No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Data-Driven Control of Converters in DC Microgrids for Bus Voltage Regulation
Abstract
This paper investigates the applicability of a Data-Driven Control (DDC) approach for power electronic converters in DC microgrids to stabilize the bus voltage under large perturbations from connection/disconnection of converters. In DC microgrids, loads and generators are interfaced through power electronic converters. When tightly regulated, these loads exhibit a constant power load (CPL) behavior. CPLs have negative incremental impedance behavior which may lead to system instability. The DDC approach is applied to an MVDC microgrid and compared with a model-based non-linear controller; specifically, the Linearizing State Feedback (LSF), that was previously investigated by the authors and is used here as benchmark. Both control techniques act on the interface converters at the generator side. The results of the simulation cases show that the DDC approach has a good performance for bus voltage regulation under system reconfigurations. As opposed to the LSF controller, the DDC does not require accurate analytical models, or exchange of information among converters.
... The behavior of power electronic components like motor drives and electrical motors in MVDC power systems is like CPL's under tight regulation. In this paper, the MVDC model under-consideration is based on IEEE 1709 bus system with radial topology (Cupelli et al., 2018). The model consists of four generators (G 1 − G 4 ) connected in cascade with buck converter (C 5 − C 8 ) through rectifiers (C 1 − C 4 ) to the MVDC bus. ...
... For analysis and modeling, the average value model is considered, where the ideal voltage generator (E i ) is used for modeling the generator connected to the RLC filter. Neglecting the series cable parameters (Cupelli et al., 2018;Xiao et al., 2021), the CPL's and MVDC buses are in parallel and can be denoted by the simplified version as shown in Fig. 1. Thus, the multi-generator system given of Fig. 1 is represented as follows: ...
Medium Voltage DC (MVDC) integrated power systems are anticipated to feed the submarines and surface combatants. The MVDC based power systems use dc/dc converters tightly regulated via high bandwidth controllers act as constant power loads (CPL’s). The negative incremental impedance of the CPL’s may result in voltage instability in the system under disturbance. Thus for the efficient and robust control of the MVDC based power system, this paper proposes a high order sliding mode control (HOSMC) paradigm based on model-free adaptive super twisting sliding mode control (ASTA-SMC). This paper initially designs a first-order sliding mode control (FOSMC) and super twisting sliding mode control (STA-SMC) based on HOSMC DC link voltage controllers. The difficult task of gain tuning in timely varying conditions is resolved by incorporating adaptive control to STA-SMC. The proposed controller is validated and compared with pre-presented control schemes under two test cases using simulation and experimental workbench carried out in dSPACE based hardware in the loop (HIL) workbench. The adaptive STA-SMC (ASTA-SMC) is compared with FOSMC and feedback linearization-based control schemes. The stability of ASTA-SMC in finite time is validated using the Lyapunov stability theorem.
... The behavior of power electronic components like motor drives and electrical motors in MVDC power systems is like CPL's under tight regulation. In this paper, the MVDC model under-consideration is based on IEEE 1709 bus system with radial topology (Cupelli et al., 2018). The model consists of four generators (G 1 − G 4 ) connected in cascade with buck converter (C 5 − C 8 ) through rectifiers (C 1 − C 4 ) to the MVDC bus. ...
... For analysis and modeling, the average value model is considered, where the ideal voltage generator (E i ) is used for modeling the generator connected to the RLC filter. Neglecting the series cable parameters (Cupelli et al., 2018;Xiao et al., 2021), the CPL's and MVDC buses are in parallel and can be denoted by the simplified version as shown in Fig. 1. Thus, the multi-generator system given of Fig. 1 is represented as follows: ...
A medium voltage dc integrated power system utilizes a dc/dc converter to supply constant power loads. In a poorly controlled system, the constant power loads can cause instability in the system due to its negatively incremental impedance. This paper presents a supertwisitng sliding mode control paradigm to control the MVDC power system operating under constant power loads. Extensive simulation carried out in Matlab/SIMULINK shows that, the proposed STSMC provides smooth and robust performance under varying CPL conditions. The STSMC is compared with previously proposed SMC and backstepping SMC techniques. The proposed techniques showed superior performance in terms of chattering and robustness under changing CPL conditions.
Increase in energy demand, Reduction in the available nonrenewable sources and need for reliable power supply has been main motivation for micro-grids. When many converters are interfaced to a common bus, there could be interaction among them which could lead to instability of the overall system. The varying efficiency due to solar source causes losses and resulting in using extra power to compensate the losses and wastage of available power. To improve the voltage regulation in the system, this paper proposes a Model reference adaptive controller (MRAC) designed with MIT (Massachusetts Institute of Technology) rule. MRAC validates the performance of the model by changing parameters in the adaptation gain. Our key contributions are: (1) To design DC Microgrid with higher efficiency and reliability with reduced fluctuations; (2) To preserve stability in the performance of Microgrid by MRAC controller and compared the results against baseline controllers; (3) To ensure power management of the system in case of islanding and maintains balanced flow of energy by effective load sharing. The performance of the controller is verified in MATLAB Simulink for different irradiations and it ensures stability of DC microgrid in all operating modes. A hardware prototype of the proposed system is implemented using DSP kit (TMS320F28335) with voltage and current sensors. Simulation and experimental results show that the proposed control strategy provides better output voltage regulation and stable response with short period of rise time, settling time of 1.5 s and overshoot of 6%.
This paper investigates the use of synergetic control for voltage stability of a DC microgrid where the loads are interfaced through power electronic converters fed from a DC power source. When tightly regulated, these loads exhibit a constant power load (CPL) behavior. CPLs have negative incremental impedance behavior which may lead to system instability. The synergetic control approach is compared with a linearizing feedback approach, which was investigated by the authors before and therefore serves as a reference case. Both of the selected approach achieve stability at large signal level and overcome hereby the limits of existing small signal approaches. Both of the selected control techniques act on the interface converters at the generator side and leave the load side converters untouched. The real time performance of the system and their controllers is observed on a MVDC microgrid which is implemented in Opal-RT.
Mirroring the terrestrial power system, naval warships have employed electrical power systems for over 100 years. The design philosophy for naval power systems is expressed well by the Naval Sea Systems Command (NAVSEA) Design Practices and Criteria Manual, Electrical Systems for Surface Ships, Chapter 300: The primary aim of the electric power system design will be for survivability and continuity of the electrical power supply. to insure continuity of service, consideration shall be given to the number, size and location of generators, switchboards, and to the type of electrical distribution systems to be installed and the suitability for segregating or isolating damaged sections of the system.
This paper investigates the influence of control hbox{bandwidth} on the stability of loads, which are interfaced through power electronic converters and are fed from a dc power source. When tightly regulated, these loads exhibit a constant power load (CPL) behavior. It is shown here that the ideal CPL assumption, prevalent in literature, may not represent the worst case in real-life applications. If the control bandwidth of the load is sufficiently high, the load behaves like a CPL, and the system stability margin decreases with the increase in output power. However, in a practical range, with a lower control bandwidth, the system stability margin is influenced critically by the converter's characteristic impedance, as well as its output power. Under these conditions, the minimum stability margin may occur at a low-power range.
Adaptive filtering algorithms operating in reproducing kernel Hilbert spaces have demonstrated superiority over their linear counterpart for nonlinear system identification. Unfortunately , an undesirable characteristic of these methods is that the order of the filters grows linearly with the number of input data. This dramatically increases the computational burden and memory requirement. A variety of strategies based on dictionary learning have been proposed to overcome this severe drawback. In the literature, there is no theoretical work that strictly analyzes the problem of updating the dictionary in a time-varying environment. In this paper, we present an analytical study of the convergence behavior of the Gaussian least-mean-square algorithm in the case where the statistics of the dictionary elements only partially match the statistics of the input data. This theoretical analysis highlights the need for updating the dictionary in an online way, by discarding the obsolete elements and adding appropriate ones. We introduce a kernel least-mean-square algorithm with-norm regularization to automatically perform this task. The stability in the mean of this method is analyzed, and the improvement of performance due to this dictionary adaptation is confirmed by simulations. Index Terms—Nonlinear adaptive filtering, reproducing kernel, sparsity, online forward-backward splitting.
Classical work in model reference adaptive control for uncertain nonlinear dynamical systems with a radial basis function (RBF) neural network adaptive element does not guarantee that the network weights stay bounded in a compact neighborhood of the ideal weights when the system signals are not persistently exciting (PE). Recent work has shown, however, that an adaptive controller using specifically recorded data concurrently with instantaneous data guarantees boundedness without PE signals. However, the work assumes fixed RBF network centers, which requires domain knowledge of the uncertainty. Motivated by reproducing kernel Hilbert space theory, we propose an online algorithm for updating the RBF centers to remove the assumption. In addition to proving boundedness of the resulting neuro-adaptive controller, a connection is made between PE signals and kernel methods. Simulation results show improved performance.
Kernel-based algorithms have been a topic of considerable interest in the machine learning community over the last ten years. Their attractiveness resides in their elegant treatment of nonlinear problems. They have been successfully applied to pattern recognition, regression and density estimation. A common characteristic of kernel-based methods is that they deal with kernel expansions whose number of terms equals the number of input data, making them unsuitable for online applications. Recently, several solutions have been proposed to circumvent this computational burden in time series prediction problems. Nevertheless, most of them require excessively elaborate and costly operations. In this paper, we investigate a new model reduction criterion that makes computationally demanding sparsification procedures unnecessary. The increase in the number of variables is controlled by the coherence parameter, a fundamental quantity that characterizes the behavior of dictionaries in sparse approximation problems. We incorporate the coherence criterion into a new kernel-based affine projection algorithm for time series prediction. We also derive the kernel-based normalized LMS algorithm as a particular case. Finally, experiments are conducted to compare our approach to existing methods.
The effects of constant power loads (CPLs) has been for long time the focus of research in isolated power systems such as automotive systems, electric ships and spacecrafts, with main focus on DC-systems. Due to the instability effect of the negative incremental resistance phenomena of CPLs, several criteria have been investigated to ensure system stability in the presence of large amount of CPLs. With an increasing share of CPLs, the same negative incremental resistance phenomena can be the source of instability in AC distribution systems, especially in the event of voltage disturbances. The literature on AC systems has not yet addressed this problem in a comprehensive way, and more research effort will be required around the topic. This paper investigates the effect of the negative incremental resistance phenomena in AC distribution systems with large share of CPLs, and proposes an approach for mitigation of instability in presence of large signal disturbances based on injection of reactive current by active rectifiers interfacing the CPLs. Results from a simulation study with Matlab suggest how stability can be improved by reactive compensation and that distributed injection of reactive current by the CPLs might be more effective than a centralized compensation by a STATCOM.
Power electronic converters and electric motor drives are being put into use at an increasingly rapid rate in advanced automobiles. However, the new advanced automotive electrical systems employ multivoltage level hybrid ac and dc as well as electromechanical systems that have unique characteristics, dynamics, and stability problems that are not well understood due to the nonlinearity and time dependency of converters and because of their constant power characteristics. The purpose of this paper is to present an assessment of the negative impedance instability concept of the constant power loads (CPLs) in automotive power systems. The main focus of this paper is to analyze and propose design criteria of controllers for automotive converters/systems operating with CPLs. The proposed method is to devise a new comprehensive approach to the applications of power electronic converters and motor drives in advanced automotive systems. Sliding-mode and feedback linearization techniques along with large-signal phase plane analysis are presented as methods to analyze, control, and stabilize automotive converters/systems with CPLs
This chapter presents the fundamental theory of model-reference adaptive control. Various types of uncertainty are defined. The composition of a model-reference adaptive control system is presented. Adaptive control theory for first-order single-input single-output (SISO) systems, second-order SISO systems, and multiple-input multiple-output (MIMO) systems is presented. Both direct and indirect adaptive control methods are discussed. The direct adaptive control methods adjust the control gains online directly, whereas the indirect adaptive control methods estimate unknown system parameters for use in the update of the control gains. Asymptotic tracking is the fundamental property of model-reference adaptive control which guarantees that the tracking error tends to zero in the limit. On the other hand, adaptive parameters are only bounded in the model-reference adaptive control setting.
The aim of the work is to present the state-of-the-art in this filed of adaptive control. Some development trends based especially on the general stability theory are analyzed. The characteristics of basic model reference adaptive control (MRAC) structures and algorithms are given in order to provide an insight into the development of the theory and to indicate the possibilities of their practical application. As the paper makes only a survey of the subject area, the stability proofs of the presented algorithms are omitted. The plant to be controlled is assumed to be described by a linear model.
This paper investigates the data-based opti- mal control for a class of networked industrial processes with double-layer architecture. Without knowing the dynam- ics of subsystems at device layer, the index prediction function is constructed via the input/output signals, and radial basis function neural networks (RBFNNs). The tuning laws for the index prediction function are obtained through optimal control strategy. Then, by treating the network- induced phenomenon as random round-trip time delay and introducing the predictive algorithm, the compensation scheme is designed at the operation layer to dynamically decompose the setpoints. Finally, two simulation examples are given to further illustrate the effectiveness of the pro- posed compensation strategy.
This paper presents the Hardware in the Loop (HiL) validation of a control method to regulate the voltage of the DC Bus fed by Distributed Generation (DG) units in an islanded MVDC microgrid. The disturbance based control, employs a decentralized Linear Quadratic Gaussian (LQG) control of individual generator side converters to be executed locally, and which does not need remote measurements of loads and sources in its control scheme. This enables “plug-and-play” capabilities. The virtual disturbance model permits changes in the reference voltage and the variation in the load parameters based on local knowledge without the need of a communication infrastructure. The testing is performed via HiL implementation with Real Time Digital Simulator (RTDS). HiL Simulations show that this method can handle switching noise, ripple and other effects that are not present in averaged models.
In future all-electric ships, which rely on a Medium Voltage DC distribution network, the power system stability can be jeopardized by the presence of constant power loads, due to their negative incremental resistance. In fact, this causes stability problems due to the negative impedance characteristic of tightly regulated converters. The Linearizing State Feedback (LSF) has shown the capability to stabilize these systems. The key modelling assumption for this control is the possibility to neglect cable parameters. In this paper we implement the LSF control in an FPGA hardware and simulate the shipboard power system in RTDS to test the validity of the cable assumption in Hardware in the Loop (HiL) tests.
A model reference adaptive control (MRAC) algorithm is proposed based on application of hyperstability theory combined with the use of constant feedback on tracking error. The algorithm provides the advantages of both robust and adaptive approaches with reference to the desired quality of tracking transient process. The crucial point is an hypothesis of a 'separate' dependence of the error equation system on the state variables and the unknown parameters.
Bus voltage stability is a key issue in future medium-voltage DC (MVDC) power systems on ships. The presence of high-bandwidth controlled load converters (Constant Power Load, CPL) may induce voltage instabilities. A control design procedure is presented which starts at the modeling level and comes to control implementation. A control method based on a Linearization via State Feedback (LSF), is proposed to face the CPL destabilizing effect and to ensure the MVDC bus voltage stability. A multiconverter shipboard DC grid is analyzed by means of a new comprehensive model, which is able to capture the overall behavior in a second-order nonlinear differential equation. Exploiting DC–DC converters that interface power sources to the bus, LSF technique is able to compensate for system nonlinearities, obtaining a linear system. Then, traditional linear control techniques can be applied to obtain a desired pole placement. With reference to system parameters mismatch, LSF control design is verified by means of a sensitivity analysis, evaluating the possibility of an over-linearization strategy. Time-domain numerical simulations are used to validate the proposed control, in presence of relevant perturbations by means of a two-way comparison (average value model and detailed switching model).
A major challenge in MVDC shipboard power systems is maintaining Medium Voltage DC (MVDC) system stability, due to the presence of the negative incremental resistance behavior of Constant Power Loads (CPLs). This paper presents a decentralized Linear Quadratic Gaussian (LQG) control approach for individual local generator side converter to regulate and stabilize the MVDC system. In this control approach, we model the non-linear CPL, which the decentralized controller is responsible for, as a virtual disturbance and consider it as an additional state to be estimated by Extended Local Kalman Filter (ELKF). Using the estimated virtual disturbance, the set-point trajectory in steady state can be obtained and applied in Linear Quadratic Regulator (LQR) with the optimal state feedback gain. Accordingly, linearization of the constant power load is achieved via this accurate estimation of the virtual disturbance; stability is guaranteed via an optimal LQR, as well as the load is shared among the generation units via droop control. Due to the fact that the estimation of each individual virtual disturbance is based only on the local measurements and the local fix model implemented in ELKF, the MVDC system adaptive set-point is achieved in local controllers without communication. This assumption holds true not only in the case of a sudden large load connection or disconnection, but also in case of generator loss.
Voltage stability in Medium-Voltage DC (MVDC) power systems on ships is a key design goal. MVDC bus voltage stability can be impaired due to the presence of high-bandwidth controlled Constant Power Load (CPL) converters, which can induce negative incremental resistance instabilities. A linearization via state-feedback control is presented in this paper, to stabilize a MVDC bus voltage in presence of destabilizing CPLs. Once a linear control system is obtained, a traditional stable pole placement is operated. The control is designed to be implemented using the controlled DC/DC converters that interface the system power sources to the MVDC bus. The proposed control is verified by means of time-domain numeric simulations based on both an average state space model and a detailed power electronics circuit model, thus providing a two-way comparison.
This paper presents a nonlinearly parameterized controller for the adaptive control of base-isolated buildings subjected to a set of near-fault earthquakes. The control scheme is based on discrete direct adaptive control, wherein the system response is minimized under parameter uncertainties. Stable tuning laws for the controller parameters are derived using the Lyapunov approach. The controller utilizes a linear combination of nonlinear basis functions, and estimates the desired control force online. The measurements that are necessary to generate the control force to reduce the system responses under earthquake excitations are developed based on the adaptive systems theory. The main novelty in this paper is to approximate the nonlinear control law using a nonlinearly parameterized neural network, without an explicit training phase. A perturbed model is used to initialize the controller parameters in order to simulate the uncertainty in the mathematical modeling that typically exists in representing civil structures. Performance of the proposed control scheme is evaluated on a full-scale nonlinear three-dimensional (3-D) base-isolated benchmark structure. The lateral-torsion superstructure behavior and the bi-axial interaction of the nonlinear bearings are incorporated. The results show that the proposed controller scheme can achieve good response reductions for a wide range of near-fault earthquakes, without a corresponding increase in the superstructure response. Copyright © 2011 John Wiley & Sons, Ltd.
This paper proposes a new approach to the control of a power network with parallel power converters. A decentralized linear quadratic Gaussian (LQG) control approach based on local models without full knowledge of the overall network is proposed. Through integrating a virtual disturbance source model the local LQG controllers can individually learn online information about the dynamics of the others. By using this information for adaptation of its local set-point the decentralized controllers could drive the system to a stable operating point and eliminate the disturbance in a cooperative way. The load is also properly shared between the power converters. This control logic could be applied to the coordination of a set of power electronics systems managing the interface of distributed power sources. The presented approach could be regarded as a first step towards the development of a “plug-in” or ad-hoc power generation concept. In this concept the generation units feed power into the network via power converter interfaces without any centralized control while nevertheless guaranteeing the network's stability and performance.
A synergetic control for parallel-connected DC-DC Buck Converters
- I Kondriatev