Electrical fault detection and classification based on multiple machine learning algorithms

High temperature superconducting (HTS) power cables are expected to be used in the future in electric ships that will have integrated power and energy systems (IPES). There have been a few studies on understanding electrical faults in power systems consisting of HTS cables; however, there are no comprehensive studies on the response of HTS cables for various types of electrical faults. We have recently initiated a research project on understanding various electrical faults in shipboard medium voltage direct current (MVDC) power systems and the duration of the fault, maximum voltage, and maximum current that the HTS cables will encounter during the fault. This paper presents investigations on the potential cable architectures being developed for MVDC power systems, and the type of faults that HTS cables will encounter in such systems. The paper will assess the relative merits of these architectures in terms of their suitability for accommodating the limitations of cryogenically cooled HTS cables and offer design suggestions for resilient power systems. Electrical and cryogenic thermal models of HTS cables that are suitable for assessing the response of HTS cables for electrical faults, as well as HTS cable designs that can endure electrical faults without catastrophic damage will also be discussed.

The faults in a power system may be of various nature and they may affect the stability of the power system by causing outage of the line, there by overloading other connected lines and also causing cascading effect in the system. In order to access the nature of fault and analyze the impact of the same in a power system, fault analysis is required to be carried out as the same will be useful for the system operation and to take quick decisions in securing the system. In this paper we discuss the various types of faults in EHV system that may occur in line or substation. These faults are identified by Numerical relays which use IEDs [Intelligent Electronic Device]. The faults identified includes symmetrical and unsymmetrical faults in power system. Further in this project, we have taken into consideration the theoretical and practical study of the fault analysis through Numerical relays. In case of occurrence of Transient faults, Auto re-closure works with dead time of 1 sec. Current, voltage, Breaker status (open, close) are graphically represented by Disturbance Recorder. Events are recorded by Event logger. The above mentioned project for faults and analysis of faults in line or substation were carried out at Power grid Corporation Limited, 400 / 230kv Sub-Station Shoolagiri.

The identification and classification of electrical faults have a great importance in power system analysis. They help in the dimensioning and the adequate choice of electrical equipment, especially, for protective and interrupting devices. This chapter describes the various faults undergone by the power system and removes some ambiguities causing confusions and difficulties for the correct classification of faults. Once the faults terminologies are well understood and properly assessed; they can be used efficiently to develop enhanced algorithms dedicated to fault detection, classification, isolation and diagnostics.

The aim of this study is to build Machine Learning models to evaluate the effect of blast furnace slag (BFS) and waste tire rubber powder (WTRP) on the compressive strength of cement mortars. In order to develop these models, 12 different mixes with 288 specimens of the 2, 7, 28, and 90 days compressive strength experimental results of cement mortars containing BFS, WTRP and BFS+WTRP were used in training and testing by Random Forest, Ada Boost, SVM and Bayes classifier machine learning models, which implement standard cement tests. The machine learning models were trained with 288 data that acquired from experimental results. The models had four input parameters that cover the amount of Portland cement, BFS, WTRP and sample ages. Furthermore, it had one output parameter which is compressive strength of cement mortars. Experimental observations from compressive strength tests were compared with predictions of machine learning methods. In order to do predictive experimentation, we exploit R programming language and corresponding packages. During experimentation on the dataset, Random Forest, Ada Boost and SVM models have produced notable good outputs with higher coefficients of determination of R2, RMS and MAPE. Among the machine learning algorithms, Ada Boost presented the best R2, RMS and MAPE values, which are 0.9831, 5.2425 and 0.1105, respectively. As a result, in the model, the testing results indicated that experimental data can be estimated to a notable close extent by the model.

The electric power system is one of the indispensable infrastructures in modern society, and its stable operation is crucial for guaranteeing economic development and social security. By calculating the correlation coefficient and drawing the correlation heat map, we find that there is a correlation between the voltage at each location and the occurrence of electrical faults, and at the same time, there is also a certain correlation between various types of electrical faults and the voltage at each location. These results indicate that there are complex and close interrelationships between different locations and between different types in the power system. In order to better identify and classify different types of electrical faults, we used the random forest model for prediction. The model is used to classify whether an electrical fault occurs or not and to determine its category. The prediction results show that the Random Forest achieved 99.63% in classifying whether an electrical fault occurred or not, and 91.26% in classifying the category of electrical faults. This shows that the Random Forest algorithm model is able to detect electrical faults well, as well as judge the category of electrical faults. In summary, this paper provides insight into the operation of the power system by studying the correlation between the voltage at various locations in the power system and the different types of electrical faults and using the random forest model to make predictions. These findings are of great significance in ensuring the stable operation of power systems.

The random forest model is a classification model whose underlying meta-classifier is the decision tree of the CART algorithm. The random forest model adopts sampling and putting back, that is, each training set is constructed based on the Bagging method and a new training set is randomly selected without additional pruning. In this paper, a financial risk assessment model based on the random forest is proposed. Firstly, six types of variables are set according to the characteristics of samples, and the outliers are eliminated by using the box graph model. After processing, the samples are modeled and tested by the cross verification method. The model adopts 600 decision trees, and the split node is 4. After calculation, the model precision is 82.27%, the recall rate is 83.55%, and the F1 value is 0.8196, which has good prediction accuracy. By comparing the accuracy of the random forest model with KNN and SVM models, it was found that the accuracy of the random forest model was 9.56% and 3.36% higher than the other two models respectively, which was obviously due to KNN and SVM in the accuracy of processing high-dimension big data samples.

The STATCOM (synchronous static compensator) operation can be adversely affected due to voltage source converter (VSC) over-currents and trips, during power system faults when its dynamic VAR support functionality is most required strict requirements of STATCOM losses and total system loss penalty preclude the use of PWM (pulse-width modulation) for VSC based STATCOM applications. This constraint of implementing VSC without PWM functionality, results in over- currents and trips of the STATCOM during and after system faults. In this paper, we propose an "emergency PWM" strategy to prevent over-currents (and trips) in the VSC during and after system faults, with slightly different B-H characteristics (due to manufacturing tolerances) for series connected transformers. Simulation results are presented for a 48-pulse VSC based plusmn100 MVAr STATCOM connected to a 2-bus power system. Simulation results validate the "emergency PWM" strategy to prevent VSC over-currents and to supply reactive power under all power system faults - single-line to ground, line-line and three-phase system faults.

Power swing blocking function is used in distance relays to distinguish power swing and fault in power system. But distance relay may mal-operate during power swing condition which can be generated due to variation of heavy load and switching of adjacent loaded line or due to any fault in the system. The new method based on equal area criteria method can successfully identify the power swing and out of step condition in the network. It is independent to change in network configuration. It needs only locally measured quantities like electrical power and power angle. The instability can be observed by monitoring the cumulative acceleration and deceleration area in power angle curve. These areas are continuously computed as well as whether or not the rotor angle reaches the maximum stable rotor angle or not it should be checked. By comparing these areas and from rotor angle, stable and out of step swings condition can be identified.

In modern buildings, increasing the probability of an electrical fault. Therefore, in the electrical fault detection, the application of fault neural network have being more and more widely. This paper presents a diagnostic method based on using Probabilistic Neural Networks (PNN) to classify forecasting. The key is training and calculating data to classify fault types, which is collected from ten different position in fault circuit. Diagnostic results show that The diagnosis results show that the PNN algorithm can have a very good diagnostic effect. The correct diagnosis rate is 90%.

Protection of power systems is important for stable and uninterrupted operation of power network. Maloperation of traditional back up protection schemes (e.g-Zone3 maloperation) causes many catastrophic phenomenon like cascaded tripping in power systems. So, some defensive mechanism are needed for taking care of such failures in power systems and to increase their security and reliability. Due to hugeness and complexity of Modern power systems, they are controlled, monitored and protected by Wide Area Measurement systems (WAMS). WAMS increases the efficiency and reliability of power networks and prevent the chance of blackouts and cascading failures in tandem. A synchronized phasor measurement-based wide-area backup protection scheme is presented in the paper where the magnitude of sequence voltages of buses at a system protection centre are used to identify the faulted area, bus closest to the fault and faulted line. The proposed method is tested for several faults including simultaneous faults in power systems with interconnections also. This technique is found to be Reliable, accurate and fast with contemporary synchronized measurement based technology. It is expected that numerous disastrous blackouts, cascaded tripping of power networks can be avoided and reliability and security of the power systems can be improved by the scheme. The study of new and improved back up protection scheme is done on a WSCC-3 machine-9 bus system. The data is simulated through EMTDC/PSCAD and MATLAB /SIMULINK software.

Bearing failures represent the most prevalent fault type in electrical machines, potentially leading to catastrophic consequences if not detected early. Conventional detection methods primarily rely on thermal, acoustic, and vibration sensors. Traditional vibration-based techniques have gained widespread adoption due to their stable and straightforward signal-processing capabilities. However, these approaches require direct motor mounting, introducing economic, temporal, and safety inefficiencies. This study presents the first investigation of contactless radar-based detection of bearing faults according to the authors' knowledge. The research employs the absolute value of complex signals derived from quadrature signals recorded by a 24 GHz radar transceiver as the vibration signal. Various defects like corrosion, improper oil levels, and scratches were deliberately introduced to the inner race, outer race, and balls of bearings, establishing 16 distinct fault classes. Classification performance was evaluated using both time-domain statistical features and frequency-domain PSD features. Multiple machine learning algorithms were applied to both approaches, consistently achieving accuracy rates exceeding 98%. This study validates the potential of radar-based systems for bearing fault diagnosis and introduces a novel paradigm for contactless bearing fault detection comprising radar signal data from 880 experiments. The results demonstrate that radar technology offers a promising alternative to traditional contact-requiring methods, enabling efficient and reliable bearing fault classification through non-invasive vibration detection.

The main target in this paper is to present, design fuzzy logic controller (FLC) applied to static var compensator (SVC) on two machine transmission system to improve transient stability and rapid damping oscillations of synchronous generators, when power generators sudden changes occur.stability that also played important role in power systems. static var compensator with fuzzy logic controller (SVCFLC) is a new control strategy can help improve transient stability.The effect of three phase fault causes instability on power system. By and large, it is very difficult to control machine speeds ,rotor angle and voltage during three-phase fault.SVCFLC is a voltage stablizer using three static var compensator which are controlled by SVC with fuzzy logic controller(FLC).The FLC is an effective device for transient stability of two-mashine transmission system. The nonlinear model dynamic formulation problem in unstable system can be solved by using artificial intelligence theorem. Fuzzy logic theory is used to improve the system stability . simulation results of three-phase fault in power system show that SVCFLC caused to increase the stability and damp out the oscillation of machine, compared with effective of SVC in the presence of power system stabilizer(PSS).

Detection of faulted phase type in distribution systems based on one end voltage measurement

2 automated electrical systems department, ural federal

Transmission line protective relaying is an important aspect of a reliable power system operation. The issue has become critical due to wide spread grid integration of distributed energy sources. The power system faults do occur in power system network and are more frequent in transmission and distribution systems. The behaviour of faults changes in the presence of renewable energy (RE) sources. This paper presents the detection of power system faults in the presence of wind power generation. The power system faults such as line to ground (LG), double line (LL), double line to ground (LLG) and three phase fault involving ground (LLLG) has been investigated. An algorithm based on discrete wavelet transform has been utilized for detection of faults under investigation. The current supplied by the wind power system has been found to be key features for the detection of power system faults. The study has been performed using a test system and results have been validated in MATLAB/Simulink environment.