic Neural Network for fault detection and classification for PV farm with string inverter technology is proposed.

yes / This paper presents a methodology for fault detection, fault prediction and fault isolation based on the integration of hidden Markov modelling (HMM) and Bayesian networks (BN). This addresses the nonlinear and non-Gaussian data characteristics to support fault detection and prediction, within an explainable hybrid framework that captures causality in the complex engineered system. The proposed methodology is based on the analysis of the pattern of similarity in the log-likelihood (LL) sequences against the training data for the mixture of Gaussians HMM (MoG-HMM). The BN model identifies the root cause of detected/predicted faults, using the information propagated from the HMM model as empirical evidence. The feasibility and effectiveness of the presented approach are discussed in conjunction with the application to a real-world case study of an automotive exhaust gas Aftertreatment system. The paper details the implementation of the methodology to this case study, with data available from real-world usage of the system. The results show that the proposed methodology identifies the fault faster and attributes the fault to the correct root cause. While the proposed methodology is illustrated with an automotive case study, its applicability is much wider to the fault detection and prediction problem of any similar complex engineered system.

Yes / This paper presents a fault detection method in three-phase induction motors using Wavelet Packet Transform (WPT). The proposed algorithm takes a frame of samples from the three-phase supply current of an induction motor. The three phase current samples are then combined to generate a single current signal by computing the Root Mean Square (RMS) value of the three phase current samples at each time stamp. The resulting current samples are then divided into windows of 64 samples. Each resulting window of samples is then processed separately. The proposed algorithm uses two methods to create window samples, which are called non-overlapping window samples and moving/overlapping window samples. Non-overlapping window samples are created by simply dividing the current samples into windows of 64 sam-ples, while the moving window samples are generated by taking the first 64 current samples, and then the consequent moving window samples are generated by moving the window across the current samples by one sample each time. The new window of samples consists of the last 63 samples of the previous window and one new sample. The overlapping method reduces the fault detection time to a single sample accuracy. However, it is computationally more expensive than the non-overlapping method and requires more computer memory. The resulting window sam-ples are separately processed as follows: The proposed algorithm performs two level WPT on each resulting window samples, dividing its coefficients into its four wavelet subbands. Infor-mation in wavelet high frequency subbands is then used for fault detection and activating the trip signal to disconnect the motor from the power supply. The proposed algorithm was first implemented in the MATLAB platform, and the Entropy power Energy (EE) of the high frequen-cy WPT subbands’ coefficients was used to determine the condition of the motor. If the induction motor is faulty, the algorithm proceeds to identify the type of the fault. An empirical setup of the proposed system was then implemented, and the proposed algorithm condition was tested under real, where different faults were practically induced to the induction motor. Experimental results confirmed the effectiveness of the proposed technique. To generalize the proposed meth-od, the experiment was repeated on different types of induction motors with different working ages and with different power ratings. Experimental results show that the capability of the pro-posed method is independent of the types of motors used and their ages.

<p dir="ltr">Resilient engine operation hugely depends on proper functioning of the engine’s sensors, enabling efficient feedback control of the engine systems operation. When the sensors on the engine measure a physical quantity incorrectly, it leads the engine control system to determine that the sensor measuring the physical quantity has failed. This failure may be attributed to a sensor stick failure, bias failure, drift failure, or failure occurring due to physical wear and tear of the sensor. Failure of crucial engine sensors may have adverse effects on engine operation, and in most cases leading into a limp home mode or a torque limitation mode. This affects the engine performance and efficiency. The engine under study in this work is a medium duty marine engine with diesel fuel. Sensor failures in the middle of a marine operation can hugely impact its mission. Therefore, fault tolerant control systems are essential to counter these challenges occurring due to sensor failures. In this thesis, an advanced nonlinear fault detection and state estimation algorithm is developed and implemented on a GT-Power engine model, employing a sophisticated co-simulation approach. The focus is on a 6.7L Cummins diesel engine, for which a detailed nonlinear state space model is constructed. This model accurately replicates critical engine parameters, such as pressures, temperatures, and engine speed, by integrating various submodels. These sub-models estimate key parameters like cylinder inlet charge flow, valve flow, cylinder outlet temperature, turbocharger turbine flow, and charge air cooler flow. To assess the model’s accuracy and reliability, it is rigorously validated against a truth reference GT-Power engine model. The results demonstrate exceptional performance, with the nonlinear model exhibiting a minimal percentage performance error of less than 5% under steady-state conditions and less than 15% during transient conditions. The core of the Fault Detection and State Estimation (FDSE) modules consists of a bank of Extended Kalman Filters (EKF). These filters are meticulously designed to estimate vital engine states, generate residuals, and assess these residuals even in the presence of process and measurement noise. This approach enables the detection of sensor faults and facilitates controller reconfiguration, ensuring the engine’s robustness in the face of unexpected sensor failures. Crucially, the nonlinear physics-based model serves as the foundation for the state transition functions utilized in the design of the observer bank. Residuals generated by the EKFs are evaluated using both fixed and adaptive thresholding techniques masking the sensor faults at the time step at which it is detected, ensuring robust performance not only in steady-state conditions but also during varying transient load conditions. To comprehensively evaluate the system’s resilience in practical scenarios, multiple sensor stuck failures are introduced into the GT-Power model. A software-in-the-loop co-simulation strategy is meticulously established, employing both the GT-Power truth reference engine model and the nonlinear Fault Detection and State Estimation (FDSE) model within the Simulink environment. This unique co-simulation approach provides a platform to assess the FDSE performance and its effect on engine performance in simulated sensor fault scenarios. The FDSE module is able to detect sensor failures which deviate at least 5% from their actual values. The percentage estimation error is less than 10% under steady state conditions and less than 20% under transient load conditions. Ultimately, this process creates analytical redundancy, not only forming the basis of state estimation but also empowering the engine to maintain its performance in the presence of sensor faults.</p>

The popularity of MPI-based middleware and applications has led to their wide deployment. Such systems, however, are not inherently reliable and cannot tolerate external faults. This thesis presents a novel model-based approach for exploiting application features and other characteristics to categorize and create AEMs (Application Execution Model). This work realizes MPI/FT(tm), a middleware derived by selective incorporation of fault-tolerant features into MPI/Pro(tm) for two relevant AEMs. This thesis proves the following hypothesis: it is possible to successfully complete select MPI applications even in the presence of external faults, and such fault-tolerance can be achieved with acceptable performance overhead. This work defines parameters to measure the impact of this middleware on performance through faultree and fault-injected overheads. The hypothesis is validated through experimentation and measurement of sample MPI applications for two AEMs.

No description available.

No description available.

Modern digital energy meters are installed between the distribution network and customers. Network operators and customers can use those meters to monitor electrical parameters, i.e., voltages and currents and to calculate statistics such as RMS value, fundamental Fourier component, etc. Observation of distribution network impedance can reveal problems in the system, such as broken neutral conductors. This project proposes a loop-theory method using variables from smart meters to estimate the network impedance, therefore giving the possibility of conductor fault detection. Specifically, loop theory uses the distribution network terminal’s voltage and current measured by smart meters and selects cases where the load appears to have varied significantly, so that there is a change of current together with a resulting change of voltage. For selected cases, the methods calculate different loop impedance and finally address single conductor impedance by the least-squares method. The project validates the proposed method by simulation, and with recorded data from a real house with varied known neutral-conductor impedance. Based on that, this method’s limitation and possible improvement are discussed for further study. / Moderna digitala energimätare används mellan distributionsnätet och kunderna. Sådana mätare kan användas för att övervaka elektriska parametrar, dvs. spänningar och strömmar och beräkna statistik som RMS-värde, grundläggande Fourier-komponent etc. Observation av distributionsnätets impedans kan avslöja problem i systemet, såsom neutrala ledare med brott eller hög impedans (’nollfel’). Detta projekt föreslår en loopteorimetod med hjälp av variabler från smarta mätare för att uppskatta nätverksimpedansen, vilket ger möjlighet till detektering av ledningsfel. Specifikt använder loopteorin spänning och ström mätt med genom smarta mätare, och väljer fall där belastningen verkar ha varierat avsevärt så att det sker en förändring av strömmen tillsammans med en resulterande spänningsförändring. För valda fall beräknar metoderna impedanser i olika slingor av nätets ledare, och slutligen gör en estimering av impedansen hos enskilda ledare genom minstakvadratmetoden. Projektet validerar den föreslagna metoden genom simulering, och med inspelade data från ett riktigt hus med varierad känd neutral ledare impedans. Baserat på detta diskuteras denna metods begränsning och möjliga förbättring för vidare studier.

Faulty manufactured alternators lead to commercial and safety concerns when installed in vehicles. Alternators have a major role in the Electrical Power Generation System (EPGS) of vehicles, and a defective alternator will lead to damaging of the battery and other important electric accessories. Therefore, fault detection and diagnosis of alternators can be implemented to quickly and accurately determine the health of an alternator during end of line testing, and not let faulty components leave the manufacturer. The focus of this research is to develop a Model Based Fault Detection and Diagnosis (FDD) strategy for detecting alternator faults during end of line testing. The proposed solution uses Extended Kalman Smooth Variable Structure Filter (EK-SVSF) to detect common alternator faults. A solution using the Dual Extended Kalman Filter (DEKF) is also discussed. The alternator faults were programmatically simulated on alternator measurements. The experimental results prove that both the EK-SVSF and DEKF strategies were very effective in alternator modeling and detecting open diode faults, shorted diode faults, and stator imbalance faults. / Thesis / Master of Applied Science (MASc)

Deep Learning Networks (DLN) is a relatively new artificial intelligence algorithm that gained popularity quickly due to its unprecedented performance. One of the key elements for this success is DL’s ability to extract a high-level of information from large amounts of raw data. This ability comes at the cost of high computational and memory requirements for the training process. Estimation algorithms such as the Extended Kalman Filter (EKF) and the Smooth Variable Structure Filter (SVSF) are used in literature to train small Neural Networks. However, they have failed to scale well with deep networks due to their excessive requirements for computation and memory size. In this thesis the concept of using EKF and SVSF for DLN training is revisited. A New family of filters that are efficient in memory and computational requirements are proposed and their performance is evaluated against the state-of-the-art algorithms. The new filters show competitive performance to existing algorithms and do not require fine tuning. These new findings change the scientific community’s perception that estimation theory methods such as EKF and SVSF are not practical for their application to large networks. A second contribution from this research is the application of DLN to Fault Detection and Diagnosis. The findings indicate that DL can analyze complex sound and vibration signals in testing of automotive starters to successfully detect and diagnose faults with 97.6% success rates. This proves that DLN can automate end-of-line testing of starters and replace operators who manually listen to sound signals to detect any deviation. Use of DLN in end-of-line testing could lead to significant economic benefits in manufacturing operations. In addition to starters, another application considered is the use of DLN in monitoring of the State-Of-Charge (SOC) of batteries in electric cars. The use of DLN for improving the SOC prediction accuracy is discussed. / Thesis / Doctor of Science (PhD) / There are two main ideas discussed in this thesis, both are related to Deep Learning (DL). The first investigates the use of estimation theory in DL network training. Training DL networks is challenging as it requires large amounts of data and it is computationally demanding. The thesis discusses the use of estimation theory for training of DL networks and its utility in information extraction. The thesis also presents the application of DL networks in an end-of-line Fault Detection and Diagnosis system for complex automotive components. Failure of appropriately testing automotive components can lead to shipping faulty components that can harm a manufacturer’s reputation as well as potentially jeopardizing safety. In this thesis, DL is used to detect and analyze complex fault patterns of automotive starters, complemented by sound and vibration measurements.