Urban user electricity utilization efficiency collaborative optimization analysis method and system based on Internet of Things

Abstract

The invention discloses an urban user power utilization efficiency collaborative optimization analysis method and system based on the Internet of Things, and relates to the technical field of power analysis, and the technical scheme is characterized in that a multi-source heterogeneous data pool containing timestamp alignment is constructed; generating a user electricity consumption characteristic matrix through a space-time dynamic weight fusion algorithm; calculating a frequency modulation potential value through an entropy-TOPSIS combinatorial algorithm based on the user power consumption characteristic matrix; inputting the frequency modulation potential value into an LSTM-GAN confrontation generative network, and dynamically generating a three-dimensional efficiency parameter vector containing a user optimization degree, a good power grid contribution degree and an environmental benefit index; and based on the three-dimensional efficiency parameter vector, an optimization target of a user and a power grid is coordinated through a Stackelberg game model, and a dynamic electricity price strategy and a load adjustment strategy are output. The method has the advantages of being low in response delay, high in energy efficiency optimization rate, good in environmental benefit and the like when being used for coping with complex power consumption scenes in urban areas.

Description

Technical Field

[0001] The present invention relates to the field of power analysis technology, and more specifically, to a method and system for collaborative optimization analysis of urban users' electricity efficiency based on the Internet of Things. Background Art

[0002] With the accelerated pace of urbanization, urban electricity loads are becoming more diverse and dynamic. The spatial and temporal differences in industrial, commercial, and residential electricity demand exacerbate the pressure on grid frequency regulation. Under the "dual carbon" goals, aligning user-side energy efficiency optimization with grid frequency regulation has become a core issue in smart grid development. The rapid development of the Internet of Things (IoT) provides technical support for the real-time collection of massive amounts of electricity consumption data. However, significant technical challenges remain in the fusion and analysis of multi-source heterogeneous data, the dynamic assessment of user frequency regulation potential, and the intelligent generation of collaborative optimization strategies.

[0003] The existing technology records static electricity price control methods based on load forecasting, but they use a single historical electricity consumption to build a forecasting model and do not consider spatiotemporal correlation factors such as geographic information and environmental perception, resulting in significant deviations in the assessment of frequency regulation potential. Some existing technologies also record peak-valley electricity price optimization solutions based on game theory, but they use a fixed weight distribution mechanism and are unable to dynamically adjust user participation according to the real-time working conditions of the power grid, which can easily cause strategy lags in the event of sudden load fluctuations. When dealing with complex electricity consumption scenarios in urban areas, the above methods generally have problems such as response delays, low energy efficiency optimization rates, and insufficient environmental benefits.

[0004] Therefore, how to research and design a collaborative optimization analysis method and system for urban users' electricity efficiency based on the Internet of Things that can overcome the above-mentioned defects is a problem that we urgently need to solve. Summary of the Invention

[0005] In order to address the deficiencies in the prior art, the purpose of the present invention is to provide a collaborative optimization analysis method and system for urban user electricity consumption efficiency based on the Internet of Things. By constructing a spatial weight matrix to characterize the topological correlation of user electricity consumption behavior, multi-source data time series alignment is achieved, and through objective quantitative data based on entropy value-Topsis, as well as through the collaboration of generative adversarial networks and game models, a two-way balance of interests between users and power grids is achieved. When dealing with complex urban electricity consumption scenarios, the method has the advantages of low response delay, high energy efficiency optimization rate and good environmental benefits.

[0006] The above technical objectives of the present invention are achieved through the following technical solutions:

[0007] First, a collaborative optimization analysis method for urban users' electricity efficiency based on the Internet of Things is provided, including the following steps:

[0008] Collect real-time electricity consumption data from industrial, commercial, and residential users through the IoT sensor network, simultaneously obtain geographic information data, environmental perception data, and grid dispatch instructions, and build a multi-source heterogeneous data pool with timestamp alignment;

[0009] Based on the geographic information data and user electricity consumption data in the multi-source heterogeneous data pool, a user electricity consumption feature matrix is ​​generated through a spatiotemporal dynamic weight fusion algorithm;

[0010] Based on the user electricity consumption characteristic matrix, the frequency regulation potential value is calculated by using the entropy-TOPSIS combined algorithm;

[0011] Input the frequency modulation potential value into the LSTM-GAN adversarial generative network to dynamically generate a three-dimensional performance parameter vector including user optimization degree, grid contribution degree and environmental benefit index;

[0012] Based on the three-dimensional efficiency parameter vector, the optimization targets of the user and the power grid are coordinated through the Stackelberg game model, and a dynamic electricity price strategy and a load regulation strategy are output.

[0013] Furthermore, the spatial weight matrix construction process in the spatiotemporal dynamic weight fusion algorithm includes:

[0014] The Delaunay triangulation method is used to determine the topological relationship between adjacent users, and the traffic correlation is introduced to make corrections to obtain the spatial weight matrix.

[0015] Furthermore, the frequency modulation potential value is calculated by the entropy-TOPSIS combined algorithm, including:

[0016] Determining the user's load elasticity coefficient, peak load contribution, and adjustable capacity based on the user's electricity consumption characteristics in the user electricity consumption characteristic matrix, wherein the user's electricity consumption characteristics include basic electricity consumption characteristics, spatiotemporal correlation characteristics, environmental coupling characteristics, and user attribute characteristics;

[0017] Based on the load elasticity coefficient, the peak regulation contribution and the adjustable capacity, the frequency regulation potential value of the corresponding user is calculated by an entropy-TOPSIS combined algorithm.

[0018] Furthermore, the FM potential value is input into the LSTM-GAN adversarial generative network to dynamically generate a three-dimensional performance parameter vector, including:

[0019] Inputting the historical performance parameter sequence into the generator network of LSTM-GAN, embedding the frequency modulation potential value as a conditional vector, and outputting the three-dimensional performance parameter vector;

[0020] The grid dispatching instruction is input into the discriminator network, and after the logical consistency of the three-dimensional performance parameter vector and the frequency regulation potential value is verified, the finally generated three-dimensional performance parameter vector is output.

[0021] Furthermore, the method further comprises:

[0022] After the dynamic electricity price strategy is executed, the frequency regulation potential value is reversely updated according to the user's actual frequency regulation response data;

[0023] Based on the updated FM potential value, a new three-dimensional performance parameter vector is generated through the generator network.

[0024] Furthermore, the generation of the dynamic electricity price strategy includes:

[0025] Determine the grid contribution and environmental benefit index of user transmission to the grid;

[0026] If the grid contribution is greater than the grid contribution threshold, determining the electricity price discount rate for the user;

[0027] If the environmental benefit index is less than the environmental benefit threshold, an environmental surcharge for the user is determined.

[0028] Furthermore, the generation of the load regulation strategy includes:

[0029] Determine the user's user optimization degree and grid contribution;

[0030] When the user optimization degree is less than the energy efficiency threshold, triggering generation of an energy efficiency upgrade instruction for the user's terminal device according to the user optimization degree;

[0031] Furthermore, the power grid allocates a frequency regulation task of load reduction to the user according to the power grid contribution.

[0032] In a second aspect, a system for collaborative optimization and analysis of electricity consumption efficiency of urban users based on the Internet of Things is provided. The system is used to implement the method for collaborative optimization and analysis of electricity consumption efficiency of urban users based on the Internet of Things as described in any one of the first aspects, including:

[0033] The data acquisition module is used to collect real-time electricity consumption data from industrial, commercial, and residential users through the Internet of Things sensor network, simultaneously obtain geographic information data, environmental perception data, and power grid dispatch instructions, and build a multi-source heterogeneous data pool with timestamp alignment;

[0034] A feature generation module is used to generate a user electricity consumption feature matrix based on the geographic information data and user electricity consumption data in the multi-source heterogeneous data pool through a spatiotemporal dynamic weight fusion algorithm;

[0035] A potential analysis module is used to calculate the frequency modulation potential value based on the user power consumption characteristic matrix by using an entropy-TOPSIS combined algorithm;

[0036] An efficiency generation module is used to input the frequency modulation potential value into an LSTM-GAN adversarial generative network to dynamically generate a three-dimensional efficiency parameter vector including user optimization degree, grid contribution degree and environmental benefit index;

[0037] The strategy optimization module is used to coordinate the optimization goals of users and power grids based on the three-dimensional efficiency parameter vector through the Stackelberg game model, and output dynamic electricity price strategy and load regulation strategy.

[0038] In a third aspect, a computer terminal is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for collaborative optimization and analysis of electricity efficiency of urban users based on the Internet of Things as described in any one of the first aspects is implemented.

[0039] In a fourth aspect, a computer-readable medium is provided, on which a computer program is stored. The computer program is executed by a processor to implement the collaborative optimization analysis method for urban user electricity efficiency based on the Internet of Things as described in any one of the first aspects.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. The IoT-based collaborative optimization analysis method for urban user electricity consumption efficiency provided by the present invention achieves time series alignment of multi-source data by constructing a spatial weight matrix to characterize the topological correlation of user electricity consumption behavior. Furthermore, through objective quantitative data based on entropy (Topsis), and through the collaboration of generative adversarial networks and game models, a two-way balance of interests between users and the power grid is achieved. When dealing with complex urban electricity consumption scenarios, it has the advantages of low response delay, high energy efficiency optimization rate, and good environmental benefits.

[0042] 2. This invention overcomes the limitations of simply overlaying traditional GIS data with electricity consumption data by constructing a spatiotemporal dynamic weight fusion algorithm and employing a dual mechanism of Delaunay triangulation and traffic correlation correction. It precisely quantifies user topological relationships into a spatial weight matrix and introduces a time attenuation factor to dynamically adjust the grid load rate, enabling the feature matrix to simultaneously capture spatial proximity, traffic correlation, and time sensitivity. Compared to existing technologies, this method improves the efficiency of constructing the user electricity consumption feature matrix and enhances the accuracy of identifying spatiotemporal correlation features, laying a precise data foundation for subsequent frequency modulation potential calculations.

[0043] 3. The entropy-Topsis combined algorithm proposed in this invention constructs an objective weighting model based on information entropy by integrating multiple indicators such as load elasticity coefficient, peak load contribution, and adjustable capacity. This method overcomes the subjective bias of traditional expert weighting methods and combines the Euclidean distance calculation between positive and negative ideal solutions to achieve dynamic calibration of frequency modulation potential values. This method significantly reduces the frequency modulation response prediction error in industrial user clusters and significantly improves the evaluation accuracy compared to existing methods.

[0044] 4. The proposed coupling architecture of LSTM-GAN and Stackelberg game generates three-dimensional performance parameters by embedding frequency modulation potential values ​​through a generative adversarial network, and uses the game model to achieve Pareto optimality for electricity pricing strategy and load regulation. This method breaks through the limitations of traditional one-way optimization models and introduces grid dispatch instructions into the discriminator to verify logical consistency, which increases the response speed of dynamic electricity pricing strategy to milliseconds. This method can significantly reduce the peak-to-valley difference rate, improve users' overall energy efficiency, and reduce carbon emissions.

[0045] 5. The present invention constructs a closed-loop optimization system of "strategy execution-data feedback-model iteration" through the linkage mechanism of reverse updating frequency regulation potential value and generator network, realizing real-time calibration of dynamic electricity price strategy and actual user response. After three months of continuous operation, the accuracy of strategy generation is improved, which is better than the performance degradation problem of existing static models. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0047] Figure 1 This is a flowchart of Example 1 of the present invention;

[0048] Figure 2 This is a system block diagram in Example 2 of the present invention. DETAILED DESCRIPTION

[0049] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0050] Example 1: A collaborative optimization analysis method for urban users' electricity efficiency based on the Internet of Things, such as Figure 1 As shown, the following steps are included:

[0051] S1: Collect real-time electricity consumption data from industrial, commercial, and residential users through the IoT sensor network, simultaneously obtain geographic information data, environmental perception data, and grid dispatch instructions, and build a multi-source heterogeneous data pool with timestamp alignment;

[0052] S2: Based on the geographic information data and user electricity consumption data in the multi-source heterogeneous data pool, the user electricity consumption feature matrix is ​​generated through the spatiotemporal dynamic weight fusion algorithm;

[0053] S3: Based on the user's electricity consumption characteristic matrix, the frequency regulation potential value is calculated using the entropy-TOPSIS combined algorithm;

[0054] S4: Input the FM potential value into the LSTM-GAN adversarial generative network to dynamically generate a three-dimensional performance parameter vector;

[0055] S5: Based on the three-dimensional efficiency parameter vector, the Stackelberg game model is used to coordinate the optimization goals of users and the power grid, and output dynamic electricity price strategy and load regulation strategy.

[0056] In step S1, to improve the coverage of electricity consumption data collection, the IoT sensor network includes a cluster of smart meters, environmental sensing units, and a grid dispatch command receiver. Smart meter clusters are deployed in industrial plants, commercial complexes, and residential communities, supporting LoRaWAN and NB-IoT dual-mode communication with an adjustable sampling frequency. Environmental sensing units include, but are not limited to, meteorological sensors and traffic flow monitors. Meteorological sensors monitor temperature, humidity, and light intensity, typically with a spacing of less than 500 meters and a data update cycle of 5 minutes. Traffic flow monitors are integrated at intersections on urban arterial roads and transmit real-time traffic data via 5G. The grid dispatch command receiver, configured with a dedicated 5G network slice, receives frequency modulation demand instructions issued by the grid dispatch center. These instructions, for example, are formatted in JSON format and contain fields such as target load, time window, and priority.

[0057] For industrial users, electricity consumption data collection generally collects three-phase voltage, current, active power, power factor and harmonic distortion rate, and RFID tags are installed on interruptible loads (such as electroplating production lines) to record the start and stop status. For commercial users, electricity consumption data collection generally includes sub-item metering data and dynamic load tracking. Sub-item metering data includes hourly electricity consumption of air-conditioning systems, lighting systems, and elevators; dynamic load tracking captures instantaneous power fluctuations through high-frequency sampling. For residential users, electricity consumption data collection generally includes non-invasive load monitoring and behavior-related data collection. Non-invasive load monitoring, such as using current ripple characteristics to identify major home appliances, and behavior-related data, through user authorization, obtains smart home APP data, such as air-conditioning set temperature and charging pile reservation time.

[0058] The acquisition of geographic information data mainly involves obtaining GIS data sources, such as accessing the urban geographic information platform, and extracting the following layers: user location vector map, power supply network topology map, and building three-dimensional model. The power supply network topology map includes the coordinates of substations, feeders, and ring network cabinets, and the building three-dimensional model includes height and volume ratio data.

[0059] Considering time deviations and missing data across data sources, data cleaning and alignment are necessary. For example, missing value filling and noise filtering are performed on anomalous data. Missing value filling uses a spatiotemporal KNN algorithm, utilizing similar data from adjacent users. Noise filtering primarily involves applying wavelet threshold denoising to power data. Timestamp alignment primarily involves running the PTP precision clock protocol on edge computing nodes to reduce synchronization errors. Virtual timeline mapping technology is used for non-real-time data, such as GIS static data.

[0060] In step S2, considering that simple geographic distance weighting ignores the actual impact of the traffic network on user electricity consumption, such as congestion on main roads in commercial areas causing delays in peak electricity consumption, the spatial weight matrix construction process in the spatiotemporal dynamic weight fusion algorithm of the present invention includes: using the Delaunay triangulation method to determine the topological relationship between adjacent users, and modifying it by introducing traffic correlation to obtain the spatial weight matrix.

[0061] Specifically, the user electricity consumption data is extracted from the multi-source heterogeneous data pool to obtain the hourly electricity consumption E of each user. i (t), unit is kWh, sampling frequency is 1 minute, covering a total of N nodes of industrial, commercial and residential users. Geographic information data includes user geographic coordinates (x i ,y i ) and power supply network topology. Traffic related data includes the commuting path overlap rate Q between users ij and road traffic volume T k .

[0062] In addition, we use a spatiotemporal KNN algorithm to fill in missing electricity consumption data, for example, based on the electricity consumption patterns of adjacent users within a spatial radius of 500 meters or less during the same time period. We also remove outliers in geographic coordinates, such as using an isolation forest algorithm to detect and remove user points that are 2 kilometers or more away from the main cluster.

[0063] (1) The process of using Delaunay triangulation to determine the topological relationship of adjacent users includes: first constructing an initial triangle containing all user points, with vertex coordinates extended to a range of three times the extreme value of the user coordinates; inserting user points according to the Bowyer-Watson algorithm to ensure that the circumscribed circle of each triangle does not contain other points; extracting all edges E = {(p i ,p j)}, define users i and j as adjacent if and only if (p i ,p j )∈E and d ij ≤2000m; finally generate the binary adjacency matrix A ij ∈{0,1}, where A ij =1 indicates that users i and j are adjacent.

[0064] (2) For adjacent user pair A ij =1, calculate the initial weight. The closer the distance, the stronger the spatial correlation of users. The weight decays with the square of the distance. The specific formula is:

[0065]

[0066] in, is the initial weight between users i and j, reflecting the impact of geographical distance on electricity consumption correlation; γ is the distance attenuation coefficient, such as 1×10 -6 m -2 , controls the speed at which the weight decays with the square of the distance; d ij is the Euclidean distance between the geographic coordinates of users i and j (unit: meter), which is used to quantify spatial proximity.

[0067] Then normalize the initial weights and ensure that the sum of the weights is 1 through row normalization. The specific formula is:

[0068]

[0069] in, is the initial weight after normalization; is the initial weight between users i and k.

[0070] (3) Define the traffic correlation C between users i and j ij for:

[0071]

[0072] Among them, Path ij is the set of road segments included in the shortest commuting path from user i to j; ij is the path overlap rate between users i and j; T k is the average daily traffic flow of road k; max(O) is the maximum path overlap rate; max(T) is the maximum average daily traffic flow.

[0073] The initial weights are modified by traffic correlation, and the modified weights are as follows:

[0074]

[0075] in, is the corrected weight between users i and k; δ is the traffic correction coefficient, which adjusts the degree of enhancement of the weight by traffic factors, such as 0.3.

[0076] (4) Adjust the time sensitivity according to the real-time load rate of the power grid β(t). The higher the load rate of the power grid (close to the peak load period), the slower the time decay rate, and the more historical data weight is retained. The specific expression is:

[0077]

[0078] Where λ(t) is the real-time time attenuation factor (unit: s -1 ), which controls the degree of retention of historical data weights; λ0 is the basic attenuation rate, the benchmark attenuation when the grid load is normal; α is the load sensitivity coefficient, which adjusts the intensity of the influence of the grid load on the attenuation speed, such as taking a value of 0.5; β avg is the historical average load rate of the power grid, which can be obtained by sliding average based on the load data of the past 30 days; β max is the historical maximum load rate of the power grid; β min is the historical minimum load rate of the power grid.

[0079] By reducing the weight of historical data over time, the decay rate slows down when the grid load is high, retaining more real-time data features, and generating the final spatiotemporal dynamic weights as follows:

[0080]

[0081] Among them, W ij (t) is the dynamic spatiotemporal fusion weight of users i and k at time t, which integrates spatiotemporal correlation and time sensitivity; t0 is the Unix timestamp at 0 o'clock on the same day, which is used to calculate the time difference.

[0082] (5) The spatiotemporal weight W ij (t) and user electricity consumption data E i (t) Fusion, generating feature matrix

[0083]

[0084] Among them, F(i,t) is the electricity consumption feature vector of user i at time t, which includes the neighborhood electricity consumption pattern and its own electricity consumption; E j (t) is the electricity consumption of user j at time t (unit: kWh); ∈ is the weight coefficient of its own electricity consumption, which balances the contribution of neighborhood influence and its own electricity consumption characteristics.

[0085] Perform Z-score normalization on the row vector of each user:

[0086]

[0087] Among them, F norm (i, t) is the normalized row vector of user i at time t; μ i The mean value of the electricity consumption characteristics of user i in the time window; σ i is the standard deviation of user i in the time window.

[0088] In step S3, considering that a single load elasticity coefficient assessment can easily lead to high-elasticity, low-capacity users being misjudged as high-quality frequency regulation resources, such as when the actual adjustable capacity is less than 5% of the rated value, this can easily lead to scheduling failure. To this end, the present invention calculates the frequency regulation potential value using an entropy-TOPSIS combined algorithm, including: determining the user's load elasticity coefficient, peak-shaving contribution, and adjustable capacity based on the user's electricity consumption characteristics in the user electricity consumption feature matrix, where the user electricity consumption characteristics include basic electricity consumption characteristics, spatiotemporal correlation characteristics, environmental coupling characteristics, and user attribute characteristics; and calculating the frequency regulation potential value of the corresponding user using the entropy-TOPSIS combined algorithm based on the load elasticity coefficient, peak-shaving contribution, and adjustable capacity.

[0089] (1) Evaluation index definition and data preprocessing

[0090] In some examples, the load elasticity coefficient represents the sensitivity of a user's electricity consumption to fluctuations in electricity prices and is calculated as follows:

[0091]

[0092] Among them, η i is the load elasticity coefficient of user i; ΔE i is the change in electricity consumption of user i (unit: kWh); E i is the benchmark electricity consumption of user i (unit: kWh); ΔP is the electricity price fluctuation range (unit: yuan / kWh); P is the benchmark electricity price (unit: yuan / kWh).

[0093] In some examples, the peak load contribution represents the response effect of the user's participation in the grid peak load regulation, and the calculation formula is as follows:

[0094]

[0095] Among them, ξ i is the peak load contribution of user i; E base (i, t) is the baseline electricity consumption of user i at time t (unit: kWh); E act (i, t) is the actual electricity consumption of user i at time t (unit: kWh); T is the number of time windows.

[0096] In some examples, the adjustable capacity represents the ratio of the user's adjustable load to its total load, and is calculated as follows:

[0097]

[0098] Among them, γ i For; E flex,i is the adjustable load capacity of user i (unit: kWh); E rated,i is the total rated capacity of electrical equipment of user i (unit: kWh).

[0099] The above indicators are standardized as follows:

[0100]

[0101] Among them, x ik is the normalized value of user i on the kth indicator; min(X k ) is the minimum value of all users on the kth index; max(X k ) is the maximum value of all users on the kth index; X ik is the original value of the kth indicator of user i, such as (η i ,ξ i ,γ i ).

[0102] (2) Calculation of indicator weights based on entropy method

[0103] First, calculate the index proportion. The specific formula is as follows:

[0104]

[0105] Among them, p ik is the probability density of household i on the kth indicator, reflecting the concentration of data distribution.

[0106] Next, calculate the information entropy. The specific formula is as follows:

[0107]

[0108] Among them, e k is the information entropy of the kth indicator, and its value range is [0,1]. The smaller the entropy value, the higher the discreteness of the indicator data and the greater the weight.

[0109] Finally, calculate the difference coefficient and weight. The specific formula is as follows:

[0110]

[0111] Among them, g k is the coefficient of difference of the kth indicator; w k is the entropy weight of the kth indicator.

[0112] (3) Calculation of frequency modulation potential value using the TOPSIS method

[0113] First, construct the weighted decision matrix Z. The specific formula is as follows:

[0114] Z=[z ik ]=[w k ·x ik ];

[0115] Among them, z ik is the weighted value of user i on the kth indicator, which is the comprehensive indicator weight and normalized features.

[0116] Then determine the positive / negative ideal solution. The specific formula is as follows:

[0117] Z + =(max(z i1 ),max(z i2 ),max(z i3 ));

[0118] Z - =(min(z i1 ),min(z i2 ),min(z i3 ));

[0119] Among them, Z + is a positive ideal solution, indicating the user feature combination with the best frequency modulation potential; Z - is a negative ideal solution, indicating the user feature combination with the worst frequency modulation potential.

[0120] Then calculate the Euclidean distance. The specific formula is as follows:

[0121]

[0122] in, is the Euclidean distance between user i and the positive ideal solution, reflecting the degree of deviation of its frequency modulation potential from the ideal state; is the positive ideal solution of the kth index; is the Euclidean distance between user i and the negative ideal solution; is the negative ideal solution of the kth index.

[0123] Finally, the frequency modulation potential value is calculated. The specific formula is as follows:

[0124]

[0125] Among them, C i is the frequency modulation potential value of user i, and its value range is [0,1]. The larger the value, the stronger the user's frequency modulation ability.

[0126] In step S4, the frequency modulation potential value is input into the LSTM-GAN generative adversarial network to dynamically generate a three-dimensional performance parameter vector, including: inputting the historical performance parameter sequence into the generator network of the LSTM-GAN, embedding the frequency modulation potential value as a conditional vector, and outputting the three-dimensional performance parameter vector; inputting the power grid dispatching instruction into the discriminator network, and after verifying the logical consistency of the three-dimensional performance parameter vector and the frequency modulation potential value, outputting the final generated three-dimensional performance parameter vector.

[0127] In the LSTM-GAN adversarial generation network, the generator input is the historical performance parameter sequence where h t =(a t ,b t ,c t ) represents the performance parameters (time dimension, space dimension, and regulation dimension) at time t. The conditional vector is the frequency modulation potential value C i ∈[0,1], converted to a conditional vector through the embedding layer The input of the discriminator is the power grid dispatch instruction (including load rate, voltage deviation, frequency deviation), and the generated three-dimensional performance parameter vector is spliced ​​and input into the discriminator. The output of the generator is a three-dimensional performance parameter vector The discriminator outputs the authenticity probability P real ∈[0,1], indicating the matching degree between the generated vector and the grid demand.

[0128] In addition, after the dynamic electricity price strategy is executed, the present invention reversely updates the frequency regulation potential value according to the user's actual frequency regulation response data; based on the updated frequency regulation potential value, a new three-dimensional performance parameter vector is generated through the generator network.

[0129] In step S5, the generation of a dynamic electricity price strategy includes: determining the grid contribution and environmental benefit index of the user's transmission to the grid; if the grid contribution is greater than the grid contribution threshold, determining the user's electricity price discount rate; if the environmental benefit index is less than the environmental benefit threshold, determining the user's environmental surcharge.

[0130] Specifically, the grid contribution G i ∈[0,1] is the quantitative value of user i’s contribution to grid frequency regulation. Environmental benefit index E i ∈[0,1] is the degree of environmental friendliness of user i’s electricity consumption behavior. The threshold parameters include the grid contribution threshold G th and environmental benefit threshold E th .

[0131] The logic for generating dynamic electricity price strategy is as follows:

[0132] (1) If G i >Gth , the electricity price discount rate is calculated as:

[0133]

[0134] Among them, r i is the electricity price discount rate of user i (unit: %), and its value range is [0,r max ]; r max is the maximum discount rate, set by grid policy.

[0135] (2) If E i <E th , then the environmental surcharge is calculated as:

[0136]

[0137] Among them, f i is the environmental surcharge for user i (unit: yuan / kWh); f base Base surcharge.

[0138] The final electricity price is:

[0139] P i (t) = P base (t)·(1-r i )+f i ;

[0140] Among them, P i (t) is the dynamic electricity price of user i at time t (unit: yuan / kWh); P base (t) is the basic time-of-use electricity price of the power grid.

[0141] The generation of the load regulation strategy includes: determining the user's user optimization degree and grid contribution; when the user optimization degree is less than the energy efficiency threshold, triggering the generation of the user's terminal equipment energy efficiency upgrade instruction based on the user optimization degree; and the grid assigning the user the frequency regulation task of load reduction based on the grid contribution degree.

[0142] Specifically, user optimization degree U i ∈[0,1] is the energy efficiency equipment upgrade potential of user i, and the calculation formula is:

[0143]

[0144] Among them, E saved,i The amount of electricity that can be saved by user i through energy efficiency upgrade (unit: kWh); E rated,i is the rated power consumption capacity of user i (unit: kWh).

[0145] The load regulation strategy generation logic is as follows:

[0146] (1)If Ui th , then the energy efficiency upgrade instruction is triggered and the instruction intensity coefficient is generated:

[0147]

[0148] Among them, α i The urgency level of the upgrade instruction, the value range is [0,1].

[0149] (2) Based on the grid contribution G i Perform frequency modulation task allocation:

[0150]

[0151] Among them, L i is the load reduction that user i needs to bear (unit: kW); L total is the total grid load reduction demand (unit: kW).

[0152] In the Stackelberg game model, the power grid aims to minimize load deviation and environmental pollution, and decides the electricity price strategy P i (t). Users respond to load regulation instructions with the goal of minimizing electricity costs.

[0153] Specifically, the objective function of the power grid is:

[0154]

[0155] Where λ is the load balancing weight coefficient, which penalizes load distribution deviation.

[0156] The user's objective function is:

[0157]

[0158] Among them, μ is the electricity comfort coefficient, which weighs the cost and electricity demand.

[0159] The above objective function is solved iteratively using the reverse induction method until the electricity price strategy and load regulation amount converge.

[0160] Experimental verification: Traditional method A adopts a static electricity price control method based on load forecasting, and traditional method B adopts a peak-valley electricity price optimization scheme based on game theory.

[0161] Table 1 shows the results of a three-month operational data analysis for a provincial capital city. Through Delaunay triangulation and traffic correlation correction, the error in constructing the spatial weight matrix was reduced to one-third of that of traditional methods. Experiments showed that in a commercial complex scenario, the accuracy of air conditioning load forecasting increased by 27%, validating the effectiveness of the spatiotemporal dynamic fusion algorithm. ​

[0162] Furthermore, after three months of operation, the error in frequency modulation potential assessment decreased from an initial 5.3% to 4.2%, demonstrating that the reverse update mechanism enables the model's ability to continuously evolve. In particular, during sudden downpours (where environmental sensing data undergoes a sudden change), the proposed strategy adjustment response speed remained within 200ms, while traditional methods experienced strategy failures of over 15 minutes.

[0163] Table 13 Months of Operation Data Results

[0164]

[0165] Taking an industrial park as an example, the data response comparison results are shown in Table 2. As can be seen, through the dynamic electricity pricing strategy, industrial users reduced their quarterly electricity costs by ¥1.26 million (a 31% year-on-year decrease) and simultaneously reduced carbon emissions by 23,000 tons. Residential user demand response participation reached 82%, and peak load shifting increased by 2.7 times compared to traditional methods.

[0166] Table 2 Data response comparison results

[0167]

[0168] Example 2: A system for collaborative optimization and analysis of electricity consumption efficiency of urban users based on the Internet of Things. The system is used to implement the method for collaborative optimization and analysis of electricity consumption efficiency of urban users based on the Internet of Things as described in Example 1. Figure 2 As shown, it includes data acquisition module, feature generation module, potential analysis module, effectiveness generation module and strategy optimization module.

[0169] Among them, the data acquisition module is used to collect real-time electricity consumption data of industrial, commercial and residential users through the Internet of Things sensor network, synchronously obtain geographic information data, environmental perception data and power grid dispatch instructions, and build a multi-source heterogeneous data pool with timestamp alignment; the feature generation module is used to generate a user electricity consumption feature matrix based on the geographic information data and user electricity consumption data in the multi-source heterogeneous data pool through a spatiotemporal dynamic weight fusion algorithm; the potential analysis module is used to calculate the frequency regulation potential value based on the user electricity consumption feature matrix through an entropy-TOPSIS combination algorithm; the efficiency generation module is used to input the frequency regulation potential value into the LSTM-GAN adversarial generative network to dynamically generate a three-dimensional efficiency parameter vector including user optimization degree, power grid contribution degree and good environmental benefit index; the strategy optimization module is used to coordinate the optimization goals of users and power grids based on the three-dimensional efficiency parameter vector through a Stackelberg game model, and output a dynamic electricity price strategy and load regulation strategy.

[0170] The present invention also describes a computer terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for collaborative optimization and analysis of electricity efficiency of urban users based on the Internet of Things as described in Example 1 is implemented.

[0171] The present invention also describes a computer-readable medium having a computer program stored thereon. The computer program is executed by a processor to implement the collaborative optimization analysis method for urban users' electricity efficiency based on the Internet of Things as described in Example 1.

[0172] Working Principle: This invention breaks through the limitation of simple superposition of traditional GIS data and electricity consumption data by constructing a spatiotemporal dynamic weight fusion algorithm and adopting the dual mechanisms of Delaunay triangulation and traffic correlation correction. The invention accurately quantifies the user topological relationship into a spatial weight matrix and introduces a time attenuation factor to dynamically adjust the grid load rate, so that the feature matrix simultaneously captures the characteristics of spatial proximity, traffic correlation and time sensitivity. Compared with the existing technology, the construction efficiency of the user electricity consumption feature matrix is ​​improved, and the accuracy of identifying spatiotemporal correlation features is improved, laying a precise data foundation for subsequent frequency regulation potential calculation.

[0173] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0174] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to 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 steps in the process. 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.

[0175] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work 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 The function specified in one or more boxes.

[0176] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0177] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A collaborative optimization analysis method for urban users' electricity efficiency based on the Internet of Things, characterized by: The following steps are involved: Collect real-time electricity consumption data from industrial, commercial, and residential users through the IoT sensor network, simultaneously obtain geographic information data, environmental perception data, and grid dispatch instructions, and build a multi-source heterogeneous data pool with timestamp alignment; Based on the geographic information data and user electricity consumption data in the multi-source heterogeneous data pool, a user electricity consumption feature matrix is ​​generated through a spatiotemporal dynamic weight fusion algorithm; Based on the user's electricity consumption characteristic matrix, the frequency regulation potential value is calculated by using an entropy-TOPSIS combined algorithm; Input the frequency modulation potential value into the LSTM-GAN adversarial generative network to dynamically generate a three-dimensional performance parameter vector including user optimization degree, grid contribution degree and environmental benefit index; Based on the three-dimensional efficiency parameter vector, the optimization targets of the user and the power grid are coordinated through the Stackelberg game model, and a dynamic electricity price strategy and a load regulation strategy are output.

2. The method for collaborative optimization analysis of urban user electricity efficiency based on the Internet of Things according to claim 1 is characterized in that: The spatial weight matrix construction process in the spatiotemporal dynamic weight fusion algorithm includes: The Delaunay triangulation method is used to determine the topological relationship between adjacent users, and the traffic correlation is introduced to make corrections to obtain the spatial weight matrix.

3. The method for collaborative optimization analysis of urban user electricity efficiency based on the Internet of Things according to claim 1 is characterized in that: The method of calculating the frequency modulation potential value by using the entropy-TOPSIS combined algorithm includes: Determining the user's load elasticity coefficient, peak load contribution, and adjustable capacity based on the user's electricity consumption characteristics in the user electricity consumption characteristic matrix, wherein the user's electricity consumption characteristics include basic electricity consumption characteristics, spatiotemporal correlation characteristics, environmental coupling characteristics, and user attribute characteristics; Based on the load elasticity coefficient, the peak regulation contribution and the adjustable capacity, the frequency regulation potential value of the corresponding user is calculated by an entropy-TOPSIS combined algorithm.

4. The method for collaborative optimization analysis of urban user electricity efficiency based on the Internet of Things according to claim 1 is characterized in that: The step of inputting the frequency modulation potential value into the LSTM-GAN generative adversarial network to dynamically generate a three-dimensional performance parameter vector includes: Inputting the historical performance parameter sequence into the generator network of LSTM-GAN, embedding the frequency modulation potential value as a conditional vector, and outputting the three-dimensional performance parameter vector; The grid dispatching instruction is input into the discriminator network, and after the logical consistency of the three-dimensional performance parameter vector and the frequency regulation potential value is verified, the finally generated three-dimensional performance parameter vector is output.

5. The method for collaborative optimization analysis of urban user electricity efficiency based on the Internet of Things according to claim 4 is characterized in that: The method further includes: After the dynamic electricity price strategy is executed, the frequency regulation potential value is reversely updated according to the user's actual frequency regulation response data; Based on the updated FM potential value, a new three-dimensional performance parameter vector is generated through the generator network.

6. The method for collaborative optimization analysis of urban user electricity efficiency based on the Internet of Things according to claim 1 is characterized in that: The generation of the dynamic electricity price strategy includes: Determine the grid contribution and environmental benefit index of user transmission to the grid; If the grid contribution is greater than the grid contribution threshold, determining the electricity price discount rate for the user; If the environmental benefit index is less than the environmental benefit threshold, an environmental surcharge for the user is determined.

7. The method for collaborative optimization and analysis of urban user electricity efficiency based on the Internet of Things according to claim 1 is characterized in that: The generation of the load regulation strategy includes: Determine the user's user optimization degree and grid contribution; When the user optimization degree is less than the energy efficiency threshold, triggering generation of an energy efficiency upgrade instruction for the user's terminal device according to the user optimization degree; Furthermore, the power grid allocates a frequency regulation task of load reduction to the user according to the power grid contribution.

8. The Internet of Things-based urban user electricity efficiency collaborative optimization analysis system is characterized by: The system is used to implement the method for collaborative optimization and analysis of urban user electricity efficiency based on the Internet of Things as described in any one of claims 1 to 7, comprising: The data acquisition module is used to collect real-time electricity consumption data from industrial, commercial, and residential users through the Internet of Things sensor network, simultaneously obtain geographic information data, environmental perception data, and power grid dispatch instructions, and build a multi-source heterogeneous data pool with timestamp alignment; A feature generation module is used to generate a user electricity consumption feature matrix based on the geographic information data and user electricity consumption data in the multi-source heterogeneous data pool through a spatiotemporal dynamic weight fusion algorithm; A potential analysis module is used to calculate the frequency modulation potential value based on the user power consumption characteristic matrix by using an entropy-TOPSIS combined algorithm; An efficiency generation module is used to input the frequency modulation potential value into an LSTM-GAN adversarial generative network to dynamically generate a three-dimensional efficiency parameter vector including user optimization degree, grid contribution degree and environmental benefit index; The strategy optimization module is used to coordinate the optimization goals of users and power grids based on the three-dimensional efficiency parameter vector through the Stackelberg game model, and output dynamic electricity price strategy and load regulation strategy.

9. A computer terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for collaborative optimization and analysis of urban user electricity efficiency based on the Internet of Things is implemented as described in any one of claims 1 to 7.

10. A computer-readable medium having a computer program stored thereon, characterized in that: The computer program is executed by a processor to implement the collaborative optimization analysis method for urban user electricity efficiency based on the Internet of Things as described in any one of claims 1 to 7.

Images