Description:[001] The present invention relates to the field of energy management systems (EMS) specifically designed for hybrid direct current (DC) and alternating current (AC) microgrids. This invention addresses the need for efficient energy management in systems that integrate multiple distributed energy resources (DERs) such as photovoltaic (PV) solar panels, wind turbines, and energy storage systems (ESS). These microgrids often operate in both AC and DC modes, which require seamless coordination between the different power networks to ensure optimal energy distribution, stability, and reliability.
[002] Hybrid DC-AC microgrids combine the advantages of DC systems, which are often used with renewable energy sources like solar and wind, and AC systems, which are the dominant form of electricity distribution in traditional grids. The invention is particularly applicable to microgrids that incorporate renewable energy generation, where power fluctuations due to variable energy sources, such as solar irradiance and wind speed, can cause significant operational challenges. Therefore, the efficient management of power generation, storage, and distribution is crucial for maintaining grid stability and meeting energy demands.
[003] This invention utilizes artificial neural networks (ANN) to control the energy flow in hybrid DC-AC microgrids. By integrating ANN-based algorithms, the system is capable of analyzing multiple inputs in real-time, including the state of charge (SOC) of the ESS, load demand, and available power from distributed generation units. The ANN enables the system to make intelligent decisions regarding power dispatch and energy storage utilization, thus enhancing the overall efficiency of the microgrid. This field of invention is particularly relevant to smart grids, renewable energy systems, and advanced power distribution technologies aimed at improving energy efficiency and sustainability.
BACKGROUND OF THE INVENTION
[004] The increasing global reliance on renewable energy sources, such as solar and wind power, has brought significant changes to traditional power distribution networks. Renewable energy sources are inherently variable, leading to challenges in ensuring a consistent and stable energy supply. To address these challenges, energy management systems (EMS) are essential, particularly in hybrid DC-AC microgrid configurations, where both alternating current (AC) and direct current (DC) energy sources are integrated.
[005] Hybrid microgrids offer significant advantages, such as reducing energy losses by minimizing power conversions and enhancing energy independence by incorporating distributed generation (DG) from renewable sources. However, managing these hybrid networks efficiently is a complex task due to fluctuating generation and demand, as well as the need to maintain power quality.
[006] Conventional energy management systems in hybrid microgrids are often insufficient in addressing the dynamic nature of renewable energy generation. These systems typically rely on predefined operation strategies that may not adapt effectively to real-time changes in load demand, generation, and the state of charge (SOC) of energy storage systems (ESS).
[007] The limitations of traditional EMS approaches, which tend to ignore the accumulated power from the grid and do not consider the ESS's SOC efficiently, lead to suboptimal performance and increased power losses. Furthermore, in many cases, complex algorithms and control strategies add to the operational difficulty of such systems, making them unsuitable for practical deployment in smaller-scale microgrids where variability is higher.
[008] Recent advances in artificial intelligence, particularly in neural networks, have opened new possibilities for enhancing the functionality of energy management systems. Artificial neural networks (ANNs) excel in recognizing patterns and adapting to changing inputs, making them ideal for managing the unpredictable nature of power generation and consumption in hybrid DC-AC microgrids.
[009] By incorporating an ANN into an energy management system, it becomes possible to optimize the operation of power converters based on real-time data, including load demand, distributed generation output, and the SOC of the ESS. This allows for more efficient power distribution, reduced losses, and improved overall performance of the microgrid.
[010] The primary focus of this invention is to address the inefficiencies of conventional EMS in hybrid microgrids by introducing an ANN-based energy management system. This system is designed to improve the operational efficiency of microgrids by dynamically adjusting power references based on real-time conditions.
[011] The neural network can make real-time decisions about charging or discharging the ESS and control the power flow between DC and AC systems to maintain grid stability. By simplifying the decision-making process and enabling a more adaptive and efficient energy management strategy, the proposed system overcomes the challenges posed by the variability of renewable energy sources and ensures optimal performance in hybrid microgrids.
SUMMARY OF THE INVENTION
[012] The present invention introduces an advanced energy management system (EMS) for hybrid DC-AC microgrids, leveraging the capabilities of artificial neural networks (ANNs) to optimize power distribution and maintain system stability. Hybrid microgrids, which combine both AC and DC networks, present unique challenges in managing energy flows, particularly in the presence of variable renewable energy sources like solar and wind. The proposed EMS addresses these challenges by continuously analyzing key parameters, such as the load demand, state of charge (SOC) of energy storage systems (ESS), distributed generation (DG) output, and grid power. By doing so, it ensures efficient energy use and optimal operation of power converters across the microgrid.
[013] At the heart of the system is a neural network that has been trained to predict and adjust the power flow within the microgrid in real-time. The ANN processes input data from various sources, including the distributed generators, ESS, and grid, to determine the most efficient operating mode. It evaluates conditions such as when to charge or discharge the ESS and when to export excess power to the grid. This intelligent decision-making minimizes energy losses, balances the power supply, and maintains voltage stability, ensuring reliable operation of the microgrid even under fluctuating power generation conditions.
[014] The invention supports three primary operational modes. In Mode 1 (Charging Mode), the ESS is charged when the DG produces more power than is required by the load and the SOC is below 50%. In Mode 2 (Power Export Mode), when the SOC exceeds 50% and generation exceeds demand, the excess energy is exported to the grid. Lastly, in Mode 3 (Discharging Mode), when the DG output is insufficient to meet load demand and the SOC is above a certain threshold, the ESS discharges to balance the power flow.
[015] The system offers significant improvements over traditional EMS approaches by integrating an ANN that reduces the complexity of conventional algorithms and enhances overall efficiency. By considering real-time data and dynamically adjusting operating modes, the ANN-based EMS allows for smoother transitions between charging, discharging, and power export modes, while ensuring that power quality and grid stability are maintained. This system is particularly beneficial for microgrids with high levels of renewable energy penetration, where rapid changes in generation output are common, and conventional energy management systems may struggle to respond effectively.
[016] In summary, the invention provides a sophisticated and intelligent approach to energy management in hybrid DC-AC microgrids, offering enhanced efficiency, reliability, and flexibility by utilizing artificial neural networks to optimize power flow across the entire system.
BRIEF DESCRIPTION OF THE DRAWINGS
[017] The accompanying figures included herein, and which form parts of the present invention, illustrate embodiments of the present invention, and work together with the present invention to illustrate the principles of the invention Figures:
[018] Figure 1, illustrates a block diagram, in accordance with an embodiment of the present invention. & Figure 2, illustrates a concept of the functional flow diagram, in accordance with an embodiment of the present invention.
[019] Figure 3, illustrates structure of the neural network used in the EMS, in accordance with an embodiment of the present invention. & Figure 4, illustrates ANN training data for various operating modes, in accordance with an embodiment of the present invention.
[020] Figure 5 shows the performance plot for training, testing and validation of the neural network, Figure 6. The regression value for training, testing and validation is around 0.98 to 0.99. & Figure 7. It has three neurons in the input layer, ten neurons in the hidden layer and one neuron in the output layer.
[021] Figure 8 shows the Simulink model of neural network energy management for Hybrid AC-DC microgrid system. & Figure 9 -14, shows the test results.
DETAILED DESCRIPTION OF THE INVENTION
[022] The present invention relates to a hybrid DC-AC microgrid energy management system (EMS) designed to optimize power flow using an artificial neural network (ANN). This invention addresses challenges in managing energy resources in hybrid microgrids, where distributed generation (DG) from renewable sources such as wind and solar energy coexist with an energy storage system (ESS) and a conventional AC grid.
[023] The invention utilizes an ANN-based control system to balance energy generation, storage, and consumption in real-time, enabling the system to adapt to fluctuating load demands and variable power generation. The goal is to ensure stable operation and efficient use of available energy resources, while minimizing losses and maintaining a reliable power supply.
Hybrid DC-AC Microgrid Structure
[024] The hybrid microgrid consists of both AC and DC distribution networks that are interconnected via an interlinking converter (ILC). This converter allows bidirectional power flow between the two networks, ensuring that energy can be transferred as needed. The microgrid is composed of several key components:
• Distributed Generation (DG) units: These include solar photovoltaic (PV) panels, wind turbines, and other renewable energy sources. The output from these DG units is predominantly DC.
• Energy Storage System (ESS): The ESS stores excess energy generated by the DG units and discharges it when required. The ESS can operate in both charging and discharging modes.
• AC Grid Connection: The microgrid is connected to the main AC grid through the interlinking converter, allowing the system to import or export power based on current operational needs.
[025] Figure 1 shows the basic architecture of the hybrid DC-AC microgrid with these components, where the AC and DC networks are connected by the interlinking converter. The system can function in various modes depending on the state of charge (SOC) of the ESS and the power generated by the DG units.
Energy Management System (EMS)
[026] The central component of the invention is the Energy Management System (EMS), which governs the operation of the entire microgrid. The EMS uses real-time data from sensors to determine the status of the distributed generation, load demand, and the SOC of the ESS. It also communicates with local controllers that manage individual power converters for DGs and ESS.
[027] The EMS operates in three main modes:
1. Mode 1 - Charging Mode: This mode is activated when the SOC of the ESS is below 50%, and excess power is available from the DG units. In this mode, the EMS directs the interlinking converter to store surplus energy in the ESS.
2. Mode 2 - Power Export Mode: When the SOC is sufficient (greater than 50%) and excess power is still being generated, the EMS allows the surplus power to be exported to the AC grid. The interlinking converter manages this power transfer.
3. Mode 3 - Discharge Mode: In the event that the power generated by the DG units is insufficient to meet load demand, and the SOC is greater than 50%, the EMS directs the ESS to discharge stored energy to support the load.
Artificial Neural Network (ANN) Implementation
[028] The Artificial Neural Network (ANN) is at the heart of the EMS. It is designed to process real-time data inputs and make decisions regarding power flow and system operation. The ANN continuously monitors the following inputs:
• State of Charge (SOC) of the ESS: This indicates the remaining energy stored in the ESS, a critical parameter for managing charging and discharging.
• Power generated by DG units (PDG): This is the total power generated by renewable sources such as solar PV and wind.
• Grid Power (Pgrid): This is the power exchanged with the AC grid, either imported or exported.
[029] The output of the ANN is the direct-axis reference current (Idref), which is used to control the operation of the interlinking converter, dictating whether to charge or discharge the ESS and how much power to import or export from the grid.
ANN Training and Operation
[030] The ANN is trained using historical data from the microgrid, including load demand, generation profiles, and the SOC of the ESS under different operating conditions. The training process uses backpropagation to adjust the weights of the neural network, minimizing the error between the predicted and actual power reference values for the ESS. The trained ANN can predict optimal operation strategies based on real-time inputs.
[031] The ANN consists of the following layers (Figure 3):
• Input Layer: The input layer receives data on SOC, DG power, and grid power.
• Hidden Layers: There are multiple hidden layers where the network processes the input data. Each layer has neurons with adjustable weights, which the ANN modifies during training.
• Output Layer: The output layer generates the direct-axis reference current (Idref), which is used to control the power converters in the microgrid.
[032] For each operating mode, the ANN has been trained with specific datasets:
• Mode 1 (Charging Mode): The ANN is trained using data that correlates SOC, DG power, and grid power to determine the ideal charging rate of the ESS.
• Mode 2 (Power Export Mode): Data for this mode helps the ANN predict how much power should be exported to the grid based on DG output and ESS SOC.
• Mode 3 (Discharge Mode): The ANN calculates the optimal discharge rate when the load demand exceeds the DG power, ensuring that the ESS is not over-discharged.
[033] Training sessions typically involve hundreds of iterations to ensure that the ANN's predictions are accurate, with the error rate kept below 10%. For example, the ANN is trained using historical SOC data and corresponding charging or discharging decisions under various load scenarios (Tables 1-3).
[034] The Levenberg–Marquardt algorithm is used to optimize the training process, and the performance of the trained ANN is evaluated using root mean square error (RMSE). As shown in Figure 5, the RMSE for the trained network is around 0.011, indicating a high level of accuracy in the system's predictions. Regression plots for training, testing, and validation demonstrate a high correlation between predicted and actual values, with an R-value of approximately 0.98 to 0.99 (Figure 6).
Control System and Converter Operation
[035] The EMS communicates with local controllers responsible for regulating the operation of the interlinking converter, power converters, and the ESS. Each power converter is controlled to either:
• Regulate voltage levels in the DC distribution network.
• Perform maximum power point tracking (MPPT) for DG units such as solar PV.
• Charge or discharge the ESS based on the ANN output.
[036] The interlinking converter is responsible for maintaining the voltage levels in both AC and DC distribution networks. When operating in grid-connected mode, it ensures that the AC grid’s power can be used to regulate the DC network's voltage, and vice versa. The ANN determines whether to charge or discharge the ESS based on grid conditions and load demand.
[037] The EMS simplifies the overall operation of the microgrid by centralizing decision-making while allowing local controllers to handle specific tasks. The centralized EMS sends high-level commands to local controllers, which adjust their operational modes accordingly. This hierarchical control ensures that the microgrid operates efficiently even during periods of high variability in power generation or load demand.
Simulation and Validation
[038] A small-scale hybrid DC-AC microgrid was simulated using the MATLAB Simulink environment to validate the proposed EMS. The microgrid model included a 40 kW ESS, distributed generators (solar PV and wind), and an interlinking converter connecting the DC and AC distribution networks.
[039] Two case studies were conducted to assess the performance of the EMS:
• Case 1: Initial SOC was set to 0.3, and solar irradiance was varied every second. The ESS remained in charging mode throughout the simulation, and power balance between the DG units and load demand was maintained.
• Case 2: Initial SOC was set to 0.7, and solar irradiance was similarly varied. The ESS discharged energy to meet load demand, demonstrating the EMS's ability to switch between charging and discharging modes based on real-time data.
[040] In both cases, the ANN-based EMS successfully managed power flow, maintaining voltage levels and ensuring efficient use of energy resources. The results, shown in Figures 9-14, illustrate the system's ability to adapt to changing conditions while keeping the power balance between sources and loads.
[041] The proposed EMS, utilizing an artificial neural network, offers a highly efficient and adaptable solution for managing hybrid DC-AC microgrids. By considering real-time data such as SOC, DG output, and grid conditions, the ANN ensures optimal operation across various modes. The invention simplifies the energy management process, improves grid stability, and reduces energy losses, making it ideal for microgrids incorporating renewable energy sources.
[042] One of the most notable aspects of the invention is its ability to adapt dynamically to varying conditions, such as fluctuating power generation from renewable sources and changing load demands. The ANN is trained to process critical inputs, including the state of charge (SOC) of the ESS, distributed generation output, and grid power. By using this information, the system predicts the optimal charging or discharging actions for the ESS and determines the best mode of operation—whether to store excess energy, export power to the grid, or discharge stored energy to meet load requirements. This adaptability ensures that the microgrid can respond effectively to both short-term changes in power conditions and long-term operational demands.
[043] The use of ANN also simplifies the operation of the EMS compared to traditional methods. Conventional energy management systems often rely on complex algorithms that are difficult to implement and maintain. In contrast, the ANN-based EMS offers a more straightforward approach, learning from historical data and continuously improving its performance. As demonstrated through simulation studies, the system is capable of maintaining a high level of accuracy with a low error rate, ensuring efficient energy distribution while minimizing losses.
[044] In conclusion, the development of an ANN-based EMS for hybrid DC-AC microgrids marks a critical step forward in the integration of renewable energy technologies. By intelligently managing the interplay between distributed generation, energy storage, and grid power, the system ensures reliable and efficient microgrid operation. The flexibility and robustness of the system make it an ideal solution for modern energy infrastructures, where renewable energy sources and decentralized power systems are becoming increasingly prominent. The proposed system holds great promise for improving energy sustainability, reducing reliance on fossil fuels, and enhancing the overall efficiency of electrical grids.
, Claims:Claim 1: A hybrid DC-AC microgrid energy management system, comprising:
• A plurality of distributed energy generators including solar, wind, and other renewable energy sources;
• An energy storage system (ESS);
• An interlinking converter configured to manage power flow between the AC and DC distribution networks;
• A neural network-based control unit designed to optimize power distribution by controlling the charging and discharging operations of the energy storage system, considering real-time load demand, distributed energy generation, and state of charge (SOC) of the energy storage system.
Claim 2: The system of claim 1, wherein the artificial neural network is trained to operate in various modes, including:
• Mode 1: Charging the energy storage system when excess power is available from the distributed generators and the state of charge is below 50%;
• Mode 2: Exporting power to the AC grid when the state of charge is at or above 50% and distributed generation exceeds load demand;
• Mode 3: Discharging the energy storage system to meet the load demand when the power generated by the distributed generators is insufficient, and the state of charge is at or above 50%.
Claim 3: The system of claim 2, wherein the neural network is configured to adjust the power flow between the AC grid and the DC microgrid by regulating the operation of the interlinking converter based on real-time data inputs from the distributed generators and the energy storage system.
Claim 4: The system of claim 1, wherein the artificial neural network is trained using a backpropagation algorithm with input data including:
• Distributed generation power output;
• Load demand data;
• State of charge of the energy storage system;
• Power exchanged with the AC grid.
Claim 5: The system of claim 4, wherein the artificial neural network further comprises:
• An input layer to receive real-time data on load demand, distributed generation power, and state of charge;
• One or more hidden layers for processing and learning from input data;
• An output layer for determining control actions related to charging, discharging, or power export operations.
Claim 6: The system of claim 1, wherein the interlinking converter performs bi-directional power flow regulation between the AC and DC distribution networks to ensure efficient power distribution and stable voltage in both networks.
Claim 7: The system of claim 1, wherein the energy management system reduces power losses by optimizing the charge/discharge cycles of the energy storage system based on load demand and distributed generation power fluctuations.
Claim 8: The system of claim 1, wherein the neural network control unit continuously monitors and updates operational parameters to ensure minimal energy loss and to maintain a stable grid by balancing generation, consumption, and storage operations in the hybrid microgrid.
Claim 9: The system of claim 2, wherein the energy management system uses predictive modeling for distributed generation and load forecasting to enhance operational efficiency by anticipating power surpluses or deficits in the microgrid.
Claim 10: The system of claim 1, wherein the energy storage system’s charge and discharge operations are optimized based on a performance threshold defined by a minimum state of charge to protect the longevity of the energy storage system.