Identification of dynamic equivalents for distribution power networks using recurrent ANNs
10 Oct 2004-pp 348-353
TL;DR: In this paper, a recurrent ANN-based dynamic equivalent for distribution networks in interconnected power systems is proposed and the implementation of such equivalents in simulation packages is outlined. But the authors focus on the dynamic performance of the original full network and that containing the equivalent model are simulated and their behaviours are compared under different disturbances in the retained network.
Abstract: This paper introduces a recurrent ANN-based dynamic equivalent for distribution networks in interconnected power systems and outlines the implementation of such equivalents in simulation packages. According to this approach, a recurrent ANN is trained in the offline mode using measurements only at boundary buses and hence it is independent of the network size and complexity. Then, a suitable dynamic model, which depends on the structure and parameters of the ANN, is developed and implemented within the simulation program. As the proposed ANN-based dynamic equivalent interacts with the retained subsystem in the online mode, it can be used for different stability analysis purposes. The proposed strategy is applied to define a dynamic equivalent for a distribution system containing a large number of active sources in a multimachine network. The dynamic performances of the original full network and that containing the equivalent model are simulated and their behaviours are compared under different disturbances in the retained network. The practical capability of the ANN in developing simple but accurate dynamic equivalents for distribution power networks is demonstrated through these comparisons.
Citations
More filters
TL;DR: A thorough survey on the academic research progress and industry practices is provided, and existing issues and new trends in load modeling are highlighted.
Abstract: Load modeling has significant impact on power system studies. This paper presents a review on load modeling and identification techniques. Load models can be classified into two broad categories: 1) static and 2) dynamic models, while there are two types of approaches to identify model parameters: 1) measurement-based and 2) component-based. Load modeling has received more attention in recent years because of the renewable integration, demand-side management, and smart metering devices. However, the commonly used load models are outdated, and cannot represent emerging loads. There is a need to systematically review existing load modeling techniques and suggest future research directions to meet the increasing interests from industry and academia. In this paper, we provide a thorough survey on the academic research progress and industry practices, and highlight existing issues and new trends in load modeling.
304 citations
Cites methods from "Identification of dynamic equivalen..."
...ANNbased black box modeling was presented in [78] to derive a nonlinear dynamic equivalent model for a MV distribu-...
[...]
01 Feb 2014
TL;DR: This paper summarizes major results of the work of the CIGRE working group on load modeling of new types of load including renewables using measurement data and historical data after two years' activities.
123 citations
TL;DR: In this paper, the grey-box approach was used for model development as it incorporates prior knowledge about the ADN structure into the model, making the model more physically relevant and intuitive than black-box or white-box models, and potentially improves the accuracy of the model.
Abstract: This paper presents the development of the dynamic equivalent model of an active distributed network (ADN) based on the grey-box approach. The equivalent model of an ADN comprises a converter-connected generator and a composite load model in parallel. The grey-box approach was chosen for model development as it incorporates prior knowledge about the ADN structure into the model, makes the model more physically relevant and intuitive than black-box or white-box models, and potentially improves the accuracy of the model. The dynamic equivalent model is presented in the seventh-order nonlinear state space format. It was initially loosely developed from the algebraic and differential equations describing assumed typical components of an ADN. Various static load models, dynamic load compositions, fault locations and a diverse range of distributed generation types and scenarios are considered in order to establish the generic range of model parameters for an ADN. The model is intended for the use in large power system stability studies.
110 citations
Cites background from "Identification of dynamic equivalen..."
...The complex voltages, power transfer and injected currents during the fault simulation were used to prepare suitable patterns for ANN training....
[...]
...Dynamic equivalents of distribution networks have also been based on a recurrent artificial neural network (ANN) [11], [12]....
[...]
...The ANN proposed to represent all active elements in the distribution network was trained by a time series....
[...]
TL;DR: In this paper, a distributed model predictive control (DMPC) based control scheme is proposed to coordinate local control actions, taken by many communicating control agents (CAs), in order to maintain multi-area power system voltages within acceptable bounds.
Abstract: This paper proposes a coordination paradigm for properly coordinating local control actions, taken by many communicating control agents (CAs), in order to maintain multi-area power system voltages within acceptable bounds. The proposed control scheme is inspired by distributed model predictive control (DMPC), and relies on the communication of planned local control actions among neighboring CAs, each possibly operated by an independent transmission system operator (TSO). Each CA, knowing a local model of its own area, as well as reduced-order QSS models of its immediate neighboring areas, and assuming a simpler equivalent PV models for its remote neighbors, performs a greedy local optimization over a finite window in time, communicating its planned control input sequence to its immediate neighbors only. The good performance of the proposed real-time model-based feedback coordinating controller, following major disturbances, is illustrated using time-domain simulation of the well-known realistic Nordic32 test system, assuming worst-case conditions.
88 citations
Cites background from "Identification of dynamic equivalen..."
...This is especially important when the external area contains a considerable number of active DG units whose power generation pattern (switching status) varies regularly and unpredictably [78]....
[...]
TL;DR: The dynamic equivalent model of ADNC is presented in a seventh-order nonlinear quasi state space format, developed from the algebraic and differential equations describing assumed typical components of the ADNC.
Abstract: Paper presents an equivalent model of an active distribution network cell (ADNC) with distributed generation for transmission system stability studies. The equivalent model of ADNC comprises a converter-connected generator and a composite load model in parallel. The gray-box approach was chosen as it enables inclusion of prior knowledge about the ADNC structure into the model development, hence making the model more physically relevant and intuitive than a black-box or white-box model. The dynamic equivalent model is presented in a seventh-order nonlinear quasi state space format, developed from the algebraic and differential equations describing assumed typical components of the ADNC. The developed equivalent model of ADNC was validated through small and large disturbance studies using the modified IEEE nine-bus transmission system model.
85 citations
Cites background from "Identification of dynamic equivalen..."
...Dynamic equivalents of distribution networks were also developed based on recurrent artificial neural networks (ANNs) [12], [13]....
[...]
References
More filters
TL;DR: In this article, the main factors that affect the quality of the reduced models are discussed and the benefits of dynamic reductions are demonstrated for three large interconnected power system models for stability studies.
Abstract: This paper documents experience with applications of dynamic reductions to large power system models for stability studies. The main factors that affect the quality of the reduced models are discussed. The quality of reduced models and the benefits of dynamic reductions are demonstrated for three large interconnected power systems.
201 citations
TL;DR: This large-signal model is formulated in continuous-time and is therefore compatible with standard models of power system components and does not postulate a particular model structure for the equivalent, hence the label nonparametric.
Abstract: The paper proposes an artificial neural network (ANN)-based strategy for identification of reduced-order dynamic equivalents of power systems. This large-signal model is formulated in continuous-time and is therefore compatible with standard models of power system components. In a departure from previous works on the subject, we do not postulate a particular model structure for the equivalent, hence the label nonparametric. The approach uses only measurements at points where internal (retained) and external (reduced) systems are interfaced, and requires no knowledge of parameters and topology of the external subsystem. The procedure consists of two conceptual steps: (1) the first ("bottleneck") ANN is used to extract "states" of the reduced-order equivalent; and (2) the second (recurrent) ANN is embedded in an ordinary differential equations (ODEs) solver, and trained to approximate the "right-hand side," using the states extracted at the first step. We also describe an extension in which a third ANN is used to synthesize missing interface measurements from a historical database of system responses to various disturbances. We illustrate the capabilities of the approach on a multimachine benchmark example derived from the WSCC system.
67 citations
"Identification of dynamic equivalen..." refers background in this paper
...ANNs have high capability to deal with complicated nonlinear problems and to resemble the behaviour of the original systems in a general frame [8, 11]....
[...]
...In some cases, the available measurements may be insufficient to develop accurate and reliable equivalents [8]....
[...]
TL;DR: In this paper, a trajectory sensitivity method is used to tune the aggregate exciter parameters of the reduced model and the optimal results are used to evaluate the aggregation from the DYNRED program and a weighted MVA method.
Abstract: Constructing a dynamic equivalent for a power system involves several steps: the partition of the system into coherent areas, the coherent area aggregation, and the aggregation of the coherent generators and their control devices. In this paper we investigate the aggregation of exciter models. A trajectory sensitivity method is used to tune the aggregate exciter parameters of the reduced model. The optimal results are used to evaluate the aggregation from the DYNRED program and a weighted MVA method. A three-machine system with one coherent area satisfying the theoretical coherency conditions is used to investigate the impact of the variations of the individual generator, network, and exciter parameters on the aggregate exciter model parameters. The results are then applied to the exciter aggregation of a larger 48-machine system.
60 citations
"Identification of dynamic equivalen..." refers background in this paper
...Hence, modelling the large number of active sources in detail will be a formidable task [3, 4]....
[...]
TL;DR: In this article, a set of measurements at boundary nodes between a subsystem that is to be modeled in detail ("retained" portion of the system) and the part that is being replaced by a simplified equivalent model is presented.
Abstract: The paper addresses practical capabilities of artificial neural networks (ANNs) in developing measurement-based continuous-time dynamic equivalents for power systems Our method is based on a set of measurements at boundary nodes between a subsystem that is to be modeled in detail ("retained" portion of the system) and the part that is to be replaced by a simplified ("equivalent") model We are particularly interested in combining standard physics-based models with signal-based models derived from measurements We utilize a color-coding scheme to distinguish between physics-based models (clear or white box) at one end, the signal-based models (opaque or black box) at the opposite end, and mixed (gray box) models in the middle The paper also proposes a way for combining classical and ANN-based equivalents in a hybrid model implemented in a standard software environment for transient analysis (in this case, ETMSP) Our conclusions are based on simulations performed on a model of a benchmark multimachine power system derived from the WSCC system
58 citations
"Identification of dynamic equivalen..." refers background in this paper
...ANNs have high capability to deal with complicated nonlinear problems and to resemble the behaviour of the original systems in a general frame [8, 11]....
[...]
TL;DR: A novel clustering method using an artificial neural network (ANN) is presented to identify the coherent generators for dynamic equivalents of power systems with rather encouraging results.
Abstract: A novel clustering method using an artificial neural network (ANN) is presented to identify the coherent generators for dynamic equivalents of power systems. First, a new frequency measure is devised to indicate the degree of coherency among system generators. Incorporating with the frequency measure, a neural network implementation of the K-means algorithm is then proposed to identify clusters of coherent generators. The rotor speeds at three selected instants in time are used as the feature patterns for the learning algorithm. To verify the effectiveness of the proposed method, extensive analyses are performed on two different power systems of varying sizes with rather encouraging results. >
55 citations
"Identification of dynamic equivalen..." refers background in this paper
...The classical nonlinear equivalency depends on the coherency concept, where a group of coherent generators is aggregated into a single equivalent one [5-7]....
[...]