New energy power optimization scheduling method and system under digital twin platform

Abstract

The invention relates to the technical field of new energy power systems, in particular to a new energy power optimization scheduling method and system under a digital twin platform, and the method comprises the steps: firstly building a new energy power system optimization scheduling scene based on the digital twin platform, building a digital twin edge network model through the scene, and carrying out the optimization scheduling of the new energy power system; accurate modeling, real-time monitoring, optimal scheduling and decision support of the new energy power system are realized. In the aspect of electric power tasks, the intelligent terminal has the option of performing local calculation or unloading the tasks to the edge server. In addition, an optimization model of intelligent terminal unloading decision and computing resource allocation is constructed, so that the target of minimizing the total energy consumption of the whole network is achieved. And finally, an on-policy deep reinforcement learning (DRL) algorithm is adopted to solve an unloading decision and a computing resource allocation strategy, so that the online real-time performance of the unloading decision and the computing resource allocation is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy power systems, and in particular to a new energy power optimization scheduling method and system under a digital twin platform. Background Art

[0002] With the demand for energy transformation in countries around the world, the power system, as an important pillar of energy supply, urgently needs to develop "low-carbon" and "green" power generation energy, and realize "intelligent" and "safe" operation mode. The rapid development of new intelligent technologies such as edge computing has provided favorable conditions for promoting the coordinated and integrated development of new energy power grids and data information networks, forming new application scenarios for energy-information interaction and integration, including new application technologies such as cloud-edge collaboration and digital twins. Edge computing places computing and data processing capabilities on edge nodes close to terminal devices and users, which can reduce data transmission delays and achieve real-time response and rapid decision-making. Digital twins can create an accurate virtual model to simulate the various components, operating states and behaviors of smart grids. Based on the new energy power grid model of digital twins, intelligent optimization and control algorithms can be developed to realize the automation and intelligent management of smart grids. By combining real-time data and model predictions, edge servers can realize real-time monitoring, optimized scheduling and intelligent control of the power grid to improve the efficiency and performance of the power grid. DRL is a machine learning method used to solve the problem of intelligent agents making decisions in the environment. In the problem of power system resource scheduling, DRL can dynamically adjust the allocation and mapping of virtualized resources by understanding the resource usage and user needs of the power system to achieve optimal resource utilization. This method has great research value and can be applied to solve power system resource scheduling problems and improve the efficiency and performance of the power system. By combining with the digital twin model, the DRL agent can continuously learn and optimize strategies through interaction with the environment to achieve optimal operation of the power grid.

[0003] Although new energy power generation technology has been widely used in the low-carbon transformation of power systems due to its environmental friendliness, its inherent randomness and volatility have brought challenges to the safe scheduling of power systems. Summary of the invention

[0004] In view of the above problems existing in the prior art, the present invention is proposed.

[0005] Therefore, the technical problem to be solved by the present invention is: optimizing offloading decisions and computing resource allocation to obtain optimal resource scheduling.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a method for optimizing and dispatching new energy power under a digital twin platform, comprising: constructing an optimized dispatching scenario of a new energy power system under a digital twin platform based on data collected by a third party; establishing a digital twin edge network model based on the optimized dispatching scenario of the new energy power system under the digital twin platform; establishing a communication model and a task model for unloading decisions and computing resource allocation based on the optimized dispatching scenario of the new energy power system under the digital twin platform; prioritizing tasks to obtain task processing delays of different levels; constructing a local computing model for unloading decisions and computing resource allocation based on the optimized dispatching scenario of the new energy power system under the digital twin platform; constructing an unloading model for user unloading decisions and computing resource allocation based on the optimized dispatching scenario of the new energy power system under the digital twin platform; solving user unloading decisions and computing resource allocation based on DRL.

[0007] As a preferred solution of the new energy power optimization dispatching method under the digital twin platform described in the present invention, the construction of the digital twin platform includes generating computing tasks according to K smart terminals in the energy power system, the tasks are not related to each other, and the environmental information data is obtained by using the data collected by the third party. The coordinates of the smart terminal are expressed as:

[0008]

[0009] There are N time slots in the period T, n represents the number of each time slot, and the smart terminal in each time slot offloads the computing task to the edge server or performs it locally. The edge server provides offloading service, and the fixed coordinates are expressed as:

[0010]

[0011] Among them, the number of edge servers is M, the number of the mth edge server is m,

[0012] As a preferred solution of the new energy power optimization scheduling method under the digital twin platform described in the present invention, wherein: the establishment of the digital twin edge network model includes a two-layer digital twin edge network model of the physical entity layer and the digital twin layer, wherein K smart terminals and M edge servers constitute the physical entity layer, and the digital twins of all physical entities and the communication environment constitute the digital twin layer; according to the characteristics of the smart terminal unloading to the edge server or local computing, the unloading decision variable is defined It is expressed as:

[0013]

[0014] Among them, when When n time slot intelligent terminal k offloads the computing task to edge server m, When indicates that the n-time slot intelligent terminal k chooses local calculation;

[0015] The intelligent terminal in the time slot is unloaded to the edge server, which is expressed as:

[0016]

[0017] At time slot n, the digital twin of the computational entity with respect to its computational frequency is represented as:

[0018]

[0019] in, represents the digital twin’s estimate of the true calculation frequency, Represents the error between the true calculated frequency and the digital twin estimate.

[0020] As a preferred solution of the new energy power optimization scheduling method under the digital twin platform described in the present invention, the communication model and task model of computing resource allocation include each intelligent terminal K generating a computing task, which is expressed as:

[0021]

[0022] Where n represents time slot n, Indicates the size of the task. Indicates the number of CPUs required for the computing task. Indicates the task delay limit;

[0023] The error between the actual calculation delay and the digital twin estimated delay is expressed as:

[0024]

[0025] Calculate the distance between different devices in time slot n, expressed as:

[0026]

[0027] Among them, the distance between the Kth intelligent terminal and the mth edge server is expressed as

[0028] When time slot t, the channel gain from the kth intelligent terminal to the mth edge server is expressed as:

[0029]

[0030] Among them, α0 represents the channel function gain when the reference distance is 1m;

[0031] when a k,m (t) = 1, user k is served by edge server m in time slot n, and the corresponding signal-to-noise ratio of UAVm is expressed as:

[0032]

[0033] Among them, p k represents the transmission power of intelligent terminal k, σ 2 represents the power of additive white Gaussian noise at the UVA receiver, represents the interference caused by all intelligent terminal transmissions in time slot n;

[0034] The unloading rate of intelligent terminal k in time slot n is expressed as:

[0035]

[0036] As a preferred solution of the method for optimizing and dispatching new energy power under a digital twin platform described in the present invention, the priority classification of tasks includes taking response measures or requiring high real-time transmission according to the priority of power business, which is defined as high priority γ=3, and there is a certain time tolerance from task generation to response, which is defined as medium priority γ=2, and subsequent analysis work and tasks that do not require response measures are defined as low priority γ=1; the average arrival amount of priority tasks is expressed as:

[0037]

[0038] Where λ represents the average arrival rate of terminal power tasks, λ y represents the arrival rate of power tasks with different priorities, a n , ar It represents the average arrival rate of the tasks that are offloaded in time slot n, expressed as:

[0039]

[0040] Satisfies λ1+λ2+λ3=λ γ ;

[0041] Obtaining different levels of task processing delays includes calculating the average processing time of tasks generated by high-priority services, the average processing time of tasks generated by medium-priority services, and the average processing time of tasks generated by low-priority services;

[0042] The average processing time of tasks generated by high-priority services is expressed as:

[0043]

[0044] in, Indicates the average processing time of tasks generated by high-priority services, Represents the digital twin’s estimate of the true computation frequency;

[0045] The average processing time of tasks generated by medium-priority services is expressed as:

[0046]

[0047] in, Indicates the average processing time of tasks generated by medium priority business;

[0048] The average processing time of tasks generated by low-priority services is expressed as:

[0049]

[0050] in, Indicates the average processing time of tasks generated by low-priority services.

[0051] As a preferred solution of the new energy power optimization scheduling method under the digital twin platform described in the present invention, the local computing model of computing resource allocation includes computing delay and energy consumption. When the smart terminal decides local computing, the computing delay is expressed as:

[0052]

[0053] The total delay of the smart terminal processing tasks on the local device is expressed as:

[0054]

[0055] Energy consumption is expressed as:

[0056]

[0057] in, is the local computing capability of the time slot intelligent terminal k, and k1 is the power consumption factor;

[0058] When the intelligent terminal is in time slot n, the computing power is expressed as:

[0059]

[0060] in, Indicates the maximum local computing capacity of the smart terminal;

[0061] When smart terminal k offloads the task to the edge server in time slot n, the offloading delay is expressed as:

[0062]

[0063] The unloading energy consumption is expressed as:

[0064]

[0065] When smart terminal k offloads the task to the edge server in time slot n, the actual computing delay is expressed as:

[0066]

[0067] The total delay is expressed as:

[0068]

[0069] The energy consumption is calculated as:

[0070]

[0071] The constraints of the edge server are expressed as:

[0072]

[0073] in, It is the maximum computing resource that can be allocated to each edge server.

[0074] As a preferred solution of the method for optimizing the dispatch of renewable energy power under the digital twin platform described in the present invention, the unloading decision includes defining Computing resources The optimization problem model is expressed as:

[0075]

[0076] Among them, the problem P represents MDP, and the optimal strategy is learned in the training environment through the DRL method to solve the problem of minimizing energy consumption. When the time step is m and the environment is in S n state, the agent performs action a n , the environment moves to the subsequent state s n+1 In the example, the agent receives a reward r n ;

[0077] Network Status n It is expressed as:

[0078]

[0079] When the state s in time slot n n , the DRL algorithm obtains the action space a according to the state n It is expressed as:

[0080]

[0081] Reward function r n It is expressed as:

[0082]

[0083] The PPO algorithm uses an Actor-Critic structure. The Actor network is divided into two parts, old and new, including parameters θ and oldθ. The Critic network parameter is ξ. The Actor network changes according to the state s. n Output action a n , interact with the environment; the Critic network calculates the state value V based on the state information ξ (s n ), expressed as:

[0084]

[0085] Where γ is the discount factor, represents the expected value, represents the reward function;

[0086] Actor network parameters θ and Critic network parameters ξ are expressed as:

[0087]

[0088] L critic (ξ)=[V ξ (s n+1 )-V ξ (s n )] 2

[0089] Among them, π θ (·)and Represents the new and old policy functions, ε represents the truncation parameter;

[0090] Replace the old θ with the new parameter θ old , expressed as:

[0091] θ old =θ.

[0092] Another object of the present invention is to provide a new energy power optimization dispatching system under a digital twin platform to achieve accurate modeling and real-time monitoring of the new energy power system, thereby improving the efficiency and reliability of the system. Specifically, the method first constructs an optimization dispatching scenario based on data collected by a third party; then, a digital twin edge network model is established to provide accurate modeling, real-time monitoring, optimization dispatching and decision support; in addition, by establishing a communication model and a task model, priorities are divided for different power task requirements, and an optimization model for intelligent terminal unloading decision and computing resource allocation is constructed; finally, an on-policy DRL algorithm is used to solve the unloading decision and computing resource allocation strategy.

[0093] In order to solve the above technical problems, the present invention provides the following technical solutions: a new energy power optimization dispatching system under a digital twin platform, comprising: a data collection module, a system scenario construction module, an edge network construction module, a communication and task module, a priority processing module, a local calculation module and a user unloading decision module;

[0094] The data collection module obtains necessary data from a third party to provide basic data support for the digital twin platform to establish an optimized dispatching scenario for the new energy power system;

[0095] The system scenario building module builds detailed digital twin platform scenarios based on data to reflect the real-time operating status of the power system and potential optimization scheduling needs;

[0096] The edge network construction module establishes an edge network model under the digital twin platform to simulate and analyze the optimization scheduling effect of the energy system on edge computing resources;

[0097] The communication and task module designs the communication model and the task model, completes the offloading decision and the computing resource allocation, and improves the system efficiency and response speed;

[0098] The priority processing module classifies tasks according to their priorities and schedules them according to different processing delay requirements to ensure that high-priority tasks are processed in a timely manner;

[0099] The local computing module establishes a local computing model, optimizes decision-making and resource allocation strategies, reduces communication delays and energy consumption, and improves overall system performance;

[0100] The user unloading decision module solves the user unloading decision and calculates resource allocation based on the DRL method to realize intelligent energy system optimization scheduling management.

[0101] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the new energy power optimization scheduling method under the digital twin platform as described above are implemented.

[0102] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of a new energy power optimization scheduling method under a digital twin platform as described above are implemented.

[0103] Beneficial effects of the present invention: The present invention is applicable to the scenario of new energy power generation system. By establishing the optimization and scheduling scenario of new energy power system under the digital twin platform, accurate modeling and real-time monitoring of the system can be achieved. This enables the system to accurately understand the state and performance of the power system, and provide optimization scheduling and decision support to improve the efficiency and reliability of the system. Secondly, for power tasks, the present invention provides a flexible choice for smart terminals to perform local calculations or offload to edge servers. This enables smart terminals to make decisions based on specific task requirements and system status to achieve the best task processing method. Furthermore, according to the needs of different power tasks, the present invention can perform priority division to ensure that critical tasks are given priority. By adopting the on-policy DRL algorithm, the present invention can solve the optimal unloading decision and computing resource allocation strategy. The application of this algorithm greatly improves the online real-time performance of unloading decisions and computing resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0105] Figure 1 A flowchart of a new energy power optimization scheduling method under a digital twin platform provided by an embodiment of the present invention.

[0106] Figure 2 An optimized scheduling scenario diagram of a new energy power optimization scheduling method under a digital twin platform provided by an embodiment of the present invention.

[0107] Figure 3 An algorithm convergence diagram of a new energy power optimization scheduling method under a digital twin platform provided by an embodiment of the present invention.

[0108] Figure 4 A comparison chart of user equipment task volume and average throughput of different algorithms for a new energy power optimization scheduling method under a digital twin platform provided by an embodiment of the present invention.

[0109] Figure 5 A graph showing the relationship between the total network energy consumption and the number of terminal devices for a new energy power optimization scheduling method under a digital twin platform provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0110] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0111] Example 1, reference Figure 1-Figure 4 , is an embodiment of the present invention, which provides a new energy power optimization scheduling method under a digital twin platform, including:

[0112] S1: Based on the data collected by the third party, build an optimized dispatching scenario for the new energy power system under the digital twin platform.

[0113] It should be noted that building a digital twin platform includes generating computing tasks based on K smart terminals in the energy and power system. The tasks are not related to each other. Environmental information data is obtained using data collected by a third party. The coordinates of the smart terminals are expressed as:

[0114]

[0115] There are N time slots in the period T, n represents the number of each time slot, and the smart terminal in each time slot offloads the computing task to the edge server or performs it locally. The edge server provides offloading service, and the fixed coordinates are expressed as:

[0116]

[0117] Among them, the number of edge servers is M, the number of the mth edge server is m,

[0118] Depend on Figure 2 It can be seen that this scenario considers a two-layer digital twin edge network model, namely the physical entity layer and the digital twin layer. Among them, smart terminals and edge servers together constitute the physical entity layer, and the digital twins of all physical entities and the communication environment together constitute the digital twin layer. Since digital twins reproduce the operation of physical entities, maintaining digital twins of massive devices will consume a lot of resources. In order to enable the digital twin network to run smoothly, all digital twins are built in the edge server. Smart terminals generate independent computing tasks, and can choose to offload tasks to edge servers or perform local computing.

[0119] S2: Establish a digital twin edge network model based on the optimized dispatching scenario of the new energy power system under the digital twin platform.

[0120] It should be noted that the establishment of a digital twin edge network model includes two layers of digital twin edge network models, namely the physical entity layer and the digital twin layer. K smart terminals and M edge servers constitute the physical entity layer, and the digital twins and communication environments of all physical entities constitute the digital twin layer. The smart terminals in the physical entity layer send the current operating status to the digital twin system in real time. The digital twin system builds a virtual model of the new energy power system based on the data of the physical entity layer. In this model, the unloading strategy of the smart terminal is simulated, and the unloading decision variables are defined according to the characteristics of the smart terminal unloading to the edge server or local computing. It is expressed as:

[0121]

[0122] Among them, when When n time slot intelligent terminal k offloads the computing task to edge server m, When indicates that the n-time slot intelligent terminal k chooses local calculation;

[0123] The intelligent terminal in each time slot is unloaded to the edge server, which is expressed as:

[0124]

[0125] For time slot n, the digital twin of a computational entity with respect to its computational frequency is represented as:

[0126]

[0127] in, represents the digital twin’s estimate of the true calculation frequency, Represents the error between the true calculated frequency and the digital twin estimate.

[0128] S3: Based on the optimized dispatching scenario of the new energy power system under the digital twin platform, establish the communication model and task model for unloading decision and computing resource allocation.

[0129] It should be noted that the communication model and task model of computing resource allocation include each intelligent terminal K generating a computing task, which is expressed as:

[0130]

[0131] Where n represents time slot n, Indicates the size of the task. Indicates the number of CPUs required for the computing task. Indicates the task delay limit;

[0132] The error between the actual calculation delay and the digital twin estimated delay is expressed as:

[0133]

[0134] Calculate the distance between different devices in time slot n, expressed as:

[0135]

[0136] Among them, the distance between the Kth intelligent terminal and the mth edge server is expressed as

[0137] Furthermore, considering the actual situation, the channel gain from the kth intelligent terminal to the mth edge server at time slot t is expressed as:

[0138]

[0139] Among them, α0 represents the channel function gain when the reference distance is 1m;

[0140] Furthermore, if user k is served by edge server m in time slot n, when a k,m When (t) = 1, user k is served by edge server m in time slot n, and the corresponding signal-to-noise ratio of UAVm is expressed as:

[0141]

[0142] Among them, p k represents the transmission power of intelligent terminal k, σ 2 represents the power of additive white Gaussian noise at the UVA receiver, represents the interference caused by all intelligent terminal transmissions in time slot n;

[0143] In this case, the unloading rate of intelligent terminal k in time slot n is expressed as:

[0144]

[0145] S4: Prioritize tasks and obtain task processing delays of different levels.

[0146] It should be noted that the priority classification of tasks includes the definition of power business priority. Too many levels of task priority division affect the efficiency of resource allocation, and too few levels cannot fully reflect the effectiveness of the priority. The present invention divides it into three levels according to the power business response time and uses the corresponding digital quantitative identification γ. Tasks that take response measures or have high requirements for real-time transmission are defined as high priority, generally referring to the business in the response time period, γ=3. There is a certain time tolerance from task generation to response, which is defined as medium priority γ=2. Subsequent analysis work and tasks that do not require response measures are defined as low priority γ=1;

[0147] Furthermore, from the perspective of optimizing the time and power consumption of the system in processing power services, the optimal processing method of the task is selected between local processing and offloading to the edge server for processing. After the task is offloaded, it needs to queue in the server buffer for processing. In the model proposed by the present invention, the final computing resource allocation coefficient of the edge server needs to be weighted by the task priority, and the processing needs to be diverted in the order of high, medium and low priority. The edge server has abundant processing resources and the data volume of the final processing result is small. Therefore, the computing power consumption of the edge server and the time for the processing result to be transmitted back to the terminal are negligible. The average arrival amount of the priority task is expressed as:

[0148] Where λ represents the average arrival rate of terminal power tasks, λ y represents the arrival rate of power tasks with different priorities, a n,ar represents the average arrival rate of tasks that are offloaded in time slot n, satisfying λ1+λ2+λ3=λ γ ,surface The average processing time of tasks, the average processing time of tasks generated by medium-priority services, and the average processing time of tasks generated by low-priority services;

[0149] The average processing time of tasks generated by high-priority services is expressed as:

[0150]

[0151] in, Indicates the average processing time of tasks generated by high-priority services, Represents the digital twin’s estimate of the true computation frequency;

[0152] The average processing time of tasks generated by medium-priority services is expressed as:

[0153]

[0154] in, Indicates the average processing time of tasks generated by medium priority business;

[0155] The average processing time of tasks generated by low-priority services is expressed as:

[0156]

[0157] in, Indicates the average processing time of tasks generated by low-priority services.

[0158] S5: Based on the optimized dispatching scenario of the new energy power system under the digital twin platform, a local computing model for unloading decision-making and computing resource allocation is constructed.

[0159] It should be noted that the local computing model of computing resource allocation includes computing delay and energy consumption. When the smart terminal decides local computing, the computing delay is expressed as:

[0160]

[0161] The total delay of the smart terminal processing tasks on the local device is expressed as:

[0162]

[0163] Energy consumption is expressed as:

[0164]

[0165] in, is the local computing capability of the time slot intelligent terminal k, and k1 is the power consumption factor;

[0166] When the intelligent terminal is in time slot n, the computing power is expressed as:

[0167]

[0168] in, Indicates the maximum local computing capability of the smart terminal.

[0169] S6: Based on the optimized dispatching scenario of the new energy power system under the digital twin platform, an unloading model for user unloading decision-making and computing resource allocation is constructed.

[0170] It should be noted that the local computing model of computing resource allocation includes computing delay and energy consumption. When the smart terminal k offloads the task to the edge server in time slot n, the offloading delay is expressed as:

[0171]

[0172] The unloading energy consumption is expressed as:

[0173]

[0174] When smart terminal k offloads the task to the edge server in time slot n, the actual computing delay is expressed as:

[0175]

[0176] The total delay is expressed as:

[0177]

[0178] The energy consumption is calculated as:

[0179]

[0180] The constraints of the edge server are expressed as:

[0181]

[0182] in, It is the maximum computing resource that can be allocated to each edge server.

[0183] S7: Solve user offloading decisions and calculate resource allocation based on DRL.

[0184] It should be noted that, considering the delay constraint, the offloading tasks are reasonably allocated to the edge servers, and the total energy consumption of the network is minimized by optimizing the allocation of computing resources. The offloading decision includes defining Computing resources The optimization problem model is expressed as:

[0185]

[0186] Among them, the problem P represents MDP, and the optimal strategy is learned in the DRL method training environment to solve the problem of minimizing energy consumption. Figure 3 It can be seen that the agent continuously interacts with the dynamic environment to optimize its own strategy. At time step n, the environment is in a certain state s n , the agent performs action a n , the environment transfers to any feasible successor state s with a certain probability n+1 In the example, the agent receives a reward r n , followed by n+1. The agent observes the state s n+1 With reward n To adjust its own strategy to maximize the accumulated rewards. In this process, the state space, action space and reward function are three key elements;

[0187] When the time step is n, the environment is in S n state, the agent performs action a n , the environment moves to the subsequent state s n+1 In the example, the agent receives a reward r n ;

[0188] Network Status n It is expressed as:

[0189]

[0190] When the state s in time slot n n , the DRL algorithm obtains the action space a according to the state n It is expressed as:

[0191]

[0192] Reward function r n It is expressed as:

[0193]

[0194] Furthermore, the PPO algorithm adopts the Actor-Critic structure. The Actor network is divided into two parts, the new and the old, including parameters θ and oldθ. The Critic network parameter is ξ. The Actor network is based on the state s. n Output action a n , and interact with the environment; the Critic network calculates the state value V based on the state information ξ (s n ), expressed as:

[0195]

[0196] Where γ is the discount factor, represents the expected value, represents the reward function;

[0197] The Actor network parameters θ and the Critic network parameters ξ are expressed as:

[0198]

[0199] L critic (ξ)=[V ξ (s n+1 )-V ξ (s n ] 2

[0200] Among them, π θ (·)and Represents the new and old policy functions, ε represents the truncation parameter;

[0201] Replace the old θ with the new parameter θ old , expressed as:

[0202] θ old =θ;

[0203] Furthermore, by Figure 4 It can be seen that between 0 and 100 rounds, the algorithm shows a large volatility. This is because the algorithm has a large exploratory nature in the initial stage. As the number of rounds increases, the exploratory nature gradually decreases, resulting in a decrease in the volatility of the algorithm. When the number of rounds reaches 300, the algorithm begins to converge, which shows that as the algorithm goes through a certain number of rounds, it gradually stabilizes and converges to a better solution.

[0204] The above is a schematic scheme of a new energy power optimization dispatching method under a digital twin platform of this embodiment. It should be noted that the technical scheme of the system of the new energy power system optimization dispatching method under the digital twin platform and the technical scheme of the new energy power optimization dispatching method under the above-mentioned digital twin platform belong to the same concept. The details not described in detail in the technical scheme of the new energy power system optimization dispatching system under the digital twin platform in this embodiment can be referred to the description of the technical scheme of the new energy power optimization dispatching method under the above-mentioned digital twin platform.

[0205] In this embodiment, a new energy power system optimization and dispatching system under a digital twin platform includes: a data collection module, a system scenario construction module, an edge network construction module, a communication and task module, a priority processing module, a local computing module, and a user unloading decision module;

[0206] The data collection module obtains necessary data from a third party to provide basic data support for the digital twin platform to establish an optimized dispatching scenario for the new energy power system;

[0207] The system scenario building module builds detailed digital twin platform scenarios based on data to reflect the real-time operating status of the power system and potential optimization scheduling needs;

[0208] The edge network construction module establishes an edge network model under the digital twin platform to simulate and analyze the optimization scheduling effect of the energy system on edge computing resources;

[0209] The communication and task module designs the communication model and the task model, completes the offloading decision and the computing resource allocation, and improves the system efficiency and response speed;

[0210] The priority processing module classifies tasks according to their priorities and schedules them according to different processing delay requirements to ensure that high-priority tasks are processed in a timely manner;

[0211] The local computing module establishes a local computing model, optimizes decision-making and resource allocation strategies, reduces communication delays and energy consumption, and improves overall system performance;

[0212] The user unloading decision module solves the user unloading decision and calculates resource allocation based on the DRL method to realize intelligent energy system optimization scheduling management.

[0213] This embodiment also provides a computing device, which is applicable to a method for optimizing the dispatching of new energy power under a digital twin platform, including:

[0214] Memory and processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement a new energy power optimization scheduling method under a digital twin platform as proposed in the above embodiment.

[0215] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, a new energy power optimization scheduling method under a digital twin platform proposed in the above embodiment is implemented.

[0216] The storage medium proposed in this embodiment and the new energy power optimization scheduling method under a digital twin platform proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0217] If the 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 methods 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 codes.

[0218] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0219] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0220] Example 2, reference Figure 5 Based on the previous embodiment, this embodiment provides a new energy power optimization scheduling method under a digital twin platform. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0221] Depend on Figure 5 It can be seen that the simulation results show that as the number of terminal devices increases, the total energy consumption of the network presents an upward trend. This is because when the number of terminal devices increases, the time slot allocated to each terminal device becomes smaller, so the terminal device needs to use a larger power to transmit tasks, so the total energy consumption of the user increases. When the number of terminal devices is the same, the total system energy consumption of the algorithm proposed in the present invention is always lower than the three benchmark schemes of the DQN algorithm, the non-optimized computing resource algorithm design, and the greedy algorithm design. This is because the algorithm proposed in the present invention adopts the on-policy strategy to optimize the offloading decision and computing resource allocation. Among all the algorithms, the greedy algorithm has the highest energy consumption, because the greedy algorithm does not optimize the offloading strategy. The DQN algorithm optimizes the offloading strategy and computing resource allocation at the same time, so the total system energy consumption is lower than the greedy algorithm and the non-optimized computing resource algorithm.

[0222] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A new energy power optimization dispatching method under a digital twin platform, characterized by: include: Based on the data collected by the third party, build the optimization dispatch scenario of the new energy power system under the digital twin platform; Establish a digital twin edge network model based on the optimized dispatching scenario of the new energy power system under the digital twin platform; According to the optimization dispatching scenario of the new energy power system under the digital twin platform, the communication model and task model for unloading decision and computing resource allocation are established; Prioritize tasks and obtain different levels of task processing delays; According to the optimized dispatching scenario of the new energy power system under the digital twin platform, a local computing model for unloading decision-making and computing resource allocation is constructed; According to the optimized dispatching scenario of the new energy power system under the digital twin platform, an unloading model for user unloading decision-making and computing resource allocation is constructed; Solve user offloading decisions and computing resource allocation based on DRL.

2. The method for optimizing and dispatching new energy power under a digital twin platform according to claim 1, characterized in that: The construction of the digital twin platform includes generating computing tasks based on K smart terminals in the energy and power system. The tasks are not related to each other. Environmental information data is obtained by using data collected by a third party. The coordinates of the smart terminals are expressed as: There are N time slots in the period T, n represents the number of each time slot, and the smart terminal in each time slot offloads the computing task to the edge server or performs it locally. The edge server provides offloading service, and the fixed coordinates are expressed as: Among them, the number of edge servers is M, and the number of the mth edge server is 3. The method for optimizing and dispatching new energy power under a digital twin platform according to claim 2, characterized in that: The digital twin edge network model established includes a two-layer digital twin edge network model of a physical entity layer and a digital twin layer, wherein K smart terminals and M edge servers constitute the physical entity layer, and the digital twins of all physical entities and the communication environment constitute the digital twin layer; according to the characteristics of the smart terminal offloading to the edge server or local computing, the offloading decision variable is defined It is expressed as: Among them, when When n time slot intelligent terminal k offloads the computing task to edge server m, When indicates that the n-time slot intelligent terminal k chooses local calculation; The intelligent terminal in the time slot is unloaded to the edge server, which is expressed as: At time slot n, the digital twin of the computational entity with respect to its computational frequency is represented as: in, represents the digital twin’s estimate of the true calculation frequency, Represents the error between the true calculated frequency and the digital twin estimate.

4. The method for optimizing and dispatching new energy power under a digital twin platform according to claim 3, characterized in that: The communication model and task model of computing resource allocation include using intelligent terminal K to generate computing tasks, which are expressed as: Where n represents time slot n, Indicates the size of the task. Indicates the number of CPUs required for the calculation task. Indicates the task delay limit; The error between the actual calculation delay and the digital twin estimated delay is expressed as: Calculate the distance between devices in time slot n, expressed as: Among them, the distance between the Kth intelligent terminal and the mth edge server is expressed as When time slot t, the channel gain from the kth intelligent terminal to the mth edge server is expressed as: Among them, α0 represents the channel function gain when the reference distance is 1m; when a k,m (t) = 1, user k is served by edge server m in time slot n, and the corresponding signal-to-noise ratio of UAVm is expressed as: Among them, p k represents the transmission power of intelligent terminal k, σ 2 represents the power of additive white Gaussian noise at the UVA receiver, represents the interference caused by the transmission of all intelligent terminals in time slot n; The unloading rate of intelligent terminal k in time slot n is expressed as:

5. The method for optimizing and dispatching new energy power under a digital twin platform according to claim 4, characterized in that: The task priority classification includes defining the tasks according to the priority of the power business, taking response measures or requiring high real-time transmission, which are defined as high priority γ=3, tasks with time tolerance from generation to response, which are defined as medium priority γ=2, and subsequent analysis work and tasks that do not require response measures, which are defined as low priority γ=1; the average arrival amount of priority tasks is expressed as: Where λ represents the average arrival rate of terminal power tasks, λ y represents the arrival rate of power tasks with different priorities, a n,ar It represents the average arrival rate of the tasks that are offloaded in time slot n, expressed as: Satisfy λ1+λ2+λ3=λ γ ; Obtaining different levels of task processing delays includes calculating the average processing time of tasks generated by high-priority services, the average processing time of tasks generated by medium-priority services, and the average processing time of tasks generated by low-priority services; The average processing time of tasks generated by high-priority services is expressed as: in, Indicates the average processing time of tasks generated by high-priority services, Represents the digital twin’s estimate of the true computation frequency; The average processing time of tasks generated by medium-priority services is expressed as: in, Indicates the average processing time of tasks generated by medium priority business; The average processing time of tasks generated by low-priority services is expressed as: in, Indicates the average processing time of tasks generated by low-priority services.

6. The method for optimizing and dispatching new energy power under a digital twin platform according to claim 5, characterized in that: The local computing model of computing resource allocation includes computing delay and energy consumption. When the smart terminal decides to perform local computing, the computing delay is expressed as: The total delay of the smart terminal processing tasks on the local device is expressed as: Energy consumption is expressed as: in, is the local computing capability of the time slot intelligent terminal k, and k1 is the power consumption factor; When the intelligent terminal is in time slot n, the computing power is expressed as: in, Indicates the maximum local computing capacity of the smart terminal; When smart terminal k offloads the task to the edge server in time slot n, the offloading delay is expressed as: The unloading energy consumption is expressed as: When smart terminal k offloads the task to the edge server in time slot n, the actual computing delay is expressed as: The total delay is expressed as: The energy consumption is calculated as: The constraints of the edge server are expressed as: in, It is the maximum computing resource that can be allocated to each edge server.

7. The method for optimizing and dispatching new energy power under a digital twin platform according to claim 6, characterized in that: The uninstallation decision includes defining Computing resources The optimization problem model is expressed as: Among them, the problem P represents MDP, and the optimal strategy is learned in the training environment through the DRL method to solve the problem of minimizing energy consumption. When the time step is n and the environment is in S n state, the agent performs action a n , the environment moves to the subsequent state s n+1 In the example, the agent receives a reward r n ; Network Status n It is expressed as: When the state s in time slot n n , the DRL algorithm obtains the action space a according to the state n It is expressed as: Reward function r n It is expressed as: The PPO algorithm uses an Actor-Critic structure. The Actor network is divided into two parts, old and new, including parameters θ and oldθ. The Critic network parameter is ξ. The Actor network changes according to the state s. n Output action a n , interact with the environment; the Critic network calculates the state value V based on the state information ξ (s n ), expressed as: Where γ is the discount factor, represents the expected value, represents the reward function; Actor network parameters θ and Critic network parameters ξ are expressed as: L critic (ξ)=[V ξ (s n+i )-V ξ (s n )] 2 Among them, π θ (·)and Represents the new and old policy functions, ε represents the truncation parameter; Replace the old θ with the new parameter θ old , expressed as: i old =θ.

8. A system for optimizing the dispatching of new energy power under a digital twin platform according to any one of claims 1 to 7, characterized in that: include: Data collection module, system scenario building module, edge network building module, communication and task module, priority processing module, local computing module and user uninstallation decision module; The data collection module obtains necessary data from a third party to provide basic data support for the digital twin platform to establish an optimized dispatching scenario for the new energy power system; The system scenario building module builds detailed digital twin platform scenarios based on data to reflect the real-time operating status of the power system and potential optimization scheduling needs; The edge network construction module establishes an edge network model under the digital twin platform to simulate and analyze the optimization scheduling effect of the energy system on edge computing resources; The communication and task module designs the communication model and the task model, completes the offloading decision and the computing resource allocation, and improves the system efficiency and response speed; The priority processing module classifies tasks according to their priorities and schedules them according to different processing delay requirements to ensure that high-priority tasks are processed in a timely manner; The local computing module establishes a local computing model, optimizes decision-making and resource allocation strategies, reduces communication delays and energy consumption, and improves overall system performance; The user unloading decision module solves the user unloading decision and calculates resource allocation based on the DRL method to realize intelligent energy system optimization scheduling management.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, it implements the steps of a method for optimizing and scheduling new energy power under a digital twin platform as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for optimizing and scheduling new energy power under a digital twin platform described in any one of claims 1 to 7 are implemented.

Images