Reliable operation of flexible AC transmission systems (FACTS) and high voltage direct current (HVDC) systems is necessary for stable operation of modern power grids. Converters used in FACTS and HVDC applications that are designed with high reliability constraints may still experience stoppages as a result of component failures. This may be a result of process variations in component manufacturing that can lead to over-stressing and failure of strategic components. Consequently, online estimation of parameters that provide information about the state-of-health (SoH) of critical components in these converters is crucial. This approach can help avoid unplanned stoppages and increase their availability.

Cascaded H-bridge converters (CHBCs) and modular multilevel converters (MMCs) are currently considered the state-of-the-art solutions for high-voltage and high-power conversion in both AC and DC applications. This thesis provides various solutions for online condition monitoring of submodule capacitors and power semiconductors, which are among the most critical components of CHBCs and MMCs. Moreover, novel methods are proposed to decouple the effect of temperature on health-indicating parameters. The proposed condition monitoring methods are broadly categorized into time-domain and frequency-domain estimation techniques. Both methods are thoroughly analyzed and the advantages and disadvantages of each method are explained. Furthermore, the robustness of the proposed solutions are shown under various load conditions, and in the presence of measurement uncertainties such as noise. 

The thesis also presents adaptive online methods to reduce the voltage stress on components estimated to be health-degraded. Stress reduction is achieved through modified modulation schemes, where the capacitor voltage of health-degraded submodules is reduced. The proposed adaptive control is shown to have a minor effect on the quality of the generated output current.

Finally, the thesis proposes novel semiconductor modules that comprise series-connected and parallel-connected semiconductor devices. The proposed solutions are intended to simplify the protection system and potentially provide fault-ride-though functionality in the event of single device failures in the module. The main challenge in the proposed design is the voltage balancing of series-connected devices. Two different active snubber circuits for overvoltage protection of each device are proposed, where either an active closed-loop-controlled design or a sensorless self-triggered design can be used to protect the series-connected devices against overvoltages. 

The different solutions presented in this thesis have been validated through a variety of approaches, including analytical methods, time-domain simulations, experimental testing, or a combination of these.

Tillförlitlig drift av flexibla växelspänningstransmissionsystem (FACTS) och högspänd likströmsöverföring (HVDC) är nödvändiga för stabil drift av moderna kraftnät. Effektomvandlare som har tagits fram med höga tillförlitlighetskrav kan fortfarande uppleva driftstörningar som resultat av komponentfel. Dessa är inte konstruktionsfel, utan snarare ett resultat av processvariationer i komponenttillverkning som kan leda till överbelastning och haverier i strategiska komponenter. Följaktligen kan online-estimering av parametrar som ger information om hälsotillståndet (SOH) för viktiga komponenter i omvandlaren vara behjälpliga i att undvika oplanerade driftstörningar och öka tillgängligheten. 

Så kallade Cascaded H-bridge converters (CHBC) och Modular Multilevel Converters (MMC) anses för närvarande vara de mest avancerade lösningarna för högspänd högeffektomvandling i både växelspännings- och likspänningstillämpningar. Denna avhandling presenterar olika lösningar för online-övervakning av hälsotillstånd för submodulkondensatorer och effekthalvledarkomponenter, vilka räknas bland de mest kritiska komponenterna i CHBC- och MMC- tillämpningar. Dessutom föreslås nya metoder att separera temperaturrelaterade variationer av hälsoparametrar från hälsorelaterade variationer av dessa parametrar. De föreslagna metoderna för tillståndsövervakning kategoriseras allmänt som tidsbaserade och frekvensbaserade tekniker. Båda metoderna analyseras noggrant och fördelarna och nackdelarna med respektive metod belyses. Vidare belyses robustheten hos de föreslagna lösningarna vid olika driftförhållanden och under inverkan av mätosäkerheter som brus.

Avhandlingen presenterar också adaptiva online-metoder för att minska spänningspåkänningen av komponenter som antas vara hälsodegraderade. Reduktion av påkänningar uppnås genom ett modifierat moduleringsförfarande, där kondensatorspänningen hos hälsodegraderade submoduler reduceras. Denna adaptiv styrning av submodulkondensatorernas spänning visar sig ha en obetydlig inverkan på kvaliteten av den genererade utgångsströmmen.

Slutligen föreslås nya effekthalvledarmoduler som nyttjar såväl serie- som parallellkopplade halvledarenheter. De föreslagna lösningarna syftar till att förenkla skyddssystemet och potentiellt erbjuda feltålighet vid enhetsfel i modulen. Den största utmaningen för den föreslagna konstruktionen är spänningsbalanseringen av de seriekopplade enheterna. Två olika aktiva snubbersystem för överspänningsskydd för varje enhet föreslås. Det ena är ett återkopplat reglersystem och det andra är ett sensorlöst självinitierande system. 

De olika lösningarna som presenteras i denna avhandling har validerats genom en rad olika tillvägagångssätt, inklusive analytiska metoder, tidsdomänsimuleringar, experiment eller en kombination av dessa.

This article presents a novel method for accurate online extraction of semiconductor ON-state resistance in the presence of measurement noise. In this method, the ON-state resistance value is extracted from the measured ON-state voltage of the semiconductors and the measured load current. The extracted ON-state resistance can be used for online condition monitoring of semiconductors. The proposed method is based on the extraction of selective harmonic content. The estimated values are further enhanced through an integral action that increases the signal-to-noise ratio, making the proposed method suitable in the presence of noisy measurements. The efficacy of the proposed method is verified through simulations in the MATLAB/Simulink environment, and experimentally. The estimated ON-state resistance values from the online setup are compared to offline measurements from an industrial curve tracer, where an overall estimation error of less than 1% is observed. The proposed solution maintains its estimation accuracy under variable load conditions and for different temperatures of the device under test.

Commercially available semiconductor devices have a limited range of operating voltages. This operating voltage can be increased through series connection of the devices. In this paper, a novel active snubber circuit (ASC) is proposed that protects series-connected semiconductor devices from overvoltages during operation. The unique advantage of the proposed solution is that it does not use additional sensors or an external controller for voltage protection. That is, each ASC is equipped with sufficient components to protect its respective device. The proposed solution is mainly developed for cascaded H-bridge (CHB) and modular multilevel converters (MMCs) intended for flexible alternating current transmission systems (FACTS) and high-voltage direct current (HVDC) applications, which typically operate at low switching frequencies. Suitable extensions of the proposed design are provided for increased current capability and for possible fault-ride-through functionality. The efficacy of the proposed solution is verified by simulations in the MATLAB/Simulink environment.

This article presents a unique method of extracting the ON-state resistance of metal-oxide-semiconductor field-effect transistors (MOSFETs) used in power electronic systems. The proposed method uses ON-state voltage measurements of the device-under-test as well as the load current. Contrary to prior-art solutions, the proposed method does not require estimating the device currents, but rather directly uses the load current measurements. This reduces the computational effort for estimation without reducing the estimation accuracy. Moreover, it is shown that unlike time-based estimation techniques such as the recursive least-squares estimation method, the proposed solution does not require ON-state voltage and current measurements to be accurately synchronized. The extracted ON-state resistance can be used for online condition monitoring of semiconductors, as well as for estimating the junction temperature. The efficacy of the proposed method is verified experimentally under constant and variable load conditions. Moreover, the extracted resistance values from the online setup are compared to offline measurements from an industrial curve tracer, where an overall estimation error of less than 1% is observed.

Grid-forming converters can emulate the behavior of a synchronous generator through frequency droop control. The stability of grid-forming modular multilevel converters can be studied via the impedance-based stability criterion. This paper presents an ac-side impedance model of a grid-forming modular multilevel converter which includes a complete grid-forming control structure. The impact of different control schemes and parameters on the closed-loop output impedance of the converter is thoroughly analyzed and the learnings have been used in mitigating undesired control interactions with the grid. The results are verified through simulations in time- and frequency-domains along with experiments on a down-scaled laboratory prototype.

Series connection of semiconductor devices facilitates the development of higher-voltage submodules in modular multilevel converters (MMCs) using commercially available lower-voltage semiconductor devices. However, if left uncontrolled, series-connected semiconductors may experience unequal voltage sharing after each switching event. This unequal distribution lead to semiconductor damage and system failure. This paper proposes a novel method for the series connection of semiconductor devices. The proposed solution utilizes an active snubber circuit (ASC) and a unique control system that ensure controlled voltage sharing among the semiconductors. The solution is mainly intended for MMCs operating at low switching frequencies. The efficacy of the proposed solution is verified through simulations in the MATLAB/Simulink environment.

In this paper, a novel approach for condition monitoring of de-link capacitors in modular multilevel converters (MMCs) is proposed. The solution is intended for flexible alternating current transmission systems (FACTS) and high voltage direct current (HVDC) applications. The proposed technique does not require any current measurements and solely relies on voltage measurements that are already available in MMCs. The independence from current measurements simplifies the estimation process and reduces the uncertainties that may arise from current sensors. Experimental results demonstrate an overall estimation error of less than 1% when applying the proposed technique, which is similar to the error produced by methods that use current measurements for condition monitoring.