Anubrata Dey

This paper presents the design and experimental validation of a laboratory prototype wind turbine emulator for a Type-IV Wind Energy Conversion System (WECS). This setup employs a DC motor coupled with a Permanent Magnet Synchronous Generator (PMSG), interfaced to the grid through a three level Neutral Point Clamped (NPC) back-to-back converter. Optimal Torque Control (OTC) combined with a Super-Twisting Sliding Mode Controller (ST-SMC) is employed for machine-side current regulation to ensure maximum power extraction. On the grid side, a robust ST-SMC-based DC-link voltage controller is designed for effective control of the Grid-Side Converter (GSC). In addition, a novel ST-SMC-based neutral-point balancing strategy is proposed for the NPC converter, offering faster convergence and stronger disturbance rejection than conventional methods. Experimental results demonstrate that the ST-SMC limits DC-link overshoot to 25% during reference step changes (compared to 80% with PI control) and to 11% under wind variations (compared to 25% with PI). The neutral-point balancing scheme achieves a convergence time of 260 ms, significantly outperforming existing approaches. Furthermore, when applied jointly for DC-link regulation and neutral-point balancing, the proposed method yields the lowest total harmonic distortion (THD) of 3.8%, confirming its superior performance over traditional PI control.

Due to high power density, efficiency, cost-effectiveness, and low filtering requirement, transformerless multilevel inverters have gained considerable interest in solar photovoltaic (PV) applications. In order to cut the flow of leakage current for a grid connection, bulky galvanic isolation based configuration is conventionally used. Instead, a transformerless configuration can effectively mitigate electromagnetic interference (EMI) issues associated to this leakage current that finds a path through parasitic capacitance existing between the PV panels and the ground. In this context, this work presents a novel resource optimized five-level transformerless inverter topology in which ac side decoupling is exploited and it is compared with few existing five-level transformerless inverter topologies. The presented inverter also exhibits reduced conduction and switching losses. A state-of-the-art modulation strategy is developed which not only ensure leakage current suppression in accordance to VDE-0126-1-1 standard but also able to control dc bus midpoint voltage by applying redundant switching states in a simple manner. The inverter’s performance is evaluated through simulations in MATLAB and PLECS environments. Furthermore, a scaled-down lab prototype has been developed to validate its functionality.

There is a growing trend of incorporating modular multilevel converter (MMC) in medium voltage applications. However, the large requirement of sensors and associated circuitry for conventional control of MMC increases the hardware complexity. Additionally, during normal operation, significant temperature imbalances occur between the top and bottom power devices within a submodule (SM). There is also a substantial temperature imbalance among the SMs due to different aging profiles and parameter mismatches. The prolonged increased thermal stress and unbalance in any particular SM may result in its ultimate failure, reducing its lifetime and thereby seriously hampering the reliability of MMC. In recent times, reduced sensor-based MMC operation for medium voltage application has been established, where all SM voltage sensors can be replaced by a single arm voltage sensor. The temperature imbalances may be aggravated for such operation, and it requires a thorough investigation. To address this issue, this work proposes a temperature regulation scheme for SMs in a three-phase MMC. This regulation is achieved by formulating a multiobjective cost function that also takes care of SM capacitor voltage balance. Extensive simulations and experimental studies on a grid connected three-phase MMC demonstrate the effectiveness of the proposed scheme.

Single switch based DC-DC converters are a fore-most requirement in many applications, so is their reliable operation. This article concerns with switch fault diagnosis using a signal based approach. Compared to many other methods such as model based or knowledge-based diagnosis, proposed signal based approach is the simplest one. This approach presents merits like fast diagnosis speed, general applicability, cost-effectiveness, and robustness. The scheme is tested for switch open-circuit fault, short-circuit fault under both continuous conduction mode (CCM) and discontinuous conduction mode (DCM) of operation. The robustness of the scheme is tested by considering variations in the duty cycle, parameter value, threshold value, and switching frequency. The scheme works fine under dynamic conditions for sudden change in load current. The suggested diagnosis method uses the information of diode current and it avoids the use of additional sensor. The fault diagnosis algorithm is capable of detecting and determining the type of fault in less than one switching time period. Validation of its effectiveness is carried out by simulation and experimental studies.

The modular multilevel converter (MMC) is a popular choice for various medium voltage applications. Recently, several works have focussed on reducing the number of sensors and thereby reducing the hardware complexities in an MMC. Although the normal operation of MMC with a reduced sensor scheme showed some decent performance, the detection of a switch or diode fault and its localization with a reduced sensor operation remains challenging. There is very few literature that deal with fault identification in MMC with reduced sensor operation, but they lack either due to complexity in fault identification technique or due to dependency on additional measurement. In this work, for the first time, both switch and diode open circuit faults are identified based on certain measurements such as arm voltages, observed arm current, and switching status. As these measurements are being used for the healthy operation of MMC, additional sensor is avoided for the proposed fault identification technique. The fault identification also avoids a computationally intensive software based approach. Extensive simulation and experimental studies are carried out to identify single and multiple switch and diode faults. It is also verified that there is no false trigger in open circuit submodule fault identification due to step change in output current.

With ever-increasing renewable power penetration, the reliable operation of grid interfaced inverter is an important criterion. The availability of a grid-connected multi-level inverter is relied upon the correctness of identifying faulty switch, if any, during operation and implementation of a fault-tolerant modulation technique to run the inverter without tripping off even under faulty conditions. Although having a large number of switches in a multi-level inverter throws challenges for the detection and localization of faults, it is a boon because power can still flow through alternative paths. In pursuit of this objective, this paper utilizes a T-type based hybrid multi-level topology where certain faulty conditions are created. The proposed fault detection scheme begins with the fast identification of faulty phase, achieved by monitoring the absolute phase error variable derived from normalized phase currents. Subsequently, the localization of faulty switch is fulfilled by intricate analysis of several current and voltage measurements. Notably, the proposed scheme does not need additional sensors to carry out all of these, where finally, a carrier-based fault tolerant scheme is devised to encounter open switch faults. The effectiveness and feasibility of the proposed scheme are evaluated by simulation studies followed by experiments on a seven-level inverter prototype.

The reduced sensor operation of a Modular Multilevel Converter (MMC) lowers hardware complexity and failure rates resulting from malfunctioning sensors. Hence, in medium voltage applications, the reduced sensor operation may increase the appeal of MMC. A large number of individual submodule capacitor voltage sensors and three-phase arm current sensors are eliminated in this work for the internal control of MMC. Nevertheless, a well-designed arm current observer is utilized in a direct leg modulation scheme appropriate for medium voltage applications. The computational complexity of the proposed observer is low and the execution time remains fixed irrespective of the increase in submodules of MMC. The submodule capacitor voltage prediction is based on a simple logical approach using a single arm voltage measurement. The proposed reduced sensor operation of MMC is validated using steady state and transient studies, based on both simulations and hardware experiments after considering certain variations in submodule capacitances. Moreover, an online arm inductance and submodule capacitance monitoring is performed to encounter the degradation of parameters over a period of time.

For a grid-connected inverter, a constant switching frequency hysteresis current control (HCC) technique based on targeted trajectory of current space vector (HCC-TTCSV) is proposed. The harmonic spectrum of inverter line-to-line voltage, produced by the proposed HCC-TTCSV, is ensured to exhibit similar characteristics to that of voltage-controlled conventional seven-segment space vector modulation (SVM) technique. The proposed current controller is a novel one (i.e., first of its kind) that allows the phase currents to be checked with online calculated current boundaries in stationary abc reference frame. Instead of following a current error boundary, phase current boundary check is performed. The phase current boundary values during every seven-segment switching cycle, however, are found out by observing the projections of vertices of targeted trajectory of current vector along abc axes using few trigonometric relations. The proposed controller is designed for three-phase two-level voltage-source inverter (2L-VSI), and it works well for both L filter and LCL filter. Compensation methods to overcome the current distraction effect caused by factors such as finite rate of current sampling, dead time, gate driver circuit's propagation time, and other nonidealities of the system are also discussed in detail. Theoretical background is established, and experimental verification is presented for validation.