Risk comprehensive assessment optimization method and device for low-voltage distribution area

Abstract

The invention provides a low-voltage distribution area risk comprehensive assessment optimization method and device, relates to the field of low-voltage distribution areas, and solves the technical problems that complex and changeable operation scenes are difficult to deal with, and a centralized optimization model cannot meet the real-time requirement due to communication delay. The method comprises the steps of obtaining multi-source data and performing data cleaning to obtain multi-source heterogeneous feature data; dimensionality reduction is carried out on the multi-source heterogeneous feature data through principal component analysis and an automatic encoder; defining a feature vector based on a dimension reduction result, mapping the feature vector to a unified interval through Min-Max standardization to generate a three-dimensional tensor, and obtaining a standardized risk feature matrix; according to the application, traditional isolated monitoring, evaluation and disposal links are integrated into an organic whole, and collaborative optimization of voltage fluctuation suppression, power supply reliability improvement and user power failure range reduction is realized on the premise of not upgrading hardware facilities on a large scale.

Description

Technical Field

[0001] This application relates to the field of low-voltage distribution transformer areas, and in particular to a method and apparatus for comprehensive risk assessment and optimization of low-voltage distribution transformer areas. Background Technology

[0002] With the acceleration of urbanization and electrification, the operating environment of low-voltage distribution transformer areas is becoming increasingly complex. The existing risk management system is no longer adequate to meet the needs of new power systems, highlighting core contradictions and technical bottlenecks. On the one hand, the proportion of impact loads such as central air conditioning and fast charging piles in commercial complexes and industrial parks is rising. Their random start-stop and large power fluctuations cause high-frequency instantaneous voltage drops and short-term over-limits in the transformer area, accelerating transformer aging and threatening power supply reliability. The large-scale integration of distributed photovoltaic and energy storage further exacerbates three-phase imbalance. Actual measurements in one industrial park showed that the proportion of zero-sequence current surged from 3%-5% to over 15%, triggering the risk of neutral line overload. Traditional handling relies on 10kV side operation, which has a 5-10 minute response delay and is prone to causing power outages for non-faulty users, contradicting the "user-centric" concept.

[0003] Existing solutions have significant limitations. Traditional SCADA systems, with their 15-minute sampling cycle, only collect the effective values ​​of electrical quantities, failing to capture millisecond-level disturbances and waveform distortions. Furthermore, they do not incorporate non-electrical quantities such as ambient temperature, resulting in blind spots in perception. In terms of assessment methods, existing risk assessment models mostly employ fixed threshold judgment mechanisms, neglecting the dynamic characteristics of the load and the coupling effects of the environment, making it difficult to pinpoint the root cause of risks. Therefore, existing technical approaches have significant shortcomings in these key areas: expert systems based on fixed rules struggle to cope with complex and ever-changing operating scenarios; centralized optimization models cannot meet real-time requirements due to communication delays; and fully distributed control strategies are prone to falling into local optima traps. Summary of the Invention

[0004] This application provides a method and apparatus for comprehensive risk assessment and optimization of low-voltage distribution transformer areas, which solves the technical problem that the centralized optimization model cannot meet the real-time requirements due to communication delays in dealing with complex and ever-changing operating scenarios that are difficult to cope with in the prior art.

[0005] To achieve the above objectives, this application adopts the following technical solution: Firstly, a comprehensive risk assessment and optimization method for low-voltage distribution transformer areas is provided, including: Acquire multi-source data and perform data cleaning to obtain multi-source heterogeneous feature data; Principal component analysis and autoencoder are used to reduce the dimensionality of multi-source heterogeneous feature data to obtain the dimensionality reduction result; feature vectors are defined based on the dimensionality reduction result, and the feature vectors are mapped to a unified interval through Min-Max standardization to generate a three-dimensional tensor, thus obtaining a standardized risk feature matrix; A standardized risk feature matrix is ​​input into a risk assessment model to obtain risk assessment results; wherein, the risk assessment model is trained based on an artificial intelligence model. Based on the risk assessment results, a tiered and coordinated approach is implemented, and the Pareto optimal solution set that maximizes the voltage qualification rate and minimizes the load rate is obtained by improving the PSO algorithm. The improved PSO algorithm introduces an adaptive adjustment mechanism for inertial weights and non-dominated sorting.

[0006] It should be noted that the inertia weight adaptive adjustment mechanism sets a larger inertia weight in the early stage of iteration to enhance the global exploration capability; and gradually reduces the inertia weight in the later stage of iteration to improve the convergence accuracy. Non-dominated sorting: For solving multi-objective problems, the entire particle swarm is sorted in each iteration, and all particles are divided into different fronts. The first layer is the Pareto front, and the front with the higher ranking is selected first to enter the next iteration.

[0007] In conjunction with the first aspect above, in one possible implementation, the multi-source data includes electrical quantities and non-electrical quantities; wherein the electrical quantities include three-phase voltage, three-phase current, and current load data; and the non-electrical quantities include ambient temperature and load forecast values. The data cleaning method for the three-phase voltage includes: Using the Hamming window function The three-phase voltage is filtered using a sliding window; where... The voltage output value at the nth sampling point. The original three-phase voltage signal is sampled at time nk. These are the weighting coefficients for the Hamming window. , This is the cosine modulation term; The cleaning method for the three-phase current and current load data is: outliers are removed by 3σ standard measurement; The cleaning method for non-electrical quantities is as follows: non-electrical quantities are screened using a customized threshold method.

[0008] In conjunction with the first aspect above, in one possible implementation, the dimensionality reduction of multi-source heterogeneous feature data through principal component analysis and autoencoder includes: Z-score normalization is performed on multi-source heterogeneous feature data to obtain a high-dimensional feature matrix X∈R. m×27 ; Calculate the covariance matrix C; where m is the number of samples, and the covariance matrix is ​​calculated as follows: ; Eigenvalue decomposition of the covariance matrix C yields 27 eigenvalues. and the corresponding vector; According to the cumulative variance contribution rate The first n principal components are selected based on the preset conditions to form the projection matrix. Where n is a positive integer; Projecting the high-dimensional feature matrix X onto the principal component space yields the PCA dimensionality reduction result. ; The high-dimensional feature matrix X is input into the autoencoder network to obtain the encoder dimensionality reduction result. ; Through calculation formula The dimensionality reduction result is calculated, where the dimensionality reduction result is... ; For fusion weighting coefficients.

[0009] In conjunction with the first aspect above, in one possible implementation, the autoencoder network includes an input layer, a first hidden layer, a second hidden layer, an encoding layer, and a decoding layer; The loss function Loss of the autoencoder network is the mean squared error; wherein, the loss function ;in, For the z-th original sample of the autoencoder, This is the z-th sample reconstructed by the decoding layer.

[0010] In conjunction with the first aspect above, in one possible implementation, the feature vector includes a voltage over-limit feature vector, a load state feature vector, and a three-phase imbalance feature vector; The voltage over-limit feature vector is: ;in, This is the measured voltage. Nominal voltage, For voltage over-limit time, Total monitoring time This is the imaginary part of the voltage. This represents the real part of the voltage. The load state feature vector: ;in, This is the load forecast value. For maximum load capacity, This is the actual load. For peak load, This represents the load trough value. Average load; The three-phase imbalance characteristic vector: ;in, It is a negative sequence current. It is a positive sequence current. It is the zero-sequence current. The power between phases A and B. This refers to the power between phases B and C. This refers to the power between phases C and A. This represents the average value of the relevant phase-to-phase power.

[0011] In conjunction with the first aspect above, in one possible implementation, the step of mapping the feature vectors to a unified interval through Min-Max normalization to generate a three-dimensional tensor includes: From the dimensionality reduction results In the process, the voltage over-limit feature vector V, the load state feature vector L, and the three-phase imbalance feature vector U are extracted; and the indices of the voltage over-limit feature vector V, the load state feature vector L, and the three-phase imbalance feature vector U are mapped to [0,1]. The dimensions of the three-dimensional blank tensor are set, and the three-dimensional blank tensor is filled to obtain the three-dimensional tensor; the dimensions of the three-dimensional tensor include the time dimension, the spatial dimension, and the index dimension.

[0012] It should be noted that the voltage over-limit feature vector V includes voltage deviation rate v1, over-limit duration v2, and phase offset angle v3; the load state feature vector L includes the predicted value / capacity threshold ratio l1, load change rate slope l2, and peak-valley difference l3; the three-phase imbalance feature vector includes the proportion of negative sequence current component q1, the proportion of zero sequence current component q2, and the phase-to-phase active power difference q3.

[0013] In conjunction with the first aspect mentioned above, in one possible implementation, the risk assessment model includes a dynamic weight allocation module, a risk entropy calculation module, and a risk level classification and tracing module. The dynamic weight allocation module is used to adjust the weights of the voltage over-limit feature vector, the load state feature vector, and the three-phase imbalance feature vector. The risk entropy calculation module is used to calculate the health index using a three-dimensional tensor and dynamic weights. Through calculation formula Quantifying uncertainty; among which, Let i be the i-th standardized eigenvalue in the eigenvector. The dynamic weights of the feature vectors, The health index is a preset constant. y represents the total number of features; The risk level classification and tracing module is used to map the health index to a risk level and locate the main cause of the risk.

[0014] In conjunction with the first aspect above, in one possible implementation, the dynamic weights of the feature vector are obtained as follows: ;in, Let be the weight of the eigenvector at time t. The eigenvector includes the voltage eigenvector, the load eigenvector, and the imbalance eigenvector. For real-time load rate, For ambient temperature, the Sigmoid function maps the variable to [0,1]. Constraints: + + , Let be the weight of the voltage eigenvector at time t. The weights of the load eigenvectors at time t. The weights of the unbalanced eigenvectors at time t; , , , , , These are the model parameters, obtained during the training of the risk assessment model.

[0015] In conjunction with the first aspect above, in one possible implementation, the tiered collaborative response based on risk assessment results includes: The coordinated processing of the energy storage system includes: using a smooth switching algorithm. The energy storage system's operating modes are switched to shutdown, constant pressure, constant power, and emergency support modes; among these... This is the new power value after the energy storage system switches over; The old power value before the energy storage system was switched over; This represents the expected power change before and after the switch. This is the time constant used to control the degree of smoothness; Set the required switching time; The time elapsed since the start of the switch; The constant voltage mode is defined as follows: when the voltage deviation... At that time, through calculation formula Adjust the output power; among which, This refers to the output power. For voltage deviation; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; The constant power mode is as follows: when the current load data is >90%, the calculation formula is used. The logic controls the discharge power; among which, The discharge power of the energy storage system. The rated power of the energy storage system. For the available capacity of the energy storage system, The duration of discharge for the energy storage system; The emergency support mode is as follows: when SOC > 30% and voltage drop exceeds 10%, the system will activate the emergency support mode through a calculation formula. To implement control; among which, This refers to the output power of the energy storage system in emergency support mode. The rated power of the energy storage system. It is a time variable. This is the time constant used to control the decay characteristics; Peak load reduction measures include: classifying interruptible load levels through K-means clustering, constructing a spatiotemporal dual-dimensional control strategy, and constraining and controlling air conditioning load through a predictive average vote (PMV) model. The spatiotemporal dual-dimensional control strategy includes: time-dimensional control and spatial-dimensional control; wherein, the time-dimensional control is to shift the electricity consumption time from peak load periods to off-peak periods without reducing the total electricity consumption; the spatial-dimensional control is to adjust the power supply point or power supply path using control means.

[0016] It should be noted that: using the Predictive Average Voting (PMV) model to constrain and control air conditioning load involves: calculating the current environmental PMV value in real time and setting an acceptable PMV range; with the goal of maintaining the PMV value within an acceptable comfort range, dynamically adjusting the air conditioning operating parameters. For example, when the power grid needs to reduce load, the air conditioning power consumption is reduced as much as possible while ensuring the PMV value does not exceed the range.

[0017] In conjunction with the first aspect mentioned above, one possible implementation method for the coordinated processing of energy storage systems also includes: constructing a cycle lifetime decay model based on the Arrhenius equation. and through the objective function Optimize the SOC; among which, This represents the maximum number of charge-discharge cycles the battery can perform at the current temperature. Reference temperature The maximum number of charge-discharge cycles for the battery at this time; This is the battery activation energy; Boltzmann's constant; The current temperature; For reference temperature, Let be the battery power at time t. For reference power, These are the weighting coefficients. This represents the change in battery health status.

[0018] In conjunction with the first aspect above, in one possible implementation, in each iteration of the improved PSO algorithm for solving the Pareto optimal solution set that maximizes voltage qualification rate and minimizes load rate, the inertia weight of each particle is dynamically adjusted according to the iteration progress; after each iteration, all solutions in the entire population are compared pairwise to classify non-dominated levels; solutions that are not dominated by any other particle according to their non-dominated levels are included in the Pareto optimal candidate solution set; after the iteration is completed, all solutions in the Pareto optimal candidate solution set are output to obtain the Pareto optimal solution set.

[0019] In a second aspect, an electronic device is provided, comprising: a communication unit and a processing unit; the communication unit is configured to acquire multi-source data; the multi-source data includes electrical quantities and non-electrical quantities; wherein the electrical quantities include three-phase voltage, three-phase current, and current load data; the non-electrical quantities include ambient temperature and load forecast values; the processing unit is configured to perform data cleaning on the multi-source data to obtain multi-source heterogeneous characteristic data; Principal component analysis and autoencoder are used to reduce the dimensionality of multi-source heterogeneous feature data to obtain the dimensionality reduction result; feature vectors are defined based on the dimensionality reduction result, and the feature vectors are mapped to a unified interval through Min-Max standardization to generate a three-dimensional tensor, thus obtaining a standardized risk feature matrix; A standardized risk feature matrix is ​​input into a risk assessment model to obtain risk assessment results; wherein, the risk assessment model is trained based on an artificial intelligence model. Based on the risk assessment results, a tiered and coordinated approach is implemented, and the Pareto optimal solution set that maximizes the voltage qualification rate and minimizes the load rate is obtained by improving the PSO algorithm.

[0020] Thirdly, this application provides an electronic device comprising: a processor and a storage medium; the storage medium includes instructions, and the processor is configured to execute the instructions to implement the methods described in the first aspect and any possible implementation thereof. The electronic device may be an electronic device or a chip within an electronic device.

[0021] Fourthly, this application provides a comprehensive risk assessment and optimization system for low-voltage distribution transformer areas, comprising: a data acquisition module, a data processing module, and an assessment and optimization module; The data acquisition module is used to acquire multi-source data, which includes electrical and non-electrical quantities. The electrical quantities include three-phase voltage, three-phase current, and current load data. The non-electrical quantities include ambient temperature and load forecast values. The data processing module is used to clean multi-source data to obtain multi-source heterogeneous feature data; Principal component analysis and autoencoder are used to reduce the dimensionality of multi-source heterogeneous feature data to obtain the dimensionality reduction result; feature vectors are defined based on the dimensionality reduction result, and the feature vectors are mapped to a unified interval through Min-Max standardization to generate a three-dimensional tensor, thus obtaining a standardized risk feature matrix; The evaluation and optimization module is used to input the standardized risk feature matrix into the risk assessment model to obtain the risk assessment result; wherein, the risk assessment model is trained based on an artificial intelligence model; Based on the risk assessment results, a tiered and coordinated approach is implemented, and the Pareto optimal solution set that maximizes the voltage qualification rate and minimizes the load rate is obtained by improving the PSO algorithm.

[0022] Fifthly, this application provides a computer-readable storage medium storing instructions that, when executed on an electronic device, cause the electronic device to perform the methods described in the first aspect and any possible implementation thereof.

[0023] In a sixth aspect, this application provides a computer program product containing instructions that, when run on an electronic device, cause the electronic device to perform the methods described in the first aspect and any possible implementation thereof.

[0024] This application provides a method and device for comprehensive risk assessment and optimization of low-voltage distribution transformer areas. By using a dynamic weighted interpretability risk assessment model and handling the conflicting relationships of multiple indicators through fuzzy membership functions, and dynamically adjusting the weight allocation in combination with load rate and ambient temperature, risk assessment results are generated. This significantly improves the timeliness and accuracy of risk cause analysis, and integrates the traditionally isolated monitoring, assessment, and handling links into an organic whole. Without large-scale upgrades to hardware facilities, it achieves synergistic optimization of voltage fluctuation suppression, improved power supply reliability, and reduced user power outage range.

[0025] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0026] Figure 1 A system architecture diagram of a low-voltage distribution transformer area risk comprehensive assessment and optimization system provided in this application embodiment; Figure 2 A flowchart illustrating a comprehensive risk assessment and optimization method for low-voltage distribution transformer substations provided in this application embodiment; Figure 3 A flowchart illustrating the dimensionality reduction process for multi-source heterogeneous feature data provided in this application embodiment; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application; Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0028] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0029] The low-voltage distribution transformer area risk comprehensive assessment and optimization method provided in this application embodiment can be applied to, for example... Figure 1 In the low-voltage distribution transformer area risk comprehensive assessment and optimization system 100 shown, such as Figure 1 As shown, the communication system includes: a data acquisition terminal 10 and a cloud computing device 20.

[0030] Among them, the acquisition terminal 10 is used to acquire multi-source data and perform data cleaning to obtain multi-source heterogeneous feature data.

[0031] The cloud computing device 20 is used to perform dimensionality reduction on multi-source heterogeneous feature data through principal component analysis and autoencoder to obtain dimensionality reduction results; based on the dimensionality reduction results, feature vectors are defined, and the feature vectors are mapped to a unified interval through Min-Max standardization to generate a three-dimensional tensor, thus obtaining a standardized risk feature matrix; A standardized risk feature matrix is ​​input into a risk assessment model to obtain risk assessment results; wherein, the risk assessment model is trained based on an artificial intelligence model. Based on the risk assessment results, a tiered and coordinated approach is implemented, and the Pareto optimal solution set that maximizes the voltage qualification rate and minimizes the load rate is obtained by improving the PSO algorithm.

[0032] To address the technical problem that existing technologies struggle to handle complex and ever-changing operating scenarios, and that centralized optimization models cannot meet real-time requirements due to communication delays, this application provides a comprehensive risk assessment and optimization method for low-voltage distribution transformer areas. This method includes: acquiring multi-source data and performing data cleaning to obtain multi-source heterogeneous feature data. Principal component analysis and autoencoder are used to reduce the dimensionality of multi-source heterogeneous feature data to obtain the dimensionality reduction result; feature vectors are defined based on the dimensionality reduction result, and the feature vectors are mapped to a unified interval through Min-Max standardization to generate a three-dimensional tensor, thus obtaining a standardized risk feature matrix; A standardized risk feature matrix is ​​input into a risk assessment model to obtain risk assessment results; wherein, the risk assessment model is trained based on an artificial intelligence model. Based on the risk assessment results, a tiered and collaborative approach is implemented, and the Pareto optimal solution set that maximizes the voltage qualification rate and minimizes the load rate is obtained by improving the PSO algorithm; based on this, the above-mentioned technical problems can be solved.

[0033] like Figure 2 As shown in the embodiments of this application, the low-voltage distribution transformer area risk comprehensive assessment and optimization method includes: S201. Acquire multi-source data and perform data cleaning to obtain multi-source heterogeneous feature data.

[0034] The multi-source data includes electrical and non-electrical quantities; the electrical quantities include three-phase voltage, three-phase current, and current load data; the non-electrical quantities include ambient temperature and load forecast values. In some implementations, the data cleaning methods for three-phase voltage include: Using the Hamming window function The three-phase voltage is filtered using a sliding window; where... The voltage output value at the nth sampling point. The original three-phase voltage signal is sampled at time nk. These are the weighting coefficients for the Hamming window. , This is the cosine modulation term; In some implementations, the cleaning method for three-phase current and current load data is to remove outliers through 3σ metric measurement.

[0035] In some implementations, the cleaning method for non-electrical quantities is to filter them using a customized threshold method.

[0036] It should be noted that the three-phase voltage and three-phase current are both collected by smart meters; the current load data are collected by current transformers; the ambient temperature is collected by wireless temperature sensors; and the load prediction value is predicted by a load prediction model, which is trained based on an artificial intelligence model. The training process of this model can be carried out using historical sample data.

[0037] S202. Dimensionality reduction of multi-source heterogeneous feature data is performed by principal component analysis and autoencoder to obtain the dimensionality reduction result.

[0038] In some implementations, this can be achieved through the following steps: S301. Perform Z-score normalization on the multi-source heterogeneous feature data to obtain a high-dimensional feature matrix X∈R. m×27 ; Calculate the covariance matrix C; where m is the number of samples, and the covariance matrix is ​​calculated as follows: ; S302. Perform eigenvalue decomposition on the covariance matrix C to obtain 27 eigenvalues. and the corresponding vector; S303, Based on cumulative variance contribution rate The first 8 principal components are selected based on the preset conditions to form the projection matrix. ; S304. Project the high-dimensional feature matrix X onto the principal component space to obtain the PCA dimensionality reduction result. ; S305. Input the high-dimensional feature matrix X into the autoencoder network to obtain the encoder dimensionality reduction result. ; S306, through calculation formula The dimensionality reduction result is calculated, where the dimensionality reduction result is... ; The weighting coefficients are used to balance the dimensionality reduction results of principal component analysis (PCA). Dimensionality reduction results with autoencoder (AE) In the dimensionality reduction fusion results The extent of its contribution.

[0039] In some implementations, the autoencoder network includes an input layer, a first hidden layer, a second hidden layer, an encoding layer, and a decoding layer; The loss function Loss of the autoencoder network is the mean squared error; wherein, the loss function ;in, For the z-th original sample of the autoencoder, This is the z-th sample reconstructed by the decoding layer.

[0040] In this embodiment, the input layer includes 27 nodes, corresponding to the original 27-dimensional features; the first hidden layer includes 16 nodes, using the ReLU activation function to extract mid-level features; the second hidden layer includes 12 nodes, using the LeakReLU activation function; the encoding layer includes 8 nodes, consistent with the dimensionality of PCA dimensionality reduction, facilitating subsequent fusion, and is used to output non-linear dimensionality reduction features; the decoding layer is symmetrically restored to 27 nodes, using the Sigmoid activation function; the training parameters of the autoencoder network are: Adam optimizer, batch size 32, and 100 iterations.

[0041] S203. Define feature vectors based on dimensionality reduction results.

[0042] The feature vectors include voltage over-limit feature vectors, load state feature vectors, and three-phase imbalance feature vectors.

[0043] It should be noted that the voltage over-limit feature vector is: ;in, This is the measured voltage. Nominal voltage, For voltage over-limit time, Total monitoring time This is the imaginary part of the voltage. This represents the real part of the voltage. Load state feature vector: ;in, This is the load forecast value. For maximum load capacity, This is the actual load. For peak load, This represents the load trough value. Average load; Three-phase imbalance eigenvectors: ;in, It is a negative sequence current. It is a positive sequence current. It is the zero-sequence current. The power between phases A and B. This refers to the power between phases B and C. This refers to the power between phases C and A. This represents the average value of the relevant phase-to-phase power.

[0044] S204. By standardizing the feature vectors to a unified interval through Min-Max standardization, a three-dimensional tensor is generated, resulting in a standardized risk feature matrix.

[0045] In some implementations, from the dimensionality reduction results In the process, the voltage over-limit feature vector V, the load state feature vector L, and the three-phase imbalance feature vector U are extracted; and the indices of the voltage over-limit feature vector V, the load state feature vector L, and the three-phase imbalance feature vector U are mapped to [0,1]. The dimensions of the three-dimensional blank tensor are set, and the three-dimensional blank tensor is filled to obtain the three-dimensional tensor; the dimensions of the three-dimensional tensor include the time dimension, the spatial dimension, and the index dimension.

[0046] Among them, the voltage over-limit feature vector V includes voltage deviation rate v1, over-limit duration v2 and phase offset angle v3; the load state feature vector L includes predicted value / capacity threshold ratio l1, load change rate slope l2 and peak-valley difference l3; the three-phase imbalance feature vector includes negative sequence current component proportion q1, zero sequence current component proportion q2 and phase-to-phase active power difference q3.

[0047] It should be noted that filling a three-dimensional blank tensor includes: For each time slice t (t=1,2,…,24) and each key node s (s=1,2,…,8), execute: Extract the nine original feature values ​​(v1, v2, v3, l1, l2, l3, q1, q2, q3) of the node at that moment; standardize each original value according to the Min-Max formula mentioned above to obtain nine standardized values; fill the nine standardized values ​​into the corresponding positions T[t, s, j] of the three-dimensional tensor in the order of "voltage class → load class → imbalance class"; (j=1,2,…9 are index numbers), and finally generate the three-dimensional tensor T∈R. 24×8×9 .

[0048] Time dimension: 24 in length, corresponding to 24 "5-minute time slices" within a day (e.g., 00:00-00:05 to 23:55-24:00); Spatial dimension: 8 in length, corresponding to 8 key monitoring nodes in the distribution area (e.g., transformer low-voltage side outlet, feeder branch points 1-6, end-user cluster); Indicator dimension: 9 in length, corresponding to 9 standardized risk indicators.

[0049] S205. Input the standardized risk feature matrix into the risk assessment model to obtain the risk assessment result; wherein, the risk assessment model is trained based on an artificial intelligence model; In some implementations, the risk assessment model includes a dynamic weight allocation module, a risk entropy calculation module, and a risk level classification and tracing module; The dynamic weight allocation module is used to adjust the weights of the voltage over-limit feature vector, load state feature vector, and three-phase imbalance feature vector; The risk entropy calculation module is used to calculate the health index using three-dimensional tensors and dynamic weights; Through calculation formula Quantifying uncertainty; among which, Let i be the i-th standardized eigenvalue in the eigenvector. The dynamic weights of the feature vectors, The health index is a preset constant. y represents the total number of features; The risk level classification and tracing module is used to map the health index to a risk level and locate the main cause of the risk.

[0050] For example, the grading is based on preset thresholds (red: S < 0.3, orange: 0.3 ≤ S < 0.6, yellow: S ≥ 0.6). Source analysis: Calculating feature contribution ;in, The three characteristics with the highest contribution from the historical health status are the main risk factors.

[0051] It should be noted that, in this embodiment, the risk assessment model is not only based on artificial intelligence model training, but also incorporates traditional fuzzy assessment methods, adopting a hybrid modeling approach that integrates data-driven methods and physical models.

[0052] Furthermore, the dynamic weights of the feature vectors are obtained as follows: ;in, Let be the weight of the eigenvector at time t. The eigenvector includes the voltage eigenvector, the load eigenvector, and the imbalance eigenvector. For real-time load rate, For ambient temperature, the Sigmoid function maps the variable to [0,1]. Constraints: + + , Let be the weight of the voltage eigenvector at time t. The weights of the load eigenvectors at time t. The weights of the unbalanced eigenvectors at time t; , , , , , These are the model parameters, obtained during the training of the risk assessment model.

[0053] S206. Implement graded and coordinated measures based on risk assessment results, and solve the Pareto optimal solution set that maximizes voltage qualification rate and minimizes load rate by improving the PSO algorithm.

[0054] Among these, the tiered and coordinated response based on risk assessment results includes: The coordinated processing of the energy storage system includes: using a smooth switching algorithm. The energy storage system's operating modes are switched to shutdown, constant pressure, constant power, and emergency support modes; among these... This is the new power value after the energy storage system switches over; The old power value before the energy storage system was switched over; This represents the expected power change before and after the switch. This is the time constant used to control the degree of smoothness; Set the required switching time; The time elapsed since the start of the switch; The constant voltage mode is defined as follows: when the voltage deviation... At that time, through calculation formula Adjust the output power; among which, This refers to the output power. For voltage deviation; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; The constant power mode is as follows: when the current load data is >90%, the calculation formula is used. The logic controls the discharge power; among which, The discharge power of the energy storage system. The rated power of the energy storage system. For the available capacity of the energy storage system, The duration of discharge for the energy storage system; The emergency support mode is as follows: when SOC > 30% and voltage drop exceeds 10%, the system will activate the emergency support mode through a calculation formula. To implement control; among which, This refers to the output power of the energy storage system in emergency support mode. The rated power of the energy storage system. It is a time variable. This is the time constant used to control the decay characteristics; Peak load reduction measures include: classifying interruptible load levels through K-means clustering, constructing a spatiotemporal dual-dimensional control strategy, and constraining and controlling air conditioning load through a predictive average vote (PMV) model.

[0055] The spatiotemporal dual-dimensional control strategy includes: time-dimensional control and spatial-dimensional control; wherein, the time-dimensional control is to shift the electricity consumption time from peak load periods to off-peak periods without reducing the total electricity consumption; the spatial-dimensional control is to adjust the power supply point or power supply path using control means.

[0056] For example, control measures include: power supply side: dynamic switching and output scheduling of power supply points; grid side: topology optimization and power flow control of power supply paths; load side: coordinated transfer of spatial loads; and energy storage side: coordinated support of energy storage in the spatial dimension.

[0057] Among them, on the power supply side: the core of dynamic switching and output scheduling of power supply points is to change the power supply source of the load (i.e., adjust the "power supply point"), and balance the power supply capacity and load demand of local areas by activating backup power, scheduling distributed power, or optimizing the power distribution of the main grid. On the grid side: The core of topology optimization and power flow control of power supply paths is to optimize the physical path of power transmission (i.e., adjust the "power supply path"). By changing the grid topology and adjusting the power flow of the lines, we can avoid overload of a single path and increase the power supply channel capacity of local areas. On the load side: Although the coordinated transfer of spatial loads does not directly "adjust the power source or path", it indirectly reduces the dependence of local areas on specific power sources / paths by guiding the optimized distribution of loads in space, in line with the spatial dimension control objectives. On the energy storage side: Spatial-dimensional energy storage synergy support can serve as a "mobile / schedulable power source". Through spatial charging and discharging synergy, it can indirectly adjust the power supply structure of local areas and assist in spatial-dimensional regulation.

[0058] It should be noted that: the constraint control of air conditioning load through the predictive average vote (PMV) model is as follows: the current environmental PMV value is calculated in real time and an acceptable PMV range is set. With the goal of maintaining the PMV value within an acceptable comfort range, the operating parameters of the air conditioner are dynamically adjusted.

[0059] For example, based on historical load curves, K-means clustering was performed on 120 users, dividing them into three groups: high elasticity (32 households, adjustable 15-25kW), medium elasticity (65 households, adjustable 5-15kW), and low elasticity (23 households, adjustable <5kW).

[0060] Furthermore, in this embodiment, the collaborative processing method of the energy storage system also includes: constructing a cycle lifetime decay model based on the Arrhenius equation. and through the objective function Optimize the SOC; among which, This represents the maximum number of charge-discharge cycles the battery can perform at the current temperature. Reference temperature The maximum number of charge-discharge cycles for the battery at this time; This is the battery activation energy; Boltzmann's constant; The current battery temperature; For reference temperature, Let be the battery power at time t. For reference power, These are the weighting coefficients. This represents the change in battery health status.

[0061] In this embodiment, the rainflow counting method is used to estimate the battery's state of health (SOH) in real time. For example: first, 5 minutes of current data (sampling rate 10Hz) is collected; after data preprocessing to remove noise and extract extreme points, then... Rainflow counting is performed, and finally based on Calculate SOH; where, Let represent the change in charge during the i-th segment of the charging and discharging process. This refers to the battery's nominal capacity.

[0062] Furthermore, in each iteration of the improved PSO algorithm for solving the Pareto optimal solution set that maximizes voltage qualification rate and minimizes load rate, the inertia weight of each particle is dynamically adjusted according to the iteration progress. After each iteration, all solutions in the entire population are compared pairwise to classify non-dominated levels. Solutions that are not dominated by any other particle according to their non-dominated levels are included in the Pareto optimal candidate solution set. After the iteration is completed, all solutions in the Pareto optimal candidate solution set are output to obtain the Pareto optimal solution set.

[0063] The foregoing primarily describes the solutions of the embodiments of this application from the perspective of device implementation. It is understood that each device, such as an electronic device, includes at least one of the hardware structures and software modules corresponding to the execution of each function in order to achieve the aforementioned functions. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0064] This application embodiment can divide the electronic device into functional units according to the above method example. For example, each function can be divided into a separate functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this application embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0065] When using integrated units, Figure 4 A possible structural schematic diagram of the electronic device (referred to as electronic device 40) involved in the above embodiments is shown. The electronic device 40 includes a processing unit 401 and a communication unit 402, and may also include a storage unit 403. Figure 4 The structural diagram shown can be used to illustrate the structure of the electronic device involved in the above embodiments.

[0066] when Figure 4 The schematic diagram shown is used to illustrate the structure of the electronic device involved in the above embodiments. The processing unit 401 is used to control and manage the operation of the electronic device, the communication unit 402 is used for the electronic device to communicate with other devices, and the storage unit 403 is used to store the program code and data of the electronic device.

[0067] For example, communication unit 402 is used to acquire multi-source data; Processing unit 401 is used to perform data cleaning on multi-source data to obtain multi-source heterogeneous feature data; Principal component analysis and autoencoder are used to reduce the dimensionality of multi-source heterogeneous feature data to obtain the dimensionality reduction result; feature vectors are defined based on the dimensionality reduction result, and the feature vectors are mapped to a unified interval through Min-Max standardization to generate a three-dimensional tensor, thus obtaining a standardized risk feature matrix; A standardized risk feature matrix is ​​input into a risk assessment model to obtain risk assessment results; wherein, the risk assessment model is trained based on an artificial intelligence model.

[0068] In one possible implementation, the processing unit 401 is also used to perform graded collaborative handling based on the risk assessment results, and to solve the Pareto optimal solution set that maximizes the voltage compliance rate and minimizes the load rate by improving the PSO algorithm.

[0069] The processing unit 401 can be a processor or a controller, and the communication unit 402 can be a communication interface, transceiver, transceiver circuit, transceiver device, etc. The term "communication interface" is a general term and may include one or more interfaces. The storage unit 403 can be a memory. When the electronic device 40 is a chip, the processing unit 401 can be a processor or a controller, and the communication unit 402 can be an input interface and / or an output interface, pins, or circuits, etc. The storage unit 403 can be a storage unit within the chip (e.g., a register, cache, etc.) or a storage unit located outside the chip (e.g., read-only memory (ROM), random access memory (RAM, etc.).

[0070] The communication unit can also be called a transceiver unit. The antenna and control circuit with transceiver functions in the electronic device 40 can be considered as the communication unit 402 of the electronic device 40, and the processor with processing functions can be considered as the processing unit 401 of the electronic device 40. Optionally, the device in the communication unit 402 that implements the receiving function can be considered as a communication unit. The communication unit is used to execute the receiving steps in the embodiments of this application, and the communication unit can be a receiver, a receiver circuit, etc. The device in the communication unit 402 that implements the transmitting function can be considered as a transmitting unit. The transmitting unit is used to execute the transmitting steps in the embodiments of this application, and the transmitting unit can be a transmitter, a transmitter, a transmitting circuit, etc.

[0071] Figure 4 If the integrated units in the process are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, in essence, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. Storage media for storing computer software products include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.

[0072] Figure 4 The units in the process can also be called modules; for example, a processing unit can be called a processing module.

[0073] This application also provides a hardware structure diagram of an electronic device (denoted as electronic device 50), see [link to diagram]. Figure 5The electronic device 50 includes a processor 501, and optionally, a memory 502 connected to the processor 501.

[0074] In the first possible implementation, see Figure 5 The electronic device 50 also includes a transceiver 503. The processor 501, memory 502, and transceiver 503 are connected via a bus. The transceiver 503 is used to communicate with other devices or communication networks. Optionally, the transceiver 503 may include a transmitter and a receiver. The device in the transceiver 503 that implements the receiving function can be considered as a receiver, which is used to perform the receiving steps in the embodiments of this application. The device in the transceiver 503 that implements the transmitting function can be considered as a transmitter, which is used to perform the transmitting steps in the embodiments of this application.

[0075] Based on the first possible implementation method Figure 5 The structural diagram shown can be used to illustrate the structure of the electronic device involved in the above embodiments.

[0076] in, Figure 5 This can also be illustrated by a system chip in an electronic device. In this case, the actions performed by the aforementioned electronic device can be implemented by this system chip; the specific actions performed can be found above and will not be repeated here.

[0077] In implementation, each step of the method provided in this embodiment can be completed by integrated logic circuits in the processor hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or being executed by a combination of hardware and software modules in the processor.

[0078] The processor in this application may include, but is not limited to, at least one of the following: a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a microcontroller unit (MCU), or an artificial intelligence processor, etc., which are various computing devices that run software. Each computing device may include one or more cores for executing software instructions to perform calculations or processing. The processor may be a separate semiconductor chip or integrated with other circuits into a single semiconductor chip. For example, it may be integrated with other circuits (such as encoding / decoding circuits, hardware acceleration circuits, or various bus and interface circuits) to form a SoC (System-on-a-Chip), or it may be integrated as a built-in processor within an ASIC. The ASIC with the integrated processor may be packaged separately or together with other circuits. In addition to the cores for executing software instructions to perform calculations or processing, the processor may further include necessary hardware accelerators, such as field-programmable gate arrays (FPGAs), PLDs (programmable logic devices), or logic circuits that implement dedicated logic operations.

[0079] The memory in the embodiments of this application may include at least one of the following types: read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions; random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions; or electrically erasable programmable-only memory (EEPROM). In some scenarios, the memory may also be a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto.

[0080] This application also provides a computer-readable storage medium including instructions that, when run on a computer, cause the computer to perform any of the methods described above.

[0081] This application also provides a computer program product containing instructions that, when run on a computer, cause the computer to perform any of the methods described above.

[0082] This application also provides a chip including a processor and an interface circuit. The interface circuit is coupled to the processor. The processor is used to run computer programs or instructions to implement the above-described method. The interface circuit is used to communicate with other modules outside the chip.

[0083] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks (SSDs)).

[0084] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0085] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of this application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications of this application fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and modifications.

Claims

1. A low-voltage power distribution area risk comprehensive assessment optimization method, characterized in that, The application relates to a risk assessment method and device for power supply systems. The method comprises the following steps: acquiring multi-source data and performing data cleaning to obtain multi-source heterogeneous feature data; performing dimension reduction on the multi-source heterogeneous feature data through principal component analysis and an automatic encoder to obtain a dimension reduction result; defining a feature vector based on the dimension reduction result, and mapping the feature vector to a unified interval through Min-Max standardization to generate a three-dimensional tensor and obtain a standardized risk feature matrix; inputting the standardized risk feature matrix into a risk assessment model to obtain a risk assessment result; wherein the risk assessment model is obtained through artificial intelligence model training; 2. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 1, characterized in that, based on the risk assessment result, hierarchical collaborative treatment is performed, and a PSO algorithm is improved to solve a Pareto optimal solution set of maximum voltage qualification rate and minimum load rate. The multi-source data comprises electrical quantities and non-electrical quantities; wherein the electrical quantities comprise three-phase voltage, three-phase current and current load data; and the non-electrical quantities comprise environmental temperature and load prediction values; by Hamming window function sliding window filtering the three-phase voltage; wherein, is the output value of the voltage at the nth acquisition point, is the sampling value of the original three-phase voltage signal at time n-k, is the weight coefficient of the Hamming window; The data cleaning method of the three-phase voltage comprises the following steps: The cleaning method of the three-phase current and current load data is to remove abnormal values through a 3sigma criterion; 3. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 1, characterized in that, The cleaning method of the non-electrical quantities is to screen the non-electrical quantities through a customized threshold method. The multi-source heterogeneous characteristic data is subjected to Z-score standardization to obtain a high-dimensional characteristic matrix X∈R m×27 ; and a covariance matrix C is calculated; wherein m is a sampling number, and the covariance matrix is calculated in the following manner: ; Eigenvalue decomposition of the covariance matrix C yields a number of eigenvalues and corresponding vectors; According to the cumulative variance contribution rate The first 8 principal components are selected according to the preset condition to form a projection matrix ; Projecting the high-dimensional feature matrix X to the principal component space to obtain a PCA dimension reduction result ; Input the high-dimensional feature matrix X into the autoencoder network to obtain the encoder dimension reduction result ; The dimension reduction result is calculated by the calculation formula The dimension reduction result is calculated, wherein the dimension reduction result ; is the fusion weight coefficient.

4. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 3, characterized in that, The dimension reduction on the multi-source heterogeneous feature data through the principal component analysis and the automatic encoder comprises the following steps: The loss function Loss of the auto-encoder network is mean square error; wherein, the loss function ; wherein, is the zth original sample of the auto-encoder, is the zth sample reconstructed by the decoding layer.

5. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 1, characterized in that, The auto-encoder network comprises an input layer, a first hidden layer, a second hidden layer, an encoding layer and a decoding layer; The voltage excursion feature vector is: ; wherein, is the measured voltage, is the nominal voltage, is the voltage excursion time, is the total monitoring time, is the voltage imaginary part, is the voltage real part; the load status feature vector: ; wherein, is a load forecast value, is a maximum load capacity, is an actual load, is a load peak value, is a load valley value, is an average load; the three-phase unbalance feature vector: ; wherein, is a negative sequence current, is a positive sequence current, is a zero sequence current, is a power between phase A and phase B, is a power between phase B and phase C, is a power between phase C and phase A, is an average of the relevant inter-phase power.

6. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 1, characterized in that, The feature vector comprises a voltage out-of-limit feature vector, a load state feature vector and a three-phase imbalance feature vector; From the dimension reduction results the voltage out-of-limit feature vector V, the load state feature vector L and the three-phase imbalance feature vector U are extracted; and the indexes of the voltage out-of-limit feature vector V, the load state feature vector L and the three-phase imbalance feature vector U are mapped to [0, 1]; The method of mapping the feature vector to the unified interval through Min-Max standardization to generate a three-dimensional tensor comprises the following steps:

7. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 6, characterized in that, The dimensions of a three-dimensional blank tensor are set, and the three-dimensional blank tensor is filled to obtain a three-dimensional tensor; wherein the dimensions of the three-dimensional tensor comprise a time dimension, a space dimension and an index dimension. The risk assessment model comprises a dynamic weight distribution module, a risk entropy value calculation module and a risk grade division and tracing module; The dynamic weight distribution module is used to adjust the weights of the voltage out-of-limit feature vector, the load state feature vector and the three-phase imbalance feature vector; By the calculation formula quantifying the uncertainty; wherein, is the i-th normalized eigenvalue in the feature vector, is the dynamic weight of the feature vector, is a preset constant; the health index , y is the total number of features; The risk entropy value calculation module is used to calculate a health index through the three-dimensional tensor and the dynamic weights; 8. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 7, characterized in that, The dynamic weight of the feature vector is obtained in the following manner: ; wherein, is the weight of the feature vector at time t, the feature vector including a voltage feature vector, a load feature vector, and an unbalance feature vector, is the real-time load rate, is the ambient temperature, the Sigmoid function mapping the variable to [0, 1], the constraint condition: + + , is the weight of the voltage feature vector at time t, is the weight of the load feature vector at time t, is the weight of the unbalance feature vector at time t; , , , , , are model parameters and obtained when the risk assessment model is trained.

9. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 1, characterized in that, The risk grade division and tracing module is used to map the health index to a risk grade and locate the main cause of the risk. The coordinated processing of the energy storage system includes: using a smooth switching algorithm. The energy storage system's operating modes are switched to shutdown, constant pressure, constant power, and emergency support modes; among these... This is the new power value after the energy storage system switches over; The old power value before the energy storage system was switched over; This represents the expected power change before and after the switch. This is the time constant used to control the degree of smoothness; Set the required switching time; The time elapsed since the start of the switch; The constant voltage mode is: when the voltage deviation is adjusted by the calculation formula output power, wherein, is the output power; is the voltage deviation; is the proportional coefficient; is the integral coefficient; is the differential coefficient; The constant power mode is: when the current load data > 90%, the discharge power is controlled by the calculation formula ; wherein, is the discharge power of the energy storage system, is the rated power of the energy storage system, is the available capacity of the energy storage system, is the discharge duration of the energy storage system; The emergency support mode is: when SOC > 30% and voltage drop is greater than 10%, the output power of the energy storage system in the emergency support mode is calculated by the formula is controlled; wherein, is the output power of the energy storage system in the emergency support mode, is the rated power of the energy storage system, is a time variable, is a time constant for controlling the attenuation characteristic; The hierarchical collaborative treatment based on the risk assessment result comprises the following steps: peak load reduction processing, comprising: dividing interruptible loads into grades through K-means clustering, constructing a time-space two-dimensional regulation strategy, and performing constraint control on air conditioner loads through a predicted average voting model; The time-space two-dimensional regulation strategy comprises time dimension regulation and space dimension regulation; wherein the time dimension regulation is to transfer the power consumption time from a load peak period to a valley period without reducing the total power consumption; and the space dimension regulation is to adjust the power supply source point or the power supply path through control means.

10. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 9, characterized in that, The cooperative processing mode of the energy storage system further comprises: constructing a cycle life attenuation model based on an Arrhenius equation , and optimizing the SOC through a target function , wherein, is the maximum charge-discharge cycle number of the battery at the current temperature; is the maximum charge-discharge cycle number of the battery at the reference temperature; is the maximum charge-discharge cycle number of the battery at the reference temperature; is the activation energy of the battery; is the Boltzmann constant; is the current battery temperature; is the reference temperature, is the power of the battery at t, is the reference power, is the weight coefficient, is the change amount of the battery state of health.

11. The low-voltage distribution area risk comprehensive assessment optimization method according to claim 1, characterized in that, In each iteration of the improved PSO algorithm for solving the Pareto optimal solution set of the maximum voltage qualification rate and the minimum load rate, the inertia weight of each particle is dynamically adjusted according to the iteration progress; after each iteration, all solutions in the entire population are compared two by two, and the non-dominated level is divided; the solution that is not dominated by any other particle according to the non-dominated level is included in the Pareto optimal candidate solution set; after the iteration is completed, all solutions in the Pareto optimal candidate solution set are output, and the Pareto optimal solution set is obtained.

12. An electronic device, comprising: The device comprises a communication unit and a processing unit; The communication unit is configured to acquire multi-source data; the multi-source data comprises electrical quantities and non-electrical quantities; the electrical quantities comprise three-phase voltage, three-phase current and current load data; the non-electrical quantities comprise environmental temperature and load prediction value; The processing unit is configured to perform data cleaning on the multi-source data to obtain multi-source heterogeneous feature data; The multi-source heterogeneous feature data is reduced in dimension through principal component analysis and an automatic encoder to obtain a dimension reduction result; a feature vector is defined based on the dimension reduction result, and the feature vector is mapped to a unified interval through Min-Max standardization to generate a three-dimensional tensor, thereby obtaining a standardized risk feature matrix; The standardized risk feature matrix is input into a risk assessment model to obtain a risk assessment result; the risk assessment model is obtained based on artificial intelligence model training; Based on the risk assessment result, hierarchical collaborative treatment is performed, and the Pareto optimal solution set of the maximum voltage qualification rate and the minimum load rate is solved through the improved PSO algorithm.

Images