Urban distributed energy resource scheduling method and system based on world model

Abstract

The invention relates to an urban distributed energy resource scheduling method and system based on a world model, and the method comprises the steps: collecting the operation data of various distributed energy resources in an urban power system in real time, inputting the operation data into the world model for processing, and outputting a distributed energy resource scheduling scheme, wherein the world model comprises a virtual simulation sub-model, a situation awareness device and an intelligent decision-making device, and the step of outputting the distributed energy resource scheduling scheme comprises the following steps: establishing the virtual simulation sub-model of the urban power system based on real-time collected operation data in combination with historical operation data; based on the virtual simulation sub-model, performing situation awareness and analysis by adopting the situation awareness device; and based on the situation awareness and analysis result, an intelligent decision maker is adopted to automatically generate a distributed energy resource scheduling scheme. Compared with the prior art, the method has the advantages of realizing virtual scheduling simulation of energy resources, quickly and accurately identifying exceptions, timely adjusting strategies and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of resource scheduling, and in particular to a method and system for scheduling urban distributed energy resources based on a world model. Background Art

[0002] With the continuous growth of urban energy demand and the popularization of distributed energy resources (such as photovoltaic power generation, wind power, energy storage systems, electric vehicles, etc.), the traditional centralized energy dispatching model can no longer cope with the complex power system of modern cities. Distributed energy resources are highly decentralized, unpredictable and diverse, making the existing power grid dispatching method incapable of dealing with uncertainty and dynamic demand.

[0003] Most of the current dispatching systems rely on centralized control architectures, which cannot fully utilize the advantages of distributed energy resources, often leading to problems such as energy waste, dispatching delays, and unbalanced loads. In addition, with the rapid development of technologies such as energy Internet and smart grids, urban energy systems have become more complex, and the difficulty of dispatching and management has further increased. Therefore, a more intelligent and adaptive distributed energy resource dispatching method is urgently needed to meet these challenges and achieve efficient dispatching and optimization of urban energy systems.

[0004] As a virtual and real fusion model based on metaverse technology, the world model has gradually been studied in energy resource scheduling, but the complexity and uncertainty of urban distributed energy resource scheduling have not yet been solved. Therefore, technicians in this field are committed to developing a method for urban distributed energy resource scheduling to deal with current problems. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for urban distributed energy resource scheduling based on a world model with a better distributed energy resource scheduling solution.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A method for dispatching urban distributed energy resources based on a world model comprises the following steps:

[0008] The operation data of various distributed energy resources in the urban power system are collected in real time, input into the world model for processing, and the distributed energy resource scheduling plan is output, wherein the world model includes a virtual simulation sub-model, a situation sensor and an intelligent decision maker, wherein the step of outputting the distributed energy resource scheduling plan includes:

[0009] Based on real-time collected operation data and combined with historical operation data, a virtual simulation sub-model of the urban power system is established;

[0010] Based on the virtual simulation sub-model, the situation sensor is used to perform situation awareness and analysis;

[0011] Based on situation awareness and analysis results, an intelligent decision maker is used to automatically generate distributed energy resource scheduling plans.

[0012] Furthermore, the operation data of the various types of distributed energy resources include power parameters of distributed energy equipment, external environmental factor parameters and power consumption of user-end equipment.

[0013] Furthermore, the virtual simulation sub-model includes a state space for simulating the dynamic changes of distributed energy resources under various working conditions and an output equation for observing the output of the urban power system, wherein the expression of the state space is:

[0014] x(t+1)=f(x(t),u(t),w(t))

[0015] The expression of the output equation is:

[0016] y(t)=h(x(t),v(t))

[0017] Where x(t) represents the state vector of the urban power system at time t, u(t) is the control input, w(t) is the external disturbance, f(·) is the state update function, which describes the change of the urban power system state over time, y(t) is the observed output of the urban power system, v(t) is the measurement noise, and h(·) is the observation function, which is used to map the state of the urban power system to the output variable.

[0018] Furthermore, the situation sensor performs situation perception and analysis based on random matrix theory, and the specific steps include:

[0019] Based on the virtual simulation sub-model, a rolling time window is used to obtain window data, wherein the window data is expressed as:

[0020]

[0021] In the formula, is the window matrix data containing the operating status of each node within a period of time, is the voltage and other operating data of each node at a certain moment;

[0022] Based on the window data Calculate its covariance matrix Ct w , then perform eigenvalue decomposition to obtain its eigenvalue set λ i , calculate the linear eigenvalue statistics and draw the statistics curve, where the calculation expression of the linear eigenvalue statistics is:

[0023]

[0024] In the formula, is the linear eigenvalue statistic, N is the number of nodes in the window matrix, For the test function;

[0025] The statistical curve is compared with the set warning threshold to obtain situation awareness and analysis results.

[0026] Furthermore, the step of performing situation awareness and analysis based on random matrix theory also includes:

[0027] Before calculating the linear eigenvalue statistics, the window data is standardized to obtain a standardized data matrix The expression of the standardization process is:

[0028]

[0029] in:

[0030] In the formula, is the standardized value of the i-th feature in the standardized matrix at the j-th time point, is the value of the i-th feature in the original window data matrix at the j-th time point, is the mean of the ith feature, is the standard deviation of the ith feature, N w is the number of features, that is, the number of different features contained in the window data, T w is the length of the time window.

[0031] Furthermore, the step of calculating the linear eigenvalue statistics includes:

[0032] Based on the standardized data matrix Calculate its covariance matrix Ct w , the expression of the covariance matrix is:

[0033]

[0034] The variance matrix Ct w Perform eigenvalue decomposition to obtain its eigenvalue set λ i , and further calculate the linear eigenvalue statistics to capture the overall situation characteristics.

[0035] Furthermore, the intelligent decision maker adopts a deep reinforcement learning algorithm to automatically generate a distributed energy resource scheduling plan, and the training process of the intelligent decision maker includes:

[0036] Define state representation, action space and reward function, wherein the state representation is a real-time multi-dimensional state vector of the urban power system obtained based on the virtual simulation sub-model and the situation awareness sensor, the action space is all choices of distributed energy resource scheduling schemes, different actions correspond to different distributed energy resource scheduling schemes, and the reward function is to measure the pros and cons of executing different actions;

[0037] A deep Q network is used for training. At each time step t, the intelligent decision maker selects the current action a(t) based on the current state s(t), and observes the new state s(t+1) and reward R(t) after executing the current action. By continuously updating the Q function, the intelligent decision maker gradually learns to select the optimal distributed energy resource scheduling plan in different scenarios.

[0038] Furthermore, the reward function is a weighted combination of multiple factors, including energy efficiency, load balance, scheduling response time and economic cost. The expression of the reward function is:

[0039] R(t)=w1·E(t)+w2·B(t)-w3·D(t)-w4·C(t)

[0040] Where R(t) is the reward at time t, w i is the weight of each factor, which is used to indicate the importance of different objectives in the optimization. i = 1, 2, 3, 4. E(t) is the system energy efficiency, B(t) is the balance of load demand, D(t) is the dispatch response time, and C(t) is the economic cost of the system.

[0041] Furthermore, it also includes a feedback mechanism and a scheduling optimization process, including the following steps:

[0042] Compare the distributed energy resource scheduling plan actually executed with the generated distributed energy resource scheduling plan to determine whether there is a deviation,

[0043] If so, an automatic analysis is performed based on the deviation to find the key factors affecting the distributed energy resource scheduling plan, update the parameters of the intelligent decision maker to update the distributed energy resource scheduling plan, and apply it in the next scheduling cycle.

[0044] The present invention provides a city distributed energy resource scheduling system based on a world model, comprising:

[0045] Real-time acquisition module: used to collect real-time operation data of various distributed energy resources in urban power systems;

[0046] Simulation modeling module: used to establish a virtual simulation sub-model of the urban power system based on real-time collected operation data combined with historical operation data;

[0047] Situation awareness module: used for performing situation awareness and analysis based on the virtual simulation sub-model using the situation awareness device;

[0048] Scheduling plan generation module: used to automatically generate distributed energy resource scheduling plans based on situation awareness and analysis results using intelligent decision makers.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] (1) The present invention integrates virtuality and reality by combining virtuality and reality, constructs a world model, simulates the operating status of the system in real time according to the virtual simulation sub-model in the world model, uses a situation sensor for state perception, and provides state characteristics for the intelligent decision maker, so that the intelligent decision maker can provide a better distributed energy resource scheduling solution according to the real-time state characteristics of the system.

[0051] (2) The situation sensor of the present invention adopts random matrix theory for situation perception and analysis, can reduce the dimensionality of high-dimensional data, calculate the state characteristics of various distributed energy resources, improve the analysis ability of high-dimensional spatiotemporal data, and can quickly and accurately identify possible anomalies in the system and adjust the scheduling strategy in time.

[0052] (3) The intelligent decision maker of the present invention adopts a deep reinforcement learning algorithm, which can not only learn the optimal scheduling solution selection under different scenarios according to the current state of the system, but also realize self-learning and dynamic optimization of the scheduling strategy, and can provide more robust solutions when facing uncertainty and complex demands.

[0053] (4) The present invention also provides a feedback mechanism and a scheduling optimization process, which compares the actual execution results with the expected generated scheduling plan to adjust the deviation in a targeted manner and update the scheduling strategy, thereby further improving the scheduling accuracy and robustness of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0055] Figure 2 It is the RMT implementation process of the present invention;

[0056] Figure 3 This is the rolling update feedback mechanism of the present invention. DETAILED DESCRIPTION

[0057] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0058] Example 1

[0059] This embodiment aims to solve the problem that the existing system is difficult to effectively schedule complex energy systems in real time through a combination of virtual and real methods, and lacks the ability to optimize dynamic resources at different time scales. It provides a method for scheduling urban distributed energy resources based on a world model to cope with the complexity, uncertainty and multi-level time and space requirements in energy system scheduling, and can achieve optimal scheduling of energy resources while ensuring system stability and efficient operation. Figure 1 As shown, the method comprises the following steps:

[0060] Step 1: Real-time operation data collection and simulation modeling based on urban distributed energy resources.

[0061] The present invention first collects the operating data of various distributed energy resources in the city in real time through the distributed sensor network in the city. These resources include photovoltaic power generation, wind power generation, energy storage systems, electric vehicles and users' power loads, etc. The specific indicators collected cover data in multiple dimensions such as voltage, current, power output, temperature, etc. Through the accurate collection and analysis of these data, the real-time operating status of distributed energy resources in the entire urban power grid is obtained, providing sufficient data support for subsequent simulation modeling. There are many sensors involved in the data collection process, including:

[0062] (1) Electrical sensors: used to measure power parameters such as voltage, current, frequency, and power factor of distributed energy equipment;

[0063] (2) Environmental sensors: monitor external environmental factors such as temperature and humidity, which have a direct impact on energy output and power load;

[0064] (3) User load sensor: real-time tracking and recording of user-side electricity consumption, especially the use of high-load equipment such as electric vehicles.

[0065] In order to integrate virtuality and reality, this implementation constructs a world model, which includes a virtual simulation sub-model, a situation sensor and an intelligent decision maker. The world model is used to realize the scheduling of urban distributed energy resources.

[0066] First, the collected data is transmitted to the general simulator in the system in real time through the Internet of Things technology. The general simulator combines historical data with real-time data to establish a virtual simulation sub-model of the urban power system.

[0067] Specifically, the sub-model is divided into two parts: state space and output equation. The state space is used to simulate the dynamic changes of distributed energy resources under various working conditions. Assuming that the state vector of the system is x(t), the control input is u(t), and the external disturbance is w(t), the change of the system state at time t can be expressed by the following equation:

[0068] x(t+1)=f(x(t),u(t),w(t)) (1)

[0069] Where: x(t) represents the state vector of the system, such as the output power and load conditions of energy resources such as photovoltaic power generation, wind power generation, energy storage system, and electric vehicles. u(t) is the control input of the system, which refers to the energy allocation plan generated by the scheduling system. w(t) is the external disturbance term, which includes unpredictable external factors (such as weather changes, user demand fluctuations, etc.).

[0070] The output of the system can be expressed by the observed output equation:

[0071] y(t)=h(x(t),v(t)) (2)

[0072] Where: y(t) is the observed system output (such as grid voltage, current, etc.). v(t) is the measurement noise. Function h() represents the observation function, which maps the system state to the output variable.

[0073] The virtual simulation sub-model can simulate the dispatch of distributed energy in a variety of scenarios, including operating scenarios under different load demands, different weather conditions, and different energy supply conditions. The virtual simulation sub-model can dynamically adjust the energy resource allocation method in the virtual environment according to real-time data to ensure the accuracy and practical operability of the simulation results. Through the above-mentioned state equations and output equations, the digital twin can simulate the power operation status of energy resources in the urban power system in real time in the virtual space, and the simulation results provide basic data support for subsequent situation awareness and decision optimization.

[0074] Step 2: Conduct situational awareness and analysis based on high-dimensional spatiotemporal data.

[0075] After data collection and simulation modeling are completed, the Situational Perceiver performs real-time analysis on the high-dimensional data in the virtual power network. In order to effectively process these complex spatiotemporal data, the present invention uses random matrix theory (RMT) to reduce the dimensionality of the data and extract key operating characteristics, including the supply and demand relationship of energy, power fluctuations, and the spatial distribution of loads.

[0076] The situation sensor can identify potential risks and abnormal conditions in the energy system, such as local overloads and imbalances in energy resources, from a large amount of historical data. The situation sensor analyzes high-dimensional data based on random matrix theory, and identifies potential risks and abnormal conditions in the energy system by calculating the linear eigenvalue statistic (LES) statistic curve. Specifically, the LES statistic reflects the cumulative changes in the system's eigenvalues. When the system operates normally, the LES curve shows a steady fluctuation; when an abnormal situation occurs in the system (such as a local overload or an imbalance in energy resources), the LES curve will show significant fluctuations or mutations. By setting the early warning threshold, the system can automatically detect abnormal LES values ​​and issue early warning signals in a timely manner to ensure the stable operation of the energy system. Specifically, if Figure 2 As shown, the RMT steps are implemented as follows:

[0077] (1) Based on the rolling time window, the window data is obtained. The formula is as follows:

[0078]

[0079] In the formula, is the window matrix data containing the operating status of each node within a period of time, is the operating data such as the voltage of each node at a certain moment.

[0080] (2) Standardize the window data. The formula is as follows:

[0081]

[0082] in:

[0083] In the formula, is the standardized value of the i-th feature in the standardized matrix at the j-th time point, is the value of the i-th feature in the original window data matrix at the j-th time point, is the mean of the ith feature, Generally, it is set to 0. Generally take 1, is the standard deviation of the ith feature, N w is the number of features, that is, the number of different features contained in the window data, T w is the length of the time window.

[0084] (3) Calculate the eigenvalue and LES statistic of the standardized window data. The formula is as follows:

[0085] After standardization, the data matrix On the top, calculate its covariance matrix Ct w, defined as:

[0086]

[0087] Next, use RMT to calculate the covariance matrix Perform eigenvalue decomposition to obtain its eigenvalue set λ i . The LES statistic is calculated to capture the overall state characteristics of the system:

[0088]

[0089] In the formula, is the linear eigenvalue statistic, N is the number of nodes in the window matrix, For the test function.

[0090] Step 3: Generate an adaptive scheduling plan based on the intelligent decision maker.

[0091] The Intelligent Decision-Maker of the present invention automatically generates an adaptive energy scheduling solution based on the perception results of the situation sensor and using a deep reinforcement learning algorithm. The Intelligent Decision-Maker can combine historical data and real-time perception results to generate the optimal energy scheduling strategy to ensure that the system can cope with various complex energy demand scenarios. The working steps of the Intelligent Decision-Maker are as follows.

[0092] (1) State Representation: The state of the system is represented as a multidimensional vector s(t), which contains the real-time state of the energy system, such as photovoltaic power generation, energy storage system power, electric vehicle charging requirements, wind power output, etc. The state vector is obtained through a simulation simulator and a situational awareness device and used as the input of the reinforcement learning algorithm;

[0093] s(t)=[s1(t),s2(t),...,s n (t)] (7)

[0094] Where: s i (t) represents the i-th state variable at time t, and n is the number of all state variables in the system. The state vector is obtained in real time by the simulator and situation awareness, and is used as the input of the deep reinforcement learning algorithm to reflect the operating status of the system.

[0095] (2) Action Space: The action space a(t) of the intelligent decision maker defines the possible choices for energy resource scheduling. For example, adjusting the charging and discharging strategy of the energy storage system, controlling the output power of photovoltaic and wind power generation, optimizing the charging period of electric vehicles, etc. Different actions correspond to different energy scheduling strategies;

[0096] a(t)=[a1(t),a2(t),…,a m (t)] (8)

[0097] Among them: a i (t) represents the i-th control action at time t, and m is the dimension of the action space, that is, the number of all possible scheduling schemes. Different actions correspond to different energy resource scheduling strategies, such as adjusting the charging and discharging strategy of the energy storage system, controlling the output power of photovoltaic and wind power generation, optimizing the charging time of electric vehicles, etc. Each action selection will affect the energy allocation and scheduling efficiency of the system.

[0098] (3) Reward Function: The reward function R(t) is the core of deep reinforcement learning and is used to measure the pros and cons of the scheduling scheme. R(t) is determined by multiple factors, such as energy efficiency, load balance, scheduling response time, and economic cost. The reward function can be expressed as a weighted combination of these factors:

[0099] R(t)=w1·E(t)+w2·B(t)-w3·D(t)-w4·C(t) (9)

[0100] Where: E(t) is the system energy efficiency (such as reducing energy waste); B(t) is the balance of load demand; D(t) is the dispatch response time; C(t) is the economic cost of the system; w i is the weight of each factor, which is used to indicate the importance of different objectives in optimization.

[0101] (4) Intelligent decision maker training process: Taking the Deep Q-Network (DQN) as an example, the process of the decision maker learning the optimal scheduling strategy is explained. At each time step t, the decision maker selects an action a(t) based on the current state s(t) and observes the system feedback after executing the action (i.e., the new state s(t+1) and reward R(t)). By continuously updating the Q function, the system can gradually learn which scheduling solution to choose in different scenarios to obtain the highest long-term return.

[0102] Step 4: Optimize the scheduling plan based on the feedback mechanism.

[0103] In order to ensure the effectiveness of the scheduling scheme and the continuous optimization of the system, the present invention also introduces a real-time feedback mechanism in the world model. Through this mechanism, the system can compare the scheduling execution results with the expected scheduling effects, analyze and adjust the scheduling strategy in real time, and ensure that the system always maintains efficient operation in a dynamically changing environment. Figure 3 As shown, the feedback mechanism includes the following steps:

[0104] (1) After executing the scheduling plan, the system monitors the effect of the scheduling execution in real time, compares the actual execution results with the expected plan, and analyzes the deviations that occur during the actual scheduling process;

[0105] (2) When the scheduling effect does not meet expectations, the feedback module automatically analyzes the data, finds the key factors affecting the scheduling results, and adjusts the parameters in the scheduling model accordingly;

[0106] (3) Based on the feedback results, the intelligent decision maker updates the scheduling strategy and applies the optimized solution in the next scheduling cycle to improve the scheduling accuracy and robustness of the system.

[0107] The technical solution of the present invention is described below with reference to a specific embodiment.

[0108] 1) Obtain real-time operating data of urban distributed energy resources, including the output power of photovoltaic power generation equipment, the power generation of wind power generation equipment, the charging and discharging status of energy storage systems, the charging requirements of electric vehicles, and the power load of users. The collected data also includes key parameters such as voltage, current, power output, and ambient temperature of the power network;

[0109] 2) The acquired data is input into the situation sensor, and the random matrix theory is used to reduce the dimensionality of high-dimensional data and extract key energy resource operation characteristic indicators. Furthermore, the state characteristics of each energy resource are calculated by the RMT method, including resource load changes, power output fluctuations, equipment operation health status, etc. Finally, the situation sensor generates situation awareness analysis results and provides state characteristics for the intelligent decision maker;

[0110] 3) Based on the situational awareness results, the intelligent decision maker uses deep reinforcement learning algorithms to optimize the scheduling strategy of distributed energy resources. Furthermore, the system combines historical data, real-time data, and situational awareness reports to dynamically generate the optimal scheduling plan. For different time scales, the system will flexibly allocate resources based on current and future load demands. For example, when the system recognizes that a load peak is approaching, the intelligent decision maker can adjust the charging status of the energy storage system in advance and reasonably allocate the charging time of electric vehicles to ensure that the power grid can operate smoothly during the load peak period;

[0111] 4) After the scheduling plan is executed, the system monitors the execution results of the scheduling in real time and compares and analyzes the actual scheduling effect with the expected plan. Through the feedback mechanism, the system can identify deviations that occur during the execution process and optimize the scheduling model based on these deviations. Furthermore, the system adjusts the parameters of the scheduling algorithm by calculating the difference between the actual operation of each distributed energy resource and the ideal scheduling plan, so that in the future scheduling cycle, it can better adapt to the dynamic changes of energy resources and improve the accuracy and reliability of scheduling.

[0112] In summary, the present invention takes the world model and intelligent scheduling as the guiding ideology, combines virtual-real fusion simulation, situational awareness and deep learning algorithms, and conducts research on new urban distributed energy resource scheduling. The present invention can effectively perceive the operating status of distributed energy resources in urban power grids, improve the flexible scheduling capability of resources and the overall optimization level of the system, and at the same time improve the stability and scheduling efficiency of the system through the combination of virtual simulation and real feedback; in addition, the present invention introduces random matrix theory (RMT) and deep reinforcement learning algorithms to realize the intelligent optimization of energy scheduling schemes, providing a new technical idea for the scheduling of distributed energy resources in modern cities, and significantly improving the management and scheduling intelligence level of energy resources.

[0113] Example 2

[0114] This embodiment provides a city distributed energy resource scheduling system based on a world model, including:

[0115] Real-time acquisition module: used to collect real-time operation data of various distributed energy resources in urban power systems;

[0116] Simulation modeling module: used to establish a virtual simulation sub-model of the urban power system based on real-time collected operation data combined with historical operation data;

[0117] Situation awareness module: used for performing situation awareness and analysis based on the virtual simulation sub-model using the situation awareness device;

[0118] Scheduling plan generation module: used to automatically generate distributed energy resource scheduling plans based on situation awareness and analysis results using intelligent decision makers.

[0119] The rest is the same as in Example 1.

[0120] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0121] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0122] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0123] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0124] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0125] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0126] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for scheduling urban distributed energy resources based on a world model, characterized in that: The following steps are involved: The operation data of various distributed energy resources in the urban power system are collected in real time, input into the world model for processing, and the distributed energy resource scheduling plan is output, wherein the world model includes a virtual simulation sub-model, a situation sensor and an intelligent decision maker, wherein the step of outputting the distributed energy resource scheduling plan includes: Based on real-time collected operation data and combined with historical operation data, a virtual simulation sub-model of the urban power system is established; Based on the virtual simulation sub-model, the situation sensor is used to perform situation awareness and analysis; Based on situation awareness and analysis results, an intelligent decision maker is used to automatically generate distributed energy resource scheduling plans.

2. The method for scheduling urban distributed energy resources based on a world model according to claim 1, characterized in that: The operating data of the various types of distributed energy resources include power parameters of distributed energy equipment, external environmental factor parameters, and power consumption of user-end equipment.

3. The method for scheduling urban distributed energy resources based on a world model according to claim 1, characterized in that: The virtual simulation sub-model includes a state space for simulating the dynamic changes of distributed energy resources under various working conditions and an output equation for observing the output of the urban power system, wherein the expression of the state space is: x(t+1)=f(x(t),u(t),w(t)) The expression of the output equation is: y(t)=h(x(t),v(t)) Where x(t) represents the state vector of the urban power system at time t, u(t) is the control input, w(t) is the external disturbance, f(·) is the state update function, which describes the change of the urban power system state over time, y(t) is the observed output of the urban power system, v(t) is the measurement noise, and h(·) is the observation function, which is used to map the state of the urban power system to the output variable.

4. The method for scheduling urban distributed energy resources based on a world model according to claim 1, characterized in that: The situation sensor performs situation perception and analysis based on random matrix theory, and the specific steps include: Based on the virtual simulation sub-model, a rolling time window is used to obtain window data, wherein the window data is expressed as: In the formula, is the window matrix data containing the operating status of each node within a period of time, It is the operating data such as the voltage of each node at a certain moment; Based on the window data, the linear eigenvalue statistics are calculated and a statistical curve is drawn, wherein the calculation expression of the linear eigenvalue statistics is: In the formula, is the linear eigenvalue statistic, N is the number of nodes in the window matrix, For the test function; The statistical curve is compared with the set warning threshold to obtain situation awareness and analysis results.

5. The method for scheduling urban distributed energy resources based on a world model according to claim 4, characterized in that: The step of performing situation awareness and analysis based on random matrix theory also includes: Before calculating the linear eigenvalue statistics, the window data is standardized to obtain a standardized data matrix The expression of the standardization process is: (i=1,2,...N w ,j=1,2,...T w ) in: In the formula, is the standardized value of the i-th feature in the standardized matrix at the j-th time point, is the value of the i-th feature in the original window data matrix at the j-th time point, is the mean of the ith feature, is the standard deviation of the ith feature, N w is the number of features, that is, the number of different features contained in the window data, T w is the length of the time window.

6. The method for scheduling urban distributed energy resources based on a world model according to claim 5, characterized in that: The step of calculating the linear eigenvalue statistics comprises: Based on the standardized data matrix Calculate its covariance matrix Ct w , the expression of the covariance matrix is: The variance matrix Ct w Perform eigenvalue decomposition to obtain its eigenvalue set λ i , and further calculate the linear eigenvalue statistics to capture the overall situation characteristics.

7. The method for scheduling urban distributed energy resources based on a world model according to claim 1, characterized in that: The intelligent decision maker uses a deep reinforcement learning algorithm to automatically generate a distributed energy resource scheduling plan. The training process of the intelligent decision maker includes: Define state representation, action space and reward function, wherein the state representation is a real-time multi-dimensional state vector of the urban power system obtained based on the virtual simulation sub-model and the situation awareness sensor, the action space is all choices of distributed energy resource scheduling schemes, different actions correspond to different distributed energy resource scheduling schemes, and the reward function is to measure the pros and cons of executing different actions; A deep Q network is used for training. At each time step t, the intelligent decision maker selects the current action a(t) based on the current state s(t), and observes the new state s(t+1) and reward R(t) after executing the current action. By continuously updating the Q function, the intelligent decision maker gradually learns to select the optimal distributed energy resource scheduling plan in different scenarios.

8. The method for scheduling urban distributed energy resources based on a world model according to claim 7, characterized in that: The reward function is a weighted combination of multiple factors, including energy efficiency, load balance, scheduling response time and economic cost. The expression of the reward function is: R(t)=w1·E(t)+w2·B(t)-w3·D(t)-w4·C(t) Where R(t) is the reward at time t, w i is the weight of each factor, which is used to indicate the importance of different objectives in the optimization. i = 1, 2, 3, 4. E(t) is the system energy efficiency, B(t) is the balance of load demand, D(t) is the dispatch response time, and C(t) is the economic cost of the system.

9. The method for scheduling urban distributed energy resources based on a world model according to claim 1, characterized in that: It also includes feedback mechanism and scheduling optimization process, including the following steps: Compare the distributed energy resource scheduling plan actually executed with the generated distributed energy resource scheduling plan to determine whether there is a deviation, If so, an automatic analysis is performed based on the deviation to find the key factors affecting the distributed energy resource scheduling plan, update the parameters of the intelligent decision maker to update the distributed energy resource scheduling plan, and apply it in the next scheduling cycle.

10. An urban distributed energy resource scheduling system based on a world model, characterized in that: include: Real-time acquisition module; Used to collect real-time operating data of various distributed energy resources in urban power systems; Simulation modeling module: used to establish a virtual simulation sub-model of the urban power system based on real-time collected operation data combined with historical operation data; Situation awareness module: used for performing situation awareness and analysis based on the virtual simulation sub-model using the situation awareness device; Scheduling plan generation module: used to automatically generate distributed energy resource scheduling plans based on situation awareness and analysis results using intelligent decision makers.

Images