Method and device for dynamically adjusting pricing in microgrid according to supply and demand

Abstract

The present disclosure provides a method and device for dynamically adjusting pricing in a microgrid according to supply and demand. According to one embodiment of the present disclosure, a device for optimizing energy transactions within a microgrid comprises: at least one memory; and at least one processor, wherein the processor may collect energy-related data within the microgrid in real time, preprocess the collected real-time energy-related data, predict energy supply and demand on the basis of the preprocessed energy-related data using an energy supply and demand prediction model, determine a dynamic price on the basis of the predicted energy supply and demand using a reinforcement learning-based pricing algorithm, and match buyers and sellers on the basis of the determined dynamic price using graph matching networks (GMN).

Description

Method and device for dynamic price adjustment of microgrids according to supply and demand

[0001] The present disclosure relates to a method and device for dynamically adjusting prices in a microgrid based on energy supply and demand. More specifically, it relates to a method and device for optimizing energy transactions within a microgrid through dynamic price adjustment based on energy supply and demand forecast results and matching of sellers and buyers.

[0002] Microgrids are small-scale power grids that operate independently from the central power grid and enable the effective use of distributed energy sources, such as renewable energy. While microgrids offer efficient energy management and supply stability, they also increase the complexity and volatility of energy transactions. Failure to accurately forecast and price based on real-time data can lead to imbalances in energy supply and demand. However, the numerous variables and complexities inherent in energy transactions make pricing difficult. Therefore, optimizing energy transactions within microgrids requires a system capable of effectively collecting and processing real-time data.

[0003] The background technology described above is technical information that the inventor possessed for the purpose of deriving the present invention or acquired in the process of deriving the present invention, and cannot necessarily be considered as publicly known technology disclosed to the general public prior to the application for the present invention.

[0004] The present disclosure provides a method and device for dynamically adjusting prices in a microgrid based on supply and demand. The problems addressed by the present disclosure are not limited to those mentioned above. Other problems and advantages of the present disclosure not mentioned above can be understood through the following description and will be more clearly understood through the embodiments of the present disclosure. Furthermore, it will be appreciated that the problems and advantages addressed by the present disclosure can be realized by the means and combinations thereof set forth in the claims.

[0005] A device for optimizing energy transactions within a microgrid according to one aspect comprises: a memory having at least one program stored therein; and at least one processor operating by executing the at least one program, wherein the processor collects energy-related data within the microgrid in real time, preprocesses the energy-related data collected in real time, predicts energy supply and demand according to the preprocessed energy-related data using an energy supply and demand prediction model, sets a dynamic price according to the predicted energy supply and demand using a pricing algorithm based on reinforcement learning, and matches a buyer and a seller based on the set dynamic price using a GMN (Graph Matching Networks).

[0006] A method for optimizing energy transactions within a microgrid according to another aspect includes the steps of: collecting energy-related data within the microgrid in real time; preprocessing the energy-related data collected in real time; predicting energy supply and demand according to the preprocessed energy-related data using an energy supply and demand prediction model; setting a dynamic price according to the predicted energy supply and demand using a reinforcement learning-based pricing algorithm; and matching buyers and sellers based on the set dynamic price using a GMN (Graph Matching Networks).

[0007] Another aspect of a computer-readable recording medium includes a recording medium having recorded thereon a program for executing the above-described method on a computer.

[0008] According to the problem solving means of the present disclosure described above, energy supply and demand prediction is performed by collecting energy generation and consumption data in real time, and dynamic pricing is set using a reinforcement learning algorithm, thereby maximizing the efficiency of energy transactions.

[0009] Additionally, real-time data from sensors monitoring energy generation, consumption patterns, and market prices can be integrated with historical data, demand trends, and weather forecasts to improve the performance of predictive models.

[0010] The effects of the embodiments are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the present invention.

[0011] FIG. 1 is a schematic diagram of an energy transaction optimization system within a microgrid according to one embodiment.

[0012] FIG. 2 is a flowchart illustrating an example of a method for optimizing energy trading within a microgrid according to one embodiment.

[0013] FIG. 3 is an exemplary drawing illustrating an example of an energy supply and demand prediction model according to one embodiment.

[0014] FIG. 4 is an exemplary drawing illustrating an example of a transformer encoder according to one embodiment.

[0015] FIG. 5 is an exemplary diagram illustrating an example of a learning method of a pricing algorithm according to one embodiment.

[0016] FIG. 6 is an exemplary drawing for explaining a GMN according to one embodiment.

[0017] FIG. 7 is a block diagram of an energy transaction optimization device within a microgrid according to one embodiment.

[0018] A device for optimizing energy transactions within a microgrid according to one aspect comprises: a memory having at least one program stored therein; and at least one processor operating by executing the at least one program, wherein the processor collects energy-related data within the microgrid in real time, preprocesses the energy-related data collected in real time, predicts energy supply and demand according to the preprocessed energy-related data using an energy supply and demand prediction model, sets a dynamic price according to the predicted energy supply and demand using a pricing algorithm based on reinforcement learning, and matches a buyer and a seller based on the set dynamic price using a GMN (Graph Matching Networks).

[0019] A method for optimizing energy transactions within a microgrid according to another aspect includes the steps of: collecting energy-related data within the microgrid in real time; preprocessing the energy-related data collected in real time; predicting energy supply and demand according to the preprocessed energy-related data using an energy supply and demand prediction model; setting a dynamic price according to the predicted energy supply and demand using a reinforcement learning-based pricing algorithm; and matching buyers and sellers based on the set dynamic price using a GMN (Graph Matching Networks).

[0020] Another aspect of a computer-readable recording medium includes a recording medium having recorded thereon a program for executing the above-described method on a computer.

[0021] The reward function of a pricing algorithm can be designed to balance multiple objectives, such as maximizing the profitability of energy transactions and ensuring a balance between supply and demand in a microgrid. Rewards provide feedback that measures the success or failure of an agent's actions. Agents can select actions within their action space to maximize rewards for the environment.

[0022] The pricing algorithm uses a multi-objective reward function to determine optimal prices, achieving multiple objectives including profit maximization and grid stability. An example of the pricing algorithm's specific learning method is described below in Figure 5.

[0023]

[0024] FIG. 5 is an exemplary diagram illustrating an example of a learning method of a pricing algorithm according to one embodiment.

[0025] A reinforcement learning-based pricing algorithm can be a Proximal Policy Optimization (PPO) model, which consists of an agent and an environment. The PPO (Proximal Policy Optimization) algorithm selects actions through a policy neural network, and the agent improves the policy as it interacts with the environment. To ensure stable learning, the PPO algorithm uses a clipping technique to prevent excessive changes during policy improvement. This allows the agent to gradually find the optimal policy while avoiding abrupt changes during learning.

[0026] The agent receives information about its current state, selects the optimal action, and applies it to the environment. The environment outputs a reward based on the agent's action and transitions to the next state. The PPO model is trained to help the agent select the optimal action from the current state, using a multi-reward function that simultaneously considers profit maximization and grid stability.

[0027] Referring to Figure 5, the learning process of the reinforcement learning-based pricing algorithm is as follows. First, the agent obtains the state information about the environment at the current time t from the environment. and compensation Received, Action based on decides. And the agent decides Sends it to the environment.

[0028] The environment receives actions from the agent. It causes a change in state through, and the resulting state is and passes it to the agent. The agent receives the new state Wow, compensation Learning continues based on this. This learning process is repeated at regular intervals, and through this, the agent learns how to set the optimal price.

[0029] In the learning process shown in Figure 5, the PPO model simulates various market scenarios using historical data combined with supply and demand forecasting models. For example, it generates various supply and demand patterns based on historical energy supply and demand data and meteorological data, and uses the simulated data to test various pricing scenarios to learn optimal policies.

[0030] The learning process is enhanced by a deep neural network that approximates a value function, enabling the system to effectively handle complex, high-dimensional state spaces. The value function is a function that predicts the expected sum of future rewards at a given state.

[0031] In one embodiment, the dynamic pricing algorithm operates continuously, receiving real-time data from the microgrid's network of sensors and IoT devices. This data feed can include minute-by-minute updates on energy production, consumption patterns, and storage levels. Based on this information, the PPO model can quickly recalculate the optimal price for each unit of energy, dynamically adjusting prices to meet immediate market demands and long-term strategic goals. This dynamic pricing allows energy prices to be adjusted in real time based on supply and demand, accurately reflecting current market conditions.

[0032]

[0033] Referring back to FIG. 2, the device (100) can match buyers and sellers based on the dynamic price determined in step 250. In this step, the device (100) optimally matches energy buyers and sellers using a Graph Matching Network (GMN).

[0034]

[0035] FIG. 6 is an exemplary drawing for explaining a GMN according to one embodiment.

[0036] Referring to Figure 6, GMN connects multiple buyer nodes (611, 612, 613) and multiple seller nodes (621, 622, 623). GMN is a network matching algorithm based on graph theory and is used for efficient matching in energy transactions. GMN represents each node as a vertex of the graph, and the potential for energy transactions between nodes is represented by edges in the graph. Each node reflects the characteristics of a buyer or seller, and the weight of the edges indicates the priority of the energy transaction. By representing a microgrid as a network graph, GMN can analyze complex network interactions beyond a simple matching algorithm.

[0037] Each buyer node can include an individual buyer's energy demand, maximum allowable price, geographic location, and time frame for energy demand. The time frame for energy demand refers to a specific time interval during which the buyer requires energy.

[0038] Each seller node can include its own energy supply capacity, minimum acceptable price, geographic location, and timeframe for energy availability. Here, the timeframe for energy availability refers to the specific time interval during which the seller can supply energy. For example, a solar power plant can produce and supply energy from sunrise to sunset, while an energy storage system (ESS) can supply a constant amount of energy throughout the day.

[0039] GMN matches buyers and sellers based on the connections between buyer and seller nodes. GMN maximizes the efficiency of energy transactions by prioritizing buyers and sellers with strong connections—those with well-matched energy supply and demand requirements. For example, in Figure 6, buyer 1 (611) is strongly connected to seller 2 (622), indicating a high likelihood of efficient energy transactions between the two nodes.

[0040] GMN can perform matching by considering dynamic prices determined by a pricing algorithm. These dynamic prices serve as a basis for coordinating transactions between buyers and sellers. Furthermore, GMN-based matching algorithms are particularly effective in situations where, despite identical purchase and sale prices, certain buyers may be given higher priority due to factors such as low supply or proximity between buyers and sellers.

[0041]

[0042] FIG. 7 is a block diagram of an energy transaction optimization device within a microgrid according to one embodiment. Meanwhile, the energy transaction optimization device (700) illustrated in FIG. 7 may correspond to the energy transaction optimization device (100) illustrated in FIG. 1.

[0043] Referring to FIG. 7, the energy transaction optimization device (700) may include a communication module (710), a processor (730), and a memory (720). Only components related to the embodiment are illustrated in the energy transaction optimization device (700) of FIG. 7. Therefore, those skilled in the art will understand that other general components may be included in addition to the components illustrated in FIG. 7.

[0044] The communication module (710) may include one or more components that enable wired / wireless communication with distributed energy resources, a buyer server, a seller server, and / or other external devices. For example, the communication module (710) may include at least one of a short-range communication unit (not shown), a mobile communication unit (not shown), and a broadcast reception unit (not shown).

[0045] The memory (720) is hardware that stores various data processed within the energy transaction optimization device (700), and can store a program for processing and controlling the processor (730).

[0046] The memory (720) may include random access memory (RAM) such as dynamic random access memory (DRAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), CD-ROM, Blu-ray or other optical disk storage, hard disk drive (HDD), solid state drive (SSD), or flash memory.

[0047] The processor (730) controls the overall operation of the energy transaction optimization device (700). For example, the processor (730) can control the input unit (not shown), the display (not shown), the communication module (710), the memory (720), etc., by executing programs stored in the memory (720). The processor (730) can control the operation of the energy transaction optimization device (700) by executing programs stored in the memory (720).

[0048] The processor (730) can control at least some of the operations of the energy transaction optimization device (700) described above in FIGS. 1 to 6.

[0049] For example, the processor (730) can collect energy-related data within a microgrid in real time, preprocess the energy-related data collected in real time, and predict energy supply and demand based on the preprocessed energy-related data using an energy supply and demand prediction model. In addition, the processor (730) can use a reinforcement learning-based pricing algorithm to set a dynamic price based on the predicted energy supply and demand, and use a GMN (Graph Matching Network) to match buyers and sellers based on the set dynamic price.

[0050] Meanwhile, a specific example of how the processor (730) operates is the same as described above with reference to FIGS. 1 to 6. Therefore, a specific description of the operation of the processor (730) is omitted below.

[0051] The processor (730) may be implemented using at least one of application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, micro-controllers, microprocessors, and other electrical units for performing functions.

[0052]

[0053] Various embodiments of the present disclosure may be implemented as software (e.g., a program) including one or more instructions stored on a machine-readable storage medium. For example, a processor of the machine may call at least one instruction among the one or more instructions stored on the storage medium and execute it. This enables the machine to operate to perform at least one function according to the at least one instruction called. The one or more instructions may include code generated by a compiler or code executable by an interpreter. The machine-readable storage medium may be provided in the form of a non-transitory storage medium. Here, "non-transitory" only means that the storage medium is a tangible device and does not contain a signal (e.g., electromagnetic waves), and this term does not distinguish between cases where data is stored semi-permanently and cases where it is stored temporarily in the storage medium.

[0054] According to one embodiment, the method according to various embodiments of the present disclosure may be provided as included in a computer program product. The computer program product may be traded as a product between a seller and a buyer. The computer program product may be distributed in the form of a machine-readable storage medium (e.g., compact disc read-only memory (CD-ROM)), or may be distributed online (e.g., downloaded or uploaded) through an application store (e.g., Play Store™) or directly between two user devices. In the case of online distribution, at least a portion of the computer program product may be temporarily stored or temporarily generated in a machine-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0055] Meanwhile, if the steps constituting the method according to the present disclosure are not explicitly described in an order or contrary to the order, the steps may be performed in any appropriate order. The present disclosure is not necessarily limited by the order in which the steps are described. The use of all examples or exemplary terms (e.g., “etc.”) in this disclosure is merely intended to describe the present disclosure in detail, and the scope of the present disclosure is not limited by the examples or exemplary terms unless otherwise defined by the claims. Furthermore, those skilled in the art will appreciate that various modifications, combinations, and variations can be configured according to design conditions and factors within the scope of the appended claims or their equivalents.

[0056] Therefore, the spirit of the present disclosure should not be limited to the embodiments described above, and all scopes equivalent to or equivalent to the scope of the following claims as well as the scope of the present disclosure are considered to fall within the scope of the spirit of the present disclosure.

Claims

1. As an energy transaction optimization device within a microgrid, memory in which at least one program is stored; and comprising at least one processor that operates by executing at least one program; The above processor, Collect energy-related data within the microgrid in real time, Preprocess the energy-related data collected in real time, Using the energy supply and demand prediction model, energy supply and demand are predicted based on the preprocessed energy-related data, Using a pricing algorithm based on reinforcement learning, dynamic pricing is set based on the predicted energy supply and demand. A device that matches buyers and sellers based on the dynamic price determined above using GMN (Graph Matching Network).

2. In paragraph 1, The above energy-related data is: A device comprising at least one of energy generation data from multiple energy sources, consumption pattern data, market price data, geographic data, and meteorological data.

3. In paragraph 1, The above energy supply and demand prediction model is, At least one Convolutional Neural Network (CNN) layer; At least one transformer encoder; At least one Long Short-Term Memory (LSTM) layer; and A device comprising at least one dense layer.

4. In paragraph 3, The above CNN layer is, Analyze the spatial patterns of preprocessed meteorological data, A device for extracting weather features from the above spatial pattern, wherein the weather features include cloud cover and wind speed.

5. In paragraph 3, The above transformer encoder, Input multiple sequence data, By using the multi-head attention mechanism, the plurality of sequence data are processed in parallel, A device in which the plurality of sequence data includes data converted into sequence form from the output of the CNN layer.

6. In paragraph 3, The above transformer encoder, It includes a multi-head attention layer and a feedforward network, The above multi-head attention layer pays attention to important events, including recent anomalies and rapid weather changes. The above feedforward network is a device that processes nonlinearity of variables included in meteorological data through a nonlinear activation function.

7. In paragraph 3, The above LSTM layer, A device for capturing temporal dependencies of the preprocessed data, wherein the temporal dependencies include past energy production rates and seasonal variations in energy production, to predict energy supply trends over time.

8. In paragraph 3, The above dense layer is, A device that synthesizes the outputs of the CNN layer, the transformer encoder, and the LSTM layer to output a final energy supply and demand prediction.

9. In paragraph 1, The above pricing algorithm is: It is configured based on the environment of the above microgrid, Based on the state information of the above microgrid, the optimal action is output in a continuous action space, The status information of the above microgrid is: Includes real-time energy supply and demand, the predicted energy supply and demand, and weather conditions. The above action space is, A device that is a set of unit energy prices that can be traded within the above microgrid.

10. In paragraph 9, The above pricing algorithm is: A device that sets an optimal price using a multi-objective reward function to achieve multiple objectives including profit maximization and grid stability.

11. In paragraph 9, The above pricing algorithm is: The agent decides on the first action based on the above state information, Applying the above first action to the environment, receiving a first reward and a changed second state from the environment, Decide on a second action based on the first reward and the second state, A device that learns behavior that maximizes the above reward and calculates an optimal price.

12. In paragraph 1, The above GMN is, Connect multiple buyer nodes and multiple seller nodes, A device that matches buyers and sellers with strong connections between the above nodes.

13. In paragraph 12, The above multiple buyer nodes are, Each includes the individual buyer's energy demand, maximum allowable price, geographical location and time frame for energy demand, The above multiple seller nodes are, A device, each including the energy supply capacity of an individual seller, minimum allowable price, geographical location and time frame for energy availability.

14. A method for optimizing energy transactions within a microgrid, A step of collecting energy-related data within a microgrid in real time; A step of preprocessing the energy-related data collected in real time; A step of predicting energy supply and demand according to the preprocessed energy-related data using an energy supply and demand prediction model; A step of dynamically setting a price according to the predicted energy supply and demand using a pricing algorithm based on reinforcement learning; A method comprising: a step of matching buyers and sellers based on the determined dynamic price using GMN (Graph Matching Networks); 15. A computer-readable recording medium recording a program for executing the method of Article 14 on a computer.

Images