In the case of rotating machine monitoring systems, classical approaches for health monitoring and fault detection become obsolete. These classical approaches require sensors to collect enough data. This data collection process is not always a viable option, especially in new and special equipment such as hosting unit in nuclear reactors, gantry cranes for shipbuilding, etc. For this purpose, an analytical model based on the modified winding function (MWF) of an induction motor is proposed to generate the data, and to simulate the different types of faults such as eccentricity and broken rotor bars under variable conditions. Correct implementation of faults in the model can be detected from the generated stator current. This model uses turn and winding functions to describe the distribution of windings, as well as an air gap function incorporating stator and rotor slot effects. These functions are used to calculate inductances and flux distribution in the air gap, and are saved offline in a lookup-table for a high number of rotor positions. This lookup-table is used to calculate currents, torque, speed, rotor position in a reduced simulation time online. This paper proposes an induction motor model for current simulation under variable regimes and different types of faults (healthy, broken rotor bar, static, dynamic and mixed eccentricities). These simulations are used for a robust health indicator (HI) extraction coupled with a regime identification and regime normalization steps to reduce class dispersion, and to separate different health states. Results are compared with the ones calculated using FEM under healthy, broken rotor bar, and static, dynamic, mixed eccentricities.

The diagnosis of bearing and gear faults solely through non-intrusive current signal analysis is an interesting approach. This paper’s primary contribution lies in the extraction of a novel indicator for identifying these faults irrespective of speed and load variations. In fact, comprehensive monitoring systems encompassing variable speed and load regimes alongside combined gearbox faults solely using electrical signals are scarce in the existing literature. To address this gap, we propose a pioneering method. The proposed method uses a regime normalization technique with different current sensors, allowing the grouping of various regimes under the same health state. By minimizing dispersion among class observations and distinguishing between different health states, including variations in speed and load, this method promises heightened diagnostic accuracy. This paper proposes also to improve the diagnosis of faults by using the Naïve Bayes classifier with the introduction of a criterion called the threshold in order to address the uncertainty. Addressing uncertainty is crucial to prevent false alarms during diagnosis. To verify the effectiveness of the proposed method, current data collected from a test bench composed of a gearbox system and operating under variable speed and load conditions is tested and validated. Also, the proposed method is compared with diverse machine learning classifiers to test the effectiveness of introducing the uncertainty to improve the diagnosis. The proposed model is performed based on accuracy of 97.55%, precision of 99.8% and F1-socre of 98.73%.

In practice, rotating machines operate under real-time variable speed and load conditions, particularly in railways, wind turbines, robotics, etc. In the presence of faults, the degradation of critical components is significantly accelerated, making the developments of robust monitoring algorithms crucial to be able to identify these faults with regard to regime variation. For this purpose, a signal processing methodology is proposed. It is based on regime normalization that allows the grouping of different regimes belonging to the same health state. It consists of a thresholding method that detect condition variation and uses distance metrics for normalization. This, allows reducing dispersion between the class observations while separating the classes representing different health states. To verify the performance of the methodology, current, voltage and vibration data from a gearbox system under variable speeds and loads are used.

Fault detection and diagnostics are important steps in the predictive maintenance of industrial systems, especially faults in the mechanical parts most susceptible to fail, such as bearings and gears in rotating machines. These two components represent more than 50% of causes of the operational downtime. Therefore, the detection of their appearance allows anticipating the total failure of the machine and schedule in advance maintenance actions. However, in the presence of a combined gear and bearing faults, it is difficult to isolate their states. To remedy this situation, this paper proposes a data processing methodology that exploits the three-phase current signals of the rotating machine and build a health indicator (HI) from each current phase that reveals the different health states. This indicator is constructed by extracting features from the collected raw data in frequency and time domains, and then they properly combined with a physical significance. After that, all health indicators (HIs) of the three phase current data are fed to a machine learning model for an online pattern recognition of the bearing and gear states, including the combined faults. The proposed approach is demonstrated through a test bench that studies bearing and gear defects of a gearbox under different operating conditions.

Despite their great improvements, reliability and availability of power electronic devices always remain a focus. In safety-critical equipment, where the occurrence of faults can generate catastrophic losses, health monitoring of most critical components is absolutely needed to avoid and prevent breakdowns. In this paper, a noninvasive health monitoring method is proposed. It is based on fuzzy logic and the neural network to estimate and predict the equivalent series resistance (ESR) and the capacitance (C) of capacitors and supercapacitors (SCs). This method, based on the neo-fuzzy neuron model, performs a real-time processing (time series prediction) of the measured device impedance and the degradation data provided by accelerated ageing tests. To prove the efficiency of the proposed method, two experiments are performed. The first one is dedicated to the estimation of the ESR and C for a set of 8 polymer film capacitors, while the second one is dedicated to the prediction of the ESR and C for a set of 18 SCs. The obtained results show that combining fuzzy logic and the neural network is an accurate approach for the health monitoring of capacitors and SCs.

Supercapacitors are devices used in wide range of applications, for example in automotive applications. Therefore, it is important to monitor and track their ageing. This paper presents a new approach for predicting the ageing of supercapacitors based on the neo-fuzzy neuron in association with the one-step ahead time series prediction. Ageing information collected from the measurement of the equivalent series resistance and the double layer capacitance are used to train the neo-fuzzy neuron. The obtained model is used as a prognostic tool in order to forecast the ageing of supercapacitors. The performance of the proposed approach is evaluated by using an experimental platform for ageing supercapacitors. The experimental results show that the neo-fuzzy prediction model can track the ageing of supercapacitors.