This paper proposes the use of a Hall effect sensor installed between two stator slots of a squirrel cage induction motor, spectral analysis of the air gap disturbances and machine learning approaches for diagnosing fully broken rotor bars in motors running at very low slip. The present work overcomes a typical drawback of broken bars diagnosis by using the traditional analysis of the motor current signature, usually when machines operate under low or no-load conditions. A Fast Fourier Transform (FFT) has been performed in Hall sensor data to identify some harmonic features and also some statistical data have been extracted to get time domain properties. These values were used as inputs for a Multilayer Perceptron Neural Network (MLP) classifier, a k-Nearest Neighbor (KNN) classifier and a support vector machine (SVM) classification technique. In addition, principal component analysis (PCA) has been applied to reduce data dimension among some original values, i.e., for both time and frequency features, and contribute to rotor failure detection. A Cross-Validation (CV) method has been used for classifier performance evaluation on each case. The efficiency of this approach was evaluated from a 7.5kW squirrel cage induction machine running at very low slip conditions and MLP classifier has achieved better results than the other intelligent methods.

Transformers are one of the most important part in a power system and, especially in key-facilities, they should be closely and continuously monitored. In this context, methods based on the dissolved gas ratios allow to associate values of gas concentrations with the occurrence of some faults, such as partial discharges and thermal faults. So, an accurate prediction of oil-dissolved gas concentrations is a valuable tool to monitor the transformer condition and to develop a fault diagnosis system. This study proposes a nonlinear autoregressive neural network model coupled with the discrete wavelet transform for predicting transformer oil-dissolved gas concentrations. The data fitting and accurate prediction ability of the proposed model is evaluated in a real world example, showing better results in relation to current prediction models and common time series techniques.

Dissolved gas analysis (DGA) is one of the most important methods to analyze fault in power transformers. In general, DGA is applied in monitoring systems based upon an autoregressive model; the current value of a time series is regressed on past values of the same series, as well as present and past values of some exogenous variables. The main difficulty is to decide the order of the autoregressive model; this means determining the number of past values to be used. This study proposes a wavelet-like transform to optimize the order of the variables in a nonlinear autoregressive neural network to predict the in oil dissolved gas concentration (DGC) from sensor data. Daubechies wavelets of different lengths are used to create representations with different time delays of ten DGC, which are then subjected to a procedure based on principal components analysis (PCA) and Pearson’s correlation to find out the order of an autoregressive model. The representations with optimal time delays for each DGC are applied as input in a multi-layer perceptron (MLP) network with backpropagation algorithm to predict the gas at the present and future times. This approach produces better results than choosing the same time delay for all inputs, as usual. The forecasts reached an average mean absolute percentage error (MAPE) of 5.763%, 1.525%, 1.831%, 2.869%, and 5.069% for C₂H₂, C₂H₆, C₂H₄, CH₄, and H₂, respectively.

Rectifier substation placement on DC traction systems has been the purpose of several works which have tried to establish mathematical models to optimise that placement. Those mathematical models were always performed to optimise one variable like distance between substations, consumed energy or peak demand. This study presents the rectifier substation placement through optimisation techniques, specifically genetic algorithm. This proposed technique differs from the early presented ones because the genetic algorithm is applied on substation placement for DC traction systems, considering the variations of train position and train power consumption during their operations, in order to keep all traction substations evenly loaded. In this study, the chosen parameters for optimal rectifier placement are peak demand and consumed energy, which is illustrated with the application of the proposed method.