New energy electric field cooperative scheduling method and system based on D2D assistance

Abstract

The invention relates to the technical field of edge computing, and discloses a D2D assistance-based new energy electric field collaborative scheduling method and system, and the method comprises the steps: constructing a new energy electric field physical topology network, designing an entropy weight method-based equipment scheduling priority formulation strategy, and determining a priority scheduling sequence of equipment; and a D2D-based adaptive collaborative data scheduling processing model is constructed, and support is provided for medium and long term planning of a power grid and collaborative scheduling optimization of multi-source data in a new energy electric field. And further constructing a random optimization problem of task data processing delay minimization, dynamically processing long-term energy consumption constraints of user equipment by adopting a Lyapunov optimization technology, and solving an optimization strategy of data scheduling and resource allocation among multiple devices by applying an MADDPG algorithm. According to the invention, by fusing D2D and edge computing technologies, efficient collaborative scheduling of multi-source data in the new energy electric field is realized, data processing time delay is effectively reduced, and important technical support is provided for power grid planning and intelligent management of the new energy electric field.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and in particular to a D2D-assisted new energy electric field collaborative scheduling method and system. Background Art

[0002] With the rapid development of renewable energy around the world, the improvement of new energy power generation capacity has put forward higher requirements for the stability and optimized dispatching of the power grid system. New energy power generation equipment, especially in wind power and photovoltaic power plants, often brings challenges to the load dispatching of the power grid due to its power generation volatility and uncertainty; in the process of medium- and long-term power grid planning and new energy power generation capacity forecasting, how to accurately predict meteorological changes and new energy power generation capacity, and based on these forecasts, conduct intelligent dispatching and resource optimization of the power grid, is an urgent problem to be solved.

[0003] Traditional grid planning and dispatching methods are often extrapolated based on static models of historical data. However, this method ignores the dynamic changes in meteorology and renewable energy power generation capacity, and cannot respond to emergencies or demand changes in the grid in real time. To address these issues, medium- and long-term meteorological forecasting systems and renewable energy power generation capacity forecasting systems have become an important basis for grid planning. Through more accurate meteorological forecasts and power generation forecasts, not only can the scientific nature and foresight of grid planning be improved, but data support can also be provided for intelligent dispatching of the grid and optimal allocation of energy resources.

[0004] Although some research progress has been made in the field of weather forecasting and new energy power generation capacity forecasting in recent years, it is still a huge challenge to effectively integrate weather data, power generation capacity data and the complex relationship between them and grid load when faced with multi-source and heterogeneous data. Existing forecasting technologies are mostly based on centralized data processing architectures, which transmit all data to the cloud for processing. However, this method has problems such as data transmission delays and centralized computing resources. Especially in the application scenarios of medium- and long-term forecasting, real-time performance and computing power have become constraints.

[0005] To overcome these challenges, edge computing technology deploys computing resources at the edge nodes of the power grid, which can offload data processing tasks from the centralized cloud to edge nodes close to the data source for processing, thereby significantly reducing data transmission latency and computing burden. In addition, D2D technology allows direct communication and data transmission between devices, reducing the latency and load of the central server, and ensuring that devices can analyze data in a timely manner and make decisions and responses quickly. Based on this, the present invention proposes a D2D-assisted new energy electric field collaborative scheduling system, which deploys an intelligent scheduling algorithm on each device, making each device an independent intelligent entity for collaborative data scheduling and resource optimization. Summary of the invention

[0006] In view of the above existing problems, the present invention is proposed.

[0007] Therefore, in view of the shortcomings of the prior art in the scheduling of new energy electric fields, the present invention provides a D2D-assisted new energy electric field collaborative scheduling method and system, which optimizes collaborative data scheduling and resource allocation by combining edge computing and D2D technology to achieve the goal of minimizing the task data processing delay.

[0008] In order to achieve the above object, the present invention provides the following technical solutions: In the first aspect, the present invention provides a D2D-assisted collaborative scheduling method for new energy electric fields, including: constructing a new energy electric field physical topology network based on the equipment location information of the new energy electric field; designing a device scheduling priority formulation strategy based on the entropy weight method based on the physical topology network of the new energy electric field, and determining the priority scheduling order of the equipment; constructing an adaptive collaborative data scheduling processing model based on D2D through the device scheduling priority formulation strategy; constructing a random optimization problem for minimizing the task data processing delay based on the adaptive collaborative data scheduling processing model; using Lyapunov optimization technology to dynamically process the long-term energy consumption constraints of user equipment, and decomposing the random optimization problem into a deterministic optimization problem by time slot; applying the MADDPG algorithm to solve the optimization strategy for data scheduling and resource allocation among multiple devices, and generate a specific data scheduling and resource allocation plan.

[0009] As a preferred solution of the D2D-assisted new energy electric field coordinated scheduling method described in the present invention, the construction of the new energy electric field physical topology network includes: Obtain the location information of user equipment, power generation equipment, transmission equipment, meteorological monitoring equipment, energy storage equipment and edge servers in the new energy electric field, and obtain the user equipment set N, energy infrastructure equipment set L, and edge server set M. Combine N and L to obtain the equipment set H. The three-dimensional coordinates of the equipment are expressed as , , the three-dimensional coordinates of the edge server are expressed as , ; Based on the three-dimensional location information of each device and edge server in the new energy electric field, combined with the spatial distribution relationship of the actual deployment scenario, a complete new energy electric field physical topology network is constructed.

[0010] As a preferred solution of the D2D-assisted new energy electric field coordinated scheduling method described in the present invention, the device scheduling priority formulation strategy designed based on the entropy weight method includes: Based on the physical topology network of the new energy electric field, the entire scheduling activity cycle is divided into F time frames using TDMA technology, and each time frame is divided into T fixed time lengths. time slots, where T is equal to the number of devices H; The task data generated by the i-th device in the t-th time slot is represented as a triple ,in Indicates the size of the generated task data. Indicates the number of CPU cycles required to process 1 bit of data. Represents the delay constraint of data processing; definition Indicates the remaining power of the device. The energy infrastructure equipment is always kept fully charged. The maximum power of the user device is expressed as ; The data cache queue is introduced to cache unprocessed data. The data cache queue of the i-th device in the t-th time slot The formula is: , in, represents the proportion of data calculated locally by device i in the tth time slot, represents the data ratio to be scheduled; there is a constraint between the two ratios of scheduling device i in the tth time slot ; The data processing delay limit is dynamically accumulated according to two ratios, expressed as: , in, express The corresponding data processing delay limit size; Construct task load, task importance, device remaining energy and distance indicators to reflect the device's task processing requirements, resource constraints and scheduling importance from multiple perspectives; task load is expressed as ,in represents the upper limit of the data cache queue; the task importance is expressed as ; The remaining energy of the equipment is expressed as ,right ,have ; The distance indicator is divided into the distance between the device and the user's device and the distance between the device and the edge server Two types; After completing the definition and normalization of each indicator, obtain the indicator set; The entropy weight method is used to assign weights to the indicator set, and the scheduling priority of the device is determined based on the calculated comprehensive score; based on the indicator matrix W of the device, the normalized weight of each device on each indicator is calculated ; Calculate the entropy value of each indicator according to the definition of information entropy; After obtaining the entropy value of each indicator, the information weight of each indicator is calculated, and then the final weight is obtained through normalization; Calculate the comprehensive score of each device based on the weight distribution , and a comprehensive score for each device By sorting, the scheduling priority of the equipment can be determined. Indicates the device number that ranks highest in a time slot and has not been scheduled in this time frame.

[0011] As a preferred solution of the D2D-assisted new energy electric field coordinated scheduling method described in the present invention, the construction of the D2D-based adaptive coordinated data scheduling processing model includes: Based on the device number, record , Indicates the device The uninstall decision represents a set of computing nodes. If , it means that the ratio is The data is unloaded to device s, and vice versa; a device can only be unloaded to one computing node in a time slot, and the unloaded computing node cannot be the device itself; only one device is scheduled in the tth time slot, and the interference between devices can be ignored; The computational delay required for device i to perform local computing tasks in the tth time slot is expressed as , the transmission delay required for device i to offload data to the computing node is expressed as , the computational delay required for computing node s to compute the data offloaded by device i is expressed as , then the total delay of task processing of device i in the tth time slot can be expressed as ; The local computing energy consumption of the i-th device in the t-th time slot is expressed as , the transmission energy consumption required by device i to offload data to the computing node is expressed as , the computing energy consumption required by computing node s to calculate the data unloaded by device i is expressed as , then the total energy consumed by the i-th user equipment in the t-th time slot is expressed as , without considering the energy consumption of energy infrastructure equipment.

[0012] As a preferred solution of the D2D-assisted new energy electric field coordinated scheduling method described in the present invention, the stochastic optimization problem of minimizing the delay of constructing task data processing includes: , , in, , , They represent task scheduling decisions, computing ratios, and resource allocation strategies, respectively. Indicates the CPU computing frequency assigned to device i when performing local computing tasks. Indicates the maximum CPU computing frequency of the user device. Indicates the CPU computing frequency that the computing node allocates to the task data offloaded from the device. Indicates the maximum CPU computing frequency of the edge server; where the constraint represents the device's offloading decision variable, constraint Indicates that in the tth time slot, the scheduling device can only unload task data to one computing node, constraining Indicates that the uninstall object cannot be the scheduling device itself, constraint Represents the constraint on the proportion of local computing tasks on user devices. Represents the constraint on the ratio of device offload tasks. Represents the constraint on the sum of the local computing ratio and the offloading ratio. Indicates the constraints on the local CPU computing frequency of the user device. Indicates the constraints on the CPU computing frequency of the computing node. It represents the long-term energy consumption constraint, i.e., the energy consumption of the user equipment during the entire scheduling activity period.

[0013] As a preferred solution of the D2D-assisted new energy electric field coordinated scheduling method described in the present invention, decomposing the stochastic optimization problem into a time slot-by-time slot deterministic optimization problem includes: Build an energy virtual queue for each user device according to the constraints ; A Lyapunov function is defined to describe the change of the energy virtual queue of all user equipments in the t-th time slot over time, and a Lyapunov drift is obtained by calculating the expectation of the difference between adjacent time slots of the Lyapunov function; In order to minimize the Lyapunov drift, the control strategy is optimized by introducing a penalty term, and the Lyapunov drift plus penalty function is obtained: , and then transform the optimization problem P1 into a deterministic optimization problem P2 without long-term user equipment energy constraints, and the formula is expressed as: .

[0014] As a preferred solution of the D2D-assisted new energy electric field coordinated scheduling method described in the present invention, the optimization strategy of using the MADDPG algorithm to solve data scheduling and resource allocation among multiple devices includes: Define the core elements of the reinforcement learning algorithm environment state , Action Space and rewards ; By deploying algorithms on each device, each device is regarded as an independent intelligent agent, where the device is the physical entity of the intelligent agent and provides Construct four groups of deep neural networks, namely Actor strategy network , Actor target network , Critic Value Network and Critic target network ,in and Respectively represent the status and actions of other devices except device i, and initialize network parameters , , and , initialize the experience replay pool ; In the tth time slot, agent i is in the environment Execute actions in , get reward And enter the next state ;remember , , , thus recording Stored in the experience replay pool middle; When the experience replay pool When the number of samples in is greater than the number of nodes, K groups of samples are randomly selected to train the agent. First, a group of samples is taken , and calculate the target value of agent i for this sample ; For each agent i, the value network parameters are updated using the batch update method; Adopt deterministic policy gradient method to update policy network parameters by gradient ascent , to maximize the policy objective function; use soft update to update the target network parameters of each agent; Repeat the above steps until the algorithm converges or reaches the predetermined number of training times.

[0015] In a second aspect, the present invention provides a D2D-assisted new energy electric field coordinated dispatching system, comprising: A topology network construction module is used to construct a physical topology network of a new energy electric field according to the equipment location information of the new energy electric field; A scheduling processing module, which is used to design a device scheduling priority formulation strategy based on the entropy weight method based on the physical topology network of the new energy electric field, and determine the priority scheduling order of the devices; and to construct an adaptive collaborative data scheduling processing model based on D2D through the device scheduling priority formulation strategy; An optimization problem processing module is used to construct a stochastic optimization problem for minimizing task data processing delay based on the adaptive collaborative data scheduling processing model; Lyapunov optimization technology is used to dynamically process the long-term energy consumption constraints of user equipment, and the stochastic optimization problem is decomposed into a deterministic optimization problem per time slot; The optimization scheme determination module is used to apply the MADDPG algorithm to solve the optimization strategy of data scheduling and resource allocation among multiple devices and generate specific data scheduling and resource allocation schemes.

[0016] In a third aspect, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the D2D-assisted new energy electric field coordinated scheduling method are implemented.

[0017] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the D2D-assisted new energy electric field coordinated scheduling method.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: ① By combining edge computing and D2D technology, the present invention can realize intelligent data collaborative scheduling and resource optimization in the new energy electric field, reduce data transmission delay, and ensure dynamic strategy adjustment of power grid load and energy storage equipment; ② Construct a random optimization problem to minimize the task data processing delay, dynamically process the long-term energy consumption constraints of user equipment through Lyapunov optimization technology, and ensure that while meeting the energy consumption constraints of user equipment, the optimization of task collaborative scheduling and resource allocation is achieved; ③ The present invention adopts the MADDPG algorithm, regards each device as an intelligent agent, and realizes collaborative optimization scheduling between devices through reinforcement learning. Each intelligent agent dynamically adjusts the scheduling strategy through learning and feedback, further improving the intelligent level of power grid scheduling, especially when the fluctuation of new energy power generation is large, it can better adapt to changes and respond quickly; ④ Through real-time data processing and intelligent collaborative scheduling, the present invention can effectively deal with the uncertainty and volatility of new energy power generation, optimize the use of power grid resources, ensure the stability of the power grid under different operating conditions, and improve the overall operating efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] 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.

[0020] Figure 1 This is a schematic diagram of the overall process logic of a D2D-assisted new energy electric field coordinated scheduling method according to an embodiment of the present invention; Figure 2 A schematic diagram of a D2D-assisted collaborative scheduling optimization model in a new energy electric field of a D2D-assisted new energy electric field collaborative scheduling method according to an embodiment of the present invention; Figure 3 A network architecture diagram of a MADDPG algorithm of a D2D-assisted new energy electric field coordinated scheduling method according to an embodiment of the present invention; Figure 4 A schematic diagram of convergence performance of solving data scheduling and resource allocation using a MADDPG algorithm based on a D2D-assisted new energy electric field coordinated scheduling method according to an embodiment of the present invention; Figure 5 This is a comparison chart of data processing delays of different algorithms of a D2D-assisted new energy electric field coordinated scheduling method under different numbers of devices according to an embodiment of the present invention; Figure 6 This is a comparison chart of data processing delays of different algorithms of the D2D-assisted new energy electric field coordinated scheduling method at different maximum computing frequencies described in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] 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.

[0022] Example 1 Reference Figure 1-Figure 3 As an embodiment of the present invention, a D2D-assisted new energy electric field coordinated scheduling method is provided, such as Figure 2 As shown in the figure, it consists of N user devices, L energy infrastructure devices, and M edge servers, which aims to optimize the scheduling and processing analysis of various types of data generated by different devices in the new energy power field to achieve the goal of minimizing the task data processing delay. Figure 1 The specific steps shown include: S100: construct a new energy electric field physical topology network according to the location information of power generation equipment, transmission equipment, meteorological monitoring equipment, energy storage equipment, user equipment and edge servers in the new energy electric field; S200: Design a strategy for equipment scheduling priority based on the entropy weight method based on the physical topology network of the new energy electric field, and determine the priority scheduling order of equipment by comprehensively considering the task load, task importance and distance indicators; S300: Building a D2D-based adaptive collaborative data scheduling processing model through device scheduling priority formulation strategy; S400: Construct a stochastic optimization problem for minimizing task data processing delay based on an adaptive collaborative data scheduling processing model; S500: Adopts Lyapunov optimization technology to dynamically handle the long-term energy consumption constraints of user equipment, decomposes the random optimization problem into a time slot-by-time slot deterministic optimization problem, and copes with the uncertainty of the dynamic environment by gradually approaching the global optimal solution, achieving a balance between system performance and optimization goals; S600: Apply the MADDPG algorithm to solve the optimization strategy of data scheduling and resource allocation among multiple devices, and generate a specific data scheduling and resource allocation plan.

[0023] It should be noted that ① by combining edge computing and D2D technology, the present invention can realize intelligent data collaborative scheduling and resource optimization in the new energy electric field, reduce data transmission delay, and ensure dynamic strategy adjustment of power grid load and energy storage equipment; ② construct a random optimization problem to minimize the task data processing delay, and dynamically process the long-term energy consumption constraints of user equipment through Lyapunov optimization technology to ensure that while meeting the energy consumption constraints of user equipment, the optimization of task collaborative scheduling and resource allocation is achieved; ③ the present invention adopts the MADDPG algorithm, regards each device as an intelligent agent, and realizes collaborative optimization scheduling between devices through reinforcement learning. Each intelligent agent dynamically adjusts the scheduling strategy through learning and feedback, further improving the intelligent level of power grid scheduling, especially when the fluctuation of new energy power generation is large, it can better adapt to changes and respond quickly; ④ through real-time data processing and intelligent collaborative scheduling, the present invention can effectively deal with the uncertainty and volatility of new energy power generation, optimize the use of power grid resources, ensure the stability of the power grid under different operating conditions, and improve the overall operating efficiency.

[0024] In the embodiment of the present application, the above step S100 specifically includes the following sub-steps A1 to A3: In A1: Based on the data collected by the third party, the location information of user equipment, power generation equipment, transmission equipment, meteorological monitoring equipment, energy storage equipment and edge servers in the new energy power field is obtained to obtain the user equipment set 、Energy infrastructure collection , Edge Server Collection ,right and Merge to get device collection ,in ; In A2: The three-dimensional coordinates of the device are expressed as , , the three-dimensional coordinates of the edge server are expressed as , ; In A3: Based on the three-dimensional location information of each device and edge server in the new energy electric field, combined with the spatial distribution relationship of the actual deployment scenario, a complete new energy electric field physical topology network is constructed.

[0025] In the embodiment of the present application, the above step S200 specifically includes the following sub-steps B1 to B5: In B1: Based on the physical topology network of the new energy electric field, the TDMA technology is used to divide the entire scheduling activity cycle into F time frames, each time frame is divided into T fixed time lengths. The time slot is expressed as , where T is equal to the number of devices H, which means that within a time frame, each device will be assigned a unique time slot for task scheduling; In B2: The task data generated by the i-th device in the t-th time slot is represented as a triple , ,in Indicates the size of the generated task data. Indicates the number of CPU cycles required to process 1 bit of data. represents the delay constraint of data processing; without loss of generality, assume It obeys normal distribution and is expressed as ,in represents a normal distribution, and Represent the mean and variance respectively; definition Indicates the remaining power of the device. The energy infrastructure equipment is always kept fully charged. The maximum power of the user device is expressed as ; In B3: Due to the limited computing power of the device, the task data may not be processed in a single time slot, so a data cache queue is introduced. Used to cache unprocessed data, the queue stores and retrieves data in FIFO mode; in the tth time slot, the data cache queue of the i-th device The derivation formula is: , in, represents the proportion of data calculated locally by device i in the tth time slot. Since the hardware design of energy infrastructure equipment usually does not have the computing power of high-energy sources, it is impossible to perform local computing. Only user devices can perform local computing. ,have ; Indicates the proportion of scheduled data, ; There is a constraint between the two ratios of scheduling device i in the tth time slot ; The data processing delay limit is dynamically accumulated according to two ratios, expressed as: , in, , express The corresponding data processing delay limit size; In B4: To formulate a reasonable device scheduling priority, task load, task importance, device remaining energy and distance indicators are constructed to reflect the device's task processing requirements, resource constraints and scheduling importance from multiple perspectives; the task load is expressed as ,in represents the upper limit of the data cache queue; the task importance is expressed as ; The remaining energy of the equipment is expressed as ,right ,have ; The distance indicator is divided into the distance between the device and the user's device and the distance between the device and the edge server Two types, by The distance matrix V is expressed as: , in, , Indicates the distance between each device and its nearest neighboring user device. Normalization can be obtained ,in Indicates the maximum distance between the device and the user's device; The distance matrix U is expressed as: , in, , Represents the distance between each device and its nearest edge server. Normalization can be obtained ,in Represents the maximum distance between the device and the edge server; after completing the definition and normalization of each indicator, the indicator set is obtained as , the indicator matrix of the equipment is expressed as: , in, It represents the normalized value of the rth indicator of the device at the tth time slot; In B5: the entropy weight method is used to assign weights to the indicator set, and the scheduling priority of the device is determined based on the calculated comprehensive score; based on the indicator matrix W of the device, the normalized weight of each device on each indicator is calculated , expressed as: , Calculate the entropy value of each indicator according to the definition of information entropy , to measure the information uniformity of the indicator, the formula is expressed as: , After obtaining the entropy value of each indicator, Calculate the information weight of each indicator, and then obtain the final weight through normalization , the formula is: , Calculate the comprehensive score of each device based on the weight distribution , the formula is: , Comprehensive score for each device By sorting, the scheduling priority of the equipment can be determined. Indicates the device number that ranks highest in a time slot and has not been scheduled in this time frame.

[0026] In the embodiment of the present application, the above step S300 is difficult to meet the processing delay requirements of task data due to the limited computing resources of the device, so D2D and edge computing technologies are introduced to construct data unloading models, communication models, delay models and energy consumption models in the coordinated scheduling of new energy power fields. Specifically, it includes the following sub-steps C1~C4: In C1: Based on the device number, record , Indicates the device The uninstall decision represents a set of computing nodes. If , it means that the ratio is The data of is uninstalled to device s, otherwise it is not uninstalled; device In a time slot, it can only be unloaded to one computing node, which is expressed as , the compute node to be uninstalled cannot be a device Itself, expressed as the following constraint: , in, Indicates when hour, ,otherwise ; In C2: only one device is scheduled in the tth time slot, and the interference between devices can be ignored. According to Shannon's formula, the device Compute Node The communication rate between can be expressed as: , in, represents the communication bandwidth, Indicates the transmit power of the device. represents the channel gain, represents the channel power gain per reference unit distance, represents the three-dimensional Euclidean distance between device i and computing node s, represents additive Gaussian white noise; In C3: The computational delay required for device i to perform local computing tasks in the tth time slot is expressed as , the formula is: , in, Indicates the CPU computing frequency assigned to device i when performing local computing tasks. Indicates the maximum CPU computing frequency of the user device. Since the energy infrastructure equipment does not have local computing capabilities, its local computing latency is set to 0; The transmission delay required for device i to offload data to the computing node is expressed as , the formula is: , in, It ensures that only the scheduled devices will have transmission delay, and the transmission delay of other devices is 0; The computational delay required for computing node s to compute the data offloaded by device i is expressed as , the formula is: , in, Indicates the CPU computing frequency allocated by the computing node for the task data offloaded to device i, subject to the following constraints: , in, Indicates the maximum CPU computing frequency of the edge server; The total delay of task processing of device i in the tth time slot can be expressed as: , In C4: In the tth time slot, the local computing energy consumption of the i-th device , calculated as follows: , in, Determined by the CPU parameters of the user device; Like the local computing latency, the local computing energy consumption of the energy infrastructure device is set to 0; The transmission energy consumption required for device i to offload data to the computing node , calculated as follows: , In the tth time slot, the computing energy required by computing node s to calculate the data unloaded by device i is , calculated as follows: , in, Determined by the CPU parameters of the edge server, Ensure that only the scheduled devices will have this part of computing energy consumption, and the rest of the devices will have 0. If the uninstall target is a device user, then , otherwise ; The remaining energy update formula of the user equipment is as follows: , in, represents the initial remaining energy of user equipment i, It represents the total energy consumed by the i-th user equipment in the t-th time slot, without considering the energy consumption of the energy infrastructure equipment.

[0027] In the embodiment of the present application, the above step S400 specifically includes the following sub-step D1: In D1: During the entire collaborative scheduling activity cycle, under the condition that idle user devices and edge servers provide computing resources, the present invention minimizes the data processing delay of each device task as much as possible, and then constructs the random optimization problem as follows: , , in, , , Respectively represent task scheduling decisions, computing ratios, and resource allocation strategies: , , , Among them, the constraints represents the device's offloading decision variable, constraint Indicates that in the tth time slot, the scheduling device can only unload task data to one computing node, constraining Indicates that the uninstall object cannot be the scheduling device itself, constraint Represents the constraint on the proportion of local computing tasks on user devices. Represents the constraint on the ratio of device offload tasks. Represents the constraint on the sum of the local computing ratio and the offloading ratio. Indicates the constraints on the local CPU computing frequency of the user device. Indicates the constraints on the CPU computing frequency of the computing node. It represents the long-term energy consumption constraint, i.e., the energy consumption of the user equipment during the entire scheduling activity period.

[0028] In the embodiment of the present application, the above step S500 specifically includes the following sub-steps E1 to E4: In E1: Build an energy virtual queue for each user device according to the constraints , expressed as: , in, represents the available energy allocated in a single time slot t, represents the energy deviation in the t-1th time slot; In E2: Define the Lyapunov function To describe the change of the energy virtual queue of all user devices in the tth time slot over time: , in, represents the energy virtual queue set of all user equipment in the tth time slot; by calculating the expectation of the difference between the Lyapunov function in adjacent time slots, the Lyapunov drift is obtained as: , In E3: In order to minimize the Lyapunov drift and reduce the backlog of the energy virtual queue, the control strategy is optimized by introducing a penalty term to obtain the Lyapunov drift plus penalty function : , Among them, V is a non-negative coefficient used to balance the weight between drift and objective function. The upper bound of Lyapunov drift penalty function can be expressed as: , in, is a constant; In E4: the optimization problem P1 is transformed into a deterministic optimization problem P2 without long-term user equipment energy constraints, and the formula is expressed as: , In the embodiments of the present application, Figure 3 The above step S600 specifically includes the following sub-steps F1 to F5: In F1: Define the core element of the reinforcement learning algorithm, the environment state , Action Space and rewards as follows: , , , in, Express Standardize the process; In F2: By deploying algorithms on each device, each device is regarded as an independent intelligent agent, where the device is the physical entity of the intelligent agent and provides Construct four groups of deep neural networks, namely Actor strategy network , Actor target network , Critic Value Network and Critic target network ,in and Respectively represent the status and actions of other devices except device i, and initialize network parameters , , and , initialize the experience replay pool ; In F3: In the tth time slot, the agent In the environment Execute actions in , get reward And enter the next state ;remember , , , thus recording Store in experience replay pool middle; In F4: When the experience replay pool The number of samples in When K groups of samples are randomly selected to train the agent, first take a group of samples , and calculate the target value of agent i for this sample : , in, Indicates that in the t+1th time slot, based on the state set ,pass The set of actions generated by the network, represents the discount factor; For each agent i, the batch update method is used to update the value network parameters, and the loss function is calculated as: , in, Indicates that the state in the tth time slot and actions Input Critic value network to generate corresponding value; Adopt deterministic policy gradient method to update policy network parameters by gradient ascent , to maximize the strategy objective function: , in, express About action decisions The gradient of express about The gradient of The target network parameters of each agent are updated using soft update: , , in, represents the soft update coefficient; In F5: Repeat the above sub-steps F3 and F4 until the algorithm converges or reaches a predetermined number of training times.

[0029] From the above, it can be seen that the method provided by the present invention realizes efficient collaborative scheduling of multi-source data in new energy power fields by integrating D2D and edge computing technologies, effectively reduces data processing delays, and provides important technical support for power grid planning and intelligent management of new energy power fields.

[0030] Example 2 Reference Figure 4~Figure 6 Based on the previous embodiment, this embodiment provides an application example of a D2D-assisted new energy electric field coordinated scheduling method and system to verify and illustrate the technical effects used in this method.

[0031] like Figure 4The figure shows the convergence performance of the MADDPG algorithm in solving data scheduling and resource allocation problems. As the number of training rounds increases, the performance indicators of the algorithm gradually improve and tend to stabilize. In the early stage of the algorithm (the first 50 rounds), due to the small number of iterations, the learning process is not yet sufficient, and the performance indicators show a certain volatility. As the number of training rounds increases (50 to 150 rounds), the performance indicators show a significant upward trend, indicating that the algorithm is gradually optimized and approaches the optimal solution during the learning process. After about 150 rounds of training, the performance indicators tend to stabilize and the MADDPG algorithm reaches a convergence state. This convergence process shows the effectiveness and stability of the MADDPG algorithm in data scheduling and resource allocation problems.

[0032] like Figure 5 The figure shows the comparison of data processing latency of different algorithms under different numbers of devices. It can be observed that as the number of devices increases, the latency of the three algorithms shows an upward trend, but there are obvious differences in the overall performance. The MADDPG algorithm always maintains the lowest latency, showing better resource allocation and data scheduling capabilities, and can still maintain good optimization effects when the number of devices is large. The latency of the MAPPO algorithm is higher than that of MADDPG, but lower than that of DDPG, indicating that it can optimize latency to a certain extent, but its performance is still not as good as MADDPG. In contrast, the DDPG algorithm has the highest latency, especially when the number of devices is large, its growth trend is most obvious, indicating that its latency optimization capability in a large-scale device environment is weak and it is difficult to effectively cope with complex scenarios. Overall, MADDPG shows higher optimization capabilities and system stability in a multi-device environment, followed by MAPPO, and DDPG has the worst performance.

[0033] like Figure 6 The figure shows the comparison of data processing latency of different algorithms at different maximum computing frequencies. It can be observed that with the increase of the maximum computing frequency, the latency of the three algorithms shows a downward trend, indicating that the increase of computing resources can effectively reduce the task processing latency. Among them, the MADDPG algorithm always maintains the lowest latency, and with the increase of computing frequency, its latency decreases the most, showing better task scheduling and resource allocation capabilities. The latency of MAPPO is higher than MADDPG but lower than DDPG, and the optimization effect is relatively good, but the downward trend is slow. In contrast, DDPG has the highest latency and still maintains a large latency value when the computing frequency is high, indicating that its utilization efficiency of computing resources is low and the optimization effect is the worst. Overall, MADDPG performs best at different computing frequencies, followed by MAPPO, while DDPG has weak resource utilization capabilities and is difficult to achieve effective optimization in a high computing resource environment.

[0034] Example 3 This embodiment provides a D2D-assisted new energy electric field coordinated scheduling system, including: A topology network construction module is used to construct a physical topology network of a new energy electric field according to the equipment location information of the new energy electric field; The scheduling processing module is used to design a device scheduling priority formulation strategy based on the entropy weight method based on the physical topology network of the new energy electric field, and determine the priority scheduling order of the equipment; through the device scheduling priority formulation strategy, an adaptive collaborative data scheduling processing model based on D2D is constructed; The optimization problem processing module is used to construct a stochastic optimization problem for minimizing the task data processing delay based on an adaptive collaborative data scheduling processing model. The Lyapunov optimization technique is used to dynamically process the long-term energy consumption constraints of user equipment and decompose the stochastic optimization problem into a time slot-by-time slot deterministic optimization problem. The optimization scheme determination module is used to apply the MADDPG algorithm to solve the optimization strategy of data scheduling and resource allocation among multiple devices and generate specific data scheduling and resource allocation schemes.

[0035] It should be noted that the technical solution of the system for coordinated scheduling of new energy electric fields based on D2D assistance and the technical solution of the above-mentioned method for coordinated scheduling of new energy electric fields based on D2D assistance belong to the same concept. For the details not described in detail in the technical solution of the system for coordinated scheduling of new energy electric fields based on D2D assistance in this embodiment, please refer to the description of the technical solution of the above-mentioned method for coordinated scheduling of new energy electric fields based on D2D assistance.

[0036] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0037] This embodiment also provides an electronic device, which includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a D2D-assisted new energy electric field collaborative scheduling method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0038] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method proposed in the above embodiment is implemented.

[0039] The storage medium proposed in this embodiment and the method 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.

[0040] Through the above description of the implementation mode, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation mode. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer's floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform the method of the embodiment of the present invention.

[0041] 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.

[0042] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt 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 scheme in the embodiments of the present application may be implemented in various computer languages.

[0043] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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.

[0044] 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.

[0045] 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.

[0046] Although the preferred embodiments of the present application 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 falling within the scope of the present application.

[0047] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A D2D-assisted new energy electric field coordinated scheduling method, characterized in that: include: Construct a new energy electric field physical topology network based on the equipment location information of the new energy electric field; Based on the physical topology network of the new energy electric field, a strategy for device scheduling priority formulation based on the entropy weight method is designed, and the priority scheduling order of the devices is determined; Constructing a D2D-based adaptive collaborative data scheduling processing model through the device scheduling priority formulation strategy; Constructing a stochastic optimization problem of minimizing task data processing delay based on the adaptive collaborative data scheduling processing model; Lyapunov optimization technology is used to dynamically process the long-term energy consumption constraints of user equipment, and the stochastic optimization problem is decomposed into a time slot-by-time slot deterministic optimization problem; The MADDPG algorithm is applied to solve the optimization strategy of data scheduling and resource allocation among multiple devices and generate specific data scheduling and resource allocation solutions.

2. A D2D-assisted new energy electric field coordinated scheduling method as claimed in claim 1, characterized in that: The construction of the new energy electric field physical topology network includes: Obtain the location information of user equipment, power generation equipment, transmission equipment, meteorological monitoring equipment, energy storage equipment and edge servers in the new energy electric field, and obtain the user equipment set N, energy infrastructure equipment set L, and edge server set M. Combine N and L to obtain the equipment set H. The three-dimensional coordinates of the equipment are expressed as , , the three-dimensional coordinates of the edge server are expressed as , ; Based on the three-dimensional location information of each device and edge server in the new energy electric field, combined with the spatial distribution relationship of the actual deployment scenario, a complete new energy electric field physical topology network is constructed.

3. A D2D-assisted new energy electric field coordinated scheduling method as claimed in claim 2, characterized in that: The device scheduling priority setting strategy designed based on the entropy weight method includes: Based on the physical topology network of the new energy electric field, the entire scheduling activity cycle is divided into F time frames using TDMA technology, and each time frame is divided into T fixed time lengths. time slots, where T is equal to the number of devices H; The task data generated by the i-th device in the t-th time slot is represented as a triple ,in Indicates the size of the generated task data. Indicates the number of CPU cycles required to process 1 bit of data. Represents the delay constraint of data processing; definition Indicates the remaining power of the device. The energy infrastructure equipment is always kept fully charged. The maximum power of the user device is expressed as ; The data cache queue is introduced to cache unprocessed data. The data cache queue of the i-th device in the t-th time slot The formula is: ， in, represents the proportion of data calculated locally by device i in the tth time slot, represents the data ratio to be scheduled; there is a constraint between the two ratios of scheduling device i in the tth time slot ; The data processing delay limit is dynamically accumulated according to two ratios, expressed as: ， in, express The corresponding data processing delay limit size; Construct task load, task importance, device remaining energy and distance indicators to reflect the device's task processing requirements, resource constraints and scheduling importance from multiple perspectives; task load is expressed as ,in represents the upper limit of the data cache queue; the task importance is expressed as ; The remaining energy of the equipment is expressed as ,right ,have ; The distance indicator is divided into the distance between the device and the user's device and the distance between the device and the edge server Two types; After completing the definition and normalization of each indicator, obtain the indicator set; The entropy weight method is used to assign weights to the indicator set, and the scheduling priority of the device is determined based on the calculated comprehensive score; based on the indicator matrix W of the device, the normalized weight of each device on each indicator is calculated ; Calculate the entropy value of each indicator according to the definition of information entropy; After obtaining the entropy value of each indicator, the information weight of each indicator is calculated, and then the final weight is obtained through normalization; Calculate the comprehensive score of each device based on the weight distribution , and a comprehensive score for each device By sorting, the scheduling priority of the equipment can be determined. Indicates the device number that ranks highest in a time slot and has not been scheduled in this time frame.

4. A D2D-assisted new energy electric field coordinated scheduling method as claimed in claim 3, characterized in that: The construction of the D2D-based adaptive collaborative data scheduling processing model includes: Based on the device number, record , Indicates the device The uninstall decision represents a set of computing nodes. If , it means that the ratio is The data is unloaded to device s, and vice versa; a device can only be unloaded to one computing node in a time slot, and the unloaded computing node cannot be the device itself; only one device is scheduled in the tth time slot, and the interference between devices can be ignored; The computational delay required for device i to perform local computing tasks in the tth time slot is expressed as , the transmission delay required for device i to offload data to the computing node is expressed as , the computational delay required for computing node s to compute the data offloaded by device i is expressed as , then the total delay of task processing of device i in the tth time slot can be expressed as ; The local computing energy consumption of the i-th device in the t-th time slot is expressed as , the transmission energy consumption required by device i to offload data to the computing node is expressed as , the computing energy consumption required by computing node s to calculate the data unloaded by device i is expressed as , then the total energy consumed by the i-th user equipment in the t-th time slot is expressed as , without considering the energy consumption of energy infrastructure equipment.

5. A D2D-assisted new energy electric field coordinated scheduling method as claimed in claim 4, characterized in that: The stochastic optimization problem of minimizing the delay of constructing task data processing includes: ， ， in, , , They represent task scheduling decisions, computing ratios, and resource allocation strategies, respectively. Indicates the CPU computing frequency assigned to device i when performing local computing tasks. Indicates the maximum CPU computing frequency of the user device. Indicates the CPU computing frequency that the computing node allocates to the task data offloaded from the device. Indicates the maximum CPU computing frequency of the edge server; where the constraint represents the device's uninstallation decision variable, constraint Indicates that in the tth time slot, the scheduling device can only unload task data to one computing node, constraining Indicates that the uninstall object cannot be the scheduling device itself, constraint Represents the constraint on the proportion of local computing tasks on user devices. Represents the constraint on the ratio of device offload tasks. Represents the constraint on the sum of the local computing ratio and the offloading ratio. Indicates the constraints on the local CPU computing frequency of the user device. Indicates the constraints on the CPU computing frequency of the computing node. It represents the long-term energy consumption constraint, i.e., the energy consumption of the user equipment during the entire scheduling activity period.

6. A D2D-assisted new energy electric field coordinated scheduling method as claimed in claim 5, characterized in that: Decomposing the stochastic optimization problem into a time slot-by-time slot deterministic optimization problem includes: Build an energy virtual queue for each user device according to the constraints ; A Lyapunov function is defined to describe the change of the energy virtual queue of all user equipments in the t-th time slot over time, and a Lyapunov drift is obtained by calculating the expectation of the difference between adjacent time slots of the Lyapunov function; In order to minimize the Lyapunov drift, the control strategy is optimized by introducing a penalty term, and the Lyapunov drift plus penalty function is obtained: , and then transform the optimization problem P1 into a deterministic optimization problem P2 without long-term user equipment energy constraints, which is expressed as: 。 7. A D2D-assisted new energy electric field coordinated scheduling method as claimed in claim 6, characterized in that: The optimization strategy of applying the MADDPG algorithm to solve data scheduling and resource allocation among multiple devices includes: Define the core elements of the reinforcement learning algorithm environment state , Action Space and rewards ; By deploying algorithms on each device, each device is regarded as an independent intelligent agent, where the device is the physical entity of the intelligent agent and provides Construct four groups of deep neural networks, namely Actor strategy network , Actor target network , Critic Value Network and Critic target network ,in and Respectively represent the status and actions of other devices except device i, and initialize network parameters , , and , initialize the experience replay pool ; In the tth time slot, agent i is in the environment Execute actions in , get reward And enter the next state ;remember , , , thus recording Stored in the experience replay pool middle; When the experience replay pool When the number of samples in is greater than the number of nodes, K groups of samples are randomly selected to train the agent. First, a group of samples is taken , and calculate the target value of agent i for this sample ; For each agent i, the value network parameters are updated using the batch update method; Adopt deterministic policy gradient method to update policy network parameters by gradient ascent , to maximize the policy objective function; use soft update to update the target network parameters of each agent; Repeat the above steps until the algorithm converges or reaches the predetermined number of training times.

8. A D2D-assisted new energy electric field coordinated dispatching system, using a D2D-assisted new energy electric field coordinated dispatching method as claimed in any one of claims 1 to 7, characterized in that: include: A topology network construction module is used to construct a physical topology network of a new energy electric field according to the equipment location information of the new energy electric field; A scheduling processing module, used to design a device scheduling priority formulation strategy based on the entropy weight method based on the physical topology network of the new energy electric field, and determine the priority scheduling order of the devices; Constructing a D2D-based adaptive collaborative data scheduling processing model through the device scheduling priority formulation strategy; An optimization problem processing module is used to construct a stochastic optimization problem for minimizing task data processing delay based on the adaptive collaborative data scheduling processing model; Lyapunov optimization technology is used to dynamically process the long-term energy consumption constraints of user equipment, and the stochastic optimization problem is decomposed into a deterministic optimization problem per time slot; The optimization scheme determination module is used to apply the MADDPG algorithm to solve the optimization strategy of data scheduling and resource allocation among multiple devices and generate specific data scheduling and resource allocation schemes.

9. An electronic device, comprising a memory and a processor, characterized in that: The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having computer-executable instructions stored thereon, characterized in that: When the computer executable instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images