Artificial intelligence technology availability evaluation method, device and medium for power distribution network reconfiguration service scenario

Abstract

The present application relates to the field of artificial intelligence, and more particularly to an artificial intelligence technology availability evaluation method for power distribution network reconfiguration business scenarios. The method evaluates the artificial intelligence technology from the availability of data, the functionality, safety, robustness and efficiency of the model, etc. dimensions for the power distribution network reconfiguration decision business scenario. Through the improved AHP algorithm, combined with the adaptive power grid operation state, dynamic weight adjustment is carried out, the comprehensive score of each evaluation dimension is given, and the qualitative and quantitative evaluation conclusion is given. The problem that the prior art cannot adjust the importance of the evaluation dimension in real time according to the change of the operation state, only focuses on a single performance index, and cannot guarantee the reliability and continuous effectiveness of the recommended result in the actual operation is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence, and in particular to an artificial intelligence technology usability evaluation method, device and medium for a power distribution network reconfiguration business scenario. BACKGROUND

[0002] In daily operation, the power distribution network is often affected by factors such as load fluctuation, equipment maintenance, planned shutdown, sudden failure, and the integration of distributed power and flexible load, and often needs to be reconfigured to isolate faults, balance power flow, restore power supply, and improve economic efficiency. With the popularity of the perception layer, the communication layer, and edge computing, the operation monitoring platform can continuously obtain multi-source measurement and event information, and form business data streams for key links such as state perception, topology analysis, and load transfer with topology models, operation procedures, and device constraints.

[0003] Under this background, artificial intelligence technology is used to generate or evaluate reconfiguration strategies, and the data sources, training and inference mechanisms, constraint processing methods, and computational efficiency of the model vary, and the power grid operating state has obvious time-varying and scenario characteristics, thus generating the need for objective measurement of the matching relationship between "data-model-scenario". In order to select a more suitable model and data combination for subsequent network reconfiguration business without interfering with on-site control, a usability evaluation method is needed that follows the laws of electricity, reflects the requirements of business links, and can adaptively adjust the focus according to the operating state, to form a unified evaluation result combining qualitative and quantitative methods and support stable and reliable business decisions. SUMMARY

[0004] The present application provides an artificial intelligence technology usability evaluation method, device and medium for a power distribution network reconfiguration business scenario, to solve the problem that the prior art cannot adjust the importance of the evaluation dimension in real time according to the change in the operating state, resulting in insufficient matching of the model and data in different business situations.

[0005] In a first aspect, an artificial intelligence technology usability evaluation method for a power distribution network reconfiguration business scenario is provided, comprising the following steps:

[0006] A data base is constructed, and a weight vector is calculated by an improved AHP algorithm; specifically including: in the improved AHP algorithm, a candidate data set is constructed based on the data base, a power grid state vector is generated based on the candidate data set, a candidate combination is generated based on the candidate data set and a candidate artificial intelligence model base, and a state potential energy is calculated based on the power grid state vector; and a weight vector is calculated based on the state potential energy;

[0007] The availability score vector of the candidate combination is calculated, the comprehensive availability score of the candidate combination is calculated based on the weight vector and the availability score vector, the worst comprehensive score of the candidate combination is calculated based on the comprehensive availability score of the candidate combination, the best candidate set in availability is generated based on the worst comprehensive score of the candidate combination, and the score result of the best candidate set in availability is provided to an operation decision system or an operation and maintenance personnel of the power distribution network network reconstruction business.

[0008] Preferably, the power grid state vector is generated based on the candidate data set, and specifically includes:

[0009] The power grid state vector composed of five comprehensive state indexes is generated based on the elements in the current candidate data set; the five comprehensive state indexes include a feeder average load rate, a key branch thermal margin index, an event intensity, a transferable channel operability, and data freshness.

[0010] Preferably, the state potential is calculated based on the power grid state vector, and specifically includes:

[0011] The power grid state vector is mapped into the state potential of each evaluation dimension by using a linear-bias structure, and the evaluation dimensions are respectively a data availability dimension, a model functionality dimension, a model safety dimension, a model robustness dimension, and a model efficiency dimension.

[0012] Preferably, the availability score vector of the candidate combination is calculated, and specifically includes:

[0013] A test scene set is constructed on each evaluation dimension, the availability score of the candidate combination in each evaluation dimension is calculated under each test scene, and the availability score vector of the candidate combination is composed of combinations.

[0014] Preferably, the comprehensive availability score of the candidate combination is calculated based on the weight vector and the availability score vector, and specifically includes:

[0015] Based on the weight vector and the availability score vector, a single mapping of robust weighting and conflict penalty is used to compress the availability scores of different evaluation dimensions of the candidate combination into the comprehensive availability score.

[0016] Preferably, the worst comprehensive score of the candidate combination is calculated based on the comprehensive availability score of the candidate combination, and specifically includes:

[0017] The disturbance vector is introduced to perform lower bound evaluation on the uncertainty of the weight vector, and the worst comprehensive score of the candidate combination is obtained.

[0018] Preferably, the disturbance vector is located in a bounded neighborhood centered on the weight vector.

[0019] Preferably, the worst comprehensive score based on the candidate combination generates the best candidate set, specifically including:

[0020] Based on the worst comprehensive score of the candidate combination, the Pareto non-dominated screening and sorting are performed to obtain the best candidate set.

[0021] In a second aspect, an electronic device is provided, comprising:

[0022] A memory having a computer program stored thereon;

[0023] A processor configured to load and execute the computer program to implement the artificial intelligence technology availability evaluation method for the power distribution network reconfiguration service scenario as described above.

[0024] In a third aspect, a computer readable storage medium having a computer program stored thereon is provided, and the computer program is executed by a processor to implement the artificial intelligence technology availability evaluation method for the power distribution network reconfiguration service scenario as described above.

[0025] The technical scheme of the present application has the following advantages:

[0026] 1. The dynamic weight mechanism based on the state potential is constructed to adjust the priority of each evaluation dimension in different operating states in real time, so that the evaluation result meets the current business requirements; the scene-based testing method is tightly coupled with the power grid state vector to make the availability scores of the five evaluation dimensions of data availability, functionality, security, robustness and efficiency fully reflect the actual performance of the key links of the power distribution network reconfiguration.

[0027] 2. The robust weighting and conflict penalty mechanism is introduced in the comprehensive availability score process to effectively avoid unacceptable combinations such as single dimension virtual high or high function low security; by calculating the worst comprehensive score in the weight disturbance range and performing screening, it is ensured that the recommended candidate set is still reliable under the fluctuation of operating state.

[0028] 3. The present application not only can quantitatively measure the pros and cons of different combinations, but also can output qualitative conclusions with guiding significance for business, so as to help the operation and maintenance personnel to quickly select the optimal combination of data resources and artificial intelligence models, improve the accuracy, stability and execution efficiency of the power distribution network reconfiguration decision, reduce the operation risk, and ensure the power supply safety and economy. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0030] Figure 1 The figure is a flow chart of an artificial intelligence technology availability evaluation method for a power distribution network reconfiguration service scenario provided by the embodiments of the present application. DETAILED DESCRIPTION

[0031] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0033] The present application provides an artificial intelligence technology availability evaluation method, device and medium for a power distribution network reconfiguration service scenario, to solve the problem that the prior art cannot adjust the importance of evaluation dimensions in real time according to the change of the running state, leading to insufficient matching degree of the model and data under different business situations; and most methods only focus on a single performance indicator, lacking a comprehensive evaluation mechanism covering data availability, functionality, security, robustness and efficiency at the same time, and easily leading to the problem of decision failure due to strong strategy function but insufficient security or low data quality; in addition, the present application also solves the problem that the prior art lacks consideration of stability under running state fluctuation, lacks worst performance guarantee under weight disturbance condition, and cannot guarantee the reliability and continuous effectiveness of the recommended results in actual operation. The technical solutions provided by the present application will be specifically described below in combination with the drawings.

[0034] The embodiments of the present application provide an artificial intelligence technology availability evaluation method for a power distribution network reconfiguration service scenario, as shown in Figure 1 The method comprises the following steps:

[0035] S1, construct a data base, and calculate a weight vector by using an improved AHP algorithm; specifically including: in the improved AHP algorithm, a candidate data set is constructed based on the data base, a power grid state vector is generated based on the candidate data set, a candidate combination is generated based on the candidate data set and a candidate artificial intelligence model base, and a state potential energy is calculated based on the power grid state vector; the weight vector is calculated based on the state potential energy;

[0036] Through the distribution network operation monitoring platform, the feeder load rate, the thermal margin calculated based on the key branch current and the rated current, the fault and maintenance event log, the available state and operation restriction of the transferable tie switch, the data arrival timestamp and the packet loss rate of each measurement point are synchronously collected in a fixed sampling period. After the above collected data are time-aligned and topologically mapped, abnormal records that do not meet the electrical balance relationship are removed, and missing values are completed by interpolation using adjacent nodes in the same period of history, to generate a data base that meets the device constraints.

[0037] Based on the data base, the state potential energy is calculated by using the improved AHP algorithm, and a weight vector is generated, and the specific process is as follows:

[0038] The candidate data set is constructed based on the data base Each element in the candidate data set is a data versioned instance configured with specific parameters, containing visible branch and node subsets, historical window length, resampling period, data freshness threshold, quality threshold, structure-preserving interpolation / denoising operator set and derived feature set, etc. is the number of elements in the candidate data set. Different represents the difference in data resources that can be used under the evaluation conditions, and directly affects the calculation results of the following five evaluation dimensions, namely data availability, functionality, security, robustness and efficiency.

[0039] At any evaluation time, five comprehensive state indicators related to weight calculation are read and normalized from the currently selected to form a power grid state vector, which is a compressed representation of the candidate data set at the current time, and is only used for state potential energy calculation and dynamic weight calculation.

[0040] The candidate artificial intelligence model base is composed of network reconfiguration algorithm entities adapted through a unified interface, represents the number of models in the candidate artificial intelligence model base; represents any model in the candidate artificial intelligence model base; the input of the artificial intelligence model is the selected The provided data and constraint information are output as a set of reconstruction strategies that satisfy radiality and operating constraints, for performance evaluation in terms of availability, security, robustness, and efficiency, without directly executing physical control. The candidate combinations thus formed In the current grid state vector Enter the subsequent multi-dimensional availability score, ranking, and other calculation processes.

[0041] Grid state vector , where is the feeder average load rate, obtained from real-time measurement and short-term prediction; is the key branch thermal margin index, calculated from the rated current and measured current; is the event intensity, obtained by normalizing the time density and severity of faults, maintenance, and overload alarms from line current alarm records; is the operability of the transferable channel, an index formed based on the current available tie switch ratio, whether there is an isolation path in the topology that meets the operating procedures and safety limits, and the operating window constraint conditions; is the data freshness, obtained by fitting the historical data of measurement arrival delay and packet loss rate, with a larger value representing a fresher value; represents the transpose. All components in the grid state vector are limited to the [0, 1] interval through Min-max normalization mapping.

[0042] The five evaluation dimensions are the data availability dimension , the model functionality dimension , the model security dimension , the model robustness dimension , and the model efficiency dimension , with the corresponding immediate priority driver being , where the model refers to the artificial intelligence model being evaluated, i.e., the calculation model / reasoning engine that maps the input grid state vector and topology constraints to reconstruction strategies in the distribution network reconstruction business. The topology constraint refers to the electrical structure requirements that must be met. The marginal driving direction and intensity of the current grid state on each dimension are represented, and must satisfy explicit monotonicity and sign constraints:

[0043] The marginal effect of event intensity on model security is non-negative, i.e., ;

[0044] The marginal effect of feeder average load rate on model efficiency is non-negative, i.e., ;

[0045] The marginal effect of data freshness on data availability and model robustness is non-positive, i.e., ;

[0046] The marginal effect of switchable channel operability on model functionality is non-negative, i.e. .

[0047] The composition adopts a linear-bias structure to maintain interpretability and calibrability, mapping the observable grid state vector to the instantaneous priority driving force of each evaluation dimension. The specific calculation is:

[0048] ;

[0049] wherein, is the five-dimensional state potential energy, corresponding to the instantaneous priority driving force of the five evaluation dimensions at time t, is the potential energy of the data availability dimension; is the potential energy of the model functionality dimension; is the potential energy of the model security dimension; is the potential energy of the model robustness dimension; is the potential energy of the model efficiency dimension; is the coefficient matrix, used to map the five grid states to the five state potential energy components, with row index as potential energy dimension and column index as grid state, and the value range is [-0.5, 0.5]; is the intercept vector, used to reflect the basic potential energy level of each dimension "in the baseline state of zero event / zero load growth", with the value range of [-0.5, 0.5], and are obtained by least squares method.

[0050] The five-dimensional state potential energy is converted into a five-dimensional weight vector, so that the weight changes in real time with the grid state and maintains physical interpretability; the weight is expressed as:

[0051] ;

[0052] wherein, is any one of the five-dimensional weight vector, used to dynamically weight the five evaluation dimensions; is the state sensitivity coefficient, used to control the contrast strength after exponential mapping, obtained by fitting the "state-weight response" curve, ; is any one of the five-dimensional state potential energy, and any evaluation dimension .

[0053] S2. Calculate the availability score vector of the candidate combination. Based on the weight vector and the availability score vector, calculate the comprehensive availability score of the candidate combination. Based on the comprehensive availability score of the candidate combination, calculate the worst comprehensive score of the candidate combination. Based on the worst comprehensive score of the candidate combination, generate the candidate set with the best availability. Provide the score results of the candidate set with the best availability to the operation decision system or operation and maintenance personnel of the distribution network reconfiguration business.

[0054] To calculate candidate combinations A usability score vector is generated across five evaluation dimensions, ensuring tight coupling with the weight vector and the grid state vector. First, in each evaluation dimension... The system constructs a set of test scenarios consistent with the business processes. Each test scenario serves as a test environment for usability across each evaluation dimension. Within these test scenarios, usability scores for candidate combinations across each evaluation dimension are calculated. The five-dimensional usability scores are obtained through a weighted average, specifically calculated as follows:

[0055] ;

[0056] in, At any moment Below, candidate combinations In the Usability scores for each evaluation dimension; Is with the first An index set of test scenarios related to each evaluation dimension; At any moment Next, the The first evaluation dimension The weighting factor for the first test scenario is used to reflect the current power grid state. The importance of each test scenario is determined by a value ranging from [0,1]. Candidate combination In the The first evaluation dimension Performance scores in each test scenario, based on candidate combinations The standardized evaluation score obtained after the network reconfiguration task supported by the joint system performs feature extraction, constraint verification and performance measurement based on power grid physical rules has a value range of [0,1].

[0057] Using a five-dimensional weight vector and a usability score vector as input, the usability scores of different evaluation dimensions at the same time are compressed into a single scalar score, i.e., a comprehensive usability score. A robust weighted and conflict-penalized single-form mapping is employed, specifically calculated as follows:

[0058] ;

[0059] in, At any moment For candidate combinations The overall usability score is normalized to (0,1); This is a smoothing compression function used to map the input to (0,1); This is the variance penalty coefficient, used to suppress artificially inflated values ​​in one dimension, obtained through expert empirical statistics. ; yes The sample variance At any moment For candidate combinations In five evaluation dimensions—data availability dimension Model functional dimension Model security dimension Model robustness dimension With model efficiency dimension The normalized score vector on; For a set of conflicting pairs, such as , From the set of conflict pairs The dimension labels extracted from, for example This indicates a conflict between the usability score of the model's functionality dimension and the usability score of the model's safety dimension. The candidate combinations are in the functional dimension of the model Usability rating The candidate combinations in terms of model safety Usability rating; This is the conflict intensity coefficient, which is applied when the penalty condition is triggered. hour, The severity of the point deduction is determined by this. It's the tolerance ratio, in comparison. Dimensions and When scoring usability in dimensions, This determines the threshold conditions for triggering penalties.

[0060] A lower bound assessment is performed on the uncertainty of the weight vector to ensure that the overall availability score remains comparable and reliable under short-term state fluctuations. The perturbation vector is assumed to lie within a bounded neighborhood centered on the weight vector, with the neighborhood radius taken from the 95th percentile of a one-step change in the weight vector to characterize the acceptable range of weight changes. The worst-case overall score is calculated within a bounded neighborhood, and a robust ranking is performed accordingly. The robust score is defined as follows:

[0061] ;

[0062] in, Candidate combination Within the allowable weight perturbation range the worst overall score, i.e. robust score, among all the weight vectors within the disturbance constraint denotes the overall availability score of the candidate combination among all the disturbance vectors that satisfy the disturbance constraint , the one that makes the overall availability score the allowed weight disturbance radius, derived from the 95% quantile of one-step change in the weight vector, i.e. the maximum single dimension change that can be tolerated in weight adjustment; is norm, i.e. the maximum component absolute value, which can be obtained by searching within the ball via gridding or projecting the sub-gradient is the disturbance vector denotes applying a disturbance to the current calculated weight vector and re-computing the overall availability score of the candidate combination with the new disturbed weight vector . Pareto non-dominated screening and ranking is performed on the scale of to obtain the best candidate set ; is the optimal data set instance selected from the candidate data set pool , and is the optimal artificial intelligence model selected from the candidate artificial intelligence model pool .

[0063] After obtaining the worst overall score , all the candidate “data set-artificial intelligence model” combinations are ranked from high to low according to , and the Pareto non-dominated combinations are screened as the best candidate set; subsequently, the score result of the candidate set is output to provide the operation decision system or the operation and maintenance personnel of the power distribution network reconfiguration business, so as to select the data set and model combination that is the most stable and meets the business requirements under the current and expected operation state, thereby completing the artificial intelligence technology availability evaluation for the power distribution network reconfiguration business scenario.

[0064] The embodiment of the present application also provides an electronic device, comprising:

[0065] a memory having a computer program stored thereon;

[0066] a processor configured to load and execute the computer program to implement the artificial intelligence technology availability evaluation method for the power distribution network reconfiguration business scenario as described above.

[0067] The embodiment of the present application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the artificial intelligence technology availability evaluation method for the power distribution network reconfiguration service scenario.

[0068] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. In addition, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0069] The present application is described with reference to flowcharts and / or block diagrams according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination 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 an apparatus that carries out the functions specified in one or more flows and / or blocks.

[0070] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction apparatus that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that carries out the functions specified in one or more flows and / or blocks.

[0071] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that carries out the functions specified in one or more flows and / or blocks.

[0072] The progressive nature of the specification and examples does not constitute a requirement that all embodiments include all of the described elements nor that the disclosed steps be performed in the order presented. In some embodiments, a step can be performed in a different order or omitted.

[0073] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments.

[0074] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalent ones. The modification or replacement does not change the essence of the corresponding technical solutions, and should be included in the protection scope of the present application.

Claims

1. A method for assessing the availability of artificial intelligence technology for distribution network reconfiguration business scenarios, characterized in that, Includes the following steps: A data foundation library is constructed, and a weight vector is calculated using an improved AHP algorithm. Specifically, the improved AHP algorithm involves: constructing a candidate data set based on the data foundation library; generating a power grid state vector based on the candidate data set; generating a candidate combination based on the candidate data set and a candidate artificial intelligence model library; calculating the state potential energy based on the power grid state vector; and calculating the weight vector based on the state potential energy. The state potential energy is constructed using a linear-bias structure to maintain interpretability and calibrability, mapping the observable grid state vector to the immediate priority driving forces of each evaluation dimension. The specific calculation is as follows: ； in, The five-dimensional state potential energy corresponds to the immediate priority driving force of each of the five evaluation dimensions at time t. It is the potential energy of the data availability dimension; It is the potential energy of the model's functional dimension; It is the potential energy of the model's safety dimension; It is the potential energy of the model's robustness dimension; It is the potential energy of the model's efficiency dimension; It is a coefficient matrix; It is the intercept vector; This is the power grid state vector; The five-dimensional state potential is transformed into a five-dimensional weight vector, with the weights represented as follows: ； in, It is any one of the five-dimensional weight vectors; It is the state sensitivity coefficient. ; It is any one of the five-dimensional potential energies, any evaluation dimension. ; Calculate the availability score vector of the candidate combinations, and calculate the comprehensive availability score of the candidate combinations based on the weight vector and the availability score vector; calculate the worst comprehensive score of the candidate combinations based on the comprehensive availability score of the candidate combinations; generate the candidate set with the best availability based on the worst comprehensive score of the candidate combinations; and provide the score results of the candidate set with the best availability to the operation decision system or operation and maintenance personnel of the distribution network reconfiguration business.

2. The method for assessing the availability of artificial intelligence technology for distribution network reconfiguration business scenarios according to claim 1, characterized in that, The generation of the power grid state vector based on the candidate data set specifically includes: Based on the elements in the current candidate data set, a corresponding power grid state vector consisting of five comprehensive state indicators is generated; the five comprehensive state indicators include average feeder load rate, critical branch thermal margin index, event intensity, transferable channel operability, and data freshness.

3. The method for assessing the availability of artificial intelligence technology for distribution network reconfiguration business scenarios according to claim 2, characterized in that, The calculation of state potential energy based on the power grid state vector specifically includes: A linear-bias structure is adopted to map the power grid state vector into the state potential energy of each evaluation dimension, namely data availability dimension, model functionality dimension, model security dimension, model robustness dimension and model efficiency dimension.

4. The method for assessing the availability of artificial intelligence technology for distribution network reconfiguration business scenarios according to claim 1, characterized in that, The calculation of the availability score vector for candidate combinations specifically includes: Construct a set of test scenarios for each evaluation dimension, calculate the usability score of the candidate combination for each evaluation dimension under each test scenario, and combine them to form the usability score vector of the candidate combination.

5. The method for assessing the availability of artificial intelligence technology for distribution network reconfiguration business scenarios according to claim 1, characterized in that, The calculation of the comprehensive usability score for candidate combinations based on the weight vector and usability score vector specifically includes: Based on the weight vector and usability score vector, a robust weighted and conflict-penalized single mapping is used to compress the usability scores of different evaluation dimensions of candidate combinations into a comprehensive usability score.

6. The method for assessing the availability of artificial intelligence technology for distribution network reconfiguration business scenarios according to claim 1, characterized in that, The comprehensive usability score based on candidate combinations, specifically calculating the worst comprehensive score for candidate combinations, includes: By introducing a perturbation vector, the uncertainty of the weight vector is evaluated by lower bound, and the worst comprehensive score of the candidate combination is obtained.

7. The method for assessing the availability of artificial intelligence technology for distribution network reconfiguration business scenarios according to claim 6, characterized in that, The perturbation vector is located in a bounded neighborhood centered on the weight vector.

8. The method for assessing the availability of artificial intelligence technology for distribution network reconfiguration business scenarios according to claim 6, characterized in that, The process of generating the best-usability candidate set based on the worst comprehensive score of candidate combinations specifically includes: Based on the worst overall score of the candidate combinations, Pareto non-dominated screening and ranking are performed to obtain the candidate set with the best usability.

9. An electronic device, characterized in that, include: A memory on which computer programs are stored; A processor is configured to load and execute the computer program to implement the artificial intelligence technology availability assessment method for distribution network reconfiguration business scenarios as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the artificial intelligence technology availability assessment method for distribution network reconfiguration business scenarios as described in any one of claims 1 to 8.

Images