Intelligent collaborative decision-making method for power distribution system

Abstract

The invention relates to the technical field of power distribution systems, in particular to an intelligent collaborative decision-making method for a power distribution system, which comprises the following steps of: S1, acquiring sampling frequency, uploading delay and data integrity identifiers of sensing nodes of each sub-region, analyzing the structural consistency of operating data, identifying a time alignment defect or a logic coupling fault, and acquiring a data integrity identifier of a sensing node of each sub-region; generating a heterogeneous drift characteristic spectrum; s2, combining historical data, a topology mechanism and a boundary rule, reconstructing control input, and generating a data consistency control input set; and S3, on the basis of the input set, the scheduling target and the system constraint, constructing and distributing a collaborative strategy to realize intelligent collaborative regulation and control. According to the method, the heterogeneous drift characteristic spectrum is constructed, the time-space consistency operation data is reconstructed, and the cooperative strategy set with the adaptive execution capability is generated, so that accurate identification, effective repair and intelligent cooperative regulation of the power distribution system under the condition of data inconsistency are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution systems, in particular to a power distribution system intelligent collaborative decision-making method. BACKGROUND

[0002] With the rapid development of new power systems, distributed power sources, energy storage devices and multi-level loads are continuously connected to the power distribution network. The traditional power distribution system operation mode mainly based on central dispatching faces the demand for intelligentization and collaboration transformation. In order to improve the adaptability of the system to the multi-source heterogeneous operation environment, more and more intelligent sensing nodes are deployed in each sub-region of the power distribution system to realize fine monitoring of the operation state and linkage control between regions, thereby assisting in building a flexible and efficient collaborative control mechanism.

[0003] However, the existing distributed sensing devices in the power distribution system have problems such as inconsistent sampling frequency, unstable data upload delay, and transmission packet loss, which leads to drift deviation of sensing data in each sub-region in the time dimension and logical structure, making it difficult to directly serve as an effective input for control strategies. Traditional strategy generation methods rely on idealized synchronous data assumptions and lack the ability to identify and repair data heterogeneous drift states, which can easily cause strategy mismatch and control failure, especially at the boundaries of multiple regions, which can easily cause coupling faults, seriously affecting the global coordination and robustness of the system. SUMMARY

[0004] The present application provides a power distribution system intelligent collaborative decision-making method, which constructs a heterogeneous drift feature map, reconstructs the spatiotemporal consistency control input, and generates a collaborative strategy set with adaptability and autonomy based on the dispatching target and system constraints, thereby realizing high-reliable intelligent collaborative control of the power distribution system in the data inconsistency scenario, and significantly improving the accuracy of strategy generation and the stability of execution.

[0005] A power distribution system intelligent collaborative decision-making method, comprising the following steps:

[0006] S1, collecting operation data from multiple sub-regions in the power distribution system, including the sampling frequency, upload delay and data integrity identification of the sensing nodes, identifying the heterogeneous drift segments with time alignment defects or logical coupling faults by analyzing the structural consistency of the operation data of each sub-region, and generating a heterogeneous drift feature map representing the data drift characteristics of each region;

[0007] S2, based on the generated heterogeneous drift feature map, combining historical redundant data, topology completion mechanism and physical boundary prior rules, dynamically generating a reconstructed consistent version of the control input of each sub-region, and forming a data consistency control input set for strategy generation;

[0008] S3, taking the data consistency control input set as a basis for strategy generation, combining the current scheduling target and system constraints, constructing a collaborative strategy set, and distributing the collaborative strategy set to each sub-area control unit to realize intelligent collaborative regulation and control of the power distribution system under data inconsistency.

[0009] Optionally, the S1 comprises:

[0010] S11, collecting operation data from multiple sub-areas in the power distribution system, the operation data including sampling frequency, upload delay and data integrity identifier of the perception node, and performing standardized processing on the operation data from different sub-areas;

[0011] S12, after completing the standardized collection of operation data, performing structural consistency analysis on the operation data of each sub-area based on the data organization structure between sub-areas, the time window alignment degree and the logical topology connectivity, identifying heterogeneous drift segments with time alignment defects or logical coupling faults, extracting the spatial distribution, drift type and influence range of the heterogeneous drift segments, and constructing a heterogeneous drift feature map reflecting the data consistency state of the sub-area.

[0012] Optionally, the S11 comprises:

[0013] S111, for each sub-area R k (k = 1, 2,..., K, K is the total number of areas) in the power distribution system, collecting the operation data output by the perception node N k,m (the mth node in the kth area) deployed in each area, including sampling frequency f k,m , upload delay Δ k,m , and data integrity identifier γ k,m (γ k,m = 1 indicates complete data within the current time window, and γ k,m = 0 indicates packet loss or incomplete upload), and constructing an original operation data set D k for each sub-area R k ;

[0014] S112, for all sub-areas R k , taking a unified observation time window T obs and a sampling reference frequency f ref as a structural alignment reference to construct a node perception state matrix S k ;

[0015] S113, performing standard processing on the operation data of each perception node to obtain a standard operation data vector, and constructing a standard operation data set D

[0016] Optionally, the S12 comprises:

[0017] S121, after completing the standardized collection of operation data, the time sequence feature extraction is performed on the uploading behavior of each sub-region R k , the time center of the uploading of the perception node and the overall missing rate are calculated, the time offset and data loss in the standard time window are analyzed, and the time center difference and alignment score between the logically adjacent region pairs (R k , R u ) are calculated to identify the region pairs that are misaligned in time and form a time type drift candidate set Ω time ;

[0018] S122, the collaborative response of the sub-regions in the logical topology is analyzed to evaluate whether there is a coupling failure phenomenon, the expected collaborative level and the actual consistency are calculated by combining the average uploading behavior of the perception nodes and the power topology coupling weight between the sub-regions, and whether the residual error exceeds the residual error judgment threshold is evaluated to identify the logically faulted region pairs and form a logical type drift candidate set Ω logic ;

[0019] S123, after the time type drift candidate set Ω time and the logical type drift candidate set Ω logic are combined, the spatial affected range, the drift type number and the adjacent fault influence number of each region are extracted to finally construct a heterogeneous drift feature graph Ψ.

[0020] Optionally, the S2 comprises:

[0021] S21, for the sub-regions with time alignment defects in the heterogeneous drift feature graph, uploading delays, missing segments and sampling misalignment problems in the operation data are identified, interpolation completion is performed by using historical redundant data, and the sampling rhythm is unified by referring to the time step of the adjacent regions to realize time consistency processing of the operation data;

[0022] S22, after the time consistency processing is completed, for the sub-regions with logical coupling faults in the heterogeneous drift feature graph, the relationship between the power flow and the control response between the sub-regions and the adjacent regions is analyzed, the missing control information is inferred, and the operation data after the time consistency processing is corrected to generate a consistent control input set with logical closure.

[0023] Optionally, the S21 comprises:

[0024] S211, for the sub-regions marked as having time alignment defects in the heterogeneous drift feature graph, the uploading delay, data missing and sampling misalignment features are extracted according to the uploading behavior matrix of each perception node, and the data segment set to be corrected is identified

[0025] S212, for the identified data segment set to be corrected Using historical redundant data, the interpolation method is used to fill in the missing points, and the continuous observation sequence of the perception node in the time window T is reconstructed;

[0026] S213, after completing the interpolation completion, the sampling frequency inconsistent perception nodes in each sub-region are remapped to the reference time step Δt ref .

[0027] Optionally, the S22 comprises:

[0028] S221, after completing the time consistency processing, based on the region pair (R k , R u ) marked as existing logical coupling fault in the heterogeneous drift feature map, the power exchange behavior and control response state in the standard observation time window are analyzed, the perception node pair with discontinuous power direction, missing response information or untransmitted boundary signal is identified, and the missing control information index set C k,u is constructed;

[0029] S222, for the missing operation data in the missing control information index set C k,u , based on the operation behavior of the same type node in the adjacent region in the same time window, the response similarity supplement method is used to generate the proxy operation data for filling the boundary logical fault;

[0030] S223, for the supplemented proxy operation data, according to the maximum or minimum operation capacity of the perception node belonging to the equipment, the load adjustment range, the physical boundary condition of the upper limit of power response, the upper and lower limit constraint correction is carried out, and finally the consistency control input set with logical closure is generated

[0031] Optionally, the S3 comprises:

[0032] S31, after obtaining the consistency control input set of each sub-region, according to the current scheduling target and system constraint condition (device capacity limit, network topology state, tie line power boundary), the collaborative strategy set is constructed;

[0033] S32, the generated collaborative strategy set is decomposed and issued according to the region granularity, according to the current operation state, boundary condition and execution ability of each sub-region, the matched sub-strategy version is automatically selected, and local autonomous execution is realized.

[0034] Optionally, the S31 comprises:

[0035] S311, after obtaining the consistency control input set of each sub-region, setting an integrated scheduling target function according to the current running scene, the target including minimum total active loss, minimum node voltage deviation and optimal load adjustable response;

[0036] S312, according to the set integrated scheduling target function, introducing system constraint conditions, including device capacity limit, network topology state (power balance) and tie-line flow boundary;

[0037] S313, outputting a cooperative strategy set based on the scheduling target function and the system constraint conditions

[0038] Optionally, the S32 comprises:

[0039] S321, according to the generated cooperative strategy set According to the sub-region number The cooperative strategy set is divided into a plurality of regional strategy sets, and each sub-strategy set corresponds to the control instruction of all perception nodes in the sub-region.

[0040] S322, after receiving the sub-strategy set, the sub-region control unit analyzes the current running state Boundary condition Γ k , available control set Ω k Select the most matched strategy sub-version in the feasible region

[0041] S323, the control unit executes the device regulation operation in the region according to the selected strategy sub-version .

[0042] The beneficial effects of the present application are:

[0043] The present application can accurately identify the data inconsistency problem caused by the uploading delay of the perception node, the sampling misposition or the logic fault in the power distribution system by constructing a heterogeneous drift feature map, realizes the systematic diagnosis of the time alignment defect and the logic coupling fault, and the multi-dimensional structure consistency analysis mechanism breaks through the limitation of the traditional judgment relying on a single data index, improves the identification accuracy and response timeliness of the system to the data drift state.

[0044] The present application realizes the spatio-temporal consistency reconstruction of the perception data by introducing the historical redundant data interpolation completion mechanism and the logic supplement method, combining the upper and lower limit constraint correction of the physical boundary condition, generates a consistency control input set with logical closure, effectively makes up the structural loss and response void in the running data, and improves the data recovery ability and running stability of the power distribution system under the background of heterogeneous drift.

[0045] The application realizes closed-loop adaptive cooperative control between strategy generation and execution by establishing a cooperative strategy generation function based on a consistent control input set, combining a set scheduling objective function and system operation constraints, outputting an optimal cooperative control strategy set for deployable regions, and supporting each sub-region to autonomously select a sub-strategy version according to execution capability, and improves the flexibility, robustness and autonomous response capability of the system as a whole. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only illustrate the application, and other drawings can also be obtained by those skilled in the art without any creative effort.

[0047] Fig. 1 The decision-making method flowchart of the embodiment of the application;

[0048] Fig. 2 The heterogeneous drift feature map construction flowchart of the embodiment of the application. DETAILED DESCRIPTION

[0049] The application will be described in detail below with reference to the drawings and specific embodiments. For some known technologies, other alternative ways can also be implemented by those skilled in the art; and the drawings are only used to more specifically describe the embodiments, and are not intended to specifically limit the application.

[0050] As shown in the drawings, Figs. 1-2 An intelligent cooperative decision-making method of a power distribution system, comprising the following steps:

[0051] S1, collecting operation data from multiple sub-regions in the power distribution system, including sampling frequency, uploading delay and data integrity identification of the perception node, identifying heterogeneous drift segments with time alignment defects or logical coupling faults by performing structural consistency analysis on the operation data of each sub-region, and generating a heterogeneous drift feature map representing the data drift characteristics of each region;

[0052] S2, based on the generated heterogeneous drift feature map, combining historical redundant data, topology completion mechanism and physical boundary priori rules, dynamically generating a reconstructed consistency version of the control input of each sub-region, forming a data consistency control input set for strategy generation;

[0053] S3, taking the data consistency control input set as the basis for strategy generation, combining the current scheduling target and system constraints, constructing a cooperative strategy set, and distributing the cooperative strategy set to each sub-region control unit, realizing intelligent cooperative regulation and control of the power distribution system under the condition of inconsistent data.

[0054] S1 comprises:

[0055] S11, collecting operation data from multiple sub-regions in the power distribution system, the operation data including sampling frequency, upload delay and data integrity identifier of the perception node, and performing standardized processing on the operation data from different sub-regions;

[0056] S12, after completing the standardized collection of operation data, performing structural consistency analysis on the operation data of each sub-region based on the data organization structure between sub-regions, time window alignment degree and logical topology connectivity, identifying heterogeneous drift segments with time alignment defects or logical coupling faults, extracting spatial distribution, drift type and influence range of the heterogeneous drift segments, and constructing a heterogeneous drift feature map reflecting the consistency state of the sub-region data.

[0057] S11 comprises:

[0058] S111, for each sub-region R k (k = 1, 2,..., K, K is the total number of regions) in the power distribution system, collecting the operation data output by the perception node N k,m (the mth node in the kth region) deployed in each region, including sampling frequency f k,m , upload delay Δ k,m , data integrity identifier γ k,m (γ k,m = 1 indicates complete data within the current time window, γ k,m = 0 indicates packet loss or incomplete upload), and constructing an original operation data set D k for each sub-region R k , denoted as:

[0059] D k = {(f k,m , Δ k,m , γ k,m ) | m = 1, 2,..., M k};

[0060] wherein M k is the number of perception nodes in the kth region;

[0061] S112, for all sub-regions R k , taking a unified observation time window T obs and a sampling reference frequency f ref as a structural alignment reference, constructing a node perception state matrix S k , denoted as:

[0062]

[0063] wherein T = Tobs ·f ref denotes the number of data points that should be in the standard time window, is the upload behavior matrix of the mth perception node in the kth sub-area at time t;

[0064] S113, standardize the operation data of each perception node to obtain a standard operation data vector, and construct a standard operation data set of the sub-area by using the standard operation data vectors of all perception nodes is represented as:

[0065]

[0066] wherein, is the standard operation data vector of the mth perception node in the kth sub-area, Δ max is the maximum allowable upload delay.

[0067] S12 includes:

[0068] S121, after completing the standardized collection of operation data, perform time sequence feature extraction on the upload behavior of the perception nodes in each sub-area R k , calculate the upload time barycenter and overall missing rate of the perception nodes, analyze the time offset and data loss in the standard time window, and calculate the time barycenter difference and alignment score between logically adjacent area pairs (R k , R u ), identify the time misaligned area pairs, and construct a time class drift candidate set Ω time , which is represented as:

[0069]

[0070] If A k,u <A thr or p k >p thr or p u >p thr , the area pair (k, u) is marked as a time alignment drift pair and added to the set Ω time ;

[0071] wherein, μ k,m is the node upload activity time barycenter, is the area upload activity time barycenter mean, p k is the area data upload missing rate, θ k,u is the time barycenter difference between the area pairs, A k,u is the area alignment score, and the closer to 1 indicates the more synchronous, A thr is the threshold value of the time alignment judgment, and p thrThreshold for data loss judgment

[0072] The threshold for time alignment judgment is determined according to the maximum time offset step θ allowed by the system max The setting is represented as:

[0073]

[0074] The threshold for data loss judgment is represented as:

[0075] p thr = 1-α cover ;

[0076] Wherein, α cover is the minimum perception coverage requirement in the area;

[0077] S122, analyze the cooperative response of sub-regions in the logical topology, evaluate whether there is coupling failure phenomenon, calculate the expected cooperation level and actual consistency through the average upload behavior of the perception node, combine the power topology coupling weight between sub-regions, and evaluate whether the residual error exceeds the residual error judgment threshold, identify the logical fault region pair, and constitute the logical class drift candidate set Ω logic , represented as:

[0078]

[0079]

[0080] ρ k,u = λ k,u -β k,u ;

[0081] If g k,u = 1 and ρ k,u > ρ thr , the region pair (k, u) is recorded as a logical coupling fault, and added to the set Ω logic ;

[0082] Wherein, g k,u is the topological connectivity flag, w k,u is the coupling strength between the region pair, λ k,u is the expected cooperation degree, is the average upload behavior of the region R k at time t, β k,u is the actual cooperation consistency, ρ k,u is the cooperation residual error of the region pair, ρ thr is the residual error judgment threshold;

[0083] The residual error judgment threshold ρ thr is represented as:

[0084] ρ thr= μ ρ + δ ρ ;

[0085] wherein μ ρ is the residual mean of the normal region pair in the whole network, δ ρ is the residual safety buffer coefficient;

[0086] S123, after merging the identified time class drift candidate set Ω time and the logical class drift candidate set Ω logic , the spatial affected range, the number of drift types and the number of adjacent fault affected of each region are extracted, and finally the heterogeneous drift feature atlas Ψ is constructed, which is represented as:

[0087]

[0088] η k = ∑ u g k,u ·1((k,u)∈Ω time ∪Ω logic );

[0089] Ψ k,u,1 = 1-A k,u , Ψ k,u,2 = max(ρ k,u ,0), Ψ k,0,3 = σ k , Ψ k,0,4 = η k ;

[0090] wherein, is the abnormal time segment set of the region R k , σ k is the failure node proportion in the region, η k is the connection number affected by the adjacent drift of the region, Ψ k,u,1 is the time alignment defect intensity of the region pair (k, u) in the heterogeneous drift atlas, Ψ k,u,2 is the topological logical coupling residual intensity of the region pair (k, u) in the heterogeneous drift atlas, Ψ k,0,3 is the spatial drift coverage proportion inside the region R k , and Ψ k,0,4 is the adjacent drift affected count of the region R k .

[0091] S2 includes:

[0092] S21, for the sub-region with time alignment defect in the heterogeneous drift feature atlas, the upload delay, missing segment and sampling misalignment problem in the running data are identified, the interpolation completion is performed by using the historical redundant data, and the sampling rhythm is unified by referring to the time step of the adjacent region, so that the time consistency processing of the running data is realized.

[0093] S22, after completing the time consistency processing, for the sub-regions in the heterogeneous drift feature map that exist logical coupling faults, analyze the relationship between power flow and control response between adjacent regions, infer the missing control information, and combine the physical boundary conditions (load capacity and equipment limitations) to modify the running data after time consistency processing, and generate a consistent control input set with logical closure.

[0094] S21 includes:

[0095] S211, for the sub-regions in the heterogeneous drift feature map marked as having time alignment defects, according to the upload behavior matrix of each perception node, extract upload delay, data loss and sampling misplacement features, and identify the data segment set to be corrected Specifically includes:

[0096] (1) Node average upload delay:

[0097] Wherein, is the actual upload time step of the node at time t, T is the total step length of the time window, is the node average upload delay;

[0098] (2) Node upload loss rate:

[0099] Wherein, q k,m is the node upload loss rate;

[0100] (3) Determine the sampling misplacement: Wherein, is a Boolean flag indicating whether the node has sampling frequency misplacement, is the actual effective sampling frequency of the node, f ref is the set reference sampling frequency, if the sampling frequency deviation exceeds ε f , then True, otherwise False, ε f is the allowable deviation threshold of the sampling frequency, if or q k,m > q thr or , mark the perception node N k,m has time alignment defects, record its time period, and add it to the data segment set to be corrected μ thr is the threshold of the average upload delay, q thr is the threshold of the upload loss rate;

[0101] The allowable deviation threshold of the sampling frequency is expressed as:

[0102] εf = δ f · f ref ;

[0103] wherein δ f is a frequency tolerance coefficient;

[0104] The threshold of the average upload delay is represented as:

[0105]

[0106] wherein Δt max is the maximum delay time allowed, Δt ref is the reference sampling period;

[0107] The threshold of the upload missing rate is represented as:

[0108] q thr = 1 - α cov ;

[0109] wherein α cov is the minimum expected perception data coverage rate;

[0110] S212, for the identified data segment set to be corrected using historical redundant data, using an interpolation method to fill in the missing points, reconstructing the continuous observation sequence of the perception node within the time window T, specifically including:

[0111] (1) Linear interpolation (for missing points ):

[0112]

[0113] wherein x is the interpolation value, x and x are the original running data of the node at time t1 and t2 respectively, t1 and t2 are the effective time points before and after interpolation reference respectively;

[0114] (2) Update the node data sequence after interpolation:

[0115]

[0116] S213, after completing the interpolation completion, the perception nodes with inconsistent sampling frequencies in each sub-region are remapped to the reference time step Δt ref , represented as:

[0117]

[0118] wherein x is the resampled time-aligned running data sequence, T' is the total number of resampled time steps.

[0119] S22 comprises:

[0120] S221, after completing the time consistency processing, based on the region pair (R k ,R u ) marked as existing logical coupling fault in the heterogeneous drift feature map, analyzing its power exchange behavior and control response state in the standard observation time window, identifying the perceptual node pair with power direction discontinuity, response information loss or boundary signal non-transmission, constructing the missing control information index set C k,u , expressed as:

[0121]

[0122] If , the boundary node pair (i,j) in the region pair is added to the missing control information index set C k,u , where, is the power injection of region R k to R u , is the power feedback of region R u to R k , is the power balance residual, is the average power inconsistency degree in the time window, is the power residual judgment threshold;

[0123] The power residual judgment threshold is expressed as:

[0124]

[0125] Where, μ ∈ is the historical average power residual of the whole system, λ P is the safety factor, and σ ∈ is the historical residual standard deviation of the whole system;

[0126] S222, for the missing operation data in the missing control information index set C k,u , based on the operation behavior of the same type nodes in the adjacent region in the same time window, using the response similarity supplement method, generate proxy operation data for filling the boundary logical fault, expressed as:

[0127]

[0128] Where, is the proxy node set used for supplement, is the actual operation data of the adjacent region proxy node n, is the operation data supplement result for node i, and ω i,na weight of similarity of behavior between node i and node n;

[0129] S223, according to the maximum or minimum operating capacity of the device to which the sensing node belongs, the load adjustment range, the physical boundary condition of the upper limit of power response, the upper and lower limit constraint correction is carried out on the agent running data of the supplementary push, and the physical feasibility is ensured, and finally a consistent control input set with logical closure is generated is expressed as:

[0130]

[0131] wherein, is the final running data of node i at time t, respectively, the upper and lower limits of the running of the device corresponding to node i.

[0132] S3 includes:

[0133] S31, after obtaining the consistency control input set of each sub-region, a collaborative strategy set is constructed according to the current scheduling target and system constraint condition (device capacity limit, network topology state, tie line power boundary) ;

[0134] S32, the generated collaborative strategy set is decomposed and issued according to the region granularity, and according to the current running state, boundary condition and execution ability of each sub-region, a matched sub-strategy version is automatically selected to realize local autonomous execution.

[0135] S31 includes:

[0136] S311, after obtaining the consistency control input set of each sub-region, a comprehensive scheduling target function is set according to the current running scene, and the target includes minimum total active loss, minimum node voltage deviation and optimal load adjustable response, and the comprehensive scheduling target function is expressed as:

[0137]

[0138] wherein, is the total amount of line active loss, is the sum of node voltage offset squares, is the load response deviation, and α1, α2 and α3 are respectively weight coefficients corresponding to the system control variable set;

[0139] S312, according to the set comprehensive scheduling target function, the system constraint condition is introduced, including device capacity limit, network topology state (power balance) and tie line power flow boundary, and is expressed as:

[0140] Device capacity limit (based on injected power) :

[0141] Interconnection line power boundary (based on power flow constraint):

[0142] Network topology power flow balance relationship:

[0143] Where, P i is the active power injection of node i, are the upper and lower limits of the device capacity of node i, P i,j (u) is the power flow from node i to j, and u is the power carrying limit of the interconnection line i, j, is the set of adjacent nodes topologically connected to node i;

[0144] S313, based on the scheduling objective function and system constraint conditions, output the cooperative strategy set is expressed as:

[0145]

[0146] Where, Φ coord is the strategy generation function, Θ target is the parameter of the comprehensive scheduling objective function, Θ bound is the system constraint condition.

[0147] S32 includes:

[0148] S321, according to the generated cooperative strategy set According to the sub-area number The cooperative strategy set is divided into multiple regional sub-strategy sets, and each sub-strategy set corresponds to the control instruction of all sensing nodes in the sub-area, which is expressed as:

[0149]

[0150] Where, is the sub-strategy set of sub-area k;

[0151] S322, after receiving the sub-strategy set, the sub-area control unit analyzes the current running state Boundary conditions Γ k , available control set Ω k Select the most matched strategy sub-version in the feasible region Specifically includes:

[0152] (1) Construct the feasible region constraint set: the control unit analyzes the current state, boundary conditions and control accessibility, determines the feasible control space at the current time, which is expressed as:

[0153]

[0154] wherein, is the current feasible strategy candidate set;

[0155] (2) Calculate the strategy adaptation matching degree: evaluate the matching degree of each candidate strategy in the sub-strategy set with the current running state, select the strategy version most suitable for the control requirement, denoted as:

[0156]

[0157] wherein, η(u) is the strategy matching deviation function value, φ is the state matching deviation weight coefficient, ζ is the control execution cost weight coefficient, Cost(u) is the expected control cost of the system control variable set u, is the target running state prediction value corresponding to the system control variable set u;

[0158] (3) Select the optimal sub-strategy version: select the final execution strategy version of the current sub-region according to the minimum matching deviation criterion, denoted as:

[0159]

[0160] S323, the control unit executes the equipment regulation and control operation in the region according to the selected strategy sub-version. S323, the control unit executes the equipment regulation and control operation in the region according to the selected strategy sub-version.

[0161] The present application encompasses any alternative, modification, equivalent method and scheme made on the essence and scope of the present application. In order to make the public have a thorough understanding of the present application, specific details are described in the following preferred embodiments of the present application, and the present application can also be fully understood without the description of these details to those skilled in the art. In addition, in order to avoid unnecessary confusion to the essence of the present application, well-known methods, processes, procedures, elements and circuits, etc. are not described in detail.

[0162] The above is only the preferred embodiment of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, and these improvements and refinements should be considered as the protection scope of the present application.​

Claims

1. A power distribution system intelligent collaborative decision-making method, characterized in that, The method comprises the following steps: S1, collecting operation data from multiple sub-regions in the power distribution system, including sampling frequency, upload delay and data integrity identification of the sensing nodes, identifying heterogeneous drift segments with time alignment defects or logical coupling faults through structural consistency analysis of the operation data of each sub-region, and generating a heterogeneous drift feature map representing the data drift characteristics of each region; S2, based on the generated heterogeneous drift feature map, combining historical redundant data, topology completion mechanism and physical boundary priori rules, dynamically generating a reconstructed consistent version of the control input of each sub-region, and forming a data consistency control input set for strategy generation; S3, taking the data consistency control input set as the basis for strategy generation, combining the current scheduling target and system constraint conditions, constructing a collaborative strategy set, and distributing the collaborative strategy set to each sub-region control unit to realize intelligent collaborative regulation and control of the power distribution system under inconsistent data conditions.

2. The intelligent collaborative decision-making method of a power distribution system according to claim 1, characterized in that, The S1 comprises: S11, collecting operation data from multiple sub-regions in the power distribution system, including sampling frequency, upload delay and data integrity identification of the sensing nodes, and standardizing the operation data from different sub-regions; S12, after completing the standardized collection of operation data, based on the data organization structure, time window alignment degree and logical topology connectivity between sub-regions, performing structural consistency analysis on the operation data of each sub-region, identifying heterogeneous drift segments with time alignment defects or logical coupling faults, extracting the spatial distribution, drift type and influence range of the heterogeneous drift segments, and constructing a heterogeneous drift feature map reflecting the data consistency state of the sub-regions.

3. The intelligent collaborative decision-making method of a power distribution system according to claim 2, characterized in that, The S11 comprises: S111, for each sub-area R in the power distribution system k Data is collected from N sensing nodes deployed in various regions. k,m The output runtime data includes the sampling frequency f. k,m Upload delay Δ k,m Data integrity identifier γ k,m And for each sub-region R k The original running dataset D k ; S112, for all sub-regions R k with a uniform observation time window T obs and a sampling reference frequency f ref construct a node-aware state matrix S k as a structural alignment reference; S113, standard processing the running data of each perception node to obtain a standard running data vector, and constructing a standard running data set of the sub-region by using the standard running data vectors of all the perception nodes 4. The intelligent collaborative decision-making method of a power distribution system according to claim 3, characterized in that, The S12 comprises: S121, after completing the standardized collection of operation data, to each sub-region R k The time sequence feature extraction is performed on the uploading behavior of the perception node, the uploading time center and the overall missing rate of the perception node are calculated, the time offset and the data loss in the standard time window are analyzed, and the time center difference and the alignment score between the logically adjacent region pairs (R k , R u ) are calculated, the region pairs that are wrong in time are identified, and a time class drift candidate set Ω time is constructed; S122, analyze the collaborative response situation among the sub-regions on the logical topology, evaluate whether there is a coupling failure phenomenon, calculate the expected collaboration level and the actual consistency through the average upload behavior of the perception nodes, combine the power topology coupling weight between the sub-regions, evaluate whether the residual exceeds the residual judgment threshold, identify the logical fault region pair, and constitute the logical class drift candidate set Ω logic ; S123, a set of identified time class drift candidates Ω time and a set of logical class drift candidates Ω logic After merging, the spatial affected range, the number of drift types, and the number of adjacent fault affected of each region are extracted, and finally the heterogeneous drift feature atlas Ψ is constructed.

5. The intelligent collaborative decision-making method of a power distribution system according to claim 1, characterized in that, The S2 comprises: S21, for sub-regions with time alignment defects in the heterogeneous drift feature map, identifying upload delay, missing segments and sampling misplacement problems in the operation data, interpolating and completing using historical redundant data, and referring to the time step of adjacent regions to unify the sampling rhythm, realizing time consistency processing of the operation data; S22, after completing the time consistency processing, for sub-regions with logical coupling faults in the heterogeneous drift feature map, analyzing the relationship between power flow and control response between adjacent regions, inferring missing control information, and combining physical boundary conditions to correct the time consistency processed operation data, generating a consistent control input set with logical closure.

6. The intelligent collaborative decision-making method of a power distribution system according to claim 5, characterized in that, The S21 comprises: S211, for the sub-regions marked as existing time alignment defects in the heterogeneous drift feature map, according to the upload behavior matrix of each perception node, extract the upload delay, data missing and sampling misalignment features, and identify the data segment set to be corrected S212, for the identified to be corrected data segment set With historical redundant data, the interpolation method is used to fill in the missing points, and the continuous observation sequence of the perception node in the time window T is reconstructed. S213, after completing the interpolation, the perceptual nodes with inconsistent sampling frequencies in each sub-region are remapped to the reference time step Δt ref .

7. The intelligent collaborative decision-making method of a power distribution system according to claim 6, characterized in that, The S22 comprises: S221, after completing the time consistency processing, based on the region marked as existing logical coupling fault in the heterogeneous drift feature map (R k ,R u ), analyze its power exchange behavior and control response state in the standard observation time window, identify the perceptual node pair with power direction discontinuity, response information loss or boundary signal non-transmission, and construct the missing control information index set C k,u ; S222, generating proxy operation data for missing operation data in the missing control information index set C k,u based on the operation behaviors of the same type of nodes in adjacent areas in the same time window, using a response similarity supplement method to generate proxy operation data for filling the boundary logical fault. S223, according to the maximum or minimum operating capacity of the device to which the perception node belongs, the load adjustment range, the physical boundary condition of the upper limit of power response, the upper and lower limit constraint correction is carried out on the agent running data of the supplementary push, and finally a consistent control input set with logical closure is generated 8. The intelligent collaborative decision-making method of a power distribution system according to claim 1, characterized in that, The S3 comprises: S31, after obtaining the consistent control input set of each sub-region, constructing a collaborative strategy set according to the current scheduling target and system constraint conditions; S32, decomposing and issuing the generated collaborative strategy set according to the region granularity, automatically selecting the matching sub-strategy version according to the current operation state, boundary conditions and execution capability of each sub-region, and realizing local autonomous execution.

9. The intelligent collaborative decision-making method of a power distribution system according to claim 8, characterized in that, The S31 comprises: S311, after obtaining the consistent control input set of each sub-region, setting a comprehensive scheduling target function according to the current operation scenario, including the minimum total active loss, the minimum node voltage deviation and the optimal load adjustable response; S312, according to the set integrated scheduling objective function, introduce system constraints, including equipment capacity limit, network topology state and tie-line flow boundary; S313, output the coordination strategy set based on the scheduling target function and the system constraint condition 10. The intelligent collaborative decision-making method of a power distribution system according to claim 9, characterized in that, The S32 includes: S321, dividing the generated coordination strategy set into a plurality of regional sub-strategy sets, each sub-strategy set corresponding to control instructions of all perception nodes in a sub-region; According to the sub-region number The coordination strategy set is divided into a plurality of regional sub-strategy sets, and each sub-strategy set corresponds to control instructions of all perception nodes in a sub-region; S322, the sub-region control unit, after receiving the sub-policy set, analyzes the current running state boundary condition Γ k , available control set Ω k to parse, select the most matching policy sub-version in the feasible region S323, the control unit determines the selected policy sub-version Performing device regulation operations within the zone.

Images