New energy grid management and analysis system for smart community

Abstract

The invention discloses a smart community new energy grid management and analysis system, and belongs to the technical field of smart power grid and energy management, and the system comprises a data collection module which is used for collecting photovoltaic power generation, energy storage equipment and user power utilization data in real time; the dynamic modeling module is used for dynamically generating a multi-level energy grid division scheme according to the collected data; the optimization decision module is used for outputting energy regulation and control parameters according to the grid division scheme; and the iterative feedback module is used for monitoring the regulation and control effect and triggering online updating of the model parameters. According to the method, a dynamic grid topology adaptive division mechanism, a multi-source adaptive collaborative optimization algorithm and a pre-verification type network splitting simulation technology are adopted, so that real-time dynamic reconstruction, multi-target joint optimization and millisecond-level security risk prevention and control of a community energy network can be realized; the problems of high transmission loss, response delay, equipment service life reduction and the like caused by traditional static division are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart grid and energy management, and in particular to a new energy grid management and analysis system for smart communities. Background Art

[0002] With the large-scale application of distributed energy sources such as photovoltaic power generation, energy storage devices, and electric vehicles in smart communities, the community energy system presents characteristics of high density and diversified access. With the increasing growth of the installed capacity of distributed photovoltaic in residential areas and the penetration rate of energy storage systems, the dynamic coupling contradiction between power generation and load has become increasingly prominent. Under this background, how to construct a new energy management system with dynamic perception, topology self-adaptation, and multi-objective coordination capabilities has become a key technical challenge to break through the high penetration barrier of community energy.

[0003] Currently, the new energy grid management generally adopts an energy partition strategy with a fixed topology and a one-way regulation mechanism, and centrally schedules devices through a central controller. However, the spatio-temporal volatility of new energy distribution in the community scenario (such as changes in light intensity, peak-valley differences in user electricity consumption) and the multi-node dynamic interaction characteristics (such as the coupling of energy storage charge and discharge states, fluctuations in grid interaction power) lead to significant bottlenecks in transmission efficiency, response speed, and operation reliability in the traditional architecture. Especially when the load suddenly changes or the energy supply fluctuates violently, the static partitioning strategy is prone to problems such as local energy overload and a sharp increase in cross-region transmission losses. There is an urgent need to develop an energy grid management technology with dynamic reconfiguration capabilities. Summary of the Invention

[0004] To solve the above problems, the present invention provides a new energy grid management and analysis system for smart communities, which adopts a dynamic grid topology self-adaptive partitioning mechanism, a multi-source self-adaptive collaborative optimization algorithm, and a pre-verified network splitting simulation technology, and can realize the real-time dynamic reconfiguration of the community energy network, multi-objective joint optimization, and millisecond-level security risk prevention and control, effectively solving problems such as high transmission losses, response delays, and equipment life reduction caused by traditional static partitioning.

[0005] The above object can be achieved by the following solutions:

[0006] A new energy grid management and analysis system for smart communities, comprising: a data acquisition module, which is distributed at each energy node in the community and is used to collect real-time data of photovoltaic power generation, energy storage devices, and user electricity consumption; a dynamic modeling module, connected to the data acquisition module, and is used to dynamically generate a multi-level energy grid partitioning scheme according to the collected data; an optimization decision module, which is used to output energy regulation parameters according to the grid partitioning scheme; and an iterative feedback module, which is used to monitor the regulation effect and trigger the online update of model parameters.

[0007] Optionally, the dynamic modeling module includes: an equilibrium degree evaluation unit, a network classification unit, and a verification evaluation unit; wherein, the equilibrium degree evaluation unit is configured to evaluate the equilibrium degree of community energy distribution according to the collected data; the network classification unit is configured to select a fully grid-based, partially grid-based, or unit individual-based modeling mode for network division according to the evaluation result; the verification evaluation unit is configured to calculate the energy transmission loss value after grid division according to the division result, and re-divide when the energy transmission loss value exceeds a predetermined threshold.

[0008] Optionally, the evaluation of the equilibrium degree of community energy distribution according to the collected data includes: collecting the power data of each regional node in real time; according to the power data, counting the number of energy interactions between adjacent regional nodes within a preset first time window and performing normalization processing; according to the power data, calculating the power Gini coefficient and performing normalization processing; according to the normalized number of energy interactions and the normalized power Gini coefficient, calculating the energy distribution equilibrium degree. For the energy distribution equilibrium degree Q, there is

[0009] Q = λ * G + (1 - λ) * T,

[0010] wherein, λ is a weight factor, G is the normalized power Gini coefficient, and T is the normalized number of energy interactions.

[0011] Optionally, the selection of a fully grid-based, partially grid-based, or unit individual-based modeling mode for network division according to the evaluation result includes: judging whether the energy distribution equilibrium degree is greater than a preset first threshold; if the energy distribution equilibrium degree is greater than the first threshold, selecting a fully networked modeling mode for network division; if the energy distribution equilibrium degree is less than or equal to the first threshold, judging whether the energy distribution equilibrium degree is greater than a preset second threshold; if the energy distribution equilibrium degree is greater than the second threshold, selecting a partially networked modeling mode for network division; if the energy distribution equilibrium degree is less than or equal to the second threshold, selecting a unit individual-based modeling mode for network division.

[0012] Optionally, the selection of the fully networked modeling mode for network partitioning includes: modeling the node connection relationship of the area to be split as a weighted undirected graph F = {V, E, W}, where the vertex set V represents power nodes, the edge set E represents physical connection lines, and the weight W is the line transmission capacity; determining physically adjacent node groups based on Delaunay triangulation to generate an initial candidate splitting plane; calculating the instantaneous variance of the load of each phase in the current area and presetting a second time window; when it is detected that the instantaneous variance within a preset number of consecutive second time windows is greater than or equal to a preset fluctuation value, performing area splitting along the initial candidate splitting plane to obtain multiple sub-areas; when the instantaneous variance of all sub-areas is less than the preset fluctuation value, stopping the splitting process and locking the topological structure; using Monte Carlo simulation to verify the dynamic stability of the topological structure after splitting under a preset load perturbation, and writing the verified network partitioning topological structure into a preset partition database.

[0013] Optionally, the selection of the partially networked modeling mode for network partitioning includes: collecting the energy storage state of charge values of each distributed node and calculating the power change rate within a preset adjacent third time window in real time; using the energy storage state of charge value and the power change rate to calculate the node weight. For the node weight w n , there is

[0014]

[0015] where α is the weight coefficient of the energy storage state of charge value, SOC n is the energy storage state of charge value of the nth distributed node, tanh is the hyperbolic tangent function, P n is the power change rate of the nth distributed node within the adjacent third time window, P max is the rated power change threshold of the system, and β is the weight coefficient of the power change rate; mapping each distributed node to two-dimensional space coordinates and using the node weight as the third-dimensional coordinate to construct a three-dimensional space; performing weighted Delaunay triangulation in the three-dimensional space and using a preset correction equation to correct the distance between distributed nodes to complete dynamic connection, obtaining multiple triangular unit meshes. The correction equation is

[0016]

[0017] where d' ij is the distance between the ith distributed node and the jth distributed node, d ij is the original geographical distance between the ith distributed node and the jth distributed node, w i is the node weight of the ith distributed node, and w j is the node weight of the jth distributed node.

[0018] Optionally, the network division in the networked modeling mode of the selection part further includes: calculating the weight difference of the edges of each triangular element, where the weight difference ε = |w i - w j |; calculating the difference threshold of the edges of each triangular element, where the difference threshold where max() is the maximum value function; when the weight difference is greater than the difference threshold, obtain the voltage levels of the two distributed nodes corresponding to the edge of the triangular element; if the difference between the voltage levels of the two distributed nodes is greater than 10% of the rated voltage, then delete the corresponding edge of the triangular element and initiate local triangular element grid reconstruction.

[0019] Optionally, the network division in the individual modeling mode of the selection unit includes: collecting the port energy exchange rate of each device unit; judging whether the energy exchange rate is greater than a preset third threshold; if the energy exchange rate is greater than the third threshold, then mark the device as an island node and perform independent modeling; if the energy exchange rate is less than or equal to the third threshold, then form an independent microgrid unit according to the maximum energy exchange path and the nearest neighbor node.

[0020] Optionally, the monitoring of the regulation effect and the triggering of the online update of the model parameters include: collecting the grid interaction power error and the execution response time after regulation in real time; judging whether there is a situation where the grid interaction power error is greater than a preset tolerance or the execution response time is greater than a preset response threshold; if so, then adjust at least one of the weight factor and the power Gini coefficient according to a preset algorithm, calculate a new energy distribution balance degree; and re-select a modeling mode for network division according to the new energy distribution balance degree.

[0021] Optionally, the judgment of whether there is a situation where the grid interaction power error is greater than a preset tolerance or the execution response time is greater than a preset response threshold further includes: if there is a situation where the grid interaction power error is less than or equal to the preset tolerance and the execution response time is less than or equal to the preset response threshold, then obtain the current parameter combination to obtain an effective configuration set; collect the environmental data and operation parameter data of the scenario corresponding to the effective configuration set to obtain a scenario label; associate the effective configuration set with the scenario label and store it in a preset stable parameter library; when the system has an abnormal fault, match the scenario label according to the current environmental data and operation parameter data, and select the corresponding effective configuration set for network division according to the matched scenario label.

[0022] Compared with the prior art, the present invention has the following advantages:

[0023] 1. The present invention evaluates the energy distribution balance in real time through a dynamic modeling module, and flexibly selects a fully meshed, partially meshed, or unit individualization mode according to thresholds, enabling the system to automatically adapt to the spatio-temporal fluctuations of new energy distribution, significantly improving the accuracy and response speed of resource allocation, and avoiding the problem of a sharp increase in transmission losses caused by traditional fixed topological structures;

[0024] 2. Based on the three-dimensional space reconstruction technology of Delaunay triangulation and weighted node weights, the present invention generates a dynamic connection scheme, verifies the topological stability through Monte Carlo simulation, and combines the feedback mechanism of energy transmission loss values to ensure the optimal grid division, greatly reducing the energy cross-region transmission loss;

[0025] 3. The iterative feedback module of the present invention monitors the regulation effect in real time. When the deviation exceeds the threshold, it automatically adjusts the model parameters (such as weight factors, Gini coefficients) to drive the system to re-divide the network; at the same time, the historical effective configurations are associated and stored with the scenario labels, supporting the quick matching of the optimal parameter combination during faults, and enhancing the system robustness and recovery efficiency;

[0026] 4. In the partially meshed mode, the present invention avoids equipment damage caused by voltage mismatch through node voltage level difference threshold detection and local triangular unit grid reconstruction, ensures the safe operation of the power grid, and reduces the failure rate;

[0027] 5. In the unit individualization mode, the present invention determines through the port energy exchange rate, automatically marks the island nodes or forms independent microgrids, adapts to the dynamic access requirements of various devices such as photovoltaic, energy storage, and electric vehicles, and enhances the system scalability;

[0028] 6. The present invention combines the continuous monitoring of load instantaneous variance and the regional splitting mechanism to prevent local overload; dynamically adjusts the node spacing using a correction equation to balance the power change rate and the energy storage state, ensuring the system stability under complex fluctuation scenarios.

[0029] Other features and advantages of the present invention will be described in the subsequent specification, and will be partially obvious from the specification, or understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures pointed out in the specification, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0031] Figure 1 It is a framework diagram of a new energy grid management and analysis system for smart communities in an embodiment of the present invention.

[0032] Figure 2 It is an execution flowchart of a new energy grid management and analysis system for smart communities in an embodiment of the present invention.

[0033] Figure 3 It is a structural schematic diagram of a new energy grid management and analysis system for smart communities in an embodiment of the present invention. Specific implementation manners

[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0035] Referring to Figure 1 , an embodiment of the present invention proposes a new energy grid management and analysis system for smart communities. Through a dynamic grid topology adaptive division mechanism, a multi-source adaptive collaborative optimization algorithm, and a pre-verification network splitting simulation technology, it can achieve real-time dynamic reconstruction, multi-objective joint optimization, and millisecond-level security risk prevention and control of the community energy network, effectively solving problems such as high transmission loss, response delay, and equipment life reduction caused by traditional static division.

[0036] The system of this embodiment specifically includes:

[0037] A data acquisition module, which is distributed and deployed at each energy node in the community and is used to collect photovoltaic power generation, energy storage device, and user electricity consumption data in real time;

[0038] Specifically, the data acquisition module flexibly arranges sensor nodes according to the physical location distribution of energy nodes (such as photovoltaic panels, energy storage batteries, distribution boxes, etc.), directly embeds them at the device end or installs them nearby to form an Internet of Things perception network covering the entire community; continuously obtains data through high-precision sensors (voltage / current transformers, power meters, etc.), supports a millisecond-level sampling frequency, and ensures the timeliness of the data; the data types include the power generation power, irradiance, inverter efficiency, etc. of photovoltaic power generation, the charge and discharge power, state of charge (SOC), state of health (SOH), etc. of energy storage devices, and the time-of-use load curve, power factor, peak-valley electricity consumption characteristics, etc. of user electricity consumption.

[0039] A dynamic modeling module, which is connected to the data acquisition module and is used to dynamically generate a multi-level energy grid division scheme according to the collected data;

[0040] Specifically, the dynamic modeling module directly receives the real-time data stream from the data acquisition module and establishes a dynamic mathematical model of the energy scenario (such as Markov chain, spatio-temporal distribution probability model); the multi-level energy grid division generates a grid topology structure in layers (community level - building level - equipment level) based on the balance degree evaluation result, including fully networked (strongly coupled interconnection), which is applicable to high-density areas with balanced energy distribution; partially networked (dynamically selectively connected), which adapts to mixed scenarios with medium volatility; unit individuation (microgrid or island mode), which is used for decentralized nodes with low energy interaction rate.

[0041] The optimization decision-making module is used to output energy regulation parameters according to the grid division scheme;

[0042] Specifically, the optimization decision-making module dynamically distributes the photovoltaic output and energy storage charge and discharge plan according to the grid load, triggers the line relay or microgrid controller to adjust the physical connection, and optimizes the user-side demand response threshold for the time-of-use electricity price model.

[0043] The iterative feedback module is used to monitor the regulation effect and trigger the online update of the model parameters.

[0044] Specifically, the iterative feedback module monitors the regulation effect in real time (such as grid interaction error, response delay time), compares it with the preset performance indicators (tolerance, threshold); adjusts the contribution ratio of each index in the balance degree evaluation through the backpropagation algorithm, such as the weight factor; adaptively matches the change characteristics of the community energy distribution (such as seasonal light differences); binds the effective parameter combination with the historical operating environment (such as weather, holidays) to form a stable parameter library to support quick call during faults.

[0045] Exemplarily, such as Figure 2As shown, in the high photovoltaic power generation scenario at noon in summer, the data acquisition module collects the reported peak output of the photovoltaic inverter as 500 kW, the average SOC of the community energy storage is 85%, and the user air conditioner loads are concentratedly turned on, resulting in the power consumption of a local area (Building A) surging to 300 kW; when the dynamic modeling module conducts the balance degree evaluation, it is found that the power Gini coefficient of Building A increases (supply-demand imbalance), triggering partial grid reconstruction; based on Delaunay triangulation, the photovoltaic arrays and energy storage nodes adjacent to Building A are dynamically interconnected to generate independent sub-grids; the optimization decision module outputs regulation parameters, restricts the energy storage discharge of Building A, preferentially calls the redundant photovoltaic power of adjacent sub-grids, turns off the unnecessary public lighting lines in Building A, and reduces the local load to 250 kW; the iterative feedback module monitors after 10 minutes and finds that the grid interaction error still exceeds the standard (due to the fluctuation of photovoltaic output); thus, the iterative feedback module automatically adjusts the weight factor, increases the proportion of the power Gini coefficient in the balance degree, and re-divides the grid; after adjustment, the error is reduced within the preset tolerance, and the current parameter combination is marked as "high irradiance scenario in summer" and stored in the stable parameter library; the present invention forms a closed loop of "perception-modeling-decision-validation", solves the lag problem of one-way control of traditional systems, and avoids the risks of "over-regulation" or "under-regulation" through dynamic parameter adjustment; the iterative learning of historical operation data accelerates the adaptation efficiency of the system to new scenarios.

[0046] Optionally, as Figure 3 shown, the dynamic modeling module includes: a balance degree evaluation unit, a network classification unit, and a verification evaluation unit; wherein,

[0047] The balance degree evaluation unit is used to evaluate the balance degree of the community energy distribution according to the collected data;

[0048] Specifically, the balance degree evaluation unit is mainly used to evaluate the balance degree of the community energy distribution; it collects the power data of each regional node in real time, counts the number of energy interaction times between adjacent regional nodes, calculates the power Gini coefficient, and then comprehensively evaluates the balance degree of energy distribution by combining these indicators.

[0049] The network classification unit is used to select a fully meshed, partially meshed, or unit individualized modeling mode for network division according to the evaluation result;

[0050] Specifically, the network classification unit selects a suitable modeling mode for network division according to the result of the balance degree evaluation unit; these modeling modes include fully meshed, partially meshed, and unit individualized.

[0051] The verification evaluation unit is used to calculate the energy transmission loss value after grid division according to the division result, and re-divide when the energy transmission loss value exceeds a predetermined threshold.

[0052] Specifically, the verification and evaluation unit is used to evaluate the effect after network division, mainly by calculating the energy transmission loss value after division; if the transmission loss value exceeds a predetermined threshold, the re-division process is triggered.

[0053] Exemplarily, the energy transmission loss value can be based on power flow analysis (such as Newton-Raphson method) or empirical formula where I i is the line current and R i is the resistance; if the energy transmission loss value L is greater than a predetermined threshold (such as 5% of the total community energy consumption), it is determined as an invalid division and the re-modeling is triggered; simulate extreme load fluctuations (such as sudden drop in PV plus energy storage failure) to ensure topological robustness, and avoid potential risks of artificial division through closed-loop feedback.

[0054] Exemplarily, the data acquisition module monitors that the community PV output suddenly drops from 200 kW to 50 kW, and the user load remains unchanged at 300 kW, and the energy storage SOC quickly drops to 40%; according to the balance degree evaluation, the unit individual modeling mode is triggered, the port energy exchange rate is calculated by region, and area A with low energy storage power and high load demand is marked as an "island node" and operates independently; the remaining areas are constructed into microgrid units according to the maximum energy path; calculate the energy transmission loss after division, the original loss is 12 kW, after division, area A is powered by a diesel generator (loss 3 kW), and the loss of the remaining areas drops to 6 kW, and the total loss is 9 kW. Assuming the predetermined threshold is 15 kW, since the total loss of 9 kW is less than the predetermined threshold of 15 kW, the division plan passes the verification; if the continuous heavy rain causes the energy storage SOC to further drop to 20%, a secondary evaluation is triggered and re-modeling is carried out; form a closed loop from "balance degree quantification, mode selection, effect verification" to cope with sudden energy fluctuations; hierarchical thresholds avoid frequent mode switching and ensure system stability; through real-time loss calculation plus stability simulation, double avoid unsafe division plans.

[0055] Optionally, the evaluation of the balance degree of community energy distribution according to the collected data includes:

[0056] Real-time collect the power data of each regional node;

[0057] Specifically, the collection scope covers all key energy nodes (PV, energy storage, load) in the community, and the monitoring units are divided according to physical regions (such as buildings, distribution units); record the real-time power data of each node (positive value represents power supply, negative value represents power consumption); realize millisecond-level data update through devices such as smart meters and embedded power sensors; provide underlying data support for subsequent balance degree evaluation to ensure the real-time and accuracy of the evaluation.

[0058] According to the power data, count the number of energy interactions between adjacent regional nodes within a preset first time window and perform normalization processing;

[0059] Specifically, within the first time window (such as 15 minutes), when the energy transferred from area A to the adjacent area B exceeds a certain threshold (such as 10 kWh), it is recorded as an effective interaction. The statistical method can be based on the integral calculation of power data, and the statistical value is mapped to the interval from 0 to 1. For example, if the maximum historical interaction times is 50 times, when the interaction times within a certain window is 20 times, the normalized result is 0.4. Quantify the activity degree of energy complementarity between regions. The more frequent the interaction, the more balanced the energy network tends to be.

[0060] According to the power data, calculate the power Gini coefficient and perform normalization processing;

[0061] Specifically, sort the power of each region (the algebraic difference between power supply and load) from small to large, calculate the cumulative percentage of the Lorenz curve, and calculate the Gini coefficient by the area method: Among them, A is the area between the Lorenz curve and the absolute equality line, and B is the area under the Lorenz curve; linearly map the original Gini coefficient (usually from 0 to 1) to the same range for easy superposition with the weight of the energy interaction times. The smaller the Gini coefficient (G approaches 0), the more balanced the power distribution of each region is.

[0062] According to the normalized energy interaction times and the normalized power Gini coefficient, calculate the energy distribution balance degree. For the energy distribution balance degree Q, there is

[0063] Q = λ * G + (1 - λ) * T,

[0064] In the formula, λ is the weight factor, G is the normalized power Gini coefficient, and T is the normalized energy interaction times.

[0065] Specifically, the weight factor λ is dynamically adjusted through historical data training or real-time optimization algorithms.

[0066] Exemplarily, in the scenario of sudden drop in photovoltaic output during community lunchtime, within a 15-minute window, the original output of photovoltaic area A drops suddenly from 150 kW to 30 kW (due to cloud occlusion), the current power supply of energy storage area B is 80 kW, and the SOC is 60%, and the load demand of user area C is 200 kW; the number of times of energy transfer from photovoltaic area A to energy storage area B is 5 times, and the normalized value is 0.5, and the number of times of energy transfer from energy storage area B to user area C is 8 times, and the normalized value is 0.8; the net power of area A is 30 kW, the net power of area B is 80 kW, and the net power of area C is -200 kW. After sorting, the net power sequence is [-200, 30, 80]. The calculated Gini coefficient is 0.52, and it remains 0.52 after normalization; assuming the weight factor is 0.7, the balance degree of photovoltaic area A is 0.514, and the balance degree of user area C is 0.604; through real-time power data and interaction times, the problem of "obsolescence" in traditional static evaluation is avoided, and at the same time, "power distribution balance" and "energy interaction activity" are considered to improve the evaluation dimension; the weight factor can be dynamically adjusted according to external conditions such as seasons and weather (for example, increasing the energy interaction weight in rainy weather to give priority to ensuring power supply in key areas).

[0067] Optionally, the selecting a fully meshed, partially meshed or unit individual modeling mode for network partitioning according to the evaluation result includes:

[0068] Judging whether the energy distribution balance degree is greater than a preset first threshold;

[0069] If the energy distribution balance degree is greater than the first threshold, select a fully meshed modeling mode for network partitioning;

[0070] If the energy distribution balance degree is less than or equal to the first threshold, judge whether the energy distribution balance degree is greater than a preset second threshold;

[0071] If the energy distribution balance degree is greater than the second threshold, select a partially meshed modeling mode for network partitioning;

[0072] If the energy distribution balance degree is less than or equal to the second threshold, select a unit individual modeling mode for network partitioning.

[0073] Optionally, the selecting a fully meshed modeling mode for network partitioning includes:

[0074] Model the node connection relationship of the area to be split as a weighted undirected graph F = {V, E, W}, where the vertex set V represents power nodes, the edge set E represents physical connection lines, and the weight W is the line transmission capacity;

[0075] Specifically, taking the connection points of equipment such as distribution boxes, photovoltaic inverters, and energy storage converters as vertices, their two-dimensional positions are determined in the community geographic coordinate system; the line transmission capacity is set as a weight value according to the wire cross-sectional area and the rated current-carrying capacity; when there is an increase in the capacity of new energy equipment, the vertex set and the edge set topology are dynamically updated.

[0076] Based on Delaunay triangulation, determine the physical adjacent node group and generate the initial candidate splitting plane;

[0077] Specifically, when calculating, first execute the standard Delaunay triangulation algorithm on the vertex set to automatically generate the minimum convex polygon coverage composed of adjacent nodes; then extract the line segments connecting adjacent node pairs as the candidate set of splitting planes, and preferentially select the line segments with high weights (transmission capacity) as the candidate planes.

[0078] Calculate the instantaneous variance of the load of each phase in the current area and preset the second time window;

[0079] Specifically, set the second time window to 30s and collect the current of each phase in real time; for the three-phase load, calculate its instantaneous variance respectively, where is the average current value within the window; when the instantaneous variance of a certain phase continuously exceeds the fluctuation threshold (such as 1.2), the splitting condition is triggered.

[0080] When it is detected that the instantaneous variances within a continuous preset number of the second time windows are greater than or equal to the preset fluctuation value, perform regional splitting along the initial candidate splitting plane to obtain multiple sub-regions;

[0081] Specifically, if the instantaneous variance of any phase within three consecutive second time windows (i.e., 90s) exceeds the preset fluctuation value, it is determined that there is long-term load fluctuation in this area. The system selects the candidate plane with the strongest correlation with the high-variance phase from the candidate set of splitting planes (such as the wire that divides the dominant load area of this phase), and performs the topological splitting operation to form sub-regions to ensure that the variance within each sub-region is limited.

[0082] When the instantaneous variances of all sub-regions are less than the preset fluctuation value, stop the splitting process and lock the topological structure;

[0083] Specifically, continuously monitor the variance index of the sub-regions after splitting. If the instantaneous variances of all sub-regions within the monitoring period (such as 5min) are less than the preset fluctuation value, it is determined that the system enters the stable state, write the current topological structure into the configuration file and lock it; otherwise, trigger recursive splitting until the conditions are met.

[0084] Use Monte Carlo simulation to verify the dynamic stability of the topological structure after splitting under the preset load disturbance, and write the network division topological structure that passes the verification into the preset partition database.

[0085] Specifically, random load disturbances are applied to each sub-region during the simulation process where σ P Take 1.5 times the historical maximum fluctuation value; in 1000 Monte Carlo iterations, if the line overload probability is lower than the preset value (such as 5%) and the node voltage violation probability is lower than the preset value (such as 3%), it is determined that the topology is stable and written into the partition database for optimization decision-making calls.

[0086] Optionally, the selection of the partial networked modeling mode for network division includes:

[0087] Collect the energy storage state of charge values of each distributed node and calculate the power change rate within a preset adjacent third time window in real time;

[0088] Exemplarily, set the third time window to 10 minutes. For the nth energy storage node, obtain its SOC value SOC n ∈[0,1].

[0089] Using the energy storage state of charge value and the power change rate, calculate the node weight. For the node weight w n , there is

[0090]

[0091] In the formula, α is the weight coefficient of the energy storage state of charge value, SOC n is the energy storage state of charge value of the nth distributed node, tanh is the hyperbolic tangent function, P n is the power change rate of the nth distributed node within the adjacent third time window, P max is the rated power change threshold of the system, and β is the weight coefficient of the power change rate;

[0092] Exemplarily, in the formula, α = 0.6 is the weight coefficient of the energy storage state of charge value, β = 0.4 is the weight coefficient of the power change rate, P max = 50kW / min is the rated power change threshold of the system; the tanh function compresses the ratio of the power change rate to the threshold to the range of [0,1) to prevent the weight from being oversaturated; nodes with higher weights (w n > 0.8) represent high SOC and stable power changes and are suitable as hub nodes.

[0093] Map each distributed node to two-dimensional space coordinates and construct a three-dimensional space with the node weight as the third dimension coordinate;

[0094] Exemplarily, the geographical coordinates (x n , y n ) are determined by GPS positioning or a preset layout diagram, and the three-dimensional coordinates are (x n , yn , w n ); By raising the "height" of the weighted nodes in the three-dimensional space, it is ensured that the high-weight nodes are preferentially interconnected during Delaunay triangulation.

[0095] Perform weighted Delaunay triangulation in the three-dimensional space, and use a preset correction equation to correct the distributed node spacing to complete dynamic connection, obtaining multiple triangular element meshes. The correction equation is

[0096]

[0097] In the formula, d' ij is the distance between the i-th distributed node and the j-th distributed node, and d ij is the original geographical distance between the i-th distributed node and the j-th distributed node, and w i is the node weight of the i-th distributed node, and w j is the node weight of the j-th distributed node.

[0098] Exemplarily, after the original two-dimensional geographical distance d ij is corrected by exponential weighting, the virtual distance between high-weight nodes is shortened, promoting the formation of grid cores; conversely, the connection of low-weight nodes is weakened; the corrected distance is input into the Delaunay algorithm to generate a three-dimensional triangular element mesh network.

[0099] Optionally, the selection of the partial networked modeling mode for network division further includes:

[0100] Calculate the weight difference of the edges of each triangular element. The weight difference ε = |w i - w j |;

[0101] Specifically, when the weight difference between the two end nodes exceeds the threshold, it may cause a problem of unidirectional power inrush.

[0102] Calculate the difference threshold of the edges of each triangular element. The difference threshold where max() is the maximum value function;

[0103] When the weight difference is greater than the difference threshold, obtain the voltage levels of the two distributed nodes corresponding to the edge of the triangular element.

[0104] If the difference between the voltage levels of the two distributed nodes is greater than 10% of the rated voltage, then delete the corresponding edge of the triangular element and initiate local triangular element mesh reconstruction.

[0105] Exemplarily, when the weight difference is greater than the difference threshold, obtain the voltage levels of the two distributed nodes corresponding to the side of the triangular unit; for example, the photovoltaic node is usually 380V AC, and the energy storage node may support a 400V DC bus, which needs to be coupled through an inverter; if the difference between the voltage levels of the two distributed nodes is greater than 10% of the rated voltage, then delete the side of the corresponding triangular unit and start the local triangular unit grid reconstruction; after that, the system re-performs the local Delaunay triangulation on the affected area (such as the node that has lost two adjacent sides), and supplements the replacement sides to maintain the grid connectivity; when reconstructing, preferentially select adjacent nodes with matching voltage levels (such as the error ≤ 5%) to establish connections to avoid the risk of AC-DC mixed connection.

[0106] Optionally, the network division in the individual modeling mode of the selection unit includes:

[0107] Collect the port energy exchange rate of each device unit;

[0108] Judge whether the energy exchange rate is greater than a preset third threshold;

[0109] Specifically, collect the port energy exchange rate of each device unit, the time window can be 1h, the port energy exchange rate reflects the energy interaction intensity between the device and the external network, the port energy exchange rate approaching 1 indicates strong exchange (frequent charging and discharging), and the port energy exchange rate approaching 0 indicates isolated operation.

[0110] If the energy exchange rate is greater than the third threshold, mark the device as an island node and perform independent modeling;

[0111] Exemplarily, if the port energy exchange rate of a certain energy storage unit is 0.85 due to frequent charging and discharging during the peak period, it is judged that its coupling degree with the main grid is too high; mark the energy storage unit as an island node and perform independent modeling; in the island mode, this node disconnects the physical connection with the main grid and enables the local controller to maintain autonomous operation (such as diesel generator backup) to avoid the cascading impact on the main grid when overloaded.

[0112] If the energy exchange rate is less than or equal to the third threshold, form an independent microgrid unit according to the maximum energy exchange path and the nearest neighbor node.

[0113] Exemplarily, if the energy exchange rate is less than or equal to the third threshold, form an independent microgrid unit according to the maximum energy exchange path and the nearest neighbor node; traverse and search the historical energy interaction records between this device and adjacent nodes, and select the 3 nodes with the largest interaction volume to form a star-shaped microgrid; the topology optimization goal is to minimize the internal line loss of the microgrid, and generate the minimum spanning tree connection scheme through the Prim algorithm.

[0114] Optionally, the monitoring of the regulation effect and the triggering of the online update of the model parameters include:

[0115] Real-time collect the grid interaction power error and execution response time after regulation;

[0116] Determine whether there is a situation where the grid interaction power error is greater than a preset tolerance or the execution response time is greater than a preset response threshold;

[0117] If so, adjust at least one of the weight factor and the power Gini coefficient according to a preset algorithm, and calculate a new energy distribution balance degree;

[0118] Re-select a modeling mode for network partitioning according to the new energy distribution balance degree.

[0119] Exemplarily, the execution response time is the time from the instruction issuance to the power reaching the standard. Real-time collect the grid interaction power error after regulation. If there is a grid interaction power error greater than 5% or the execution response time greater than 2 s, trigger parameter update, adjust the weight factor according to a preset algorithm, or reset the calculation benchmark of the power Gini coefficient G; meanwhile, the adjustment strategy can also include increasing the weight coefficient of the energy storage state of charge value when the grid interaction power error exceeds the standard, and reducing the weight coefficient of the power change rate when the execution response time exceeds the standard to weaken the sensitivity of the power change factor. The weight coefficient of the energy storage state of charge value can be iteratively optimized by the gradient descent method.

[0120] Optionally, the determination of whether there is a situation where the grid interaction power error is greater than a preset tolerance or the execution response time is greater than a preset response threshold further includes:

[0121] If there is a situation where the grid interaction power error is less than or equal to the preset tolerance and the execution response time is less than or equal to the preset response threshold, obtain the current parameter combination to obtain an effective configuration set;

[0122] Collect the environmental data and operation parameter data of the scenario corresponding to the effective configuration set to obtain a scenario label;

[0123] Associate the effective configuration set with the scenario label and store it in a preset stable parameter library;

[0124] When the system has an abnormal fault, match the scenario label according to the current environmental data and operation parameter data, and select the corresponding effective configuration set for network partitioning according to the matched scenario label.

[0125] Exemplarily, if there is a grid interaction power error less than or equal to a preset tolerance and an execution response time less than or equal to a preset response threshold, obtain the current parameter combination to obtain an effective configuration set; collect the environmental data (such as temperature, light intensity) and operating parameter data (average SOC, load peak-valley ratio) of the scenarios corresponding to the effective configuration set to obtain scenario labels; for example, the label is "winter evening peak - low light"; associate the effective configuration set with the scenario label and store it in a preset stable parameter library; support the subsequent system to automatically call historical optimal parameters in similar scenarios (by matching environmental feature vectors with KNN) to shorten the convergence time; when the system has an abnormal fault (such as data loss caused by communication interruption), match the scenario label according to the current environmental data and operating parameter data, and select the corresponding effective configuration set for network division according to the matched scenario label; for example, when the photovoltaic is offline due to thunderstorm weather, call the pre-stored parameters under the label of "rainy - energy storage dominant" to quickly reconstruct a stable network.

[0126] It should be noted that the electrical connections between the above-mentioned various units do not necessarily mean direct connection of the lines. Indirect connection methods can be applied to the embodiments of the present invention as long as the purpose of the present invention is achieved. The above are only exemplary embodiments of the present invention and cannot be used to limit the scope of the present invention.

[0127] That is, any equivalent changes and modifications made in accordance with the teachings of the present invention still fall within the scope covered by the present invention. Those skilled in the art will easily think of other implementation schemes of the present invention after considering the specification and the disclosure of the practical truth. This application aims to cover any variations, uses, or adaptive changes of the present invention, and these variations, uses, or adaptive changes follow the general principles of the present invention and include the common general knowledge or conventional technical means in the technical field not recorded in the present invention.

Claims

1. A smart community new energy grid management and analysis system, characterized by: The system comprises: The data collection module is deployed at various energy nodes in the community to collect real-time data on photovoltaic power generation, energy storage equipment and user electricity consumption; A dynamic modeling module, connected to the data acquisition module, for dynamically generating a multi-level energy grid division scheme based on the acquired data; An optimization decision module, used for outputting energy control parameters according to the grid division scheme; Iterative feedback module is used to monitor the control effect and trigger the online update of model parameters.

2. According to claim 1, a smart community new energy grid management and analysis system is characterized in that: The dynamic modeling module includes: a balance evaluation unit, a network classification unit, and a verification evaluation unit; wherein, The balance evaluation unit is used to evaluate the balance of community energy distribution based on the collected data; The network classification unit is used to select a full gridding, partial gridding or unit individualization modeling mode for network division according to the evaluation result; The verification and evaluation unit is used to calculate the energy transmission loss value after grid division according to the division result, and to re-divide the grid when the energy transmission loss value exceeds a predetermined threshold.

3. According to claim 2, a smart community new energy grid management and analysis system is characterized in that: The method of evaluating the balance of community energy distribution based on the collected data includes: Collect power data of nodes in each area in real time; According to the power data, the number of energy interactions between nodes in adjacent regions within a preset first time window is counted and normalized; According to the power data, a power Gini coefficient is calculated and normalized; According to the normalized energy interaction times and the normalized power Gini coefficient, the energy distribution balance is calculated. For the energy distribution balance Q, we have Q = λ*G+(1-λ)*T, In the formula, λ is the weight factor, G is the normalized power Gini coefficient, and T is the normalized number of energy interactions.

4. According to claim 2, a smart community new energy grid management and analysis system is characterized in that: The selection of full gridding, partial gridding or unit individualization modeling mode for network division according to the evaluation results includes: Determining whether the energy distribution balance is greater than a preset first threshold; If the energy distribution balance is greater than the first threshold, a fully networked modeling mode is selected for network division; If the energy distribution balance degree is less than or equal to the first threshold, determining whether the energy distribution balance degree is greater than a preset second threshold; If the energy distribution balance is greater than the second threshold, selecting a partial network modeling mode for network division; If the energy distribution balance degree is less than or equal to the second threshold, the unit individualization modeling mode is selected to perform network division.

5. According to claim 4, a smart community new energy grid management and analysis system is characterized in that: The selecting of the fully networked modeling mode for network partitioning includes: The node connection relationship of the area to be split is modeled as a weighted undirected graph F = {V, E, W}, where the vertex set V represents the power nodes, the edge set E represents the physical connection line, and the weight W is the line transmission capacity; Determine the physical neighboring node group based on Delaunay triangulation and generate the initial candidate splitting plane; Calculate the instantaneous variance of each phase load in the current area and preset the second time window; When it is detected that the instantaneous variance in the second time window for a continuous preset number of times is greater than or equal to the preset fluctuation value, performing region splitting along the initial candidate splitting plane to obtain a plurality of sub-regions; When the instantaneous variance of all sub-regions is less than the preset fluctuation value, the splitting process is stopped and the topology is locked; Monte Carlo simulation is used to verify the dynamic stability of the split topology under the preset load disturbance, and the verified network partition topology is written into the preset partition database.

6. According to claim 4, a smart community new energy grid management and analysis system is characterized in that: The selecting of a partial network modeling mode for network partitioning includes: Collect the energy storage charge state value of each distributed node, and calculate the power change rate within the adjacent preset third time window in real time; The node weight is calculated using the energy storage charge state value and the power change rate. n ,have In the formula, α is the weight coefficient of the energy storage state of charge value, SOC n is the energy storage charge state value of the nth distributed node, tanh is the hyperbolic tangent function, P n is the power change rate of the nth distributed node in the third adjacent time window, P max is the rated power change threshold of the system, β is the weight coefficient of the power change rate; Mapping each distributed node into a two-dimensional space coordinate, and using the node weight as a third-dimensional coordinate to construct a three-dimensional space; A weighted Delaunay triangulation is performed in the three-dimensional space, and the distributed node spacing is corrected using a preset correction equation to complete the dynamic connection, thereby obtaining a plurality of triangular unit meshes. The correction equation is: Where, d' ij is the distance between the i-th distributed node and the j-th distributed node, d ij is the original geographical distance between the i-th distributed node and the j-th distributed node, w i is the node weight of the i-th distributed node, w j is the node weight of the jth distributed node.

7. A smart community new energy grid management and analysis system according to claim 6, characterized in that: The selecting of a partial network modeling mode for network partitioning also includes: Calculate the weight difference of the edges of each triangular unit, the weight difference ε = |w i -w j |; Calculate the difference threshold of the edge of each triangular unit, the difference threshold max(w i ,w j ), where max() is the maximum value function; When the weight difference is greater than the difference threshold, obtaining voltage levels of two distributed nodes corresponding to the edges of the triangular unit; If the difference between the voltage levels of two distributed nodes is greater than 10% of the rated voltage, the corresponding triangular element edges are deleted and local triangular element mesh reconstruction is started.

8. According to claim 4, a smart community new energy grid management and analysis system is characterized in that: The selecting unit individualized modeling mode for network partitioning includes: Collect the port energy exchange rate of each equipment unit; Determining whether the energy exchange rate is greater than a preset third threshold; If the energy exchange rate is greater than the third threshold, marking the device as an island node and independently modeling it; If the energy exchange rate is less than or equal to the third threshold, an independent microgrid unit is established with the nearest neighbor node according to the maximum energy exchange path.

9. The smart community new energy grid management and analysis system according to claim 3 is characterized in that: The monitoring of the regulation effect and triggering the online update of the model parameters include: Real-time collection of grid interaction power error and execution response time after regulation; Determine whether the grid interaction power error is greater than a preset tolerance or the execution response time is greater than a preset response threshold; If yes, at least one of the weight factor and the power Gini coefficient is adjusted according to a preset algorithm to calculate a new energy distribution balance; Reselect the modeling mode to divide the network according to the new energy distribution balance.

10. A smart community new energy grid management and analysis system according to claim 9, characterized in that: The determining whether the grid interaction power error is greater than a preset tolerance or the execution response time is greater than a preset response threshold further includes: If the grid interaction power error is less than or equal to the preset tolerance and the execution response time is less than or equal to the preset response threshold, the current parameter combination is obtained to obtain a valid configuration set; Collecting environmental data and operating parameter data of the scene corresponding to the valid configuration set to obtain a scene label; Associating the valid configuration set with the scene tag and storing them in a preset stable parameter library; When the system fails abnormally, the scenario label is matched according to the current environment data and operation parameter data, and the corresponding valid configuration set is selected according to the matched scenario label for network division.

Images