A bayesian dynamic game intelligent electric meter edge data transmission method and system

Abstract

The application relates to the technical field of edge data transmission, and discloses a smart meter edge data transmission method and system based on Bayesian dynamic game, which comprises the following steps: mapping channel state data of a meter to a frequency band set according to a center frequency of the channel, so as to serve as a frequency band resource pool of the meter; using a Bayesian probability graph model to infer time and space distribution characteristics of power consumption behaviors of a user, and outputting a power consumption activity probability distribution graph of the user; constructing a non-cooperative game model facing frequency band allocation and solving the model to obtain an optimal frequency band use strategy of each meter, performing meter working channel allocation, and using the allocated meter working channel to perform edge data transmission. The meter channel state and historical power consumption data of a user are collected, a Bayesian graph model is constructed to infer a user activity probability distribution, the user activity probability distribution is introduced into a non-cooperative game model, an optimal channel allocation strategy of each meter is realized through a Bayesian Nash equilibrium solver, and the edge data transmission efficiency is improved.

Description

Technical Field

[0001] This invention relates to the technical field of edge data transmission, and in particular to a method and system for edge data transmission of smart meters using Bayesian dynamic game theory. Background Technology

[0002] With the continuous growth of global energy demand, traditional power systems are facing increasing pressure. Energy efficiency and the sustainability of power transmission have become key issues in modern power systems. The emergence of smart grid technology offers a solution to this problem, and smart meters, as a core component of smart grids, are being widely deployed globally. By collecting real-time electricity consumption data, smart meters provide power companies with detailed load information, thereby enabling load forecasting, energy efficiency optimization, and dynamic scheduling of power resources.

[0003] However, in the widespread application of smart meters, information transmission has become a problem that cannot be ignored. Smart meters need to transmit a large amount of electricity data between power companies and users via wireless communication networks, and factors such as the channel quality, transmission distance, transmission rate of the wireless communication network, and user behavior all directly affect the efficiency and stability of smart meter data transmission.

[0004] In existing research, CN113533809A proposes a multimedia smart meter and data transmission system, which realizes the parsing and display of existing display data, and receives the meter button response to realize the switching between carousel display and meter display. CN113625031B involves a smart meter terminal and its data reporting system and method, which optimizes the meter data reporting mechanism and improves the efficiency and reliability of data transmission. However, it lacks a meter channel optimization method that comprehensively considers the meter channel utility and user utility, thereby improving the efficiency of edge data transmission.

[0005] Therefore, optimizing the channels of smart meters to enable them to transmit data effectively in complex communication environments has become a pressing technical challenge. Summary of the Invention

[0006] In view of this, the present invention provides a method and system for edge data transmission of smart meters using Bayesian dynamic game theory. The method collects the current channel state of the meter and maps it to a frequency band resource pool. Combined with the user's historical electricity consumption data, the method uses a Bayesian probabilistic graphical model to infer the activity distribution, and then constructs a non-cooperative game model to jointly model user activity and channel state. Through iterative solution using Bayesian Nash equilibrium, an efficient and conflict-controllable meter channel optimization strategy is obtained. Edge data transmission is then performed using the optimized meter working channel.

[0007] To achieve the above objectives, this invention provides a Bayesian dynamic game-based method for edge data transmission in smart meters, comprising the following steps:

[0008] S1: Collect channel status data of the electricity meter and map the channel status data to a frequency band set according to the center frequency of the channel, which serves as the frequency band resource pool of the electricity meter;

[0009] S2: Collect users' historical electricity consumption data, use Bayesian probabilistic graphical models to infer the spatiotemporal distribution characteristics of users' electricity consumption behavior, and output the probability distribution map of users' electricity consumption activity.

[0010] S3: Using the frequency band resource pool of the electricity meter and the probability distribution map of the user's electricity consumption as strategy resources, construct a non-cooperative game model for frequency band allocation;

[0011] S4: Based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each meter, allocate the meter working channels, and use the allocated meter working channels for edge data transmission.

[0012] As a further improvement of the present invention:

[0013] Optionally, an embedded communication module is deployed in the electricity meter, and the embedded communication module is used to scan the available channels of the electricity meter and evaluate the status of the available channels, including:

[0014] The embedded communication module includes a wireless receiver, a received signal strength indicator, and a signal-to-noise ratio measuring device;

[0015] The transmitter periodically transmits signals to all channels, and the wireless receiver receives the channel signals of each channel one by one, and records the signal strength of the channel signal using a received signal strength indicator as the signal strength of the channel.

[0016] Channels with signal strength higher than a preset strength threshold are designated as available channels. A signal-to-noise ratio (SNR) measurement device is used to record the SNR of the channel signal in the available channels. The signal strength and SNR of the available channels are used as the status of the available channels. The SNR measurement device is embedded with a low-pass filter to perform low-pass filtering on the channel signal to obtain a denoised channel signal. The changes in the channel signal before and after denoising are compared as noise signals in the channel. The signal power P1 and noise signal power P2 are calculated, and the SNR of the channel, 10lg(P1 / P2), is obtained.

[0017] Extract the status of all available channels in the meter to form the meter's channel status data.

[0018] Optionally, the channel state data is mapped to a set of frequency bands according to the center frequency of the channel, serving as the frequency band resource pool for the meter, including:

[0019] Extract the center frequency of available channels from the channel state data, and map the state of available channels to different frequency bands according to the center frequency. The set of frequency bands is: {B d |d∈[1,D]}, where B d This refers to the d-th frequency band in the frequency band set, where D represents the total number of frequency bands in the set. The frequency band B... d The frequency range is If the center frequency of the available channel Then the state of the available channels will be mapped to frequency band B. d ,in Indicates the d-th frequency band B d The lowest frequency in Indicates the d-th frequency band B d The highest frequency in;

[0020] The frequency band resource pool of the electricity meter includes frequency band resources of D frequency bands, and each frequency band resource consists of the state of the available channels after mapping.

[0021] Specifically, each frequency band includes several channels. Taking the 2.4 GHz band as an example, there are 14 channels, each with a bandwidth of 22 MHz.

[0022] Optionally, a Bayesian probabilistic graphical model is used to infer the spatiotemporal distribution characteristics of users' electricity consumption behavior, including:

[0023] A Bayesian probabilistic graphical model is constructed based on the user's historical electricity consumption data. The user's historical electricity consumption data refers to the power consumption recorded by the user's associated electricity meter in different time periods over the past G days. The Bayesian probabilistic graphical model is a graph model structure, consisting of graph nodes and the relationships between graph nodes. The graph nodes include time periods and users, and the relationships between graph nodes include the user's electricity consumption activity status in different time periods, historical electricity consumption data, and prior information about the user's electricity consumption behavior characteristics.

[0024] A Bayesian probabilistic graphical model is used to calculate the relationships between graph nodes. A joint probabilistic model is constructed with user electricity activity levels at different time periods as latent variables, user electricity behavior characteristics as prior driving variables, and historical electricity consumption data as observed variables. The latent variables are unobservable random variables.

[0025]

[0026] Where A represents the latent variable in the joint probability model, U represents the prior driving variable, Y represents the observed variable, and Pr(user) n ) represents the prior characteristics of the electricity consumption behavior of the nth user, user nThis represents the electricity consumption behavior characteristics of the nth user, where N represents the total number of users, and A... n,g,b Y represents the electricity activity status of the nth user in the b-th time period on day g, where b∈[1,B], B represents the total number of time periods, n∈[1,N], N represents the total number of users, g∈[1,G], and Y n,g,b Pr(A) represents the power consumption of the nth user during the b-th time period on day g. n,g,b user n ,Y n,g,b ) represents the prior of the active electrical state, Pr(Y) n,g,b |A n,g,b ) represents a given active electricity consumption state A n,g,b The observed power consumption Y n,g,b Prior observational data;

[0027] Using the complete factor assumption, i.e., all latent variables and variables satisfy variational independence, a variational distribution q(A) of latent variable A is generated. This variational distribution is applied in the active electricity consumption state A. n,g,b The variational probability is q(A) n,g,b A n,g,b ∈A, where the variational probability is a value between 0 and 1, and the higher the value, the more active the electricity consumption;

[0028] The variational inference method is used to iterate the variational probability and the prior of electricity consumption behavior features until the variational probability and the prior of electricity consumption behavior features do not change. The variational probability of all active electricity consumption states is extracted as the spatiotemporal distribution feature of the user's electricity consumption behavior.

[0029] Optionally, based on the spatiotemporal distribution characteristics of the user's electricity consumption behavior, a probability distribution map of the user's electricity consumption activity is output, including:

[0030] Calculate the mean variational probability of a user's electricity activity state in each time period to construct the probability distribution map of the user's electricity activity: (A(n,b)) N×B , where A(n,b) represents the mean variational probability of the active electricity consumption state of the nth user in the bth time period, n∈[1,N], N represents the total number of users, b∈[1,B], and B represents the total number of time periods.

[0031] Optionally, the frequency band resource pool of the electricity meter and the probability distribution map of the user's electricity consumption activity are used as policy resources to construct a non-cooperative game model for frequency band allocation, including:

[0032] Specifically, each user is associated with an electricity meter. Steps S1 and S2 respectively collect the frequency band resource pool of the electricity meter associated with the user and the probability distribution of the user's electricity consumption activity.

[0033] The electricity activity probability distribution map is composed of the electricity activity probability distributions of N users. The non-cooperative game model uses the electricity activity probability distribution of each user and the frequency band resource pool of the electricity meter associated with the user as the user's strategy resources, and the frequency band allocation coefficient as the strategy variable. It also constructs a meter utility function based on the frequency band resource pool and a user utility function based on the electricity activity probability distribution, where the expression for the meter utility function is:

[0034]

[0035] C(i)=L(i)·log2(1+SINR(i));

[0036] Where F1(·) represents the meter utility function, the input value of which is the frequency band resource allocated to the meter, A(n) represents the frequency band resource allocated to the meter associated with the nth user, the frequency band resource consists of the states of the available channels after mapping, i∈A(n), i represents the state of the available channels in the frequency band resource A(n), SINR(i) represents the signal-to-noise ratio in state i, S(i) represents the signal strength in state i, and max{1+C(i)+S(i)|i∈A(n)} represents 1+C(i)+

[0037] The maximum value in S(i), C(i) represents the Shannon capacity corresponding to state i, f(A(n)) represents the center frequency of the frequency band associated with the frequency band resource A(n), f(A(n,near)) represents the center frequency of the frequency band associated with the frequency band resource allocated to the nearest user of the nth user, λ represents the adjacent frequency band suppression factor, Δf represents the unit center frequency, n∈[1,N], N represents the total number of users, L(i) represents the bandwidth of the available channel associated with state i of the available channel, C_Max represents the preset maximum Shannon capacity, and S_Max represents the preset maximum signal strength;

[0038] The expression for the user utility function is:

[0039]

[0040] Where F2(·) represents the user utility function, which takes the probability distribution of the user's electricity activity and the frequency band resource pool of the user's associated electricity meter as input, A (n) Let d represent the mean variational probability of the nth user's active electricity consumption state during the current time period. n Let d represent the frequency band resources associated with the nth user's meter, where d ∈ [1, D], and D represents the total number of frequency band resources for each meter. Indicates frequency band resources d n The normalized mean signal-to-noise ratio of the available channels. SINR_MAX represents the preset maximum signal-to-noise ratio. Indicates frequency band resources d n The average signal-to-noise ratio of available channels. This represents the frequency band resource d allocated to the electricity meter associated with the nth user. n The distribution probability, j∈[1,N], A (j) Let d represent the mean variational probability of the j-th user's active electricity consumption status in the current time period. j This represents the j-th frequency band resource associated with the electricity meter of the j-th user.

[0041] Optionally, based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each meter, including:

[0042] Based on Bayesian Nash equilibrium theory, a leader and followers are generated in a non-cooperative game model. The leader, based on the user utility function, prioritizes allocating frequency band resources to highly active users. The followers are electricity meters, which select channels that maximize their own utility from the frequency band resources allocated by the leader based on their utility function. Specifically, the electricity meters, as participants in an incomplete information game, can only make strategy choices based on their own observations (such as signal strength and signal-to-noise ratio). Each meter can only independently decide whether to connect to a certain channel based on the principle of maximizing its own interests. This mechanism is suitable for large-scale deployment of electricity meter networks, reduces the system's dependence on centralized control, makes the overall communication more distributed and autonomous, and enables the electricity meters to continuously adjust their strategies based on feedback data during long-term operation using Bayesian Nash equilibrium theory, gradually approaching the optimal real-time channel, thus constructing a closed-loop channel game optimization method of perception-decision-feedback.

[0043] Based on the user utility function and the meter utility function, a dynamic game theory approach is used to iterate the allocation probability and result of frequency band resources for each user's associated meter until both converge, yielding the frequency band allocation coefficient for each meter, where the frequency band allocation coefficient represents the frequency band resources allocated to the meter. Specifically, the iterative strategy of the dynamic game theory approach is the alternating direction multiplier method, which divides the frequency band resource allocation method into two sub-problems: one led by the meter side, optimizing its frequency band access performance; the other led by user activity, guiding the priority of frequency band resource allocation. Multiplier terms are constructed for the user utility function and the meter utility function. These two sub-problems are coupled and coordinated through multiplier terms. In each iteration, the meter updates its frequency band usage strategy based on the current frequency band resources and multiplier feedback, while the user updates their demand allocation strategy based on their current activity status and resource response. The multiplier is then dynamically adjusted based on the difference between the two, gradually approaching the equilibrium solution of resource allocation, where the multiplier term represents the utilization rate of frequency band resources.

[0044] Optionally, the optimal frequency band usage strategy is used to allocate working channels for the electricity meter, including:

[0045] The channel with the highest signal strength and the highest signal-to-noise ratio, which is higher than a preset first strength threshold, is selected from the optimal frequency band usage strategy and used as the working channel of the electricity meter.

[0046] To address the aforementioned problems, this invention provides a smart meter edge data transmission system, comprising a server and a data acquisition device. The server includes an electricity consumption activity quantification module and a channel optimization module.

[0047] The electricity activity quantification module is used to infer the spatiotemporal distribution characteristics of users' electricity consumption behavior using a Bayesian probabilistic graphical model, and outputs a probability distribution map of users' electricity activity.

[0048] The channel optimization module is used to construct a non-cooperative game model for frequency band allocation by using the frequency band resource pool of the electricity meter and the probability distribution map of the user's electricity consumption as strategy resources. Based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each electricity meter, allocate the working channel of the electricity meter, and use the allocated working channel of the electricity meter for edge data transmission.

[0049] The data acquisition device is used to collect channel status data of the electricity meter and the user's electricity consumption history data, and to map the channel status data to a frequency band set according to the center frequency of the channel, which serves as the frequency band resource pool of the electricity meter. The channel status data includes the signal strength and signal-to-noise ratio of different channels.

[0050] To achieve any of the aforementioned Bayesian dynamic game-based edge data transmission methods for smart meters.

[0051] To address the above problems, the present invention provides an electronic device, the electronic device comprising:

[0052] Memory, storing at least one instruction;

[0053] Communication interfaces enable communication between electronic devices; and

[0054] The processor executes the instructions stored in the memory to implement the Bayesian dynamic game-based edge data transmission method for smart meters described above.

[0055] To address the aforementioned issues, the present invention also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor in an electronic device to implement the aforementioned Bayesian dynamic game-based edge data transmission method for smart meters.

[0056] Compared with existing technologies, this invention proposes a Bayesian dynamic game-based method and system for edge data transmission in smart meters, which has the following advantages:

[0057] Firstly, existing frequency band allocation mechanisms mostly ignore the usage behavior characteristics of end users, optimizing only based on static channel characteristics (such as signal-to-noise ratio). This easily leads to a disconnect between resource allocation and user behavior, resulting in low frequency band utilization and high communication conflicts. This application introduces a collaborative modeling of user behavior activity distribution and multi-parameter channel utility functions, achieving for the first time a refined game-driven frequency band allocation at the meter level, effectively alleviating communication congestion and improving allocation fairness and adaptability. The proposed meter utility function comprehensively considers parameters such as bandwidth capacity, signal strength, signal-to-noise ratio under channel conditions, and frequency offset between the meter and the frequency band center, forming a comprehensive evaluation mechanism for frequency band resource quality. This function not only considers the quality of a single channel but also designs a nearest neighbor interference evaluation function. This approach aims to limit the clustering and allocation of frequency bands, thereby improving the overall communication reliability of the electricity meter system. The proposed user utility function uses the probability distribution of user electricity consumption activity as a key input variable to dynamically drive the matching and optimization of frequency band resources. It innovatively introduces the probability distribution of user electricity consumption activity into the communication optimization process. Unlike the traditional allocation logic centered on traffic, it no longer uses the current traffic demand as the sole basis. Instead, it introduces the probability of electricity consumption activity of the user's associated electricity meter at different times as an important reference variable, realizing a "predictive" and "prophetic" resource scheduling logic. This helps to seize clean frequency bands, avoid interference, and improve overall optimization efficiency.

[0058] Meanwhile, current channel optimization mechanisms are mostly static planning or one-off game theory, which are difficult to adapt to multi-user, multi-time-period, and dynamic access environments, and lack policy evolution and feedback mechanisms. This application overcomes the shortcomings of existing methods, such as poor real-time performance, slow convergence, and poor user adaptability, by introducing a dynamic Stackelberg game and frequency band probability vector evolution mechanism. It achieves adaptive and collaborative distributed optimization of frequency band resources, greatly improving the intelligence level and robustness of frequency band allocation. By guiding frequency band resources towards the utility-optimal direction through the Stackelberg leader strategy, a game optimization process with controllable upstream and downstream convergence is constructed. The game evolution result is mapped to the frequency band allocation strategy through the allocation probability update formula, and smooth adjustments are made based on historical strategies in each iteration to avoid getting trapped in local optima. A perturbation factor is introduced into the formula. While improving efficiency, the exploratory nature of the policy space is enhanced. During the update of frequency band resource allocation results, L1 norm distance constraints are introduced to ensure that the frequency band allocation results remain relatively balanced with the user's historical policies and control drastic policy fluctuations. This mechanism helps to reduce channel interference problems caused by sudden centralized access and can improve system stability. Attached Figure Description

[0059] Figure 1 This is a flowchart illustrating a Bayesian dynamic game-based edge data transmission method for smart meters, as provided in an embodiment of the present invention.

[0060] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0061] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0062] This application provides a method for edge data transmission of smart meters using Bayesian dynamic game theory. The execution entity of this method includes, but is not limited to, at least one electronic device that can be configured to execute the method provided in this application, such as a server or a terminal. In other words, the method can be executed by software or hardware installed on a terminal device or a server device, where the software may be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0063] Reference Figure 1 Embodiment 1 of the present invention is as follows:

[0064] A Bayesian dynamic game-based method for edge data transmission in smart meters includes the following steps:

[0065] S1: Collect channel status data of the electricity meter and map the channel status data to a frequency band set according to the center frequency of the channel, which serves as the frequency band resource pool for the electricity meter.

[0066] An embedded communication module is deployed in the electricity meter. This embedded communication module is used to scan the meter's available channels and assess the status of the available channels, including:

[0067] The embedded communication module includes a wireless receiver, a received signal strength indicator, and a signal-to-noise ratio measuring device;

[0068] The transmitter periodically transmits signals to all channels, and the wireless receiver receives the channel signals of each channel one by one, and records the signal strength of the channel signal using a received signal strength indicator as the signal strength of the channel; as an embodiment of the present invention, the transmitter is a signal base station;

[0069] Channels with signal strength higher than a preset strength threshold are designated as available channels. A signal-to-noise ratio (SNR) measurement device is used to record the SNR of the channel signal in the available channels. The signal strength and SNR of the available channels are used as the status of the available channels. The SNR measurement device is embedded with a low-pass filter to perform low-pass filtering on the channel signal to obtain a denoised channel signal. The changes in the channel signal before and after denoising are compared as noise signals in the channel. The signal power P1 and noise signal power P2 are calculated, and the SNR of the channel, 10lg(P1 / P2), is obtained.

[0070] Extract the status of all available channels in the meter to form the meter's channel status data.

[0071] The channel state data is mapped to a set of frequency bands according to the center frequency of the channel, forming a frequency band resource pool for the electricity meter, including:

[0072] Extract the center frequency of available channels from the channel state data, and map the state of available channels to different frequency bands according to the center frequency. The set of frequency bands is: {B d |d∈[1,D]}, where B d This refers to the d-th frequency band in the frequency band set, where D represents the total number of frequency bands in the set. The frequency band B... d The frequency range is If the center frequency of the available channel Then the state of the available channels will be mapped to frequency band B. d ,in Indicates the d-th frequency band B d The lowest frequency in Indicates the d-th frequency band B d The highest frequency in;

[0073] The frequency band resource pool of the electricity meter includes frequency band resources in D frequency bands, and each frequency band resource consists of the state of the available channels after mapping. As an embodiment of the present invention, the channels of all electricity meters are divided into D frequency bands, and the frequency band resources of the same frequency band under each electricity meter are not necessarily the same;

[0074] Specifically, each frequency band includes several channels. Taking the 2.4 GHz band as an example, there are 14 channels, each with a bandwidth of 22 MHz.

[0075] S2: Collect users' historical electricity consumption data, use a Bayesian probabilistic graphical model to infer the spatiotemporal distribution characteristics of users' electricity consumption behavior, and output the probability distribution map of users' electricity consumption activity.

[0076] The spatiotemporal distribution characteristics of users' electricity consumption behavior are inferred using a Bayesian probabilistic graphical model, including:

[0077] A Bayesian probabilistic graphical model is constructed based on the user's historical electricity consumption data. The user's historical electricity consumption data refers to the power consumption recorded by the user's associated electricity meter in different time periods over the past G days. The Bayesian probabilistic graphical model is a graph model structure, consisting of graph nodes and the relationships between graph nodes. The graph nodes include time periods and users, and the relationships between graph nodes include the user's electricity consumption activity status in different time periods, historical electricity consumption data, and prior information about the user's electricity consumption behavior characteristics.

[0078] A Bayesian probabilistic graphical model is used to calculate the relationships between graph nodes. A joint probabilistic model is constructed with user electricity activity levels at different time periods as latent variables, user electricity behavior characteristics as prior driving variables, and historical electricity consumption data as observed variables. The latent variables are unobservable random variables.

[0079]

[0080] Where A represents the latent variable in the joint probability model, U represents the prior driving variable, Y represents the observed variable, and Pr(user) n ) represents the prior characteristics of the electricity consumption behavior of the nth user, user n This represents the electricity consumption behavior characteristics of the nth user, where N represents the total number of users, and A... n,g,b Y represents the electricity activity status of the nth user in the b-th time period on day g, where b∈[1,B], B represents the total number of time periods, n∈[1,N], N represents the total number of users, g∈[1,G], and Y n,g,b Pr(A) represents the power consumption of the nth user during the b-th time period on day g. n,g,b user n ,Y n,g,b ) represents the prior of the active electrical state, Pr(Y) n,g,b |A n,g,b ) represents a given active electricity consumption state A n,g,b The observed power consumption Y n,g,b Prior observational data;

[0081] Using the full factorial assumption, i.e., all latent variables and variables satisfy variational independence, the variational distribution q(A) of latent variable A is generated, where the electricity consumption activity state A n,g,b The variational probability is q(A) n,g,b A n,g,b ∈A, where the variational probability is a value between 0 and 1, and the higher the value, the more active the electricity consumption;

[0082] The variational inference method is used to iterate the variational probability and the prior of electricity consumption behavior features until the variational probability and the prior of electricity consumption behavior features do not change. The variational probability of all active electricity consumption states is extracted as the spatiotemporal distribution feature of the user's electricity consumption behavior.

[0083] As an embodiment of the present invention, a mean-field variational update method is used to perform EM loop iteration on the variational probability and the prior characteristics of electricity consumption behavior, wherein the variational probability q(A) n,g,b The iterative update formula for ) is:

[0084]

[0085] Where, q * (A n,g,b ) represents the variational probability q(A) n,g,b The updated results of ) Pr(user) represents the prior characteristics of fixed electricity consumption behavior. n The formula for calculating the expected value of )

[0086] The prior Pr(user) of electricity consumption behavior characteristics n The update formula for ) is:

[0087]

[0088] Among them, Pr * (user n Pr(user) represents the prior characteristics of electricity consumption behavior. n The iterative update results of ) Represents the fixed variational probability q * (A n,g,b The formula for calculating the constant expected value;

[0089] Specifically, the user's electricity consumption behavior characteristics include the user's average daily electricity consumption and the variance of electricity consumption, wherein Pr(user) n Let be a Gaussian distribution representing the electricity consumption behavior characteristics of the nth user;

[0090] As an embodiment of the present invention, the prior of the active electricity consumption state follows a Bernoulli distribution, the prior of the observed data follows a Gaussian distribution, and the probability distribution of the prior of the electricity consumption behavior features follows a Gaussian distribution.

[0091]

[0092] Pr(A n,g,b user n ,Y n,g,b ) = Bernouli(A n,g,b |sigmoid(w T [usern ,Y n,g,b ]);

[0093] in, The active electrical state is indicated as A. n,g,b The average power consumption of all users during the specified time period. The active electrical state is indicated as A. n,g,b The variance of power consumption of all users over a given time period The mean is μ n The variance is The distribution is Gaussian, w represents the feature map vector, and T represents the transpose.

[0094] sigmoid(·) represents the Sigmoid function, Bernouli(·) represents the Bernoulli distribution, and μ n This represents the average daily power consumption of the nth user. This represents the variance of the daily power consumption of the nth user.

[0095] Specifically, this application integrates Bayesian networks with user historical electricity consumption behavior modeling to establish a joint probability model. It performs variational inference on each user's "electricity consumption behavior feature vector" and "behavior probability distribution" in the spatiotemporal dimensions, creating a refined probabilistic profile of user active behavior. This structure explicitly expresses the activity level estimate of each user at different times, greatly improving the temporal and spatial resolution of user modeling. It employs a variational EM iterative optimization method, approximating latent variables and jointly modeling the prior and posterior of electricity consumption behavior features. The prior distribution of user active states is Gaussian, and posterior modeling is performed using observed variables. The observed variables are connected to the Bernoulli output using a Sigmoid function, facilitating the handling of activity classification tasks. Compared to conventional static modeling, this method is structurally scalable, adaptable to massive amounts of time-series electricity consumption data and user heterogeneity, and forms the basis for achieving high-precision spectrum scheduling and channel selection.

[0096] Based on the spatiotemporal distribution characteristics of the user's electricity consumption behavior, an output probability distribution map of the user's electricity consumption activity is generated, including:

[0097] Calculate the mean variational probability of a user's electricity activity state in each time period to construct the probability distribution map of the user's electricity activity: (A(n,b)) N×BWhere A(n,b) represents the variational probability mean of the electricity consumption activity state of the nth user in the b-th time period. In this embodiment of the invention, the variational probability mean A(n,b) is the sum of the variational probabilities of the electricity consumption activity state of the nth user in the b-th time period of G days divided by G, n∈[1,N], N represents the total number of users, b∈[1,B], and B represents the total number of time periods.

[0098] S3: Using the frequency band resource pool of electricity meters and the probability distribution map of users' electricity consumption as strategy resources, construct a non-cooperative game model oriented towards frequency band allocation.

[0099] Using the frequency band resource pool of electricity meters and the probability distribution map of users' electricity consumption activity as policy resources, a non-cooperative game model for frequency band allocation is constructed, including:

[0100] Specifically, each user is associated with an electricity meter. Steps S1 and S2 respectively collect the frequency band resource pool of the electricity meter associated with the user and the probability distribution of the user's electricity consumption activity.

[0101] The electricity activity probability distribution map is composed of the electricity activity probability distributions of N users. The non-cooperative game model uses the electricity activity probability distribution of each user and the frequency band resource pool of the electricity meter associated with the user as the user's strategy resources, and the frequency band allocation coefficient as the strategy variable. It also constructs a meter utility function based on the frequency band resource pool and a user utility function based on the electricity activity probability distribution, where the expression for the meter utility function is:

[0102]

[0103] C(i)=L(i)·log2(1+SINR(i));

[0104] Where F1(·) represents the meter utility function, the input value of which is the frequency band resource allocated to the meter, A(n) represents the frequency band resource allocated to the meter associated with the nth user, the frequency band resource consists of the states of the available channels after mapping, i∈A(n), i represents the state of the available channels in the frequency band resource A(n), SINR(i) represents the signal-to-noise ratio in state i, S(i) represents the signal strength in state i, and max{1+C(i)+S(i)|i∈A(n)} represents 1+C(i)+

[0105] The maximum value in S(i), C(i) represents the Shannon capacity corresponding to state i, f(A(n)) represents the center frequency of the frequency band associated with the frequency band resource A(n), f(A(n,near)) represents the center frequency of the frequency band associated with the frequency band resource allocated to the nearest user of the nth user, λ represents the adjacent frequency band suppression factor, Δf represents the unit center frequency, n∈[1,N], N represents the total number of users, L(i) represents the bandwidth of the available channel associated with state i of the available channel, C_Max represents the preset maximum Shannon capacity, and S_Max represents the preset maximum signal strength;

[0106] The expression for the user utility function is:

[0107]

[0108] Where F2(·) represents the user utility function, which takes the probability distribution of the user's electricity activity and the frequency band resource pool of the user's associated electricity meter as input, A (n) Let d represent the mean variational probability of the nth user's active electricity consumption state during the current time period. n Let d represent the frequency band resources associated with the nth user's meter, where d ∈ [1, D], and D represents the total number of frequency band resources for each meter. Indicates frequency band resources d n The normalized mean signal-to-noise ratio of the available channels. SINR_MAX represents the preset maximum signal-to-noise ratio. Indicates frequency band resources d n The average signal-to-noise ratio of available channels. This represents the frequency band resource d allocated to the electricity meter associated with the nth user. n The distribution probability, j∈[1,N], A (j) Let d represent the mean variational probability of the j-th user's active electricity consumption status in the current time period. j This represents the j-th frequency band resource associated with the j-th user's meter. As an embodiment of the present invention, the smaller |f(A(n))-f(A(n,near))| is, the greater the probability that adjacent meters are allocated the same frequency band resource, the greater the probability that adjacent meters use the same channel, and thus the channels of adjacent meters interfere with each other.

[0109] S4: Based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each meter, allocate the meter working channels, and use the allocated meter working channels for edge data transmission.

[0110] Based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each meter, including:

[0111] Based on Bayesian Nash equilibrium theory, a leader and followers are generated in a non-cooperative game model. The leader allocates frequency band resources to users with high activity based on the user utility function. The followers are electricity meters. The followers select the channel that maximizes their own utility from the frequency band resources allocated by the leader based on the electricity meter utility function according to the frequency band resource allocation result.

[0112] Based on the user utility function and the meter utility function, a dynamic game approach is used to iterate the allocation probability and allocation result of the frequency band resources associated with each user's meter until both converge, thus obtaining the frequency band allocation coefficient for each meter, where the frequency band allocation coefficient represents the frequency band resources allocated to the meter.

[0113] The optimal frequency band usage strategy is used to allocate working channels for electricity meters, including:

[0114] The channel with the highest signal strength and the highest signal-to-noise ratio, which is higher than a preset first strength threshold, is selected from the optimal frequency band usage strategy and used as the working channel of the electricity meter.

[0115] Example 2:

[0116] This embodiment provides a smart meter edge data transmission system, which includes a server and a data acquisition device. The server includes an electricity activity quantification module and a channel optimization module.

[0117] The electricity activity quantification module is used to infer the spatiotemporal distribution characteristics of users' electricity consumption behavior using a Bayesian probabilistic graphical model, and outputs a probability distribution map of users' electricity activity.

[0118] The channel optimization module is used to construct a non-cooperative game model for frequency band allocation by using the frequency band resource pool of the electricity meter and the probability distribution map of the user's electricity consumption as strategy resources. Based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each electricity meter, allocate the working channel of the electricity meter, and use the allocated working channel of the electricity meter for edge data transmission.

[0119] The data acquisition device is used to collect channel status data of the electricity meter and the user's electricity consumption history data, and to map the channel status data to a frequency band set according to the center frequency of the channel, which serves as the frequency band resource pool of the electricity meter. The channel status data includes the signal strength and signal-to-noise ratio of different channels.

[0120] To implement the Bayesian dynamic game-based edge data transmission method for smart meters in Example 1.

[0121] Example 3:

[0122] This scheme compares the Bayesian dynamic game-based smart meter edge data transmission method, the greedy channel allocation method, the round-robin channel allocation method, and the traditional game-based channel optimization method through comparative experiments.

[0123] In the greedy channel allocation method, each meter always selects the channel with the highest signal-to-noise ratio, and there is no global conflict coordination mechanism.

[0124] The rotating channel allocation method uses a channel rotation mechanism to avoid long-term fixed occupation without considering user state examples;

[0125] Traditional game theory channel optimization methods use game theory to make a one-time channel selection, without dynamic updates or user behavior awareness;

[0126] The results of the comparative experiment are shown in Table 1:

[0127] Table 1

[0128]

[0129] As shown in Table 1, the Bayesian dynamic game-based smart meter edge data transmission method achieves a higher communication signal-to-noise ratio by dynamically learning the channel state and considering user activity. By utilizing game equilibrium optimization and activity distribution guidance, it significantly suppresses frequency band conflicts among multiple meters. Bayesian modeling can identify highly active users, prioritize their frequency band quality resources, and improve the efficiency of edge data transmission.

[0130] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0131] It should be noted that the sequence numbers of the above embodiments of the present invention are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0132] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0133] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A Bayesian dynamic game-based method for edge data transmission in smart meters, characterized in that, The method includes: S1: Collect channel status data of the electricity meter and map the channel status data to a frequency band set according to the center frequency of the channel, which serves as the frequency band resource pool of the electricity meter. The channel status data includes the signal strength and signal-to-noise ratio of different channels. S2: Collect users' historical electricity consumption data, use Bayesian probabilistic graphical models to infer the spatiotemporal distribution characteristics of users' electricity consumption behavior, and output the probability distribution map of users' electricity consumption activity. S3: Using the frequency band resource pool of the electricity meter and the probability distribution map of the user's electricity consumption as strategy resources, construct a non-cooperative game model for frequency band allocation; S4: Based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each meter, allocate the meter working channels, and use the allocated meter working channels for edge data transmission. Using the frequency band resource pool of electricity meters and the probability distribution map of users' electricity consumption activity as policy resources, a non-cooperative game model for frequency band allocation is constructed, including: The electricity consumption activity probability distribution map is composed of the electricity consumption activity probability distribution of N users; The non-cooperative game model uses the probability distribution of each user's electricity consumption activity and the frequency band resource pool of the user's associated electricity meter as the user's strategy resources, and the frequency band allocation coefficient as the strategy variable. It constructs a meter utility function based on the frequency band resource pool and a user utility function based on the probability distribution of electricity consumption activity, where the expression for the meter utility function is: ； ； in, This represents the meter utility function, whose input value is the frequency band resources allocated to the meter. This represents the frequency band resources allocated to the electricity meter associated with the nth user, whereby the frequency band resources consist of the state of the available channels after mapping. , Indicates frequency band resources The status of available channels in the system. Representing state The signal-to-noise ratio in the middle, Representing state Signal strength in express The maximum value in, Representing state Corresponding Shannon capacity, Indicates frequency band resources The center frequency of the associated frequency band, This represents the center frequency of the frequency band associated with the frequency band resources allocated to the nearest neighboring user of the nth user. Indicates the suppression factor of adjacent frequency bands. Indicates the unit center frequency. N represents the total number of users. Indicates the status of available channels The bandwidth of the associated available channels, This indicates the preset maximum Shannon capacity. This indicates the preset maximum signal strength.

2. The method for edge data transmission of smart meters using Bayesian dynamic game theory as described in claim 1, characterized in that, An embedded communication module is deployed in the electricity meter. This embedded communication module is used to scan the meter's available channels and assess the status of the available channels, including: The embedded communication module includes a wireless receiver, a received signal strength indicator, and a signal-to-noise ratio measuring device; The transmitter periodically transmits signals to all channels, and the wireless receiver receives the channel signals of each channel one by one, and records the signal strength of the channel signal using a received signal strength indicator as the signal strength of the channel. Channels with signal strength higher than a preset strength threshold are designated as available channels. The signal-to-noise ratio (SNR) of the channel signal in the available channels is recorded using a signal-to-noise ratio (SNR) measurement device. The signal strength and SNR of the available channels are used as the status of the available channels. Extract the status of all available channels in the meter to form the meter's channel status data.

3. The method for edge data transmission of smart meters using Bayesian dynamic game theory as described in claim 2, characterized in that, The channel state data is mapped to a set of frequency bands according to the center frequency of the channel, forming a frequency band resource pool for the electricity meter, including: The center frequency of the available channels in the channel state data is extracted, and the state of the available channels is mapped to a frequency band containing the center frequency to form the frequency band resource pool of the meter. The frequency band resource pool contains frequency band resources of D frequency bands, and each frequency band resource consists of the state of the available channels after mapping, where D represents the total number of frequency bands in the frequency band set.

4. The edge data transmission method for smart meters using Bayesian dynamic game theory as described in claim 1, characterized in that, The spatiotemporal distribution characteristics of users' electricity consumption behavior are inferred using a Bayesian probabilistic graphical model, including: Construct a Bayesian probabilistic graphical model based on users' historical electricity consumption data; The user's historical electricity consumption data refers to the power consumption recorded by the user's associated electricity meter at different time periods over the past G days. The Bayesian probabilistic graphical model is a graphical model structure, consisting of graph nodes and the relationships between them. The graph nodes include time periods and users, and the relationships between them include the user's electricity consumption activity status in different time periods, historical electricity consumption data, and prior knowledge of the user's electricity consumption behavior characteristics. A joint probability model is constructed by using a Bayesian probabilistic graphical model to calculate the relationships between graph nodes, with the user's electricity consumption activity status in different time periods as latent variables, the user's electricity consumption behavior characteristics as prior driving variables, and historical electricity consumption data as observed variables. The latent variables are unobservable random variables. The variational inference method is used to iterate the variational probability and the prior of electricity consumption behavior features until the variational probability and the prior of electricity consumption behavior features do not change. The variational probability of all active electricity consumption states is extracted as the spatiotemporal distribution feature of the user's electricity consumption behavior.

5. The method for edge data transmission of smart meters using Bayesian dynamic game theory as described in claim 4, characterized in that, Based on the spatiotemporal distribution characteristics of the user's electricity consumption behavior, an output probability distribution map of the user's electricity consumption activity is generated, including: Calculate the mean variational probability of a user's electricity activity status in each time period to construct a probability distribution map of the user's electricity activity: ,in Let represent the mean variational probability of the active electricity consumption state of the nth user in the b-th time period. N represents the total number of users. B represents the total number of time periods.

6. The method for edge data transmission of smart meters using Bayesian dynamic game theory as described in claim 5, characterized in that, Based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each meter, including: Based on Bayesian Nash equilibrium theory, a leader and followers are generated in a non-cooperative game model. The leader allocates frequency band resources to users with high activity based on the user utility function. The followers are electricity meters. The followers select the channel that maximizes their own utility from the frequency band resources allocated by the leader based on the electricity meter utility function according to the frequency band resource allocation result. Based on the user utility function and the meter utility function, a dynamic game approach is used to iterate the allocation probability and allocation result of the frequency band resources associated with each user's meter until both converge, thus obtaining the frequency band allocation coefficient for each meter, where the frequency band allocation coefficient represents the frequency band resources allocated to the meter.

7. The Bayesian dynamic game-based edge data transmission method for smart meters as described in claim 6, characterized in that, The optimal frequency band usage strategy is used to allocate working channels for electricity meters, including: The channel with the highest signal strength and the highest signal-to-noise ratio, which is higher than a preset first strength threshold, is selected from the optimal frequency band usage strategy and used as the working channel of the electricity meter.

8. A smart meter edge data transmission system, characterized in that, The smart meter edge data transmission system includes a server and a data acquisition device. The server includes an electricity activity quantification module and a channel optimization module. The electricity activity quantification module is used to infer the spatiotemporal distribution characteristics of users' electricity consumption behavior using a Bayesian probabilistic graphical model, and outputs a probability distribution map of users' electricity activity. The channel optimization module is used to construct a non-cooperative game model for frequency band allocation by using the frequency band resource pool of the electricity meter and the probability distribution map of the user's electricity consumption as strategy resources. Based on Bayesian Nash equilibrium theory, a solver is used to iterate the non-cooperative game model to obtain the optimal frequency band usage strategy for each electricity meter, allocate the working channel of the electricity meter, and use the allocated working channel of the electricity meter for edge data transmission. The data acquisition device is used to collect channel status data of the electricity meter and the user's electricity consumption history data, and to map the channel status data to a frequency band set according to the center frequency of the channel, which serves as the frequency band resource pool of the electricity meter. The channel status data includes the signal strength and signal-to-noise ratio of different channels. To achieve the edge data transmission method for smart meters using Bayesian dynamic game theory as described in any one of claims 1-7; Using the frequency band resource pool of electricity meters and the probability distribution map of users' electricity consumption activity as policy resources, a non-cooperative game model for frequency band allocation is constructed, including: The electricity consumption activity probability distribution map is composed of the electricity consumption activity probability distribution of N users; The non-cooperative game model uses the probability distribution of each user's electricity consumption activity and the frequency band resource pool of the user's associated electricity meter as the user's strategy resources, and the frequency band allocation coefficient as the strategy variable. It constructs a meter utility function based on the frequency band resource pool and a user utility function based on the probability distribution of electricity consumption activity, where the expression for the meter utility function is: ； ； in, This represents the meter utility function, whose input value is the frequency band resources allocated to the meter. This represents the frequency band resources allocated to the electricity meter associated with the nth user, whereby the frequency band resources consist of the state of the available channels after mapping. , Indicates frequency band resources The status of available channels in the system. Representing state The signal-to-noise ratio in the middle, Representing state Signal strength in express The maximum value in, Representing state Corresponding Shannon capacity, Indicates frequency band resources The center frequency of the associated frequency band, This represents the center frequency of the frequency band associated with the frequency band resources allocated to the nearest neighboring user of the nth user. Indicates the suppression factor of adjacent frequency bands. Indicates the unit center frequency. N represents the total number of users. Indicates the status of available channels The bandwidth of the associated available channels, This indicates the preset maximum Shannon capacity. This indicates the preset maximum signal strength.

Images