Method and system for evaluating power load regulation potential based on main incoming line

Abstract

The invention provides a power load regulation potential assessment method and system based on a total incoming line, and relates to the field of power data analysis, and the method comprises the steps: obtaining the total incoming line load data of a target power distribution system, and extracting feature factors; a causal directed graph with feature factors as nodes is constructed, the causal relationship between the nodes is determined through an improved PC algorithm, and the improvement comprises time delay expansion, chi-square statistic correction, time sequence constraint and dynamic adjacency optimization; performing anti-fact reasoning based on a causal diagram to obtain a predicted value of the total load; according to the multi-dimensional adjustment potential index, calculating the load adjustment potential under virtual intervention through the predicted value of the total load; according to the method, non-intrusive, explainable and dynamic evaluation of the user load flexibility adjustment potential is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electric power data analysis, in particular to a method and system for evaluating the load regulation potential of electric power based on total incoming lines. BACKGROUND

[0002] The interaction of "source, network, load and storage" has become an important issue in the management of power systems. With the increasing challenges of grid dispatching, fully evaluating and utilizing the rich flexible resources on the load side has become an important means to alleviate the pressure of the whole grid operation and achieve safe and stable dispatching.

[0003] Current flexible resource evaluation techniques on the load side can be roughly classified into: (1) Methods based on load curve decomposition, including various clustering algorithms, which extract the shape features of the load curve and establish a statistical correlation model between load adjustability and feature values. The main limitation of this method is that it can capture the correlation of load data, but it cannot identify the true causal effect and lacks the ability to identify implicit regulation capacity; (2) Decomposition methods based on physical modeling, which build a physical characteristic model library of electrical equipment (such as air conditioner COP curve, lighting dimming characteristics), perform device decomposition and potential calculation. The main limitation is the reliance on high-precision monitoring equipment (requiring a sampling rate of 1 MHz or higher), high deployment cost, and low generalizability of the model, making it difficult to adapt to new power electronic devices; (3) Aggregated evaluation methods based on game optimization, which establish user response behavior models and quantify aggregated potential through Stackelberg game, auction mechanism, etc. The main limitation of this method is that it ignores the physical constraints of individual devices, resulting in an overestimation of potential evaluation results, and the behavior change process of users in the model is approximated as a black box, making it difficult to ensure regulation reliability.

[0004] In summary, the current evaluation techniques for electric power load regulation potential mainly have three bottlenecks, namely causal confusion (such as misjudging the temporal correlation between price signals and load changes as causal correlation, leading to the failure of demand response strategies), lack of evaluation ability for implicit regulation (such as insufficient evaluation of the thermal inertia energy storage potential of temperature-controlled loads), and the separation of data and knowledge (such as the lack of analysis framework for the fusion of multi-source knowledge such as user behavior patterns and environmental parameters). SUMMARY

[0005] To solve the above problems, the present application proposes a method and system for evaluating the load regulation potential of electric power based on total incoming lines, which realizes non-intrusive, interpretable and dynamic evaluation of the flexible regulation potential of user loads.

[0006] According to some embodiments, the present application adopts the following technical solutions: An evaluation method of power load regulation potential based on total incoming line, comprising: Obtaining total incoming line load data of a target power distribution system, and extracting characteristic factors; Building a causal directed graph with the characteristic factors as nodes, determining the causal relationship between the nodes through an improved PC algorithm, and the improvement includes time delay extension, modified chi-square statistic, time sequence constraint and dynamic adjacency optimization; Based on the causal graph, counterfactual reasoning is performed to obtain the predicted value of the total load; According to the multi-dimensional regulation potential index, the load regulation potential under virtual intervention is calculated based on the predicted value of the total load.

[0007] Further, the total incoming line load data is the three-phase voltage and current signals synchronously sampled by the wideband acquisition device deployed at the total incoming line of the power distribution system.

[0008] Further, the extraction of characteristic factors is specifically: Convert the three-phase physical quantity into components in the stationary orthogonal coordinate system through Clarke transformation; Calculate the voltage amplitude characteristic, current imbalance characteristic, current change rate characteristic and harmonic distortion rate to form the characteristic factors; The characteristic factors are denoised and abnormal value corrected to obtain multi-dimensional time sequence characteristics.

[0009] Further, the time delay extension is to introduce the maximum time delay to the multi-dimensional time sequence characteristics to construct an extension matrix. The modified chi-square statistic is modified based on the parent node set. The time sequence constraint is a condition that the causal relationship must comply with, which ensures that the causal direction of the power load meets the time sequence priority.

[0010] Further, the dynamic adjacency optimization has the following objective function:

[0011] Wherein, Z represents the candidate condition set; The adjacency set of The adjacency set of The sparsity penalty coefficient is represented by The maximum condition set size is represented by

[0012] Further, the multi-dimensional regulation potential index includes the adjustable range, response speed and fluctuation range of adjustable power.

[0013] According to some embodiments, the present application adopts the following technical solutions: An evaluation system of power load regulation potential based on total incoming line, comprising: The feature extraction module is configured to: acquire total incoming line load data of a target power distribution system, and extract characteristic factors; The graph construction module is configured to: construct a causal directed graph with the characteristic factors as nodes, and determine causal relationships between the nodes by using an improved PC algorithm, wherein the improvement includes time delay extension, modified chi-square statistics, time sequence constraint, and dynamic adjacency optimization; The load prediction module is configured to: perform counterfactual reasoning based on the causal graph to obtain a predicted value of the total load; The potential calculation module is configured to: calculate a load regulation potential under virtual intervention according to a multi-dimensional regulation potential index and the predicted value of the total load.

[0014] According to some embodiments, the present application adopts the following technical solutions: A computer program product comprises a computer program, which, when executed by a processor, implements the following steps: Acquire total incoming line load data of a target power distribution system, and extract characteristic factors; Construct a causal directed graph with the characteristic factors as nodes, and determine causal relationships between the nodes by using an improved PC algorithm, wherein the improvement includes time delay extension, modified chi-square statistics, time sequence constraint, and dynamic adjacency optimization; Perform counterfactual reasoning based on the causal graph to obtain a predicted value of the total load; Calculate a load regulation potential under virtual intervention according to a multi-dimensional regulation potential index and the predicted value of the total load. The present application realizes an evaluation method for power load regulation potential based on total incoming lines.

[0015] According to some embodiments, the present application adopts the following technical solutions: A non-transitory computer readable storage medium is used to store computer instructions, which, when executed by a processor, implement the following steps: Acquire total incoming line load data of a target power distribution system, and extract characteristic factors; Construct a causal directed graph with the characteristic factors as nodes, and determine causal relationships between the nodes by using an improved PC algorithm, wherein the improvement includes time delay extension, modified chi-square statistics, time sequence constraint, and dynamic adjacency optimization; Perform counterfactual reasoning based on the causal graph to obtain a predicted value of the total load; Calculate a load regulation potential under virtual intervention according to a multi-dimensional regulation potential index and the predicted value of the total load. The non-transitory computer readable storage medium is used to store computer instructions, which, when executed by a processor, implement an evaluation method for power load regulation potential based on total incoming lines.

[0016] According to some embodiments, the present application adopts the technical solutions as follows: An electronic device comprises a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements the following steps when executing the program: the processor, the memory, and the computer program; wherein the processor is connected with the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the evaluation method of the total line-based power load regulation potential.

[0017] Obtain the total line load data of the target power distribution system, and extract characteristic factors; Construct a causal directed graph with the characteristic factors as nodes, and determine the causal relationship between the nodes through an improved PC algorithm, wherein the improvement includes time delay extension, correction of chi-square statistics, time sequence constraint, and dynamic adjacency optimization; Based on the causal graph, perform counterfactual reasoning to obtain the predicted value of the total load; According to the multi-dimensional regulation potential index, calculate the load regulation potential under virtual intervention through the predicted value of the total load.

[0018] Compared with the prior art, the present application has the following beneficial effects: The present application proposes a total line characteristic analysis power load regulation potential evaluation method based on causal graph reasoning, extracts characteristic factors from non-intrusive total line load data, constructs the characteristic factors of the load into a causal directed graph, and through the improved PC algorithm of the present application, can eliminate false causal relationships, so as to accurately grasp the interaction relationship between the characteristic factors; based on the causal graph, using counterfactual reasoning prediction, the load adjustable potential and response speed can be accurately evaluated, and the present application combined with demand response can solve the problem of load side resource utilization under a given dispatching instruction. BRIEF DESCRIPTION OF DRAWINGS

[0019] The drawings accompanying the specification of the present application form part of the present application and serve to provide further understanding of the present application, and the illustrative embodiments of the present application and their descriptions serve to explain the present application, and do not constitute an improper limitation on the present application.

[0020] Figure 1 Flow chart of the evaluation method of the total line-based power load regulation potential of embodiment 1. DETAILED DESCRIPTION The present application will be further described below in combination with the drawings and embodiments.

[0021] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0022] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0023] Example 1 One embodiment of the present invention provides a method for assessing the power load regulation potential based on the main incoming line. Through equipment-level causal mechanism analysis evolution, multi-modal data collaboration, and online dynamic simulation, it achieves a non-intrusive, interpretable, and dynamic assessment of the user load flexibility regulation potential. Figure 1 As shown, it includes: Step S1: Obtain the total incoming load data of the target power distribution system and extract characteristic factors.

[0024] To perform non-intrusive data acquisition, a broadband data acquisition device should be deployed at the main incoming line of the low-voltage power distribution system (such as the 380V bus). The specific data sampling settings are as follows: Sampling rate: This frequency was set to cover the characteristic frequency of the main load. Sampling accuracy: 16-bit ADC, range ±600V (voltage), ±200A (current). Synchronous sampling: Voltage and current signals are strictly synchronized (delay error <10μs).

[0025] After completing the sampling deployment, the instantaneous values ​​of the three-phase voltage and the three-phase current are monitored. These two physical quantities are expressed by the formula: (1) in, These are the instantaneous values ​​of the three-phase voltages. , , Let A, B, and C represent the instantaneous values ​​of the three-phase voltages at time t.

[0026] (2) in, These are the instantaneous values ​​of the three-phase currents. , , ABC three-phase current at t time respectively represents the instantaneous value.

[0027] Because three-phase imbalance will bring noise interference, it is necessary to convert three-phase physical quantity into the component in the stationary orthogonal coordinate system, for this, Clarke transformation is carried out, and the orthogonal voltage and orthogonal current are obtained, which are expressed by formula as follows: (3) Among them, and represent the orthogonal voltage component in the stationary coordinate system; and represent the orthogonal current component in the stationary coordinate system.

[0028] The orthogonal voltage and orthogonal current after Clarke transformation are sampled, and the following characteristics are obtained at each sampling point: (4) (5) (6) (7) (8) Among them, the multi-dimensional characteristics of the load include four characteristic factors, i.e. four components, the first component of represents the voltage amplitude characteristic, reflecting the system voltage level; the second component characterizes the current imbalance characteristic; the third component characterizes the current rate of change characteristic, which can be used to capture the load switching transient process, represents the time increment, which can be set as , the sampling rate; the fourth component represents the harmonic distortion rate, which is used to identify nonlinear loads such as frequency converters and LED lighting, represents the voltage peak value in the hth harmonic frequency band.

[0029] Set the time window , the step , then the feature matrix output by each window is: (9) Because the obtained feature matrix may contain high-frequency noise, the improved LMS filter is used to reduce the noise of the feature matrix, and the improvement here is to introduce the convergence factor, which is expressed by formula as follows: (10) wherein, denotes the feature matrix after noise reduction; is a convergence factor; denotes the error signal; denotes the reference noise input.

[0030] Then, for the feature matrix that may exist, Tukey correction is performed on each feature factor, and the correction value is: (11) wherein, and are the first and third quartiles (i.e., 25% and 75%).

[0031] Finally, the multi-dimensional time-series feature matrix after noise reduction and outlier correction is obtained, wherein, is the tth time period.

[0032] Step S2: A causal directed graph with feature factors as nodes is constructed, and the causal relationship between nodes is determined by an improved PC algorithm, and the improvement includes time delay extension, correction of chi-square statistics, time series constraint and dynamic adjacency optimization.

[0033] In order to avoid misjudgment caused by ignoring the causal relationship in the traditional method, the causal contribution of each feature factor is identified from the total incoming line mixed signal, and the adjustable potential is quantified. Based on the electrical coupling relationship and power topology relationship of the target power distribution system, an initial causal directed graph is constructed, and an improved Peter-Clarke algorithm is used to simplify the initial causal directed graph, that is, to exclude false causal edges, to obtain the final causal directed graph.

[0034] Specifically, the obtained multi-dimensional time-series feature matrix is taken as input, and the initial causal directed graph at the total incoming line is simplified, wherein, is the end point of the directed graph, which refers to the feature (i.e., the voltage amplitude feature, the current imbalance feature, the current rate of change feature, and the harmonic distortion rate); is the edge of the directed graph, which refers to the causal relationship between nodes; in order to exclude false causal edges and generate a low-dimensional causal directed graph, an improved Peter-Clarke algorithm is used, and the specific steps are as follows: Step (1) The multi-dimensional time-series feature matrix is introduced into the maximum time delay, and the following extended matrix is constructed: (12) wherein, denotes the maximum time delay multiple; denotes a single time increment, and the value is still .

[0035] Step (2) for the expansion matrix, construct the following chi-square statistics for testing the relevance of the characteristic factors: (13) In the formula, denotes the observed frequency of the i-th row and j-th column element in , that is, the observed frequency between the characteristics and time points; denotes the expected frequency of the i-th row and j-th column element in .

[0036] The above chi-square distribution is improved as follows: (14) In the formula, denotes a coefficient for adjusting the complexity, and its value should be such that the first degree of freedom of the chi-square distribution is optimal, denotes the parent node set of the node in the causal graph, and in the causal graph, if there is a directed edge , then is called the parent node of , and the definition of is: (15) where, denotes the directed edge in the causal graph.

[0037] Since in physical laws, cause must precede effect, the causal direction of the power load needs to satisfy the time sequence priority, that is, for any edge, the points at both ends have a causal relationship, and need to comply with: there is k greater than 0, so that the time period t-k The condition for obtaining in the t time period can be expressed in a formula as: (16) where k refers to the sampling period, and in the causal graph generated by the PC algorithm, all edges that do not satisfy formula (16) are removed.

[0038] Since the traditional PC algorithm may have problems such as fixed condition set size and dimension disaster, in order to reduce the calculation difficulty, the optimal key causal relationship is found from various causal relationships, and the dynamic adjacency optimization thereof is performed in this embodiment. The problem is to determine the optimal condition set, and the objective function to be satisfied is represented as: (17) In the formula, Z represents the candidate condition set. denotes the set of neighbors of ; denotes the sparsity penalty coefficient; denotes the maximum conditional set size, determined by (18) where L is the maximum time-lag multiplier in equation (12).

[0039] The specific steps of dynamic neighbor optimization are as follows: (1) Initialization: starting from an empty conditional set ; (2) Perform the following greedy selection strategy process: traverse all candidate variables , where denotes set subtraction; select the feature element that maximizes the decrease of the objective function (equation (17)) to join Z: (19) (3) Terminate when the objective function no longer decreases or .

[0040] For ease of understanding, an embodiment is provided below for illustration, and the scenario is to analyze the causal relationship between air conditioner frequency and voltage sag: Under the traditional method, all third-order combinations need to be tested, such as the conditional set ; The method of the embodiment is to simplify the conditional set to two orders, i.e. , through dynamic neighbor optimization, and only the causal relationship between and , i.e., the voltage amplitude time-lag term and the current change rate in the two sampling periods, need to be analyzed.

[0041] After the above steps, all false causal edges are excluded, and the final causal graph about the load causal relationship is obtained.

[0042] Step S3: Based on the causal graph, perform counterfactual reasoning to obtain the predicted value of the total load.

[0043] After constructing the causal graph of the load based on the total incoming line data, counterfactual reasoning is performed through the causal graph. For example, the load is determined by the product of voltage and current, and the influence of current and voltage on the load will affect the load. This influence is not simply that one item affects another item, but it is possible that one item and another item jointly affect a third item, such as voltage sag and harmonic distortion jointly affecting the current change rate.

[0044] Counterfactual reasoning aims to answer the question “If a specific control is applied to a device or feature, how will the total load of the system change?” For example, how will reducing the power of an air conditioner affect the frequency of the system.

[0045] Let the actual observed variable be (i.e. the specific observed value of the characteristic variable given by the part before), the total load be , and the mathematical form of the counterfactual problem be: (20) In the above formula, represents the assumption that is changed to ; represents how much the total load will change under the condition that .

[0046] Now assume that a total incoming line load data set has been obtained, and a semi-parametric model is constructed as follows: (21) In the above formula, represents the intervention variable; represents the causal effect of the intervention variable; represents the confounding variable, i.e. the influence of other factors on the explained variable; represents the nonlinear influence of the confounding variable; represents the residual error.

[0047] The data set is divided into and , and is trained (i.e. the is estimated by machine learning), and the nuisance function is used to estimate as follows: (22) In the above formula, is the conditional expectation, is used to estimate the direct influence of X on Y; is used to estimate the mixed effect of on Y.

[0048] The residual error is estimated on the data set , specifically: (23) The final causal effect can be estimated according to the following formula: (24) In the above formula, represents the covariance; represents the variance.

[0049] After obtaining the above items, the result of the counterfactual prediction is: (25) the observed value of the total load; the observed value of the characteristic; the observed value of the total load; the adjustment target value of the hypothetical intervention variable.

[0050] Step S4: Calculate the load adjustment potential under virtual intervention by the predicted value of the total load according to the multi-dimensional adjustment potential index.

[0051] After the above work is completed, the question of "when adjusting a characteristic quantity that affects the system load level, how will the total load level of the system change" can be answered. In order to quantify the adjustment potential on the basis of counterfactual prediction, the following multi-dimensional adjustment potential index is constructed: (1) Adjustable power range The adjustable range represents the maximum adjustable range of power under a given intervention action, specifically (26) (2) Response speed After adding the time dimension, the response speed of the load can be given by the following formula (27) (3) Adjustable power fluctuation range Since the adjustable range gives a fixed scalar, but the result has uncertainty when counterfactual prediction is performed, the uncertainty range of the result needs to be explained. For this purpose, Bootstrap is used to give the fluctuation range corresponding to the adjustable range.

[0052] First, resample B times (B ), and the mean value is (28) wherein represents the adjustable range value calculated by the bth resampling. Sort according to the order from small to large, and given the confidence level , the confidence interval is (29) wherein and respectively represent the quantile of and in B resamplings.

[0053] To facilitate understanding of the practical application of the present embodiment, a specific example is provided as follows: (1) Scenario setting: An industrial park power distribution system needs to evaluate the regulation potential of the frequency converter driven water pump unit, that is, the following target equipment is taken as a whole to evaluate the power load regulation potential for participating in grid demand response, and is configured as: 1) Monitoring point: 380V bus total incoming line 2) Target equipment: 3 sets of 22kW water pump units (equipped with frequency converters) 3) Data acquisition: 4) Sampling rate: 10kHz 5) Collection duration: 24 hours (including start-stop, speed regulation and other working conditions) (2) Causal diagram construction results: Frequency converter command → carrier frequency (k=1) Carrier frequency → current THD (k=2) Current THD → voltage fluctuation (k=3).

[0054] (3) Counterfactual prediction: Intervention setting: The original value of the carrier frequency is 5kHz, and now the dispatching instruction needs to be adjusted to 4kHz.

[0055] Counterfactual prediction results:

[0056] (4) Actual verification: The actual adjusted power is 63.5kW, and the relative error is 1.1%.

[0057] Example 2 In an embodiment of the present application, an evaluation system for power load regulation potential based on total incoming line is provided, comprising: A feature extraction module configured to obtain total incoming line load data of a target power distribution system and extract characteristic factors; A graph construction module configured to construct a causal directed graph with characteristic factors as nodes, and determine the causal relationship between nodes through an improved PC algorithm, the improvement including time delay extension, modified chi-square statistic, time series constraint and dynamic adjacency optimization; A load prediction module configured to perform counterfactual reasoning based on the causal graph to obtain a predicted value of the total load; A potential calculation module configured to calculate the load regulation potential under virtual intervention based on the predicted value of the total load according to multi-dimensional regulation potential indicators.

[0058] Example 3 An embodiment of the present application provides a computer program product comprising a computer program which, when executed by a processor, implements the following steps: Total incoming line load data of a target power distribution system are acquired, and characteristic factors are extracted; A causal directed graph with the characteristic factors as nodes is constructed, and a causal relationship between the nodes is determined through an improved PC algorithm, the improvement including time delay extension, correction of chi-square statistics, time sequence constraint and dynamic adjacency optimization; Based on the causal graph, counterfactual reasoning is performed to obtain a predicted value of the total load; Based on the predicted value of the total load, a load regulation potential under virtual intervention is calculated according to a multi-dimensional regulation potential index, and the evaluation method of the power load regulation potential based on the total incoming line is realized.

[0059] Embodiment 4 An embodiment of the present application provides a non-transitory computer readable storage medium for storing computer instructions, the computer instructions being executed by a processor to implement the following steps: Total incoming line load data of a target power distribution system are acquired, and characteristic factors are extracted; A causal directed graph with the characteristic factors as nodes is constructed, and a causal relationship between the nodes is determined through an improved PC algorithm, the improvement including time delay extension, correction of chi-square statistics, time sequence constraint and dynamic adjacency optimization; Based on the causal graph, counterfactual reasoning is performed to obtain a predicted value of the total load; Based on the predicted value of the total load, a load regulation potential under virtual intervention is calculated according to a multi-dimensional regulation potential index, and the evaluation method of the power load regulation potential based on the total incoming line is realized.

[0060] Embodiment 5 An embodiment of the present application provides an electronic device comprising a processor, a memory and a computer program; wherein the processor is connected with the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory to implement the following steps: Total incoming line load data of a target power distribution system are acquired, and characteristic factors are extracted; A causal directed graph with the characteristic factors as nodes is constructed, and a causal relationship between the nodes is determined through an improved PC algorithm, the improvement including time delay extension, correction of chi-square statistics, time sequence constraint and dynamic adjacency optimization; Based on the causal graph, counterfactual reasoning is performed to obtain a predicted value of the total load; According to the multi-dimensional adjustment potential index, the load adjustment potential under the virtual intervention is calculated through the predicted value of the total load, so as to make the electronic device execute the total line-based power load adjustment potential evaluation method.

[0061] The present application is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus generate a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 means for carrying out the functions specified in the flowchart

[0062] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be executed on the computer or other programmable data processing apparatus to produce a computer-implemented process, so that the instructions executed by the computer or other programmable data processing apparatus provide a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 steps for carrying out the functions specified in the flowchart

[0063] The above describes the specific embodiments of the present application in conjunction with the accompanying drawings, but is not a limitation on the scope of protection of the present application. Those skilled in the art should understand that various modifications or variations made by those skilled in the art on the basis of the technical solutions of the present application without creative labor are still within the scope of protection of the present application.

Claims

1. A method for evaluating total-feeder-based power load regulation potential, characterized by, The method comprises: obtaining total incoming line load data of a target power distribution system and extracting characteristic factors; building a causal directed graph with the characteristic factors as nodes, determining the causal relationship between the nodes by an improved PC algorithm, the improvement including time delay extension, correction of chi-square statistics, time sequence constraint and dynamic adjacency optimization; performing counterfactual reasoning based on the causal graph to obtain a predicted value of the total load; calculating the load regulation potential under virtual intervention according to a multi-dimensional regulation potential index and the predicted value of the total load.

2. The method of claim 1, wherein, The total incoming line load data is obtained by synchronously sampling three-phase voltage and current signals through a wideband acquisition device deployed at the total incoming line of the power distribution system.

3. The method of claim 1, wherein, The characteristic factors are extracted by: converting the three-phase physical quantities into components in the stationary orthogonal coordinate system through Clarke transformation; calculating the voltage amplitude characteristics, current imbalance characteristics, current change rate characteristics and harmonic distortion rate to form the characteristic factors; performing noise reduction and outlier correction on the characteristic factors to obtain multi-dimensional time sequence characteristics.

4. The method of claim 1, wherein, The time delay extension is to introduce a maximum time delay to the multi-dimensional time sequence characteristics to build an extension matrix. The correction of chi-square statistics is to correct the chi-square statistics based on the parent node set. The time sequence constraint is to set conditions that the causal relationship must comply with to ensure that the causal direction of the power load meets the time sequence priority.

5. The method of claim 1, wherein, The objective function of the dynamic adjacency optimization is: wherein Z denotes a candidate condition set; denotes an adjacency set of denotes a sparsity penalty coefficient; denotes a maximum condition set size.

6. The method of claim 1, wherein, The multi-dimensional regulation potential index includes the adjustable range, response speed and fluctuation range of the adjustable power.

7. A system for evaluating total-feeder-based power load regulation potential, comprising: The method comprises: a characteristic extraction module configured to obtain total incoming line load data of a target power distribution system and extract characteristic factors; a graph construction module configured to build a causal directed graph with the characteristic factors as nodes, determine the causal relationship between the nodes by an improved PC algorithm, and the improvement includes time delay extension, correction of chi-square statistics, time sequence constraint and dynamic adjacency optimization; a load prediction module configured to perform counterfactual reasoning based on the causal graph to obtain a predicted value of the total load; a potential calculation module configured to calculate the load regulation potential under virtual intervention according to a multi-dimensional regulation potential index and the predicted value of the total load.

8. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the evaluation method of the power load regulation potential based on the total incoming line according to any one of claims 1-6.

9. A non-transitory computer-readable storage medium, comprising: The non-transitory computer readable storage medium is used to store computer instructions, which are executed by the processor to implement the evaluation method of the power load regulation potential based on the total incoming line according to any one of claims 1-6.

10. An electronic device, comprising: The method comprises: a processor, a memory and a computer program; wherein the processor is connected with the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory to make the electronic device execute the evaluation method of the power load regulation potential based on the total incoming line according to any one of claims 1-6.

Images