Special Section

Modern drive systems with AC motors require high-quality control under varying operating conditions and the highest possible level of drive reliability. This idea results in the dynamic development of methods for diagnosing failures in the entire drive, starting with the power supply to the electric motor and ending with the control system supported by measurement systems. The application of fault detection and classification techniques is a key point in drive applications based on the concept of fault-tolerant control (FTC) systems. The task of these systems is to ensure that the drive continues to operate even if there is a failure in the components or interference with information within the control structure. It should be emphasized that asymmetries in the drive system are caused by faults in the motor, power electronics, measurement system, and control structure, which adversely affect the operation of the electromechanical system. However, taking steps to compensate for failures requires a precise determination of the type, location, and level of the fault, which explains the implications of detection and classification techniques. Such extended tasks posed to FTC systems result in an increasingly interdisciplinary design approach that considers both basic signal-analysis techniques and advanced artificial intelligence methods.
Therefore, the topic of this Special Section focuses on the detection and classification of failures in AC motor drive system components that operate according to the FTC concept using classical methods and artificial intelligence techniques. The papers prepared for this Special Section must include original material that has not been submitted to or published in any other journal. These topics include, but are not limited to:

• Sensor fault detection with the use of real or virtual signal processing,
• Data and signal processing in fault-tolerant control applications,
• Extraction of AC drive fault symptoms based on mathematical modeling,
• Algorithmic methods of fault detection and classification in AC drive systems,
• Fault diagnostics of electrical machines based on classic and deep neural networks,
• Optimization of an artificial intelligence-based diagnostic application.

Renewable power generation and electromotion have come to the forefront of technological evolution during the last years. In the energy sector, there is a surge for reducing dependence on fossil fuels and for progressive transition to renewable energy sources. Thus, there is need for nonlinear control, estimation and fault diagnosis methods that will enable the connection and synchronization of distributed and heterogeneous AC and DC renewable energy sources with the main electricity grid. In the transportation sector, the minimization of polluting gas emissions from vehicle combustion engines has become a priority, and this has imposed transition to electric vehicles and the deployment of electromotion. Thus, there is need for nonlinear control, estimation and fault diagnosis methods that will optimize the traction system of electric vehicles and will make electromotion more efficient.
This Special Section aims at analyzing recent advances in nonlinear control, estimation and fault diagnosis for renewable power generation and electromotion. Nonlinear control and estimation techniques are essential for the reliable functioning and early fault diagnosis of (a) renewable power generation units, including AC power sources (wind power, hydropower, wave energy conversion or biomass power units), DC power sources (photovoltaics or fuel cell power units), and hybrid microgrids (b) electromotion based on AC motors (such as three-phase and multi-phase PMSMs, induction motors, synchronous reluctance motors, PMBLDC motors), powertrains of electric vehicles and electric propulsion systems (c) power electronics (comprising DC/DC and AC/DC converters and DC/AC inverters, or VSC-HVDC transmission systems) that achieve synchronization of renewable energy units with the power grid or optimize energy management in electric vehicle powertrains. Topics include, but are not limited to:

• Nonlinear control based on global linearization (feedback linearization, differential flatness theory-based or Lie algebra-based control)
• Nonlinear control based on approximate linearization (optimal feedback control with linearization around specific operating points)
• Lyapunov theory-based control methods (adaptive control, sliding-mode control and neurofuzzy control)
• Nonlinear control based on sequential optimization (NMPC or sequential quadratic programming)
• Nonlinear control based on backstepping techniques (backstepping or multi-loop flatness-based control)
• Nonlinear state estimators based on global linearization (observers and filtering with differential flatness theory and Lie algebra)
• Nonlinear observers based on Lyapunov stability conditions (adaptive and sliding-mode observers)
• Nonlinear state estimators based on approximate linearization (Extended Kalman Filtering or nonlinear H-infinity Kalman Filtering)
• Nonlinear state estimators based on statistical concepts (Unscented Kalman Filtering or Particle Filtering)
• Disturbance observers (Kalman Filter-based disturbance estimators for condition monitoring)
• Fault diagnosis (fault detection and isolation with model-based and model-free techniques) statistical methods for fault threshold selection
• Machine learning methods for modelling, system identification and for large-scale optimization