Distribution Internet of Things routing selection method and system based on end-to-end value learning
By employing an end-to-end value learning-based routing method, the problem of unstable data transmission caused by electrical equipment failures or off-grid access in distribution networks with a high proportion of renewable energy has been solved by traditional routing methods, achieving higher communication reliability and lower packet loss rate.
Patent Information
- Application Number
- CN202310764482.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-06-26
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2043-06-26
AI Technical Summary
When a high proportion of new energy sources are connected to the distribution network, traditional routing methods fail to effectively consider the link breakage caused by electrical equipment failures or disconnection from the grid, resulting in unstable data transmission and high packet loss rate, which affects the communication reliability of the distribution Internet of Things.
We adopt an end-to-end value learning-based routing method. By constructing a routing value model that considers the coupling characteristics of dual networks, we predict the link breakage probability and select the link with the highest transmission quality for data transmission. We then optimize the routing by combining the link transmission quality and the breakage probability.
It improves the reliability of power distribution IoT communication, reduces data packet loss due to electrical equipment failure or off-grid operation, and enhances service forwarding performance.
Smart Images

Figure CN116708269B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of power distribution Internet of Things communication, and in particular to a power distribution Internet of Things routing selection method based on end-to-end value learning. Background Art
[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.
[0003] In the context of a high proportion of renewable energy being widely connected to the distribution network, the operating status of the distribution network is complex, and the transmission quality of the power line carrier communication (PLC) link responsible for data transmission is inconsistent. Even a link needs to be disconnected when encountering an emergency in the power grid. The distribution Internet of Things urgently needs to adopt more advanced routing selection technology to ensure the reliability of data transmission.
[0004] At present, the research on routing selection mainly involves single indicators such as latency. The communication network architecture of traditional power grids faces problems such as high packet loss rate and circuit breaking in emergencies, which seriously affect the routing selection and information transmission of power grid services. Under the premise of massive access to new energy terminals, traditional routing optimization methods do not take into account the switching characteristics of new energy equipment and the impact of complex working environments. The various complex factors of distribution network operation are not considered in the effect of terminal selection and forwarding routing. The limited routing resources are difficult to directly meet the transmission needs of terminal data, and the transmission quality between routing ends is different, and there is a lack of methods to select the optimal transmission link. Summary of the invention
[0005] In order to solve at least one technical problem existing in the above-mentioned background technology, the present invention provides a distribution Internet of Things routing selection method and system based on end-to-end value learning, which reduces the problem of data packet loss caused by electrical equipment failure or disconnection as much as possible on the basis of improving service forwarding performance by considering the end-to-end routing value modeling of dual-network coupling characteristics. The link with the highest transmission quality is selected for data transmission, and the subsequent routing selection performance is improved by the differentiated performance perception of different power services, so as to meet the communication reliability requirements of the distribution Internet of Things.
[0006] In order to achieve the above object, the present invention adopts the following technical solution:
[0007] The first aspect of the present invention provides a power distribution Internet of Things routing selection method based on end-to-end value learning, comprising the following steps:
[0008] Get all device nodes of the power distribution IoT network;
[0009] Based on all device nodes of the power distribution Internet of Things network and the power distribution Internet of Things routing selection model, a power distribution Internet of Things routing selection scheme is obtained, and data transmission is performed based on the scheme;
[0010] The construction process of the power distribution Internet of Things routing selection model is as follows:
[0011] When the distribution IoT terminal selects the next-hop routing node, it calculates the end-to-end routing value corresponding to the hop; the routing node with the largest end-to-end routing value is used as the next-hop routing, and data transmission is performed to determine whether the communication link is broken. If there is a break, the communication link break probability at the current moment is updated, and the current node routing value is transferred to the next time slot; otherwise, the current node routing value is updated, and the data packet is transmitted to the next-hop routing node, and the above steps are repeated until the destination node is reached or the packet is lost.
[0012] A second aspect of the present invention provides a power distribution Internet of Things routing selection system based on end-to-end value learning, comprising:
[0013] A node acquisition module is used to acquire all device nodes of the power distribution Internet of Things network;
[0014] A routing selection module is used to obtain a routing selection scheme for the power distribution Internet of Things based on all device nodes in the power distribution Internet of Things network and a routing selection model for the power distribution Internet of Things, and to perform data transmission based on the scheme;
[0015] The construction process of the power distribution Internet of Things routing selection model is as follows:
[0016] When the distribution IoT terminal selects the next-hop routing node, it calculates the end-to-end routing value corresponding to the hop; the routing node with the largest end-to-end routing value is used as the next-hop routing, and data transmission is performed to determine whether the communication link is broken. If there is a break, the communication link break probability at the current moment is updated, and the current node routing value is transferred to the next time slot; otherwise, the current node routing value is updated, and the data packet is transmitted to the next-hop routing node, and the above steps are repeated until the destination node is reached or the packet is lost.
[0017] A third aspect of the present invention provides a computer-readable storage medium.
[0018] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in the distribution Internet of Things routing selection method based on end-to-end value learning as described in the first aspect.
[0019] A fourth aspect of the present invention provides a computer device.
[0020] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for routing selection of a power distribution Internet of Things based on end-to-end value learning as described in the first aspect are implemented.
[0021] Compared with the prior art, the present invention has the following beneficial effects:
[0022] 1. The present invention fully considers the coupling characteristics between the communication network and the power grid. In addition to considering the impact of the communication network's own transmission performance on the power service data packet forwarding performance, it also considers the link break problem caused by electrical equipment failure or disconnection, and predicts the link break probability based on a multi-channel artificial neural network. In combination with the link transmission quality value, the routing value of different nodes is further modeled, so as to reduce the data packet loss problem caused by electrical equipment failure or disconnection as much as possible on the basis of improving the service forwarding performance.
[0023] 2. The traditional routing method only selects the next-hop route based on the current status, and considering the actual topological structure of the power distribution Internet of Things, the present invention will take into account all possible routing results from the current result to the destination result when selecting the next-hop route, thereby achieving a comprehensive consideration of the link transmission quality and the probability of link disconnection, and re-updating the link disconnection probability when a link disconnection occurs, thereby improving the reliability of business transmission. In terms of routing value learning at different nodes, by learning QoS performance such as delay, energy consumption, and reliability, and guiding the update of routing value through performance learning, the requirements of communication reliability of the power distribution Internet of Things are met.
[0024] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.
[0026] Figure 1 It is a flow chart of routing selection of power distribution Internet of Things based on end-to-end value learning provided by an embodiment of the present invention;
[0027] Figure 2 It is a flow chart of a method for predicting the probability of chain breakage based on a dual-channel artificial neural network provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0028] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.
[0029] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.
[0030] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.
[0031] In view of the technical problems mentioned in the background technology, and facing the requirements of delay processing and data transmission reliability in the complex working environment of the power distribution Internet of Things, the present invention comprehensively examines the operation of the distribution network, such as the historical time of branch connection / off-grid behavior, whether it is in the power outage operation / repair period, the three-phase imbalance of the power grid, the source-load imbalance, the weather, the power grid failure and other data, so as to establish a probability model to predict the probability of PLC link disconnection. At the same time, based on the electromagnetic interference, noise, packet loss rate, transmission power and other parameters when the link is not disconnected, the value of the link transmission quality is predicted, so as to construct the value of each routing node taking into account the influence of dual-network coupling. Then, the routing node with the largest end-to-end value is selected as the next hop node to transmit data.
[0032] Embodiment 1
[0033] like Figure 1 As shown, this embodiment provides a distribution Internet of Things routing selection method based on end-to-end value learning, including the following steps:
[0034] Step 1: Get all device nodes of the power distribution IoT network;
[0035] Step 2: Based on all device nodes of the power distribution Internet of Things network and the power distribution Internet of Things routing selection model, a power distribution Internet of Things routing selection scheme is obtained;
[0036] The construction process of the power distribution Internet of Things routing selection model is as follows:
[0037] When the power distribution IoT terminal selects the next-hop routing node, it calculates the end-to-end routing value corresponding to the hop; the routing node with the largest end-to-end routing value is used as the next-hop routing, and data transmission is performed to determine whether the communication link is broken. If there is a break, the communication link break probability at the current moment is updated, and the current node routing value is transferred to the next time slot; otherwise, the current node routing value is updated, and the data packet is transmitted to the next-hop routing node, and the above steps are repeated until the destination node is reached or the packet is lost.
[0038] Step 3: Data transmission based on the routing scheme.
[0039] In order to explain the technical content of the present invention more clearly, it is described below in conjunction with specific embodiments.
[0040] The power distribution Internet of Things involves a complex access environment, and various sensor devices are deployed in a scattered manner. The use of power line carrier communication (PLC) can fully reuse the original lines, without the need for additional wiring, so the cost is low. However, PLC cannot penetrate the equipment for transmission and is affected by the electromagnetic interference emitted by the electrical equipment. Therefore, long-distance stable communication is impossible. At this time, multiple device nodes are required to relay communication to ensure the transmission reliability of power business data. The present invention first takes into account the switching characteristics of new energy equipment and the impact of complex working environment, and conducts routing value learning taking into account the dual-network coupling characteristics.
[0041] Based on the graph theory method, the set G = {V, E} is used to represent the distribution Internet of Things network, where the set V = {v1,…,v i ,…,v I} is used to represent the power distribution IoT device node, where v1 is the source node from which the power service data packet is sent, and v I It is the destination node to which the power business data packet needs to be transmitted. The remaining nodes are routing nodes that provide relays for data packet forwarding. The nodes are connected by power lines, and the communication method is PLC.
[0042] According to the shortest path method, we can obtain the minimum number of nodes Z that each node needs to pass through to reach the destination node. i , considering the actual connection relationship of electrical equipment in the distribution network, the distribution Internet of Things topology is usually a hybrid network of tree topology, bus topology, and tree topology. In order to avoid excessive forwarding of power business data packets, when the current node performs next-hop routing selection, it will only select the route closer to the destination node. Therefore, the available link set in the network is defined as E = {e i,j |(v i ,v j ∈V,Z i >Z j )}, indicating node v i With node v j It can be directly connected through the PLC link, and the minimum number of nodes from the destination node after transmission is smaller.
[0043] The quasi-static time slot model is introduced, and the overall routing optimization cycle is divided into several time slots. The time slot set is defined as T = {1,...t,...T}. At the beginning of each time slot, the source node generates a power business data packet for transmission in the network until it is transmitted to the destination node, or when the packet is lost due to link interruption, the time slot ends and the next time slot begins. In addition, to simplify the model, it is assumed that the link transmission quality remains unchanged within a single time slot, but changes dynamically between time slots.
[0044] The routing value table of the power distribution Internet of Things is constructed for the equipment. For the source node, the transmission delay, packet loss rate and other performance of the power business data packet in the network should be ensured. Its performance guarantee capability is related to the link transmission quality of each routing node itself. Therefore, the routing value of all routing nodes in the network is defined as U i (t), each device will collect the routing values of the other routing nodes connected to it to form a routing value table for the power distribution Internet of Things to guide it to optimize routing forwarding. On the one hand, the communication performance of the power distribution Internet of Things based on the PLC communication method is related to the forwarding capability of the node itself. On the other hand, the working environment of the distribution network also has a certain impact on it. For example, the electromagnetic interference generated by electrical equipment during operation will increase the probability of bit errors and packet loss in data transmission. The link interruption problem caused by the dynamic switching of new energy equipment will even have a direct impact on the network topology.
[0045] Therefore, the routing value modeling of each node needs to fully consider the dual-network coupling characteristics. Based on the actual operation of the distribution network, events that may lead to link disconnection include power outage maintenance / emergency repair, equipment node failure and grid withdrawal, active / passive switching of new energy equipment, etc. For example, when the source and load power on the distribution network side are unbalanced, new energy power generation devices or flexible loads need to be connected to and disconnected from the grid, which will cause all PLC links related to the node to no longer be able to transmit corresponding business data. If the routing optimization at this time still indicates forwarding through this node, data packet loss will occur.
[0046] Therefore, the probability P of each link disconnection is calculated based on the historical time of branch connection / off-grid behavior, whether it is in the power outage operation / repair period, three-phase imbalance of the power grid, source-load imbalance, weather, power grid failure and other data. i (t) To make predictions, we first divide the factors that affect whether a link is broken into three categories: equipment failure factors, equipment off-grid factors, and meteorological factors. The characteristics of link interruptions caused by different reasons are significantly different, so they need to be analyzed separately. In addition, meteorological factors can be classified as one of the causes of equipment failure, so the two are combined to further simplify the prediction model. Considering the diversity and uncertainty of various types of information, based on the sample data accumulated from historical observations,
[0047] This embodiment proposes a method for predicting the probability of chain breakage based on a dual-channel artificial neural network, and its corresponding principle is as follows: Figure 2 As shown, the functions and data processing operations of different layers are as follows:
[0048] 1) Input layer: For channel 1, the input sample data set is the data of the causes of node equipment failure, such as equipment aging, power outage repair, continuous precipitation, etc. The number of neurons in the input layer is the same as the dimension of the input sample set. For channel 2, the input sample data set is the data of the causes of equipment off-grid, such as three-phase unbalance, source-load unbalance, etc. The number of neurons in the input layer is the same as the dimension of the input sample set.
[0049] 2) Hidden layer: The number of hidden layers is usually set based on experience and can be set to twice the number of neurons in the corresponding input layer.
[0050] 3) Output layer: The number of neurons in the output layer of channel one and channel two is the same as the dimension of the input sample set. The corresponding output results are used as the input of the second layer of the neural network, and after training, the predicted probability of link break is finally output.
[0051] Taking channel 1 as an example, the training process of the neural network is as follows:
[0052] 1) When the input sample data scale is 1×M, the number of neurons in the input layer is M, and the sample data after normalization corresponding to the mth input layer neuron is x m , the number of neurons in the hidden layer is N (N = 2M), the number of neurons in the output layer is M, and the connection weight between the mth input layer neuron and the nth hidden layer neuron is W m,n , the hidden layer threshold is β n , select the sigmoid function as the hidden layer transfer function, the connection weight between the nth hidden layer neuron and the mth output layer neuron is W n,m , the output layer threshold is β m .
[0053] 2) The hidden layer input is The output of the hidden layer is U n '=f(U n ), where f represents the sigmoid function, and the output layer corresponds to the output result:
[0054] 3) Calculate the output layer error as in For the expected output, if the output layer error is greater than the preset error threshold, the weights and thresholds of each layer are adjusted accordingly, and the weights of the hidden layer and the output layer and the output layer threshold are updated as follows:
[0055]
[0056] In the formula, η represents the learning rate, represents the generalized error of the output layer, i.e.
[0057] The input layer and hidden layer weights and hidden layer thresholds are updated as follows:
[0058]
[0059] In the formula, represents the generalized error of the hidden layer, that is
[0060] 4) Continue to import the next set of sample data until the error threshold meets the requirements.
[0061] The training process of channel two is the same as that of the next layer of neural networks until all neural network outputs meet the preset threshold conditions. The accuracy of the output link break probability is considered to be in line with expectations and is used to guide subsequent routing selection.
[0062] The link transmission quality is represented by the current state of the power business data packet and the selected next-hop routing node. The current state indication space mainly includes the routing node position of the power business data packet and all feasible paths to the destination node, as well as the electromagnetic interference, noise power, transmission power and other key elements corresponding to different PLC links at the current node position that will have a direct impact on the business transmission performance. That is, the state space is S i (t);
[0063]
[0064] In the formula, R i (t) is the feasible path set, which represents the set of all feasible paths from a node to a destination node, and is represented by R i,l (t) represents a specific path, δ i,j (t) is electromagnetic interference, indicating that the link e i,j The electromagnetic interference in the network will affect the link transmission quality, and the terminal will reduce the transmission performance after being interfered. i,j (t) is the noise power, indicating that the link e i,j The inherent additive Gaussian white noise power, P i TX (t) represents the transmission power, and represents the node v i The power consumption of the IoT terminal when transmitting data.
[0065] Secondly, the action vector A consists of the selectable next-hop routing nodes in the current state, denoted as A i (t), A i (t) = {vj |{v i ,e i,j ∈E}}, indicating that at the current node v i The set of next-hop routes that can be selected in this case. Therefore, the corresponding link transmission quality value is defined as Q(S i (t),A i (t)), its value is initialized to 0.
[0066] Therefore, the routing value of each routing node taking into account the influence of dual-network coupling can be defined as the product of the link transmission quality value and the link reliable transmission probability, that is, U i (t) is expressed as:
[0067] U i (t) = Q(S i (t),A i (t))·(1-P i (t))
[0068] In this embodiment, the route selection method based on end-to-end value specifically includes:
[0069] (1) The power distribution IoT terminal at the edge transmits data to the destination node hop by hop. When the power distribution IoT terminal selects the next-hop routing node, it first accumulates the routing values of the routing nodes of the feasible path between the next-hop routing node and the destination node, and then sums up the node values of all paths as the end-to-end routing value, that is,
[0070]
[0071] (2) According to the calculation results, the routing node with the largest end-to-end value is selected as the next hop route and data is transmitted, that is,
[0072]
[0073] (3) After the data is transmitted to the next hop, the probability of link disconnection at the current moment is updated based on the communication link disconnection probability model. And observe the next state S after the action is executed i '(t), and select action a according to the same rule i '(t).
[0074] (4) When a link is broken, the next-hop router cannot obtain the data packet. It is considered that the prediction of the link-break probability at the current moment is still inaccurate. Then, the sample data set that has not been trained up to the current time slot is input to further train the neural network, and the link-break probability P is updated accordingly. i (t).
[0075] Let Q(S i (t+1),Ai (t+1)) = Q(S i (t),A i (t)), and move to the next time slot.
[0076] Otherwise, update the Q value according to the following formula
[0077] Q(S i (t),a i (t)) = Q(S i (t),a i (t))+α[γQ(S i '(t),a i '(t))-Q(S i (t),a i (t))+H(S i (t),a i (t))]
[0078] In the formula, H(S i (t),a i (t)) represents the learning reward for executing the action in the current state. Due to the complex working environment of the distribution network Internet of Things, choosing different routes will have a direct impact on the service transmission performance. Considering the electromagnetic interference, noise power, transmission power and other elements associated with the previous state space, their main impact on the service forwarding performance is reflected in the delay, energy consumption and bit error rate. Therefore, the learning reward is set to the actual transmission delay τ(S i (t),a i (t)), energy consumption λ(S i (t),a i (t)), bit error rate φ(S i (t),a i (t)) is the inverse of the weighted sum of performance, i.e.
[0079]
[0080] In the formula, μ1, μ2 and μ3 are between 0 and 1, representing the weights of delay, energy consumption and bit error rate performance respectively, and satisfy μ1+μ2+μ3=1. The weights of the three can be adaptively adjusted according to the actual needs of different types of power services. However, if packet loss occurs due to the instability of the PLC communication link, the learning reward at this time is set to 0, and Q(S) is set after the Q value is updated. i (t+1),A i (t+1)) = Q(S i (t),A i (t)), and then move to the next time slot.
[0081] α is the learning rate, γ is the discount factor for learning, and multiplying this factor by future rewards can weaken the cumulative impact on the current action selection, thereby avoiding the impact of poor performance in the early learning stage.
[0082] (5) Update the routing value of each routing node according to the routing value formula, and repeat steps (1) to (5) until the data is transmitted to the destination node, and after the Q value is updated, let Q{S i (t+1),A i (t+1)}=Q{S i (t),A i (t)} and move to the next time slot.
[0083] Embodiment 2
[0084] This embodiment provides a power distribution Internet of Things routing selection system based on end-to-end value learning, including:
[0085] A node acquisition module is used to acquire all device nodes of the power distribution Internet of Things network;
[0086] A routing selection module is used to obtain a routing selection scheme for the power distribution Internet of Things based on all device nodes in the power distribution Internet of Things network and a routing selection model for the power distribution Internet of Things, and to perform data transmission based on the scheme;
[0087] The construction process of the power distribution Internet of Things routing selection model is as follows:
[0088] When the distribution IoT terminal selects the next-hop routing node, it calculates the end-to-end routing value corresponding to the hop; the routing node with the largest end-to-end routing value is used as the next-hop routing, and data transmission is performed to determine whether the communication link is broken. If there is a break, the communication link break probability at the current moment is updated, and the current node routing value is transferred to the next time slot; otherwise, the current node routing value is updated, and the data packet is transmitted to the next-hop routing node, and the above steps are repeated until the destination node is reached or the packet is lost.
[0089] Embodiment 3
[0090] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the steps in the distribution Internet of Things routing selection method based on end-to-end value learning as described above are implemented.
[0091] Embodiment 4
[0092] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the distribution Internet of Things routing selection method based on end-to-end value learning as described above are implemented.
[0093] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program code.
[0094] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0095] 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.
[0096] 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.
[0097] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.
[0098] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A routing selection method for power distribution Internet of Things based on end-to-end value learning, characterized in that: The steps include: Get all device nodes of the power distribution IoT network; Based on all device nodes of the power distribution Internet of Things network and the power distribution Internet of Things routing selection model, a power distribution Internet of Things routing selection scheme is obtained, and data transmission is performed based on the scheme; Among them, the construction process of the distribution Internet of Things routing selection model is as follows: when the distribution Internet of Things terminal selects the next-hop routing node, the end-to-end routing value corresponding to the hop is calculated, and the distribution Internet of Things terminal located at the edge transmits data to the destination node hop by hop. When the distribution Internet of Things terminal selects the next-hop routing node, the routing values of the routing nodes of the feasible paths between the next-hop routing node and the destination node are first accumulated, and the node values of all paths are summed up and regarded as the end-to-end routing value; the routing node with the largest end-to-end routing value is used as the next-hop routing, and data transmission is performed to determine whether the communication link is broken. If there is a break, the probability of the communication link breaking at the current moment is updated, and the current node routing value is transferred to the next time slot; otherwise, the current node routing value is updated, and the data packet is transmitted to the next-hop routing node, and the above steps are repeated until the destination node is reached or the packet is lost; Among them, the routing value of each routing node is defined as U i (t): U i (t)=Q(S i (t),A i (t))·(1-P i (t)), In the formula, Q(S i (t),A i (t)) is the link transmission value, S i (t) is the state space, including the routing node location of the power service data packet and all feasible paths to the destination node, as well as the electromagnetic interference, noise power and transmission power of different PLC links corresponding to the current node location; A i (t) indicates that at the current node v i A set of next-hop routes that can be selected in the case; Among them, the update formula of the current node routing value is: Q(S i (t),a i (t))=Q(S i (t),a i (t))+α[γQ(S i '(t),a i '(t))-Q(S i (t),a i (t))+H(S i (t),a i (t))], In the formula, α is the learning rate, γ is the learning discount factor, S i '(t) and a i '(t) represents the next state and action after executing the action; a i (t) is the routing node with the largest end-to-end routing value, that is where ω j (t) is the end-to-end routing value, v j Indicates that the node is v j ; H(S i (t),a i (t)) represents the learning reward for executing the action in the current state, and the calculation formula is: Among them, μ1, μ2 and μ3 are between 0 and 1, representing the weights of delay, energy consumption and bit error rate performance respectively, and satisfy μ1+μ2+μ3=1, τ(S i (t),a i (t)) is the transmission delay, λ(S i (t),a i (t)) represents energy consumption, φ(S i (t),a i (t)) represents the bit error rate.
2. The method for routing selection of power distribution Internet of Things based on end-to-end value learning according to claim 1, characterized in that: The process of determining whether a communication link is disconnected includes: Obtain sample data accumulated from historical observations; Based on the sample data accumulated from historical observations and the trained communication link disconnection probability model, a judgment result is obtained; wherein the construction process of the communication link disconnection probability model is as follows: A dual-channel artificial neural network is used. The first channel inputs the data of the causes of node equipment failure, and the second channel inputs the data of the causes of equipment disconnection. The outputs of the first channel and the second channel are used as the input of the second-layer neural network. After training, the outputs of the first channel and the second channel meet the set thresholds, and the final output is the predicted probability of chain disconnection.
3. The method for routing selection of power distribution Internet of Things based on end-to-end value learning as claimed in claim 2, characterized in that: When training the communication link disconnection probability model, the first set of sample data is input, and the error of the output layer is calculated. If the output layer error is greater than a preset error threshold, the weights and thresholds of each layer are adjusted accordingly, including: the weights of the hidden layer and the output layer, the output layer threshold update, the weights of the input layer and the hidden layer, and the hidden layer threshold update. The next set of sample data is continuously imported until the error threshold meets the requirements.
4. The method for routing selection of power distribution Internet of Things based on end-to-end value learning according to claim 1, characterized in that: All device nodes of the power distribution Internet of Things network include: a source node from which the power business data packet is sent, a destination node to which the power business data packet needs to be transmitted, and a routing node between the source node and the destination node that provides relays for data packet forwarding.
5. The method for routing selection of power distribution Internet of Things based on end-to-end value learning according to claim 1, characterized in that: During the data transmission process, a quasi-static time slot model is introduced. The link transmission quality remains unchanged within a single time slot, but changes dynamically between time slots. The overall routing optimization cycle is divided into several time slots. At the beginning of each time slot, the source node generates a power business data packet for transmission in the network until it is transmitted to the destination node or the packet is lost due to link interruption, at which time the time slot ends.
6. A power distribution Internet of Things routing selection system based on end-to-end value learning, adopting a power distribution Internet of Things routing selection method based on end-to-end value learning as claimed in any one of claims 1 to 5, characterized in that: include: A node acquisition module is used to acquire all device nodes of the power distribution Internet of Things network; A routing selection module is used to obtain a routing selection scheme for the power distribution Internet of Things based on all device nodes in the power distribution Internet of Things network and a routing selection model for the power distribution Internet of Things, and to perform data transmission based on the scheme; The construction process of the power distribution Internet of Things routing selection model is as follows: When the power distribution IoT terminal selects the next-hop routing node, it calculates the end-to-end routing value corresponding to the hop. The power distribution IoT terminal at the edge transmits data to the destination node hop by hop. When the power distribution IoT terminal selects the next-hop routing node, it first accumulates the routing values of the routing nodes of the feasible path between the next-hop routing node and the destination node, and then sums up the node values of all paths and regards them as the end-to-end routing value. The routing node with the largest end-to-end routing value is used as the next-hop routing for data transmission. It is determined whether the communication link is broken. If there is a break, the communication link break probability at the current moment is updated and the current node routing value is transferred to the next time slot; otherwise, the current node routing value is updated and the data packet is transmitted to the next-hop routing node. The above steps are repeated until the destination node is reached or the packet is lost.
7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the power distribution Internet of Things routing selection method based on end-to-end value learning as described in any one of claims 1-5 are implemented.
8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the power distribution Internet of Things routing selection method based on end-to-end value learning as described in any one of claims 1-5 are implemented.
