Battery cell temperature is a key parameter in battery life degradation, safety and dynamic performance. Intense charging-discharging operation and high ambient temperature escalate battery cell temperature, which accelerates its degradation. Therefore, accurate battery cell temperature estimation can play a significant role in ensuring optimal operation of a battery energy storage system (BESS) considering its degradation. In order to estimate battery cell temperature as accurate as possible, non-linear models are necessary because of the non-linear nature of the battery operation. The main objective of this paper is to propose a data-driven model based on a Non-linear Autoregressive Exogenous (NARX) neural network to estimate battery cell temperature in a utility-scale BESS, considering strongly correlated independent variables (i.e., charging-discharging current and ambient temperature). Due to different temperature and weather characteristics in each season, seasonal NARX model has also been derived and compared with the universal one. The proposed model has been evaluated using field data collected from a grid-connected BESS within a PV plant. The simulation results show high accuracy of the model compared to the measured data for both seasonal and universal models without considering the complexity of the large-scale battery and container thermal dynamics. In particular, more than 95% of the time, the estimated values yield root mean squared errors (RMSE) below 1C in different conditions, which confirms the validity and accuracy of the model. Moreover, seasonal models show better performance with 18% to 50% less RMSE on average (for 1 hour to 24 hours ahead estimation) compared to the universal model.
The replacement of conventional synchronous generators by renewable energy sources reduce the mechanical inertia of power systems. Reduction in the system mechanical inertia may jeopardize reliable and stable operation, especially in restoring frequency following a disturbance in a power system with high penetration of renewable energy. Energy storage can be utilized in such situations for providing inertial support and frequency regulation. Several energy storage technologies are available in market and each offers different power, energy, and economic characteristics and they are subjected to different limitations, including cost. Thus, it is crucial to find out the optimum storage technology for given power system. In this paper, an analytical technique is developed to estimate the size of the storage system to provide inertial response (IR) and primary frequency regulation (PFR) in power systems with high penetrations of renewable energy sources. Different types of storage technologies including hybrid storages suitable for IR and PFR services are sized while considering their dynamic characteristics and limitation. An economic analysis is also carried out to find the optimum storage technology for IR and PFR services. The theoretical estimated results are compared and corroborated by non-linear simulations.
Battery energy storage systems (BESS) are going to be an inevitable part of the smart grids. As a result, different aspects of the BESS operation and application have been investigated in numerous papers in recent years, e.g., optimal sizing and long-term operation. In such studies, different formulations have been used for battery efficiency during charging and discharging, which leads to different sets of results. In addition, most of the battery sizing studies ignored degradation over its useful lifetime. In this paper, we classify various battery efficiency formulations and compare their performance through an extensive simulation study. Moreover, a linear cyclic and calendar ageing formulation is proposed based on counting energy throughout and idle time that can be conveniently integrated with any battery optimal sizing and long-term management algorithms. The simulation results reveal a significant impact of degradation on the optimal solutions obtained in sizing studies.
This paper studies the feasibility of peer-to-peer(P2P) energy trading in a grid-tied network. The main objectives are to understand the impact of a P2P energy trading model onthe network operation, and thus demonstrate the importance of taking various issues related to power network into account while designing a P2P trading scheme. To do so, firstly a simple mechanism is developed for energy trading among prosumers without considering any network constraints, as done by many existing studies. Once the trading outcome is finalised, the developed scheme is tested on the low-voltage (LV) network model to check its feasibility in real world. It is shown that while the considered trading scheme is economically beneficial to the prosumers compared to the current incentive mechanisms (such as feed-in-tariff), 1) it could be unfit for real deployment due to violating voltage limits in some scenarios, and 2) the grid may experience financial losses for compensating the path losses during P2P trading at a given interval.
Demand response (DR) will be an inevitable part of the future power system operation to compensate for stochastic variations of the ever-increasing renewable generation. A solution to achieve DR is to broadcast dynamic prices to customers at the edge of the grid. However, appropriate models are needed to estimate the potential exibility of different types of consumers for day-ahead and real-time ancillary services provision, while accounting for the rebound effect (RE). In this study, two RE models are presented and compared to investigate the behaviour of flexible electrical consumers and quantify the aggregate exibility provided. The stochastic nature of consumers' price response is also considered in this study using chance constrained (CC) programming.
Battery cell temperature is an important factor in battery capacity degradation, performance, and safety. Elevated battery cell temperature, due to intense battery operation with high charging-discharging current and ambient temperature, accelerates battery capacity degradation as well as causing extra cooling cost. Therefore, it is indispensable to estimate battery cell temperature accurately for optimal BESS operation considering capacity degradation and its associated costs. The main objectives of this paper are to propose a linear model to estimate battery cell temperature using Autoregressive Integrated Moving Average with eXogenous variables (ARIMAX) considering strongly correlated independent variables and propose a cost function of PV plant with and without battery operation. The simulation results using field data show high accuracy of the proposed temperature estimation model without considering the complex thermal dynamics of the entire system. In addition, comprehensive cost functions are developed to show the benefit of integrating battery storage into a PV plant and determine influential factors to consider in any optimal battery operation systems.
This paper presents a new methodology to exploit consumers’ flexibility for the provision of ancillary services (AS). The proposed framework offers a control-based approach that adopts price signals as the economic driver to modulate consumers’ response. In this framework, various system operators broadcast price signals independently to fulfil their AS requirements. Appropriate flexibility estimators are developed from the transmission system operator (TSO) and distribution system operator (DSO) perspectives for price generation. An artificial neural network (ANN) controller is used for the TSO to infer the price-consumption reaction from pools of consumers in its territory. A PI controller is preferred to represent the consumers’ price-response and generate time-varying electricity prices at the DSO level for voltage management. A multi-timescale simulation model is built in MATLAB to assess the proposed methodology in different operational conditions. Numerical analyses show the applicability of the proposed method for the provision of AS from consumers at different levels of the grid and the interaction between TSO and DSOs through the proposed framework.
To be included later ...
Utility-scale energy storage is becoming the most viable and widespread solution to safely accommodate large-scale renewable generation without harming the power system operation. Different storage technologies have been considered for such applications, where various Li-based battery technologies gained the largest share of the market, after pump hydro storage. While numerous research studies addressed optimal operation of battery storage using mathematical modelling, there are few examples of battery performance analysis in such applications. The University of Queensland owns and operates one of the largest PV/battery research facility in the world at the Gatton campus. It includes 3.275 MWp solar PV modules with three different solar tracking technologies, and 600kW/760 kWh Li-Polymer battery system. The plant is primarily responsible for meeting the campus load demand, and the excess energy, if available, can be pushed back to the local network. This presentation will provide valuable insights into the utility-scale battery operation in connection with a PV plant and local network based on 1.5 years of actual data. The focus will be on evaluating the operation and performance of the battery system from different perspectives. I will start with briefly introducing the plant, battery system configuration, cooling mechanism, and inverter. Then, I will briefly review battery operation rules in the plant which is managed by a central supervisory controller. Battery operation and performance for reactive power support will be reviewed next. Afterwards, voltage agreement with the local network operator will be reviewed, and the impact of the battery will be quantified in that regard. Finally, the performance of the battery for demand charge management (or peak shaving), excess PV storage, PV ramp-rate control, and additional services to the local network will also be examined based on operational data. Data analysis will show that the battery operation had positive impacts on the plant operation.
The total energy bill for commercial/industrial (C/I) loads consists of two parts: Energy charge which is proportional to the total energy consumption, and demand charge which is proportional to the peak power consumption. The demand charge is known to be more than 50% of the total cost. To reduce demand charge, C/I customers are equipped with behind-the-meter (BTM) battery storage. However, battery technologies are still expensive and the battery in such applications sits idle for most of the day. In order to create extra revenue for the C/I customer, which further increase the economic benefit of BTM devices, we proposed a two–layer incentive –based controller to aggregate small BTM devices for participation in the wholesale energy market. In the aggregator level, incentive signals will be generated and communicated to each customer based on predicted values of wholesale energy market prices. At the local level, a local controller is designed to optimally create battery operation profile for the next day based on the Time-of-Use (ToU) and demand charge prices as well as incentive signal received from the aggregator. Eventually, the local controller commit to a certain level of generation to consume/provide in each hour of the day ahead. Simulation results show an average of 12% extra revenue for the C/I customers considering battery degradation cost, and uncertainties related to predicted values.
A system and method are provided. The system includes a processor. The processor is configured to receive power related data relating to power usage of power consuming devices at a customer site from a plurality of sources. The processor is further configured to generate object function inputs from the power related data. The processor is additionally configured to apply the generated object function inputs to an objective function to determine an optimal capacity for a battery storage system powering the power consuming devices at the customer site while minimizing a daily operational power cost for the power consuming devices at the customer site. The processor is also configured to initiate an act to control use of one or more batteries of the battery storage system in accordance with the optimal capacity for the battery storage system.
Computer-implemented methods and, a system are provided. A method includes constructing by an Energy Management System (EMS), one or more optimization-based techniques for resilient battery charging based on an optimization problem having an EMS cost-based objective function. The one or more optimization-based techniques are constructed to include a battery degradation metric in the optimization problem. The method further includes charging, by the EMS, one or more batteries in a power system in accordance with the one or more optimization-based techniques.
Aspects of the present disclosure describe a single battery degradation model and methods that considers both CYCLING and CALENDAR aging and useful in both energy management and battery management systems that may employ any of a variety of known battery technologies.
A method and system are provided. The method includes determining, by a processor, a set of battery aging modeling parameters that include battery capacity for a battery based on a statistical analysis applied to experiment data. The experiment data is obtained from measurements of a set of battery parameters that include battery capacity and that are taken by a hardware-based battery parameter monitoring device during a plurality of experiments which vary another set of battery parameters. The set and the other set have at least some different members. The method further includes generating, by the processor, a battery aging neural network based model for the battery that includes the set of battery aging modeling parameters. The method also includes storing the battery aging neural network based model in a memory device.
A management framework is disclosed that achieves maximum energy storage device lifetime based on energy storage device life estimation and the price of energy
This book is a comprehensive solution manual and briefing courses for Electrical Circuits of first and second-order. Every chapter starts with a short course and notes about a specific topic in Electrical Circuits which is followed by solving numerous questions. Every question is explained in details and the equations are dervied step-by-step. This is a valuable source to learn this difficult toipc in Electrical Engineering in depth.
A multi-objective optimization algorithm is proposed in this paper to increase the penetration level of renewable energy sources (RESs) in distribution networks by intelligent management of plug-in electric vehicle (PEV) storage. The proposed algorithm is defined to manage the reverse power flow (PF) from the distribution network to the upstream electrical system. Furthermore, a charging algorithm is proposed within the proposed optimization in order to assure PEV owner’s quality of service (QoS). The method uses genetic algorithm (GA) to increase photovoltaic (PV) penetration without jeopardizing PEV owners’ (QoS) and grid operating limits, such as voltage level of the grid buses. The method is applied to a part of the Danish low voltage (LV) grid to evaluate its effectiveness and capabilities. Different scenarios have been defined and tested using the proposed method. Simulation results demonstrate the capability of the algorithm in increasing solar power penetration in the grid up to 50%, depending on the PEV penetration level and the freedom of the system operator in managing the available PEV storage.
Power management is an essential tool for microgrid (MG) safe and economic operation, particularly in the islanded operation mode. In this paper, a multi-timescale costeffective power management algorithm (PMA) is proposed for islanded MG operation targeting generation, storage, and demand management. Comprehensive modeling, cost, and emission calculations of the MG components are developed in this paper to facilitate high accuracy management. While the MGs overall power management and operation is carried out every several minutes to hours, depending on the availability of the required data, simulation for highly dynamic devices, such as batteries and electric water heaters (EWHs) used for demand response (DR), are performed every minute. This structure allows accurate, scalable, and practical power management taking into consideration the intrainterval dynamics of battery and EWHs. Two different on/off strategies for EWH control are also proposed for DR application. Then, the PMA is implemented using the two different DR strategies and the results are compared with the no-DR case. Actual solar irradiation, ambient temperature, nonEWH load demand, and hot water consumption data are employed in the simulation studies. The simulation results for the MG studied show the effectiveness of the proposed algorithm to reduce both MGs cost and emission.
Demand response (DR) has shown to be a promising tool for balancing generation and demand in the future power grid, specifically with high penetration of variable renewable generation, such as wind. This paper evaluates thermostat setpoint control of aggregate electric water heaters (EWHs) for load shifting, and providing desired balancing reserve for the utility. It also assesses the economic benefits of DR for the customers through time-of-use pricing. Simulation results reveal the achievement of the economic benefits to the customers while maintaining their comfort level and providing a large percentage of desired balancing reserve at the presence of wind generation.
Demand response (DR) has proved to be an inevitable part of the future grid. Much research works have been reported in the literature on the benefits and implementation of Dr However, little works have been reported on the impacts of DR on dynamic performance of power systems, specifically on the load frequency control (LFC) problem. This paper makes an attempt to fill this gap by introducing a DR control loop in the traditional LFC model (called LFC-DR) for a single-area power system. The model has the feature of optimal operation through optimal power sharing between DR and supplementary control. The effect of DR communication delay in the controller design is also considered. It is shown that the addition of the DR control loop increases the stability margin of the system and DR effectively improves the system dynamic performance. Simulation studies are carried out for single-area power systems to verify the effectiveness of the proposed method.
Providing ancillary services for future smart microgrid can be a challenging task because of lack of conventional automatic generation control (AGC) and spinning reserves, and expensive storage devices. In addition, strong motivation to increase the penetration of renewable energy in power systems, particularly at the distribution level, introduces new challenges for frequency and voltage regulation. Thus, increased attention has been focused on demand response (DR), especially in the smart grid environment, where two-way communication and customer participation are part of. This paper presents a comprehensive central DR algorithm for frequency regulation, while minimizing the amount of manipulated load, in a smart microgrid. Simulation studies have been carried out on an IEEE 13-bus standard distribution system operating as a microgrid with and without variable wind generation. Simulation results show that the proposed comprehensive DR control strategy provides frequency (and consequently voltage) regulation as well as minimizing the amount of manipulated responsive loads in the absence/presence of wind power generation.
As the objective of this study, artificial neural network (ANN) and Markov chain (MC) are used to develop a new ANN–MC model for forecasting wind speed in very short-term time scale. For prediction of very short-term wind speed in a few seconds in the future, data patterns for short-term (about an hour) and very short-term (about minutes or seconds) recorded prior to current time are considered. In this study, the short-term patterns in wind speed data are captured by ANN and the long-term patterns are considered utilizing MC approach and four neighborhood indices. The results are validated and the effectiveness of the new ANN–MC model is demonstrated. It is found that the prediction errors can be decreased, while the uncertainty of the predictions and calculation time are reduced.
A new hybrid algorithm using linear prediction and Markov chain is proposed in order to facilitate very short-term wind speed prediction. First, the Markov chain transition probability matrix is calculated. Then, linear prediction method is applied to predict very short-term values. Finally, the results are modified according to the long-term pattern by a nonlinear filter. The results from proposed method are compared by linear prediction method, persistent method and actual values. It is shown that the prediction-modification processes improves very short-term predictions, by reducing the maximum percentage error and mean absolute percentage error, while it retains simplicity and low CPU time and improvement in uncertainty of prediction.
Energy sustainability of hybrid energy systems is essentially a multiobjective, multiconstraint problem, where the energy system requires the capability to make rapid and robust decisions regarding the dispatch of electrical power produced by generation assets. This process of control for energy system components is known as energy management. In this paper, the application of particle swarm optimization (PSO), which is a biologically inspired direct search method, to find real-time optimal energy management solutions for a stand-alone hybrid wind-microturbine (MT) energy system, is presented. Results demonstrate that the proposed PSO-based energy management algorithm can solve an extensive solution space while incorporating many objectives such as: minimizing the cost of generated electricity, maximizing MT operational efficiency, and reducing environmental emissions. Actual wind and end-use load data were used for simulati on studies and the well-established sequential quadratic programming optimization technique was used to validate the results obtained from PSO. Promising simulation results indicate the suitability of PSO for real-time energy management of hybrid energy systems.
Ownership cost calculation plays an important role in optimal operation of distributed energy resources (DERs) and microgrids (MGs) in the future power system, known as smart grid. In this paper, a general framework for ownership cost calculation is proposed using uncertainty and risk analyses. Four ownership cost calculation approaches are introduced and compared based on their associated risk values. Finally, the best method is chosen based on a series of simulation results, performed for a typical diesel generator (DiG). Although simulation results are given for a DiG (as commonly used in MGs), the proposed approaches can be applied to other MG components, such as batteries, with slight modifications, as presented in this paper. The analyses and proposed approaches can be useful in MG optimal design, optimal power flow, and market-based operation of the smart grid for accurate operational cost calculations.
In this paper, a two-layer controller is proposed to aggregate a fleet of behind-the-meter (BTM) energy storage devices based on the Transactive Energy (TE) concept. In the proposed model, aggregator offers an incentive to consumers to purchase power from and/or sell the excess power back to the grid. To do so, controller at the aggregator’s side determines optimal incentive which has to be offered to consumers by maximizing its own profit. Then, local controller at the consumer’s location optimizes battery operation by calculating purchased/sold power from/to the grid based on the local demand, PV generation, retail time-of-use (ToU) prices and demand charge, and the incentive offered by the aggregator to maximize its own profit. Different optimization problems are formulated in the two layers, and the profit of aggregator and consumers in the day-ahead energy market under perfect and imperfect prediction scenarios are compared.
Grid-scale energy storage systems are attracting more attention because of increased public-awareness and declining prices. However, there is still one question which needs to be answered: when utilization of Battery Storage System (BSS) is economical? To address this question, problems of simultaneously sizing BSSs and optimal power sharing –with an objective of decreasing daily operational cost– is investigated to assess the economic viability of BSSs. The assessment is carried out by specializing the problem formulation to mid-sized C&I customers associated with PG&E and by simulating scenarios that differ in the size of load, PV installation, cost of BSS and participation in Demand Response (DR). Simulation results indicate that, using price projections from DOE and Navigant, BSSs can be used to shift loads economically (savings of 10%) around the year 2019. Furthermore, the effective daily savings, when participating in DR programs, is noted to be independent of the load, and that participating in DR does not require a significantly up-sized BSS.
Cyclic and Calendar agings are the two primary sources of degradation in a battery. An accurate battery degradation model can only be achieved when both processes are considered. In this paper, a novel framework is proposed to integrate Cyclic and Calendar aging processes. The proposed framework is able to accommodate different individual Cyclic and Calendar aging models only with slight modifications. It also can work conveniently as a universal degradation framework in different applications, such as large-scale battery storage systems in microgrids (MGs) and electric vehicles (EVs).
Dynamic demand response (DR) is an integral part of ancillary services markets. The integration of dynamic DR control loop into the conventional load frequency control (LFC) model is presented by the authors in  . Extensive analytical analyses were carried out on single-area power system in previous study. In this paper, the idea is expanded to a general multi-area interconnected power system. Then, impacts of the proposed LFC-DR on the dynamic performance of the multi-area power systems in different conditions are simulated. Simulation results show a superior performance of the LFC-DR model for different conditions and power system models.
Tremendous penetration of renewable energy in electric networks, despite its valuable opportunities, such as balancing reserve and ancillary service, has raised concerns for network operators. Such concern stems from grid operating conditions. Such huge penetration can lead to violation in the grid requirements, such as voltage and current limits. This paper proposes an optimization method for determining the number of photovoltaic (PV) panels together with their arrangement in the grid in order to maximize ancillary service, without violating grid operation limits. The proposed optimization method is based on genetic algorithm. To do so, single-objective and multi-objective optimization have been considered. The proposed method is implemented on a model of a part of a Danish distribution grid to verify its effectiveness. The simulation results prove the viability of the method, while keeping the grid requirements within standard operating limits.
Increasing the penetration of variable generation sources, such as wind and solar, potentially threatens the stability of the power systems. Past studies have shown that 15-20% renewable penetration is the largest amount the power systems can handle using conventional control. To overcome the challenge that variable renewable generation presents, this paper proposes that real-time demand response (DR) be used for ancillary services (AS). The impact of varying the amount of DR on the performance of a microgrid-configured distribution feeder is evaluated in this study. Simulation results have shown that a proper amount of DR resources can help to achieve higher penetrations of renewable generation while maintaining the desired system frequency.
This paper presents an evaluation of the impact of various levels of photovoltaic (PV) power penetration in a distribution feeder connected to a simplified grid model (SGM). PV generation is implemented in second-by-second iterations with power output based on actual solar radiation and air temperature data. High penetration levels of intermittent PV generation (15% and 30%) are employed in a feeder-configured microgrid to evaluate grid frequency and voltage characteristics. In this study, only governor droop control is included in the proposed SGM without the secondary control action (known as load frequency control). Two different grid models (fast and slow grid), PV generation configurations (concentrated and distributed), and PV penetration levels (15% and 30%) are considered in the simulation studies. Simulation results indicate the impact of the aforementioned parameters on the system frequency and voltage. Results also reveal that distributed PVs in a wide geographical area with different weather regime have less impact on the frequency and voltage.
The purpose of this paper is to design an optimal CHP islanded microgrid, through technology selection and unit sizing software, to be used for further research on real-time energy management. Two software packages, HOMER and WebOpt, originally developed at the National Renewable Energy Laboratory (NREL) and Lawrence Berkley Laboratory (LBL), respectively, are utilized. Using these programs, different cases are created and compared to justify the selected technologies and their respective prices. The final microgrid design contains renewable and alternative energy generation, hydrogen as an energy carrier, and electric storage.
Real-time demand response (DR) in smart μgrid has been shown to be an effective tool for frequency regulation with increased penetration of renewable energy resources into the grid. Since DR is recognized as an incentive or direct payment to the participants, it is consequently desired to minimize the cost of DR for the utility. This paper presents an optimal DR strategy for minimizing the cost of DR for the utility in smart grid era. The economic model developed by Pennsylvania/New Jersey/Maryland (PJM) utility in the USA is used on an IEEE 13-bus standard system. Simulation results verify the effectiveness of the proposed approach to minimize the cost of DR for the utility. It is also shown that the DR, with or without optimization, decreases the overall cost of frequency regulation for the utility compared to the conventional spinning reserve, without sacrificing system stability.
Because of different technical and economical concerns, battery is happened to be an inevitable part of a microgrid as well as the most expensive component. This fact brings up the necessity of a real-time power management to guarantee the maximum possible battery lifetime based on the final cost of energy. In this way, this study attempts to present a real-time management framework for a grid-tied microgrid based on battery life and cost estimation. In order to verify the effectiveness of the proposed framework, a grid-tied commercial microgrid, which is equipped with wind turbine, PV solar panels and Li-Ion battery package, is optimally sized by HOMER and dynamic models of different components have been developed in MATLAB/Simulink. Then, simulation study has been carried out for a year on the system. All data such as load demand, wind, temperature, solar radiation, and time-based electricity tariff are grasped from different places for a year. Results show that the proposed framework effectively extends the battery lifetime while slightly decreases the cost of energy for customer.
This paper presents a demand response (DR) algorithm for regulating system frequency using responsive customer loads, while minimizing the amount of manipulated loads. The dynamic model for a small islanded microgrid and an improved hill climbing controller are developed in MATLAB/Simulink to show the proof of concept. Simulation results show that the improved DR control strategy provides frequency and voltage regulation while minimizing the amount of manipulated responsive loads. As a result, customer quality-of-service (QoS) is not compromised, while a higher percentage of responsive loads (more non-spinning reserve) would be available for additional control for responding to unexpected disturbances.
This study is an attempt to address the frequency and voltage regulation inside of an islanded microgrid. Central demand response along with an adaptive hill climbing methodology is applied to a small islanded microgrid powered by a diesel generator. All dynamic models are developed in MATLAB/Simulink. Simulation results show that the proposed method has the potential to suppress the frequency variations and stabilize the voltage of the microgrid.
Microgrids are an emerging technology which promises to achieve many simultaneous goals for power system stakeholders, from generator to consumer. The microgrid framework offers a means to capitalize on diverse energy sources in a decentralized way, while reducing the burden on the utility grid by generating power close to the consumer. As a critical component to enabling power system diversity and flexibility, microgrids encompass distributed generators and load centers with the capability of operating islanded from or interconnected to the macrogrid. To make microgrids viable, new and innovative techniques are required for managing microgrid operations given its multi-objective, multi-constraint decision environment. In this article, two example computational intelligence methods, particle swarm optimization (PSO) and ant colony optimization (ACO), for application to the microgrid power management problem are introduced. A mathematical framework for multi-objective optimization is presented, as well as a discussion of the advantages of intelligent methods over traditional computational techniques for optimization. Finally, a three-generator microgrid with an ACO-based power management algorithm is demonstrated and results are shown.
Since 2000, the increase of the installed wind energy capacity all over the world (mainly in Europe and United States) attracted the attention of electricity companies, wind farm promoters and researchers towards the short term prediction, mainly motivated by the necessity of integration into the grid of an increasing 'unknown' (fluctuating) amount of wind power. Besides, in a deregulated system, the ability to trade efficiently, make the best use of transmission line capability and address concerns with system frequency, accurate very short-term forecasts are motivated more than ever. In this study, very short term wind speed forecasting is developed utilizing Artificial Neural Networks (ANN) in conjunction with Markov chain approach. Artificial neural networks predict short term values and the results are modified according to the long term patterns due to applying Markov chain. For verification purposes, the integrated proposed method is compared with ANN. The results show the effectiveness of the integrated method.
The growing revolution in wind energy encourages for more accurate models for wind speed forecasting. In this study, a new integrated approach, which contains ANN, Markov chains and linear regression, is used due to very short-term prediction of wind speed. In this method, First ANN is used for primary prediction of wind speed. Then, second-order Markov chain is applied to calculate transition probability matrix for predicted wind speed in the first step. Finally, a linear regression among ANN primary prediction and calculated probability with Markov chain is used for the final prediction .The results of proposed method in comparison with the ANN results shows lower error of wind speed prediction particularly in the case of higher prediction horizons .The results are based on real wind speed data in an area of Denmark with 2.5 second resolution.
Very short term wind speed forecasting is necessary for wind turbine control system. In this study, a new integrated approach using Wavelet-based networks and PSO is proposed for very short term wind speed forecasting. PSO algorithm is used for training a Wavelet networks and the whole integrated approach is applied for wind speed prediction. As a case study, the wind speed data from a site in Denmark with 2.5 s measured resolution is used for training and test of the network. Proposed approach is compared to multi layer perceptron networks with Back Propagation training algorithm. Results show that the new approach improve Mean Absolute Percentage Error (MAPE) and Maximum error of prediction.
The introduction of distributed generation (DG) onto distribution networks has a significant effect on losses. This effect cannot be characterized as a detrimental or beneficial but is dependent on the allocation of DG on each distribution network. This paper proposes a new method to calculate the optimal size and to identify the corresponding optimal location for DG placement (allocation) for minimizing the total power losses in distribution networks. The proposed (presented) algorithm is an evolutionary algorithm named Particle Swarm Optimization (PSO). The method is implemented and tested on a sample distribution network. The results show the importance of placement of DGs for minimizing losses.
In this paper a new method for wind prediction in wind turbine application is proposed. Considering increase in share of electric power generation from wind across the world and random variations in wind, it is so vital to predict wind speed in different ranges. In this study, a multilayer perceptron artificial neural network trained by Particle Swarm Optimization (PSO) is used for wind speed prediction. Aperiodic and stochastic structure of wind causes those conventional training methods which use gradient tools could not train the network properly. In the other hand, the objective of training a neural network is finding weights and biases so that minimizes the training error. Hence, we can approach problem of training neural networks as an optimization problem. Since wind prediction using ANN is a sophisticated and nonlinear function, use of optimization methods is considered. The proposed method is applied on filtered and real wind data. Filtering ignores improper frequencies in wind frequency spectrum that will be eliminated in wind turbine blades.
In this study, we used PSO algorithm and ANN to predict annual electricity consumption in Iranian agriculture sector. The economic indicators used in this paper are price, value added, number of customers and consumption in the previous periods. To predict the future values, a linearlogarithmic model of electrical energy demand is considered. The PSO algorithm applied in this study has been tuned for all its parameters and the best coefficients with minimum error are identified, while all parameter values are tested concurrently. Consumption in the previous periods has been used for testing estimated model. The estimation errors of PSO algorithm are less than that of estimated by genetic algorithm and regression method. In addition, ANN is used to forecast each independent variable and then electricity consumption is forecasted up to year 2010. Electricity consumption in Iranian agriculture sector from 1981 to 2005 is considered as the case for this study.
The main object of training of artificial neural networks is founding weights and biases so that minimize training error. Hence we can approach problem of ANN training as an optimization problem. Conventional methods for ANN training uses back propagation algorithm and other gradient algorithms. When the target function is severely nonlinear and sophisticated, the conventional methods has a lot of weak points. Using PSO algorithm in ANN training and comparing with back propagation algorithm shows that in sophisticated problems, the new algorithm has higher performance. At the end, new algorithm is applied on two problems and results of comparison with back propagation algorithm is presented.
This paper describes dynamic modeling and simulation results of a Hybrid Wind-PEM fuel cell System. The system consists of a proton exchange membrane fuel cell (PEMFC),a Wind turbine, ultracapacitors, an electrolyzer, and a power converter. The output fluctuation of the load voltage due to load variation is reduced using a fuel cell stack. At mid-night, when the amount of demand is low, excess energy form Wind Turbine is converted to hydrogen using an electrolyzer for later use in the fuel cell. Ultracapacitors and a power converter unit are proposed to minimize voltage fluctuations in the system and generate AC voltage. Dynamic modeling of various components of this small isolated system is presented. Dynamic aspects of temperature variation and double layer capacitance of the fuel cell are also included. PID type controllers are used to control the fuel cell system. MATLAB/SIMULINK is used for the simulation of this highly nonlinear energy system. System dynamics are studied to determine the voltage variation throughout the system. Transient responses of the system to step changes in the load current are presented. Analysis of simulation results and limitations of the fuel cell energy system are discussed. The voltage variation at the output was found to be within the acceptable range. The proposed system does not need conventional battery storage. It may be used for off-grid power generation in remote communities.