Charging scheduling method, product and equipment based on joint priority and RNN (Recurrent Neural Network)

Abstract

The invention discloses a charging scheduling method, product and equipment based on joint priority and RNN, and relates to the technical field of wireless rechargeable scheduling. The construction method comprises the steps of defining a charging sensor network model, enhancing a K-means division network, dividing time slices in a time range, determining whether a current charging demand is overloaded or not, and formalizing a multi-objective joint optimization problem and a path planning strategy. According to the method, more accurate, efficient and reliable charging management service is provided for the wireless rechargeable sensor network, through the whole-process design of space partitioning, time management and control, priority screening and intelligent optimization, the overall performance and stability of the network are improved, the operation and maintenance cost is also reduced, and the system is suitable for popularization and application. Powerful support is provided for sensor networks in various application scenes, and further development and application of the Internet of Things technology are promoted.

Description

Technical Field

[0001] This invention relates to the field of wireless rechargeable scheduling technology, and in particular to a charging scheduling method, product, and device based on joint priority and RNN. Background Technology

[0002] Thanks to breakthroughs in wireless power transfer technology in recent years, the energy-limiting bottleneck that has hindered the widespread deployment of wireless sensor networks (WSNs) has been resolved. Therefore, the concept of wireless rechargeable sensor networks (WRSNs) is gradually becoming a reality.

[0003] In power transmission systems, cables are critical infrastructure, and their continuity and stability are essential for ensuring the safe and efficient operation of the power grid. Typically, the standard length of cables is designed to be around 100 meters to accommodate considerations of laying, maintenance, and cost-effectiveness.

[0004] However, as a critical point in cable connection, the health of cable joints directly affects the safety and reliability of the entire transmission link. Therefore, deploying sensor nodes at the joints of two cables for real-time monitoring has become an important measure to prevent potential risks such as joint aging and breakage, and to avoid large-scale power outages.

[0005] Given the vast territory and complex terrain of Xinjiang, traditional manual inspection methods face challenges in terms of efficiency, cost, and response speed. To improve the intelligence level of power grid operation and maintenance and achieve the goal of building a "smart grid," introducing drone technology for wireless charging or maintenance of sensor nodes has become an innovative and necessary solution.

[0006] With their high flexibility, wide coverage, and high operational efficiency, drones can quickly reach cable joint locations in complex environments to accurately locate and charge sensor nodes, effectively solving the problem of difficult power facility maintenance in remote areas. Furthermore, by integrating advanced navigation, communication, and autonomous control technologies, drones can also achieve remote operation and autonomous operation, further reducing manpower and material costs and improving the intelligence and automation level of power grid operation and maintenance.

[0007] In conclusion, using drones to provide charging services for sensor nodes at cable joints is not only an effective means of solving the power grid operation and maintenance problems in vast areas such as Xinjiang, but also a key link in promoting the construction of "smart grids" and improving the overall security and reliability of the power grid.

[0008] However, due to their small size, drones have limited battery capacity, making it impossible for them to perform long-distance charging tasks directly and cover a wide range of sensor nodes. At the same time, although drone hardware is inexpensive, their limited power reserves make it difficult for them to continuously carry out multiple charging operations and meet the full charging needs of the entire sensor network independently. They must work in conjunction with other devices.

[0009] Therefore, in the charging scheduling scenario of wireless sensor networks, how to simultaneously reduce the ineffective movement loss of wireless charging vehicles and the redundant flight energy consumption of drones by optimizing task allocation strategies, and extend the continuous working time of sensor nodes and the overall life cycle of the network through the design of efficient energy replenishment mechanisms, has become a key technical challenge that urgently needs to be overcome in the current field. Summary of the Invention

[0010] The technical problem to be solved by this invention is to overcome the shortcomings of the existing technology and design a new real-time charging scheduling strategy that utilizes the availability of multiple WCVs and allows them to make optimal decisions based on a joint metric of the spatial and temporal requirements of charging requests.

[0011] The solution adopted by this invention to solve its technical problem is as follows:

[0012] A charging scheduling method based on joint priority and RNN,

[0013] The charging vehicles include a wireless charging vehicle (WCV) and a wirelessly charging drone mounted on the WCV.

[0014] Its features are,

[0015] The method includes the following steps:

[0016] Step S1: Define the charging sensor network model.

[0017] It includes a base station, m charging vehicles, and n sensor nodes. The WCV and sensors are all equipped with GPS modules and batteries.

[0018] The base station is used to receive location information and charging requests sent by each sensor node, is responsible for data aggregation and fusion, and predicts events in the entire monitoring area;

[0019] The charging carrier is used to accurately locate the sensor and complete the refueling task.

[0020] The task of the sensor nodes is to monitor, collect, and transmit messages from the environment.

[0021] Step S2: Enhance the K-means partitioning network, including:

[0022] The base station selects m WCVs as center points, and the sensor nodes Its two-dimensional location information The data is sent to the base station, which then calculates the sensor node's... With the center point Distance between The sensor nodes are classified sequentially based on their proximity to the center point, ultimately resulting in all sensor nodes being grouped into clusters centered on the WCV.

[0023] Step S3: Divide the time slice within the time range.

[0024] During the charging request collection phase, when the power of the sensor node... Below the threshold At that time, the sensor node sends a charging request to the base station. The base station allocates the charging request to the corresponding WCN according to the subnet divided in step S2. Within each time slice interval, the base station collects the charging request only once before making a scheduling strategy, analyzes the collected charging request, and makes a strategy plan once.

[0025] Step S4: Determine if the current charging demand is overloaded.

[0026] During the charging process, the WCV periodically selects a charging target and drives towards it, replenishing energy upon arrival. The workload of the WCV during charging is... Represents a time slice The ratio of effective charging time within the time limit is expressed as:

[0027] ,

[0028] in, Indicates the number of charging tasks in the current round. Represented as the sensor node The time required for charging, The value represents the energy consumption rate of the sensor node, and the value represents the battery capacity of the sensor node. This represents the effective charging time for m sensor nodes. The time slice size is determined by the current round. When the value is greater than 1, the MCV is overloaded.

[0029] Step S5: Formalize the multi-objective joint optimization problem.

[0030] Step S6: Path planning strategy

[0031] when When the value is less than 1, the MCV's charging task is a normal load, and path planning is performed based on priority.

[0032] when When the value is greater than or equal to 1, the charging task of MCV is overloaded, and path planning is performed based on RNN.

[0033] As a preferred embodiment of the present invention

[0034] The time range in step S3 for:

[0035] ,

[0036] In the formula, This indicates the minimum amount of energy remaining in the sensor node after the previous charging cycle. This indicates the power level of the sensor node. This indicates the amount of charge the sensor receives per unit of time.

[0037] This represents the maximum lifetime of the sensor node with the least remaining energy.

[0038] This represents the time required to charge a specific sensor node.

[0039] When the time slice exceeds At that time, a node that has not issued a charging request in the current time slice will die.

[0040] When the time slice is less than When the WCV is unable to complete any charging task, the base station will replan the charging route for it.

[0041] As a preferred embodiment of the present invention

[0042] The formalized multi-objective joint optimization problem in step S5 includes:

[0043] ,

[0044] ,

[0045] in, It performs a minimize operation. It is a sequence of charging sensor nodes. It is the service time of the i-th sensor node. It is about dividing the time slice size. It is the distance between the WCV and the sensor node. It is the movement speed of WCV. It refers to WCV charging efficiency. It is the number of sensor nodes. This represents the summation of individual terms over n sensor nodes.

[0046] As a preferred embodiment of the present invention

[0047] Step S6, which involves priority-based path planning, includes the following steps:

[0048] Step S61: First, calculate the charging cutoff time of the sensor node based on its remaining energy and energy consumption rate. :

[0049] ,

[0050] Step S62: In the current charging task queue, mark the earliest task arrival time as... The latest task arrival time is marked as Time priority is defined as:

[0051] ,

[0052] in, This represents the arrival time of the task at the i-th sensor node. Indicates the number of sensor nodes.

[0053] Step S63: Define For sensor nodes With the j-th WCV The distance between them is marked as the distance to the nearest sensor node to the WCV. The distance to the sensor node farthest from WCV is marked as Calculate the space priority:

[0054] ,

[0055] Step S64: Combine time priority and spatial priority to obtain joint priority:

[0056] ,

[0057] The joint priority Sensor nodes with lower values ​​are given higher priority for service.

[0058] As a preferred embodiment of the present invention

[0059] Step S6, which involves path planning based on RNN, includes the following steps:

[0060] Step S65: Define the problem objective and boundary constraints:

[0061] ,

[0062] ,

[0063] in, It is a vector of all 1s. for The transpose of , It is a piecewise linear function, for ,when hour, ;when hour, ,

[0064] Step S65: Define the input parameters of the RNN model.

[0065] Define matrix C:

[0066] ,

[0067] Where P is the charging power of WCV, and I represents an identity matrix.

[0068] Define column vectors ;

[0069] ,

[0070] Define matrix A:

[0071] ,

[0072] Define column vector b:

[0073] ,

[0074] Step S66: Define the state equation and activation function of the RNN model.

[0075] ,

[0076] in, It is a positive constant. It is the transpose of matrix C. Let A be the transpose of matrix A. It is a non-negative gain parameter. Defined as: when hour, ,when hour, , Defined as: when hour, ,when hour, ,

[0077] Step S67: Determine the conditions for the value of the gain parameter σ:

[0078] ,

[0079] in, , This indicates performing a Cartesian product operation on p one-dimensional closed intervals. It is the lower bound of the i-th dimension. It is the upper bound of the i-th dimension. yes The maximum compact set, where max represents finding the maximum value. Representing vectors The Euclidean norm, min denotes finding the minimum value. Representing vectors The Euclidean norm,

[0080] Step S68: Input A, b, C, and d into the RNN model, perform calculations using the activation function and state equation in step S66, and output the path planning result sequence x.

[0081] As a preferred embodiment of the present invention

[0082] The WCV is configured with a service sequence for storing sensor nodes that are about to be charged.

[0083] As a preferred embodiment of the present invention

[0084] The charging request includes the unique identifier of the sensor node. Node location, current battery level and information.

[0085] As a preferred embodiment of the present invention

[0086] The sensor nodes communicate using the default protocol.

[0087] A computer program product includes a computer program that, when executed by a processor, implements the steps of the charging scheduling method based on joint priority and RNN.

[0088] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to implement the steps of the charging scheduling method based on joint priority and RNN.

[0089] Compared with the prior art, the present invention has the following beneficial effects:

[0090] 1. Priority Ranking: This invention assigns priority to each charging request by comprehensively considering multiple factors such as the urgency of the charging request, the remaining energy of the sensor node, and the geographical distribution, ensuring that the most urgent requests can obtain charging services first, thereby maximizing the overall stability and lifespan of the network.

[0091] 2. Dynamic Adjustment: The recurrent structure of the RNN model enables it to process sequential data, receiving and analyzing the latest data from sensor nodes in real time, and dynamically adjusting the charging plan to cope with sudden changes in the network. This real-time dynamic adjustment capability ensures that the charging scheduling scheme remains efficient and flexible in the face of complex and ever-changing network environments.

[0092] 3. System Robustness: This method can effectively identify and eliminate erroneous or abnormal charging requests, avoiding resource misallocation caused by erroneous information. Furthermore, the simulation of various scenarios considered during model training enhances the system's adaptability and stability under different environmental conditions.

[0093] This invention presents a charging scheduling method based on joint priority and RNN, providing more accurate, efficient, and reliable charging management services for wireless rechargeable sensor networks. It not only improves the overall performance and stability of the network but also reduces operation and maintenance costs, providing strong support for sensor networks in various application scenarios and promoting the further development and application of IoT technology. Attached Figure Description

[0094] Figure 1 This is a flowchart of a charging scheduling method based on joint priority and RNN proposed in this invention;

[0095] Figure 2 This is a network architecture diagram of a wireless rechargeable sensor network in a charging scheduling method based on joint priority and RNN proposed in this invention.

[0096] Figure 3 This is a diagram illustrating the network partitioning process in a charging scheduling method based on joint priority and RNN proposed in this invention.

[0097] Figure 4 This is a detailed design diagram of the RNN model in the charging scheduling method based on joint priority and RNN proposed in this invention. Detailed Implementation

[0098] The specific embodiments of the present invention are described below with reference to the accompanying drawings and examples:

[0099] It should be noted that the structures, colors, proportions, sizes, etc. shown in the accompanying drawings are only used to complement the content disclosed in the specification, so that those skilled in the art can understand and read them, and are not intended to limit the conditions under which the present invention can be implemented. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0100] like Figure 1 As shown, this invention provides a charging scheduling method based on joint priority and RNN, wherein the charging vehicle includes a wireless charging vehicle (WCV) and a wireless charging drone mounted on the WCV. The method includes the following steps:

[0101] Step S1: Define the charging sensor network model.

[0102] The system includes a base station, m charging stations, and n sensor nodes. The WCV (Wireless Video Controller) and sensors are all equipped with GPS modules and batteries. The base station receives location information and charging requests from each sensor node, is responsible for data aggregation and fusion, and predicts events throughout the monitoring area. The charging stations accurately locate the sensors and recharge them. The sensor nodes monitor, collect, and transmit messages from the environment. Figure 2 The diagram shown is a network architecture diagram of a wireless rechargeable sensor network with four wireless charging carriers.

[0103] Step S2: Enhance the K-means partitioning network, such as... Figure 3 As shown, the process includes: (a) initializing the network, (b) selecting the domain center, (c) starting the partitioning, and (d) completing the partitioning. Specifically, it includes:

[0104] The base station selects m WCVs as center points, and the sensor nodes Its two-dimensional location information The data is sent to the base station, which then calculates the sensor node's... With the center point Distance between The sensor nodes are classified according to their proximity to the center point, and finally all sensor nodes are classified into clusters centered on WCV.

[0105] By clustering sensor nodes according to their proximity to the Wireless Controller (WCV), WCVs do not need to move long distances across clusters when performing charging tasks. Each cluster naturally corresponds to an independent WCV service area, allowing subsequent path planning to be split by cluster. WCVs only need to plan their paths within their own cluster, without considering all nodes in the network. This naturally adapts to charging scheduling requirements, significantly reducing WCV movement costs and simplifying path planning complexity. It can also dynamically adapt to changes in WCV location to improve network robustness and achieve load balancing by adjusting the number of WCVs, preventing charging task overload.

[0106] Step S3: Divide the time slice within the time range. The method of this invention does not set a specific size for the time slice, but only provides a range. Users can define an appropriate time slice size within the range according to their needs.

[0107] During the charging request collection phase, when the power of the sensor node... Below the threshold When the sensor node sends a charging request to the base station, the base station allocates the charging request to the corresponding WCN according to the subnet divided in step S2, so that the base station can efficiently allocate the charging request to the corresponding WCN based on the subnet division.

[0108] Within each time slice interval, the base station collects charging requests only once before making a scheduling strategy, analyzes the collected charging requests, and makes a strategy plan once. Collecting charging requests and planning strategies only once per time slice interval reduces the computational overhead of frequent scheduling, ensures that the charging strategy accurately responds to the urgent needs of nodes, and improves the overall scheduling efficiency and adaptability of the system.

[0109] Step S4: Determine if the current charging demand is overloaded.

[0110] Each charging task has a different response time, which is the time interval between a node issuing a request and the request being executed. During the charging process, the WCV periodically selects a charging target and travels towards it, replenishing energy upon arrival. The workload of the WCV during the charging process is... Represents a specific time slice after the division. The ratio of effective charging time within the time limit is expressed as:

[0111] ,

[0112] in, Indicates the number of charging tasks in the current round. Represented as the sensor node The time required for charging, The value represents the energy consumption rate of the sensor node, and the value represents the battery capacity of the sensor node. This represents the effective charging time for m sensor nodes. The time slice size is determined by the current round. When the value is greater than 1, the MCV is overloaded.

[0113] Determine if charging demand is overloaded by calculating the WCV workload ratio: when When the value is greater than 1, it indicates that the effective charging time of WCV in the current time slice exceeds the time slice length, and it cannot complete all charging tasks, which is considered an overload.

[0114] By quantifying the matching relationship between WCV workload and time slices, the risk of task backlog can be accurately identified, providing data support for dynamically adjusting charging strategies, avoiding node response time extension or even failure due to task overload, and ensuring the efficiency of charging scheduling and network stability.

[0115] Step S5: Formalize the multi-objective joint optimization problem.

[0116] Step S6: Path planning strategy

[0117] when When the value is less than 1, the MCV's charging task is a normal load, and path planning is performed based on priority.

[0118] when When the value is greater than or equal to 1, the charging task of MCV is overloaded, and path planning is performed based on RNN.

[0119] The charging scheduling method proposed in this invention, based on joint priority and RNN, constructs a precise and efficient charging management system for wireless rechargeable sensor networks through multi-module collaborative design.

[0120] First, the network partitioning centered on WCV using K-means is enhanced by clustering nodes according to spatial distance and charging demand, directly reducing WCV movement costs and coordinating with subsequent path planning. Combined with a charging request mechanism triggered by a power threshold, this ensures that base stations minimize computational overhead and avoid the risk of node death due to fixed time slices. Second, a dual-path planning strategy based on a joint priority formula (integrating time and dimension) and a single-layer RNN model is adopted. This strategy quickly identifies high-urgency nodes through priority ranking and shortens the total scheduling time by optimizing the path using RNN, resolving the contradiction between urgency and optimality that traditional heuristic algorithms struggle to balance. Furthermore, an overload judgment mechanism based on workload ratio dynamically filters feasible task sets, avoiding resource waste in WCV due to task backlog.

[0121] The method of this invention, through the whole process design of "spatial partitioning-time control-priority screening-intelligent optimization", not only significantly improves the charging efficiency and stability of sensor networks, but also effectively reduces system operation and maintenance costs by reducing invalid WCV movement and dynamic resource scheduling. It provides a highly robust charging solution for IoT scenarios such as industrial monitoring and environmental sensing, and promotes the practical application and development of wireless charging and sensor network fusion technology.

[0122] In another embodiment, the time range in step S3 for:

[0123] ,

[0124] In the formula, This indicates the minimum amount of energy remaining in the sensor node after the previous charging cycle. This indicates the power level of the sensor node. This indicates the amount of charge the sensor receives per unit of time.

[0125] This represents the maximum lifetime of the sensor node with the least remaining energy.

[0126] This represents the time required to charge a specific sensor node.

[0127] When the time slice exceeds When a node does not issue a charging request in the current time slice, it will die; when the time slice is less than [a certain value], the node will die. If the WCV is unable to complete any charging task, the base station will replan the charging route for it. Therefore, the size of the time slice should be within the range The time constraint is determined internally. This ensures that WCV can complete the charging of at least one node within a single time slice, while preventing uncharged nodes from dying due to excessively long time slices, thus providing a feasible time constraint for charging scheduling.

[0128] In another embodiment, as the network operates, nodes consume battery power at varying rates. When a node's remaining power falls below a threshold, it broadcasts a charging request, which is forwarded to neighboring WCVs. This request carries the node ID, node location, remaining power level, and current power consumption rate. Once received, it is inserted into a service queue for later service provision. If the requesting node does not charge in time, its power will eventually be depleted. These nodes will cease providing any service in the network. Because unpredictable events can occur anytime and anywhere in real-time WRSNs, resource depletion at any node can lead to event detection failures, potentially causing catastrophic events. Therefore, this type of real-time scheduling problem requires close attention.

[0129] Therefore, the core problem to be solved in the method of this invention is: in a WRSN (Wireless Rechargeable Sensor Network) comprising a base station (BS), multiple wireless charging vehicles (WCVs), and a series of sensor nodes, how can the WCVs simultaneously consider the spatiotemporal interdependence of charging tasks and select nodes for charging according to the order of charging requests, so as to maximize energy utilization efficiency η and support the survival of the maximum number of sensor nodes?

[0130] To address this problem, the formalized multi-objective joint optimization problem in step S5 includes:

[0131] Objective function, over-optimized service sequence

[0132] This minimizes the sum of the time waiting cost and the space travel cost.

[0133] The constraints avoid the logical contradiction of forcing service to start before WCV has arrived. At the same time, through reverse derivation, nodes that can be served in a timely manner are indirectly selected, providing boundary guarantees for maximizing the number of surviving nodes.

[0134] in, It performs a minimize operation. It is a sequence of charging sensor nodes. It is the service time of the i-th sensor node. It is about dividing the time slice size. It is the distance between the WCV and the sensor node. It is the movement speed of WCV. It refers to WCV charging efficiency. It is the number of sensor nodes. This represents the summation of individual terms over n sensor nodes.

[0135] By optimizing service sequences This maximizes energy efficiency (minimizes total energy consumption); at the same time, by limiting the earliest time a node is served, it ensures that WCV has enough time to complete, indirectly achieving the survival of the maximum number of sensor nodes.

[0136] In another embodiment,

[0137] Step S6, which involves priority-based path planning, includes the following steps:

[0138] Step S61: First, calculate the charging cutoff time of the sensor node based on its remaining energy and energy consumption rate. :

[0139] .

[0140] The ratio of remaining energy to energy consumption rate determines the remaining lifespan of a node without charging, such as the charging cutoff time of a node. The charging time is 2 hours, meaning it will fail if not charged within 2 hours. (This is based on the charging deadline.) This provides a hard time constraint for priority-based path planning, ensuring that high-risk nodes are prioritized and preventing critical sensor nodes from running out of power and becoming dead, thus avoiding network function interruptions. Priority scheduling driven by charging cutoff times significantly reduces unexpected sensor node failures due to power depletion. Simultaneously, it avoids the additional recovery costs associated with network function loss due to node death, effectively reducing overall system costs.

[0141] Step S62: In the current charging task queue, mark the earliest task arrival time as... The latest task arrival time is marked as Time priority is defined as:

[0142] ,

[0143] in, This represents the arrival time of the task at the i-th sensor node. Indicates the number of sensor nodes.

[0144] Step S63: Define For sensor nodes With the j-th WCV The distance between them is marked as the distance to the nearest sensor node to the WCV. The distance to the sensor node farthest from WCV is marked as Calculate the space priority:

[0145] ,

[0146] Step S64: Combine time priority and spatial priority to obtain joint priority:

[0147] ,

[0148] The joint priority Sensor nodes with lower joint priority numbers are assigned higher priority for service. The smaller the joint priority number, the higher the priority the node is given by WCV, ensuring that urgent tasks are not delayed while avoiding excessive scheduling costs.

[0149] Joint prioritization transforms the time-dimensional arrival time of tasks and the distances to nodes in both time and space dimensions into integer indices with the same value range through discretization. This allows time urgency and spatial distance costs, which were previously difficult to compare directly, to be measured on the same scale. It eliminates the comparison barriers caused by different physical dimensions, providing a solid foundation for subsequent priority fusion and task ranking.

[0150] By weighting time priority and spatial priority proportionally with a coefficient of 0.5, fairness in the joint priority calculation of these two key dimensions is ensured. This avoids the problems of WCV frequently traveling long distances due to focusing solely on time, or neglecting urgent nodes due to focusing solely on spatial priority. Simultaneously, the introduction of a logarithmic enhancement term further strengthens the priority of nodes that are both time-urgent and geographically close, giving them a smaller joint priority number. This allows these nodes to receive charging services first, achieving efficient and precise resource allocation.

[0151] In another embodiment,

[0152] Step S6, which involves path planning based on RNN, includes the following steps:

[0153] Step S65: Define the problem objective and boundary constraints:

[0154] ,

[0155] ,

[0156] in, It is a vector of all 1s. for The transpose of , It is a piecewise linear function, for ,when hour, ;when hour, ,

[0157] Step S65: Define the input parameters of the RNN model.

[0158] Define matrix C:

[0159] ,

[0160] Where P is the charging power of WCV, and I represents an identity matrix.

[0161] Define column vectors ;

[0162] ,

[0163] Define matrix A:

[0164] ,

[0165] Define column vector b:

[0166] ,

[0167] Step S66: Define the state equation and activation function of the RNN model.

[0168] ,

[0169] in, It is a positive constant. It is the transpose of matrix C. Let A be the transpose of matrix A. It is a non-negative gain parameter. Defined as: when hour, ,when hour, , Defined as: when hour, ,when hour, ,

[0170] Step S67: Determine the conditions for the value of the gain parameter σ:

[0171] ,

[0172] in, , This indicates performing a Cartesian product operation on p one-dimensional closed intervals. It is the lower bound of the i-th dimension. It is the upper bound of the i-th dimension. yes The maximum compact set, where max represents finding the maximum value. Representing vectors The Euclidean norm, min denotes finding the minimum value. Representing vectors The Euclidean norm,

[0173] Step S68: Design of a single-layer RNN model as follows Figure 4 As shown. In the RNN model, inputs A, b, C, and d are processed through the activation function and state equation in step S66, resulting in the output path planning result sequence x.

[0174] The RNN-based path planning process achieves efficient optimization of the charging path through a closed-loop design of problem modeling, parameter mapping, dynamic evolution, convergence guarantee, and result output. Its core value lies in:

[0175] Step S65 first transforms the path planning into a mathematical problem with piecewise linear objectives and linear constraints, clarifying the core objective of minimizing the weighted deviation and the resource boundary. At the same time, through matrix C (corresponding to charging power and battery capacity), vector d (charging cutoff time), matrix A and vector b (constraints), the physical parameters are accurately mapped to the model input, building a bridge between the actual scenario and the algorithm.

[0176] The RNN state equation and piecewise activation function designed in step S66 integrate the objective function and constraints, and utilize the dynamic evolution characteristics of RNN to adapt to the spatiotemporal coupling of path planning (such as the mutual influence between WCV movement and charging time). The piecewise activation function accurately captures the differentiated effects of overcharging / undercharging.

[0177] Step S67 ensures that the model converges to a feasible solution within a finite time by strictly controlling the value of the gain parameter σ, thus avoiding invalid output caused by algorithm divergence.

[0178] The final step, S68, outputs the specific path sequence, which directly serves the actual scheduling.

[0179] It not only ensures the fit of the RNN model to the real-world scenario, but also leverages the dynamic optimization capabilities of RNN to handle complex dependencies, while ensuring the reliability of the solution through convergence, thus significantly improving the accuracy and execution efficiency of path planning.

[0180] In another embodiment, the WCV is configured with a service sequence for storing sensor nodes that are about to be charged. The WCV's service sequence is used to store sensor nodes to be charged in an orderly manner, connect scheduling decisions and execution, prioritize tasks, coordinate path planning to optimize mobile energy consumption, and dynamically adjust to ensure timely charging of high-urgent nodes, thereby ensuring scheduling orderliness and node survival.

[0181] In another embodiment, the charging request includes the unique identifier of the sensor node. Node location, current battery level and Information. It provides crucial information for base station task allocation, WCV path planning, and node urgency assessment, ensuring that charging scheduling responds to node needs.

[0182] In another embodiment, the sensor nodes communicate under a default protocol. The default protocol provides a unified standard for information exchange between nodes and between nodes and the base station / WCV, ensuring that charging requests are transmitted accurately and without errors, avoiding data loss or transmission chaos due to protocol incompatibility. In addition, the default protocol is usually lightweight and optimized to suit the limited computing power and energy consumption of sensor nodes. While realizing communication functions, it can reduce additional energy consumption, extend the single-charge endurance of nodes, indirectly reduce the charging scheduling pressure, and provide underlying support for the efficient coordination of the entire network.

[0183] A charging scheduling system based on joint priority and RNN includes an enhanced K-means partitioning network module, a time slice partitioning module, a module for determining whether the charging task is overloaded, a module for formalizing the multi-objective joint optimization problem, and a path planning strategy module.

[0184] The enhanced K-means partitioning network module includes a unit for collecting node location information.

[0185] The time-slicing module includes a charging request receiving unit and a time-slicing unit.

[0186] The path planning strategy module includes path planning units based on joint priority and path planning units based on a single-layer RNN model.

[0187] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform steps of a charging scheduling method based on joint priority and RNN.

[0188] A computer program product, such as computer program instructions, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the present invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0189] The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

[0190] Many other changes and modifications can be made without departing from the concept and scope of this invention. It should be understood that this invention is not limited to the specific embodiments, and the scope of this invention is defined by the appended claims.

Claims

1. A charging scheduling method based on joint priority and RNN, wherein, The charging vehicles include WCVs and wirelessly charging drones mounted on the WCVs. Its features are, The method includes the following steps: Step S1: Define the charging sensor network model. It includes a base station, m charging vehicles, and n sensor nodes. The WCV and sensors are all equipped with GPS modules and batteries. The base station is used to receive location information and charging requests sent by each sensor node, is responsible for data aggregation and fusion, and predicts events in the entire monitoring area; The charging carrier is used to accurately locate the sensor and complete the refueling task. The task of the sensor nodes is to monitor, collect, and transmit messages from the environment. Step S2: Enhance the K-means partitioning network, including: The base station selects m WCVs as center points, and the sensor nodes Its two-dimensional location information The data is sent to the base station, which then calculates the sensor node's... With the center point Distance between The sensor nodes are classified sequentially based on their proximity to the center point, ultimately resulting in all sensor nodes being grouped into clusters centered on the WCV. Step S3: Divide the time slice within the time range. During the charging request collection phase, when the power of the sensor node... Below the threshold At that time, the sensor node sends a charging request to the base station. The base station allocates the charging request to the corresponding WCN according to the subnet divided in step S2. Within each time slice interval, the base station collects the charging request only once before making a scheduling strategy, analyzes the collected charging request, and makes a strategy plan once. Step S4: Determine if the current charging demand is overloaded. During the charging process, the WCV periodically selects a charging target and drives towards it, replenishing energy upon arrival. The workload of the WCV during charging is... Represents a time slice The ratio of effective charging time within the time limit is expressed as: ， in, Indicates the number of charging tasks in the current round. Represented as the sensor node The time required for charging, The value represents the energy consumption rate of the sensor node, and the value represents the battery capacity of the sensor node. This represents the effective charging time for m sensor nodes. The time slice size is determined by the current round. When the value is greater than 1, the MCV is overloaded. Step S5: Formalize the multi-objective joint optimization problem. Step S6: Path planning strategy when When the value is less than 1, the MCV's charging task is a normal load, and path planning is performed based on priority. when When the value is greater than or equal to 1, the charging task of MCV is overloaded, and path planning is performed based on RNN.

2. The charging scheduling method based on joint priority and RNN as described in claim 1, Its features are, The time range in step S3 for: ， In the formula, This indicates the minimum amount of energy remaining in the sensor node after the previous charging cycle. This indicates the power level of the sensor node. This indicates the amount of charge the sensor receives per unit of time. This represents the maximum lifetime of the sensor node with the least remaining energy. This represents the time required to charge a specific sensor node. When the time slice exceeds At that time, a node that has not issued a charging request in the current time slice will die. When the time slice is less than When the WCV is unable to complete any charging task, the base station will replan the charging route for it.

3. The charging scheduling method based on joint priority and RNN as described in claim 2, Its features are, The formalized multi-objective joint optimization problem in step S5 includes: ， ， in, It performs a minimize operation. It is a sequence of charging sensor nodes. It is the service time of the i-th sensor node. It is about dividing the time slice size. It is the distance between the WCV and the sensor node. It is the movement speed of WCV. It refers to WCV charging efficiency. It is the number of sensor nodes. This represents the summation of individual terms over n sensor nodes.

4. The charging scheduling method based on joint priority and RNN as described in claim 1, Its features are, Step S6, which involves priority-based path planning, includes the following steps: step S61: First, calculate the charging cutoff time of the sensor node based on its remaining energy and energy consumption rate. : ， Step S62: In the current charging task queue, mark the earliest task arrival time as... The latest task arrival time is marked as Time priority is defined as: ， in, This represents the arrival time of the task at the i-th sensor node. Indicates the number of sensor nodes. Step S63: Define For sensor nodes With the j-th WCV The distance between them is marked as the distance to the nearest sensor node to the WCV. The distance to the sensor node farthest from WCV is marked as Calculate the space priority: ， Step S64: Combine time priority and spatial priority to obtain joint priority: ， The joint priority Sensor nodes with lower values ​​are given higher priority for service.

5. The charging scheduling method based on joint priority and RNN as described in claim 1, Its features are, Step S6, which involves path planning based on RNN, includes the following steps: Step S65: Define the problem objective and boundary constraints: ， ， in, It is a vector of all 1s. for The transpose of , It is a piecewise linear function, for ,when hour, ;when hour, , Step S65: Define the input parameters of the RNN model. Define matrix C: ， Where P is the charging power of WCV, and I represents an identity matrix. Define column vectors ; ， Define matrix A: ， Define column vector b: ， Step S66: Define the state equation and activation function of the RNN model. ， in, It is a positive constant. It is the transpose of matrix C. Let A be the transpose of matrix A. It is a non-negative gain parameter. Defined as: when hour, ,when hour, , Defined as: when hour, ,when hour, , Step S67: Determine the conditions for the value of the gain parameter σ: ， in, , This indicates performing a Cartesian product operation on p one-dimensional closed intervals. It is the lower bound of the i-th dimension. It is the upper bound of the i-th dimension. yes The maximum compact set, where max represents finding the maximum value. Representing vectors The Euclidean norm, min denotes finding the minimum value. Representing vectors The Euclidean norm, Step S68: Input A, b, C, and d into the RNN model, perform calculations using the activation function and state equation in step S66, and output the path planning result sequence x.

6. The charging scheduling method based on joint priority and RNN as described in claim 1, Its features are, The WCV is configured with a service sequence for storing sensor nodes that are about to be charged.

7. The charging scheduling method based on joint priority and RNN as described in claim 1, Its features are, The charging request includes the unique identifier of the sensor node. Node location, current battery level and information.

8. The charging scheduling method based on joint priority and RNN as described in claim 1, Its features are, The sensor nodes communicate using the default protocol.

9. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the charging scheduling method based on joint priority and RNN as described in any one of claims 1-8.

10. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the charging scheduling method based on joint priority and RNN as described in any one of claims 1-8.

Images