Flexible resource regulation capability assessment method and system based on label and risk analysis

Abstract

The invention discloses a flexible resource adjustment capability assessment method and system based on labels and risk analysis, and relates to the technical field of power system demand side management. The method comprises the following steps: acquiring real-time operation parameters of load side flexibility resources, and generating a multi-dimensional capability envelope of individual resources based on a multi-state finite automaton model; constructing a multi-dimensional value label, clustering and grouping the flexible resources in combination with a multi-dimensional capability envelope, and constructing a virtual polymer; quantizing the aggregation adjustment capability of the virtual aggregate by adopting an opportunity constraint robust aggregation method to generate a multi-dimensional vector; and fusing the multi-dimensional vector with geographic information, realizing visualization through a thermodynamic diagram and an interactive interface, and generating a system-level capability map. According to the method, the full-dimensional capability evaluation of the flexible resources from individuals to aggregation is realized, the value attributes and the operation risk are considered, and a visual and reliable decision support is provided for power grid dispatching.

Description

Technical Field

[0001] This invention relates to the field of demand-side management technology for power systems, specifically to a method and system for assessing flexible resource regulation capabilities based on tagging and risk analysis. Background Technology

[0002] Load-side flexibility resources, as an important carrier of distributed regulation potential, are a key component of the power system to provide regulation services and absorb renewable energy fluctuations. Accurately assessing the adjustability of such resources is a prerequisite for implementing efficient and reliable demand response.

[0003] Current quantitative assessment methods for load-side flexibility resources in power systems mostly focus on single-type resources such as temperature-controlled loads, employing linear or threshold-simplified models for resource description. This makes it difficult to uniformly characterize the operational status and constraints of complex and heterogeneous load-side flexibility resources such as industrial processes and electric vehicles. Furthermore, traditional assessment methods only provide static indicators such as maximum adjustable power, neglecting dynamic characteristics like ramp rate and duration. This fails to provide effective support for real-time power system dispatch at the minute or second level. In assessing resource clusters, they also fail to fully consider uncertainties such as random user behavior and communication failures, leading to overly optimistic assessment results and increasing the risk of various problems during power system dispatch. Moreover, the quantitative results obtained from traditional assessment methods are often presented in the form of data tables or simple curves, lacking integrated visualization of dimensions such as spatiotemporal distribution and confidence levels. This hinders virtual power plants, load aggregators, and power system dispatchers from quickly understanding and implementing the assessment results. Summary of the Invention

[0004] This invention addresses the technical problems in existing technologies, such as the lack of a unified descriptive model for heterogeneous flexible resources on the load side, the single evaluation dimension ignoring dynamic characteristics, and the failure to fully consider the uncertainty risks of resource adjustment. It provides a method and system for evaluating flexible resource adjustment capabilities based on labeling and risk analysis.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides a method for assessing flexible resource adjustment capabilities based on labeling and risk analysis, including: Collect real-time operational data of load-side flexibility resources, construct a multi-state finite automaton model to describe the operational status of the resources, predict the adjustable capabilities of each flexibility resource in future scheduling cycles based on the multi-state finite automaton model, and generate a multi-dimensional capability envelope for individual resources. Based on the multi-dimensional value tags of each flexible resource, and combined with the multi-dimensional capability envelope of the individual resources, the flexible resources are clustered and grouped to construct virtual aggregates. The aggregation and adjustment capabilities of each virtual aggregate are robustly quantified to generate a multi-dimensional vector characterizing the overall adjustment capability of the virtual aggregate. The multidimensional vectors are visualized to generate a system-level capability map to support scheduling decisions.

[0006] Secondly, this invention provides a flexible resource adjustment capability assessment system based on tagging and risk analysis, including: The individual resource capability analysis module is used to collect real-time operating data of load-side flexibility resources, construct a multi-state finite automaton model to describe the operating state of resources, predict the adjustable capability of each flexibility resource in the future scheduling cycle based on the multi-state finite automaton model, and generate a multi-dimensional capability envelope of individual resources. The virtual aggregate construction module is used to cluster and group the flexible resources based on the multi-dimensional value tags of each flexible resource and the multi-dimensional capability envelope of the individual resources to construct virtual aggregates, and to robustly quantify the aggregation and adjustment capabilities of each virtual aggregate to generate a multi-dimensional vector characterizing the overall adjustment capability of the virtual aggregate. The visualization decision support module is used to visualize the multi-dimensional vectors and generate a system-level capability map to support scheduling decisions.

[0007] The beneficial effects of this invention are: Compared to existing technologies, this invention first constructs a multi-state finite automaton model to uniformly describe the operating states and constraints of various load-side flexibility resources, accurately capturing their dynamic characteristics and solving the problem of inconsistent descriptions of heterogeneous resources. Secondly, it generates a multi-dimensional capability envelope containing power regulation capability, duration, and ramp rate, achieving multi-dimensional dynamic quantification of resource adjustability and overcoming the shortcomings of traditional static index evaluation. Thirdly, it constructs a virtual aggregate based on multi-dimensional value label clustering, and combines robust quantification methods to account for various uncertainties and risks, generating a multi-dimensional vector of regulation capability, mitigating scheduling risks and adapting to multi-objective scheduling needs. Finally, it performs spatiotemporal visualization processing on the multi-dimensional vector to generate a system-level capability map, transforming heterogeneous quantified data into intuitive decision-making information, improving the cognitive and decision-making efficiency of dispatchers, providing reliable support for precise power grid scheduling and virtual power plant operation, and contributing to the safe and efficient operation of the power system. Attached Figure Description

[0008] Figure 1 A flowchart illustrating the flexible resource adjustment capability assessment method based on tagging and risk analysis provided by this invention; Figure 2This is a schematic diagram of the structure of the flexible resource adjustment capability assessment system based on tagging and risk analysis provided by the present invention.

[0009] In the attached diagram, the components represented by each number are as follows: Individual resource capability analysis module 11, virtual aggregate construction module 12, and visualization decision support module 13. Detailed Implementation

[0010] Example 1, as Figure 1 As shown, embodiments of the present invention provide a method for assessing flexible resource adjustment capabilities based on tagging and risk analysis, including: S10: Collect real-time operating data of load-side flexibility resources, construct a multi-state finite automaton model to describe the operating state of the resources, predict the adjustable capacity of each flexibility resource in the future scheduling cycle based on the multi-state finite automaton model, and generate a multi-dimensional capacity envelope of individual resources. First, real-time operational data of load-side flexibility resources is collected. Load-side flexibility resources refer to power resources with power regulation capabilities, such as industrial and commercial adjustable loads, electric vehicles, and energy storage systems. In the dispatch and operation scenario of a regional distribution network, the real-time operational data of load-side flexibility resources consists of dynamic monitoring data such as the operating mode, instantaneous power, state of charge, and indoor temperature of various resources, representing the current actual operating status and physical characteristics of the resources.

[0011] Based on real-time operational data, a multi-state finite automaton model is constructed to describe the operational state of resources. This multi-state finite automaton model is an algorithmic model capable of characterizing different dynamic operational states of resources and state transition rules. It defines the transition conditions between various dynamic operational states of a resource based on the resource's state switching nodes and parameter change thresholds, accurately simulating the full-process operational characteristics of the resource, thus providing standardized model support for predicting resource adjustability. Based on this multi-state finite automaton model, the adjustability of each flexible resource in future scheduling cycles can be predicted. This adjustability is the power adjustment potential and related dynamic characteristics that the resource can achieve at various future times. Finally, a multi-dimensional capability envelope for individual resources is generated.

[0012] Specifically, the resulting multidimensional capability envelope is a multidimensional capability data curve that integrates power regulation capability, duration, and ramp rate at discrete moments within the future scheduling cycle. It is used to intuitively characterize the complete adjustable capability features of individual resources at different future moments and quantify the dynamic change pattern of resource regulation capability.

[0013] Specifically, real-time operational data of load-side flexibility resources are collected, a multi-state finite automaton model is constructed to describe the operational state of the resources, and the adjustable capabilities of each flexibility resource in future scheduling cycles are predicted based on the multi-state finite automaton model, generating a multi-dimensional capability envelope for each individual resource, including: The collected load-side flexibility resources include at least one of industrial and commercial adjustable loads, electric vehicles, and energy storage systems. By extracting the state switching nodes and parameter change thresholds of each flexibility resource during operation, a multi-state finite automaton model is constructed. The multi-state finite automaton model contains multiple dynamic operating states, and the transition conditions between each dynamic operating state include at least one of time threshold, power change value, or external excitation signal. After instantiating the multi-state finite automaton model for each of the aforementioned flexible resources, the physical constraint parameters, dynamic operating state, and response constraint parameters of each of the aforementioned flexible resources are identified and stored. The physical constraint parameters include at least the rated power, the upper limit of power adjustment, and the lower limit of power adjustment. The dynamic operating state includes at least the operating mode and instantaneous power at the current moment. The response constraint parameters include at least the maximum ramp rate, the minimum stable operating time, and the minimum downtime. The multi-state finite automaton model of each of the aforementioned flexible resources is set to autonomous operation mode, and the predicted external driving data sequence is input. The power output sequence from the current time to the end of the scheduling cycle is obtained through simulation, and the baseline load curve of each of the aforementioned flexible resources is generated. Based on the baseline load curve, for each discrete moment in the future scheduling cycle, the maximum upward adjustment potential and the maximum downward adjustment potential of each flexibility resource at that discrete moment are calculated using a constrained instantaneous power boundary search algorithm. Forward simulation is performed based on the multi-state finite automaton model to calculate the maximum duration for which the adjustment potential can be continuously invoked. At the same time, based on the static parameters in the multi-state finite automaton model and the adjustment space of the current running point, the maximum available ramp rate at that discrete moment is calculated. For each discrete moment discretized at a set time resolution within the future scheduling period, iterative calculations are repeatedly performed to generate the multidimensional capability envelope of each flexibility resource. The multidimensional capability envelope includes at least a power regulation capability value, a duration value, and a ramp rate value.

[0014] First, relevant data on the operation of at least one type of load-side flexibility resource in industrial and commercial adjustable loads, electric vehicles, and energy storage systems are collected. State switching nodes and parameter change thresholds generated by each load-side flexibility resource during operation are extracted. Based on the extracted state switching nodes and parameter change thresholds, a multi-state finite automaton model is constructed.

[0015] Optionally, the state switching nodes for industrial and commercial adjustable loads are the time nodes for equipment start-up and shutdown, operation gear switching, and process adjustment, representing key time points when the equipment's operating state changes; the state switching nodes for electric vehicles are the time nodes for charging start-up and shutdown, charging power switching, and vehicle connection and disconnection from charging piles, representing key time points when the vehicle's charging and discharging state changes; and the state switching nodes for energy storage systems are the time nodes for switching between charging, discharging, and standby modes, representing key time points when the energy conversion state of energy storage devices changes.

[0016] The parameter change threshold is the critical value at which core operating parameters such as power, state of charge, and temperature undergo effective changes during the operation of various load-side flexible resources. This parameter change threshold is obtained through statistical analysis of historical operating data, calibration of equipment factory technical parameters, and on-site measurement, and represents the critical judgment standard for triggering the switching of resource operating status.

[0017] The multi-state finite automaton model is an algorithmic model that can accurately describe the dynamic operational characteristics of load-side flexibility resources throughout the entire process. It defines different operating states of resources and the rules for state transitions. It contains multiple dynamic operating states, and the transition conditions between each dynamic operating state include at least one of the following: a time threshold, a power change value, and an external excitation signal. Specifically, the time threshold is the minimum or maximum time limit for a resource to maintain a certain operating state, serving as a quantitative indicator to determine whether the resource meets the state transition time condition; the power change value is the specific numerical value of the increase or decrease in the resource's operating power, serving as a quantitative indicator to determine whether the resource has reached the state transition power condition; and the external excitation signal is a control signal triggered by grid dispatching instructions, user operation instructions, or environmental changes, serving as a command-type condition that triggers the active transition of the resource state.

[0018] By comprehensively judging and triggering various transfer conditions, the mutual conversion between different dynamic operating states can be realized, thereby accurately matching the actual operating state change patterns of load-side flexibility resources.

[0019] Secondly, for each load-side flexibility resource, the instantiation of a multi-state finite automaton model is performed. Instantiation refers to matching the general multi-state finite automaton model with the specific operating characteristics of the load-side flexibility resource, assigning the model exclusive operating parameters and state characteristics for that resource, so as to form a personalized model that can accurately describe the actual operating state of the resource.

[0020] Based on the instantiated multi-state finite automaton model, the physical constraint parameters, dynamic operating states, and response constraint parameters corresponding to each load-side flexibility resource are comprehensively identified, and the identified parameters and state information are stored.

[0021] The physical constraint parameters include at least the rated power, the upper limit of power regulation, and the lower limit of power regulation. The rated power is the standard power value for the load-side flexibility resource during its design and operation, serving as the basic power reference indicator for normal resource operation. The upper limit of power regulation is the maximum power limit that the load-side flexibility resource can reach during regulation, representing the physical boundary for upward power regulation. The lower limit of power regulation is the minimum power limit that the load-side flexibility resource can reach during regulation, representing the physical boundary for downward power regulation. These physical constraint parameters represent the inherent and insurmountable physical operating boundaries of the load-side flexibility resource itself.

[0022] Dynamic operating status includes at least the current operating mode and instantaneous power. The current operating mode refers to the actual operating condition of the load-side flexibility resource at the current point in time, which directly reflects the current operating status of the resource; instantaneous power is the real-time output power value of the load-side flexibility resource at the current point in time, which accurately reflects the current power output level of the resource. Dynamic operating status represents the actual operating characteristics of the load-side flexibility resource at the current moment.

[0023] Response constraint parameters include at least the maximum ramp rate, minimum stable operating time, and minimum downtime. The maximum ramp rate is the maximum power change rate that load-side flexible resources can achieve per unit time, and is a core indicator for measuring the power regulation speed of resources. The minimum stable operating time is the shortest time limit that a load-side flexible resource needs to maintain after switching to a certain operating state; it is a time constraint to ensure stable operation of the resource. The minimum downtime is the shortest time limit that a load-side flexible resource needs to maintain a stopped state before restarting; it is a time constraint to protect the resource equipment. Response constraint parameters represent the dynamic response constraint requirements that load-side flexible resources must follow when performing state transitions and power regulation after receiving regulation commands.

[0024] By identifying and storing the above parameters and states, complete parameter support can be provided for predicting the adjustability of load-side flexibility resources.

[0025] Furthermore, the operational control strategy for the multi-state finite automaton model corresponding to each load-side flexibility resource is set to autonomous operation mode. Autonomous operation mode refers to the operation of the multi-state finite automaton model according to the inherent operational flow and self-constraints of the load-side flexibility resource, without receiving external scheduling and adjustment commands. By inputting a pre-predicted external driving data sequence into the multi-state finite automaton model under this autonomous operation mode, and simulating the operation process of the load-side flexibility resource based on the multi-state finite automaton model, the power output sequence of the load-side flexibility resource from the current time to the end of the scheduling cycle can be simulated.

[0026] The external driving data sequence is a time-series data set predicted in advance for various external environmental factors affecting the operational status of load-side flexibility resources within future scheduling cycles. It includes external impact data related to resource operation, such as ambient temperature, light intensity, and electricity demand forecasts. By inputting this external driving data sequence into an instantiated multi-state finite automaton model, simulation can reproduce the power change process of load-side flexibility resources under natural operating conditions, accurately simulating the operational characteristics of resources as the external environment changes, and obtaining the power output data corresponding to each time point within future scheduling cycles.

[0027] Specifically, the timeframe from the current moment to the end of the scheduling cycle is a pre-defined time range for predicting resource adjustability, serving as the time period for assessing resource adjustment capabilities. This time range is set based on the actual needs of power grid dispatch and the timescale characteristics of resource adjustment; for example, it is set to the next 24 hours in short-term dispatch scenarios and to the next 15 minutes to 1 hour in ultra-short-term dispatch scenarios. The power output sequence is a dataset formed by sequentially representing the power output values ​​of load-side flexibility resources at each discrete moment within this time range under autonomous operation mode. It is used to characterize the power output variation pattern of load-side flexibility resources in future scheduling cycles without external regulatory intervention.

[0028] Based on this power output sequence, a baseline load curve can be generated for each load-side flexibility resource. The baseline load curve is a curve obtained by visually fitting the time-series power data in the power output sequence. It can intuitively reflect the natural power change trend of the load-side flexibility resource without external adjustment commands, and serve as a benchmark for subsequent calculation of the resource power adjustment potential, providing a basis for accurately determining the upward and downward adjustment capability boundaries of the resource.

[0029] Furthermore, based on the generated baseline load curve, for each discrete moment within the future scheduling cycle divided according to set rules, a constrained instantaneous power boundary search algorithm is used to calculate the maximum upward adjustment potential and maximum downward adjustment potential that each load-side flexibility resource can achieve at that discrete moment. Here, discrete moment refers to a series of continuous time points with fixed time intervals obtained after equally dividing the future scheduling cycle according to a preset time resolution; it serves as the time reference for point-by-point calculation of resource adjustability.

[0030] The constrained instantaneous power boundary search algorithm is an algorithm that combines various physical and response constraints of load-side flexibility resources to traverse and optimize the instantaneous power adjustment boundary of resources within the feasible solution space. Based on this constrained instantaneous power boundary search algorithm, the physical constraint parameters, response constraint parameters, and current dynamic operating state of the resources can be incorporated to limit the feasible range of power adjustment, enabling precise search and calculation of the power adjustment boundary at discrete moments to obtain the limit value of resource power adjustment at that moment. Specifically, the maximum upward adjustment potential is the maximum increment of power relative to the baseline load power that the load-side flexibility resources can achieve at that discrete moment under all constraint conditions; the maximum downward adjustment potential is the maximum reduction of power relative to the baseline load power that the load-side flexibility resources can achieve at that discrete moment under all constraint conditions.

[0031] Meanwhile, based on a multi-state finite automata model, forward simulation is conducted on the adjustment process of load-side flexibility resources. Forward simulation takes the resource state after the adjustment command is applied at a discrete moment as the starting point of the simulation, and advances forward according to a preset time step, such as 1 minute or 15 minutes, continuously inputting predictive external driving data and maintaining the validity of the adjustment command, while simultaneously monitoring whether the resource operating state meets the constraint requirements.

[0032] Based on the simulation results, the maximum duration for which the regulation potential can be continuously utilized at this discrete moment can be calculated. Specifically, the maximum duration for which the regulation potential can be continuously utilized is the longest time that the load-side flexibility resources can maintain a regulated state at this discrete moment, provided that all operating parameters satisfy physical and response constraints.

[0033] Furthermore, static parameters are extracted from the multi-state finite automaton model and combined with the adjustment space of the load-side flexibility resources at the current operating point of this discrete moment. Through comprehensive calculation, the maximum available ramp rate of the load-side flexibility resources at this discrete moment is obtained. Static parameters are inherent equipment technical parameters of the load-side flexibility resources that do not change with operating states, providing a fixed reference standard for calculating resource operation and adjustment capabilities. The adjustment space of the load-side flexibility resources at the current operating point of this discrete moment represents the effective range of power that can be increased or decreased at this discrete moment, provided all constraints are met.

[0034] The maximum available ramp rate is the maximum power change rate that the load-side flexibility resources can achieve per unit time at this discrete moment, taking into account the inherent physical limits of the equipment and the adjustment space of the current operating point. It is the core indicator for measuring the response speed of resource power regulation.

[0035] Finally, the future scheduling cycle is discretized according to the set time resolution. The time resolution is the standard for dividing the future scheduling cycle into a number of discrete moments. This time resolution is set according to the precision requirements of power grid scheduling and the response characteristics of load-side flexibility resources. For example, it is set to 1 minute or 5 minutes in minute-level scheduling scenarios, and 15 minutes or 60 minutes in hour-level scheduling scenarios.

[0036] For each discrete moment obtained after discretization, the calculation steps for maximum adjustment potential, maximum duration, and maximum available ramp rate are repeated sequentially. Through iterative calculation, the power adjustment capability, duration, and ramp rate values ​​corresponding to each load-side flexibility resource at all discrete moments within the future scheduling cycle are obtained. Based on the above three types of values, a multi-dimensional capability envelope corresponding to each load-side flexibility resource is generated. This multi-dimensional capability envelope can comprehensively characterize the complete adjustable capability features of the load-side flexibility resource at different moments within the future scheduling cycle.

[0037] S20: Based on the multi-dimensional value tags of each flexible resource, and combined with the multi-dimensional capability envelope of the individual resources, the flexible resources are clustered and grouped to construct virtual aggregates. The aggregation and adjustment capabilities of each virtual aggregate are robustly quantified to generate a multi-dimensional vector characterizing the overall adjustment capability of the virtual aggregate. Secondly, based on the multi-dimensional value tags of each flexible resource, and combined with the multi-dimensional capability envelope of individual resources, virtual aggregates are constructed by clustering and grouping each flexible resource. Among them, the multi-dimensional value tags are quantifiable value identifiers set for each flexible resource from dimensions such as social attributes, spatial attributes, technological attributes, and environmental attributes, which can comprehensively characterize the comprehensive attribute characteristics of flexible resources.

[0038] Since the multidimensional capability envelope represents the complete adjustable capability characteristics of each flexible resource at different times within a future scheduling cycle, it can intuitively reflect the resource's adjustment capability level and dynamic change pattern. Therefore, multidimensional value labels and multidimensional capability envelopes can be used together as the core basis for clustering and grouping, clustering and grouping various flexible resources to construct virtual aggregates. Specifically, this clustering and grouping process involves first grouping flexible resources according to the priority of multidimensional value labels, then converting the feature information of multidimensional value labels and multidimensional capability envelopes into multidimensional feature vectors, and performing a secondary clustering and division of flexible resources based on the comprehensive similarity of feature vectors. By merging and integrating flexible resources with similar features, a virtual aggregate is constructed. This virtual aggregate is a resource cluster composed of a group of flexible resources with similar features and matched adjustment capabilities, enabling the aggregated presentation of the adjustment capabilities of individual resources.

[0039] Furthermore, robust quantification is performed on the aggregation and adjustment capabilities of each virtual aggregate. This aggregation and adjustment capability is the overall adjustment capability formed by aggregating the adjustment capabilities of all flexible resources within the virtual aggregate, and it is a core indicator reflecting the overall adjustment potential of the resource cluster. Robust quantification essentially involves comprehensively considering various uncertainties and risks such as individual resource state prediction bias, user response compliance rate, and communication reliability, and quantifying the overall adjustment capability of the virtual aggregate under a set confidence level to obtain a reliable aggregation and adjustment capability that can ensure invocation. Finally, a multi-dimensional vector representing the overall adjustment capability of the virtual aggregate is generated.

[0040] First, based on the multi-dimensional value tags of each flexibility resource, and combined with the multi-dimensional capability envelope of each individual resource, the flexibility resources are clustered and grouped to construct a virtual aggregate, including: At least one label dimension is selected as a component of the multi-dimensional value label system, wherein the label dimension includes at least one of social importance label, spatial attribute label, technological attribute label and environmental attribute label, and a quantifiable label level is set for each label dimension; All the aforementioned flexibility resources are initially grouped according to the priority of the aforementioned social importance tags to ensure that the flexibility resources within the same priority group have similar levels of social importance. The multidimensional capability envelope and multidimensional value label of each individual resource of the flexibility resource are used together as the input features for clustering and grouping. The label information and capability information of each flexibility resource after the initial grouping at any time are represented as a multidimensional feature vector. Using the multidimensional feature vector as input, a clustering algorithm is used to perform secondary clustering on the flexibility resources within the same priority group. Based on the comprehensive similarity of resources in the multidimensional feature space, resources with similar features are assigned to the same virtual aggregate, thus completing the construction of the virtual aggregate.

[0041] The social importance label is used to characterize the social function of the industry to which the resource belongs and the severity of the consequences of power outages, and the ordinal value reflecting the priority of scheduling outages is allocated according to the social function importance of the resource; the spatial attribute label is used to characterize the geographical area where the resource is located, and the spatial distance between resources is calculated using a method based on geographical coordinates; the technical attribute label is used to characterize the response characteristics and regulation accuracy of the resource, and the resource is classified into levels based on the response time and regulation accuracy; the environmental attribute label is used to characterize the carbon emission impact when the resource participates in regulation, and the resource is classified into levels based on the equivalent carbon emissions when providing regulation services.

[0042] First, select at least one of the following labels: social importance, spatial attribute, technological attribute, and environmental attribute, as a component of the multi-dimensional value label system. Then, assign a quantifiable label level to each selected label dimension to form a standardized and hierarchical multi-dimensional value label system. This multi-dimensional value label system is used to provide a unified attribute evaluation basis for subsequent resource clustering.

[0043] The core representational roles and quantification rules of various label dimensions are as follows: The social importance label characterizes the social function of the industry to which the flexible resource belongs and the severity of power outage consequences. Based on the importance of the resource's social function, ordinal values ​​reflecting the priority of dispatch interruptions are assigned to different resources, and these ordinal values ​​match the dispatch interruption priority. The spatial attribute label characterizes the geographical area where the flexible resource is located. A calculation method based on geographical coordinates is used to accurately measure the spatial distance between different flexible resources, thereby quantifying the geospatial attributes of the resource. The technical attribute label characterizes the response characteristics and regulation accuracy of the flexible resource. Based on the actual response time and regulation accuracy indicators of the resource, corresponding technical attribute levels are assigned to the resource, achieving hierarchical differentiation of response and regulation characteristics. The environmental attribute label characterizes the carbon emission impact of flexible resources participating in grid regulation. Based on the equivalent carbon emissions generated during the provision of regulation services, corresponding environmental attribute levels are assigned to the resource, quantifying the environmental impact of the resource's regulation behavior.

[0044] Secondly, all load-side flexibility resources are grouped according to the priority corresponding to the social importance label. This initial grouping operation ensures that all flexibility resources in the same priority group have similar levels of social importance, achieving preliminary classification of resources from the perspective of social function and avoiding excessive mixing of resources with different levels of social importance in subsequent clustering.

[0045] Furthermore, the multidimensional capability envelope and multidimensional value label corresponding to each flexible resource are used as the core input features for resource clustering. Based on these two core features, the label information and capability information of each flexible resource after the initial grouping are extracted and numerically transformed at any time in the future scheduling cycle. Finally, the two types of information are fused and represented into a multidimensional feature vector, realizing the integrated and numerical expression of resource attribute information and adjustment capability information, and providing standardized input data for subsequent clustering algorithm operations.

[0046] Furthermore, using the generated multidimensional feature vectors as input data, a clustering algorithm is employed to perform a secondary clustering operation on all flexibility resources within the same priority group after the initial grouping. The clustering algorithm is an algorithm that classifies and merges samples based on data feature similarity. Specifically, it quantifies the degree of feature similarity between resources by calculating the distance between the multidimensional feature vectors corresponding to different flexibility resources in the feature space, and groups resources whose feature similarity meets a preset threshold into the same class. For example, clustering algorithms based on feature distance, such as K-means, can be used.

[0047] Specifically, this clustering process uses a multi-dimensional feature space as the basis for judgment, accurately identifies the comprehensive similarity of different resources in the feature space, and based on the judgment result of comprehensive similarity, uniformly classifies flexible resources with similar features into the same resource cluster, thereby completing the construction of a virtual aggregate and realizing the intensive aggregation of flexible resources with matching features.

[0048] Furthermore, the aggregation and regulation capabilities of each virtual aggregate are robustly quantized to generate a multi-dimensional vector characterizing the overall regulation capability of the virtual aggregate, including: Obtain the multidimensional capability envelope of the individual resources of all the flexibility resources in the virtual aggregate, and calculate the nominal aggregation capability of each virtual aggregate. The nominal aggregation capability is obtained by summing the power adjustment capability values ​​of all the flexibility resources in the virtual aggregate at each time in the future scheduling period, and the nominal upward aggregation capability and the nominal downward aggregation capability are obtained respectively. A robust aggregation method based on chance constraints is used to calculate the reliable aggregation capability of each virtual aggregate, wherein the reliable aggregation capability characterizes the power value that the virtual aggregate can ensure the call at a given confidence level after comprehensively considering the uncertainty risk, and obtains the reliable up-aggregation capability and the reliable down-aggregation capability respectively. Calculate the duration aggregation value for each of the virtual aggregates, wherein the duration aggregation value is calculated by aggregating the duration values ​​in the multidimensional capability envelope of all the flexibility resources in the virtual aggregate using a power-weighted average method, and obtain the up-adjusted duration aggregation value and the down-adjusted duration aggregation value respectively. Calculate the ramp rate aggregation value for each of the virtual aggregates, wherein the ramp rate aggregation value is obtained by summing the ramp rate values ​​in the multidimensional capability envelope of all the flexibility resources in the virtual aggregate, and obtain the ramp rate aggregation value for upward adjustment and the ramp rate aggregation value for downward adjustment respectively. The nominal upward aggregation capability, nominal downward aggregation capability, reliable upward aggregation capability, reliable downward aggregation capability, upward sustainable aggregation value, downward sustainable aggregation value, upward climbing rate aggregation value, and downward climbing rate aggregation value are combined to generate a multi-dimensional vector of adjustment capability for each virtual aggregate.

[0049] First, the multidimensional capability envelopes corresponding to all flexible resources within the virtual aggregate are retrieved. Using these multidimensional capability envelopes as the data foundation, the nominal aggregation capability of each virtual aggregate is calculated. The nominal aggregation capability represents the overall power regulation potential formed by the aggregation of all flexible resources within the virtual aggregate under ideal conditions, without considering any uncertainties or risks. It represents the theoretically maximum range of power that the virtual aggregate can call upon. The calculation method involves summing the power regulation capability values ​​extracted from the multidimensional capability envelopes of all flexible resources within the virtual aggregate at each discrete moment in the future scheduling cycle. Based on the differences in the direction of power regulation, the nominal upward aggregation capability and nominal downward aggregation capability of the virtual aggregate at each moment are calculated respectively.

[0050] The nominal upward aggregation capability is the sum of the maximum upward power adjustment potential that all flexible resources within the virtual aggregate can provide at each discrete moment, representing the maximum power increase capability that the virtual aggregate can achieve under the theoretically optimal state. The nominal downward aggregation capability is the sum of the maximum downward power adjustment potential that all flexible resources within the virtual aggregate can provide at each discrete moment, representing the maximum power reduction capability that the virtual aggregate can achieve under the theoretically optimal state. The nominal aggregation capability can intuitively reflect the overall power adjustment potential of the virtual aggregate under the theoretical state.

[0051] Secondly, a robust aggregation method based on chance constraints is employed to calculate the reliable aggregation capability of each virtual aggregate. This method comprehensively considers various uncertainties related to the actual availability of flexible resources, treating the actual availability as a random variable and analyzing its probability distribution characteristics. By setting a confidence level threshold to define the acceptable risk level for scheduling, a quantitative method is used to determine the aggregation adjustment capability of the virtual aggregate that can be guaranteed to be called under this confidence level.

[0052] Reliable aggregation capability is the actual callable adjustment capability after eliminating various risk interferences. It is used to characterize the power value that the virtual aggregate can ensure the call is realized under the confidence level set by the scheduling agency after comprehensively considering various uncertain risks such as individual resource status prediction deviation, user response compliance rate, and communication reliability.

[0053] Specifically, a robust aggregation method based on chance constraints is used to calculate the reliable aggregation capability of each virtual aggregate, including: The actual callable capacity of each of the aforementioned flexibility resources is regarded as a random variable that follows a specific probability distribution, wherein the distribution parameters of the random variable are estimated based on historical response data and real-time operational status prediction deviations. For each virtual aggregate, its actual callable capacity after aggregation is the sum of random variables of all the flexibility resources in the virtual aggregate. According to the central limit theorem in probability theory, when the number of resources in the virtual aggregate is large enough, the actual callable capacity after aggregation approximately follows a normal distribution. A confidence level threshold is set, wherein the confidence level threshold is used to characterize the risk level acceptable to the scheduling agency, and the value of the confidence level threshold is greater than zero and less than or equal to one; Based on the probability distribution function of the actual callable capability after aggregation, the quantiles at the confidence level threshold are calculated, where the power value corresponding to the quantile is the reliable aggregation capability that can be guaranteed to be called at the confidence level. The reliable upward aggregation capability and the reliable downward aggregation capability are calculated separately for the upward and downward adjustment directions.

[0054] First, the actual available capacity of each flexibility resource is treated as a random variable following a specific probability distribution. The distribution parameters of this random variable are determined by combining historical response data statistical analysis with real-time operational status prediction deviation estimation. The actual available capacity of flexibility resources is affected by various uncertain factors such as user response compliance rate, equipment operational status fluctuations, and communication transmission reliability, and cannot be accurately represented by a fixed value. Therefore, it is defined as a random variable. Simultaneously, by combining historical response data from resource past adjustments and the deviation between the current real-time operational status and the predicted value, the parameters of the probability distribution followed by this random variable are accurately estimated, providing a data foundation for subsequent probabilistic analysis of aggregated capacity.

[0055] For each virtual aggregate, its actual callable capacity after aggregation is defined as the sum of the random variables corresponding to all flexible resources within the virtual aggregate. According to the Central Limit Theorem in probability theory, when the number of flexible resources contained in a virtual aggregate reaches a sufficient scale, the actual callable capacity after aggregation approximately follows a normal distribution. Specifically, the overall adjustment capacity of a virtual aggregate is jointly determined by the actual callable capacity of all its individual resources. Therefore, by summing the random variables of individual resources, the probabilistic characteristics of the actual callable capacity after aggregation can be obtained. The application of the Central Limit Theorem can simplify the complex aggregation probability distribution to a normal distribution when the number of resources is sufficient, reducing the computational complexity of subsequent reliable aggregation capacity.

[0056] Furthermore, a confidence level threshold is pre-defined. This threshold characterizes the acceptable risk level for the power grid dispatching agency during dispatching decisions, and its value is limited to a range greater than 0 and less than or equal to 1. The magnitude of the confidence level threshold is positively correlated with the dispatching agency's risk tolerance. The closer the value is to 1, the lower the dispatching agency's tolerance for dispatching risks and the higher the reliability requirements for aggregated capacity deployment; conversely, the closer the value is to 0, the higher the dispatching agency's tolerance for dispatching risks and the relatively lower the reliability requirements for aggregated capacity deployment. By defining the value range and meaning of this confidence level threshold, a unified risk assessment standard can be established for the calculation of reliable aggregated capacity.

[0057] Secondly, based on the probability distribution function of the actual callable capacity of the virtual aggregate after aggregation, the quantile of this probability distribution function at a preset confidence level threshold is calculated. The power value corresponding to this quantile is the reliable aggregation capacity that the virtual aggregate can reliably call at that confidence level. The above calculations are performed separately according to different adjustment directions: power increase and power decrease, ultimately yielding the reliable upward and downward aggregation capabilities of the virtual aggregate. The probability distribution function fully reflects all possible values ​​of the actual callable capacity of the virtual aggregate and their corresponding probabilities. By calculating the quantile at a specific confidence level, the minimum guaranteed value of the actual callable capacity of the virtual aggregate within that risk tolerance range can be determined. This value represents the aggregation adjustment capacity that the dispatching agency can stably call after excluding the influence of various uncertainties. Calculating separately according to the adjustment direction allows for precise quantification of the reliable capacity of the virtual aggregate across different adjustment dimensions.

[0058] Furthermore, an aggregation calculation is performed on the duration of the virtual aggregate's adjustment capability to obtain an aggregated value of the duration. This aggregated value of the duration is calculated using a power-weighted average method, with the power adjustment capability value of each flexible resource within the virtual aggregate as a weighting factor. The duration values ​​in the multi-dimensional capability envelope of all flexible resources at each time point within the future scheduling cycle are weighted and aggregated. Based on different adjustment directions (power increase and decrease), the aggregated values ​​of the virtual aggregate's upward and downward adjustment durations are obtained respectively, to accurately reflect the continuous duration of the virtual aggregate's overall adjustment capability.

[0059] Specifically, calculating the duration aggregation value for each of the virtual aggregates includes: Extract the up-adjustment duration value and down-adjustment duration value from the multi-dimensional capability envelope of each flexibility resource in the virtual aggregate, and extract the corresponding power adjustment capability value as a weighting factor. For the upward adjustment direction, the upward adjustment duration value of each of the aforementioned flexibility resources is multiplied by the corresponding power regulation capability value to obtain the weighted upward adjustment duration of each of the aforementioned flexibility resources. The weighted upward adjustment durations of all the aforementioned flexibility resources are summed and then divided by the sum of the power regulation capability values ​​of all the aforementioned flexibility resources to obtain the aggregated value of the upward adjustment duration. For the downward adjustment direction, the downward adjustment duration value of each flexibility resource is multiplied by the corresponding power regulation capability value to obtain the weighted downward adjustment duration of each flexibility resource. The weighted downward adjustment durations of all flexibility resources are summed and then divided by the sum of the power regulation capability values ​​of all flexibility resources to obtain the aggregated downward adjustment duration value.

[0060] First, from the multidimensional capability envelope corresponding to each flexible resource within the virtual aggregate, the duration value related to the adjustment direction is extracted, namely the up-adjustment duration value and the down-adjustment duration value. Simultaneously, the power adjustment capability value corresponding to the duration value at the same discrete moment is extracted. This power adjustment capability value is used as a weighting factor to differentiate the contribution of different resources' durations in subsequent aggregation calculations, ensuring that the aggregation results more closely reflect the weighted proportions of the actual resource adjustment capabilities.

[0061] Specifically, a weighted aggregation calculation is performed for the power adjustment direction. First, the adjustable duration value of each flexible resource within the virtual aggregate is multiplied by the power adjustment capability value of that resource at the corresponding time, i.e., the weighting factor, to obtain the weighted adjustable duration value of each flexible resource. Then, the weighted adjustable duration values ​​of all flexible resources within the virtual aggregate are summed to obtain the total weighted adjustable duration value. Finally, this total weighted adjustable duration value is divided by the sum of the power adjustment capability values ​​of all flexible resources within the virtual aggregate at the corresponding time, and the aggregated adjustable duration value of the virtual aggregate is obtained through this weighted average calculation method. This aggregated adjustable duration value comprehensively reflects the average duration for which the virtual aggregate can sustain its adjustment capability when adjusting upwards, and the weight allocation matches the actual adjustment capability of the resources.

[0062] Specifically, regarding the power reduction adjustment direction, firstly, the reduction duration value of each flexible resource within the virtual aggregate is multiplied by the power adjustment capability value of that resource at the corresponding time, i.e., the weighting factor, to obtain the weighted reduction duration value of each flexible resource. Then, the weighted reduction duration values ​​of all flexible resources within the virtual aggregate are summed to obtain the total weighted reduction duration value. Finally, this total weighted reduction duration value is divided by the sum of the power adjustment capability values ​​of all flexible resources within the virtual aggregate at the corresponding time, and a weighted average is used to obtain the aggregated reduction duration value of the virtual aggregate. This aggregated reduction duration value comprehensively reflects the average duration for which the virtual aggregate can sustain its adjustment capability during downward adjustment, and also uses weighting factors to reflect the differences in the contributions of different resources in the downward adjustment.

[0063] In summary, the above weighted average aggregation calculation method takes into account the individual characteristics of the duration of each flexible resource within the virtual aggregate. At the same time, by weighting the power adjustment capability value, it highlights the dominant influence of resources with stronger adjustment capabilities on the overall duration. The final aggregated duration value can comprehensively and accurately characterize the overall sustainable adjustment capability of the virtual aggregate under different adjustment directions.

[0064] Furthermore, the ramp rate of the virtual aggregate is aggregated to obtain an aggregated ramp rate value. Specifically, the calculation method for this aggregated ramp rate value is to sum the ramp rate values ​​extracted from the multi-dimensional capability envelope of all flexible resources within the virtual aggregate at each moment in the future scheduling cycle. Depending on the power adjustment direction, the aggregated ramp rate values ​​for upward adjustment and downward adjustment of the virtual aggregate are obtained respectively, so as to intuitively reflect the overall ability level of the virtual aggregate to achieve rapid power adjustment.

[0065] Finally, the eight indicators calculated above—nominal upward aggregation capability, nominal downward aggregation capability, reliable upward aggregation capability, reliable downward aggregation capability, upward sustainable aggregation value, downward sustainable aggregation value, upward ramp rate aggregation value, and downward ramp rate aggregation value—are integrated into a single dataset. This dataset generates a multi-dimensional vector of adjustment capability for each virtual aggregate. This multi-dimensional vector comprehensively and holistically characterizes the overall adjustment capability of the virtual aggregate within future scheduling cycles.

[0066] S30: Visualize the multidimensional vector to generate a system-level capability map to support scheduling decisions.

[0067] Finally, the multi-dimensional vectors of the regulation capabilities of all virtual aggregates are standardized and feature-mapped. Using spatiotemporal visualization technology as the core, and combined with the actual needs of power grid dispatching decisions, the quantitative indicators contained in the multi-dimensional vectors, such as nominal aggregation capability, reliable aggregation capability, sustainable aggregation value, and ramp rate aggregation value, are transformed into intuitive graphical representations. This generates a system-level capability map to support dispatching decisions, realizing an integrated display of the overall regulation capability of virtual aggregates in the time and space dimensions, and providing power grid dispatchers with a clear and comprehensive reference for resource regulation capabilities.

[0068] Specifically, the multidimensional vectors are visualized to generate a system-level capability map to support scheduling decisions, including: The adjustment capability multidimensional vector of all the virtual aggregates is fused with geographic information system data to associate corresponding spatial location information for each virtual aggregate, wherein the spatial location information includes at least latitude and longitude coordinates or administrative division codes; In the time dimension, the multidimensional vector of adjustment capability is segmented according to the time resolution within the future scheduling period to form the time-series data of the adjustment capability of each virtual aggregate in different time periods; Using an electronic map as the visualization base, a heat map is used to display the distribution of schedulable resources in different geographical areas within a specific time period. The color intensity of the heat map is proportional to the magnitude of reliable aggregation capability. A timeline control is integrated on the electronic map, and the heat map of different time periods can be dynamically switched by sliding the timeline control, so as to realize the dynamic evolution of the system-level capability map in the time dimension. An interactive click event is set on the electronic map. When an operator clicks on any geographical area corresponding to the virtual aggregate, an information panel pops up to display a detailed multi-dimensional vector of the virtual aggregate's adjustment capabilities.

[0069] First, the multi-dimensional vectors of regulation capabilities corresponding to all virtual aggregates are fused with data from an external geographic information system (GIS). The external GIS data is a standardized dataset containing spatially relevant information such as latitude and longitude coordinates, administrative division codes, and geographic boundaries of each geographic region, representing the spatial location characteristics and geographic attribute information of the geographic regions where each virtual aggregate is located. Since the multi-dimensional vectors of regulation capabilities only represent the overall quantitative indicators of the regulation capabilities of virtual aggregates and do not include spatial location information, while the system-level capability map needs to demonstrate the correlation between regulation capabilities and geographic distribution to meet the decision-making needs of power grid dispatch regarding the spatial distribution of resources, fusing the multi-dimensional vectors of regulation capabilities of virtual aggregates with the external GIS data allows for the organic combination of regulation capability data and spatial location information, supplementing the spatial attributes of the regulation capability data, thus laying a complete data foundation for subsequent geospatial visualization.

[0070] Specifically, this fusion process assigns and binds corresponding spatial location information to each virtual aggregate, which includes at least the latitude and longitude coordinates or administrative division code of the virtual aggregate's location. Through the above fusion and binding operations, a one-to-one correspondence between the virtual aggregate's adjustment capability data and geospatial information can be achieved, providing basic data support for subsequent geospatial-based visualization.

[0071] Secondly, at the time dimension, the multi-dimensional vector of adjustment capability corresponding to each virtual aggregate is processed by time-series segmentation according to the pre-set time resolution within the future scheduling cycle. Through time-series segmentation, time-series data of adjustment capability corresponding to different time periods within the future scheduling cycle for each virtual aggregate are formed, so that the adjustment capability data has both time and spatial attributes, realizing the refined decomposition and display of adjustment capability in the time dimension.

[0072] Furthermore, using a standard electronic map as the visualization base, a heatmap is employed to display the overall distribution of schedulable resources in a specific geographical area within a selected time period. The intensity of the heatmap's color is positively correlated with the reliable aggregation capability of virtual aggregates within the corresponding geographical area; the higher the reliable aggregation capability value, the darker the heatmap color for that area, thus visually reflecting the varying strengths and weaknesses of schedulable resource capabilities across different regions.

[0073] Secondly, a timeline control is integrated into the electronic map interface used for display. Operators can slide this timeline control to dynamically switch and display the heatmap distribution results corresponding to different time periods within future scheduling cycles, thereby realizing the dynamic evolution display of the system-level capability map in the time dimension, enabling schedulers to intuitively observe the overall trend of changes in schedulable resource capabilities over time.

[0074] Meanwhile, interactive click events are set for the geographical area corresponding to each virtual aggregate in the electronic map interface. When the dispatcher clicks on the geographical area corresponding to any virtual aggregate in the electronic map, an information panel automatically pops up. This information panel displays all detailed multi-dimensional vector information of the adjustment capabilities corresponding to the clicked virtual aggregate, including quantitative indicators such as nominal aggregation capability, reliable aggregation capability, sustainable aggregation value, and ramp rate aggregation value, thereby providing dispatchers with accurate and comprehensive information on local resource adjustment capabilities.

[0075] In summary, through the above steps, the visualization of the multi-dimensional vector of regulation capability is completed, ultimately generating a system-level capability map that integrates spatial distribution, temporal evolution, and interactive query, providing power grid dispatchers with an intuitive, efficient, and comprehensive decision support tool.

[0076] In addition, using electronic maps as the visualization base, heat maps are used to display the distribution of schedulable resources in different geographical areas over a specific time period, followed by: A confidence level adjustment control is set in the visual interface, wherein the confidence level adjustment control is used to receive confidence level adjustment instructions input by the operator; When the value of the confidence level adjustment control changes, the reliable aggregation capability of each virtual aggregate is recalculated in real time, and the color distribution of the heatmap is updated synchronously. At the same time, the reliable aggregation capability value displayed in the information panel is updated, enabling operators to intuitively compare the distribution of schedulable resources under different risk tolerance levels.

[0077] Specifically, using an electronic map as the visualization base, a heat map is used to display the distribution of schedulable resources in different geographical areas over a specific time period, followed by: First, a confidence level adjustment control is added to the visual interface. This confidence level adjustment control is used to receive confidence level adjustment instructions input by dispatching operators, providing operators with an interactive interface to manually adjust risk tolerance. This allows operators to independently set the corresponding confidence level threshold according to the dispatching scenario, dispatching safety level, and actual operational needs.

[0078] When an operator changes the confidence level value using the confidence level adjustment controls, the visualization system re-executes the reliable aggregation capability calculation process for each virtual aggregate in real time based on the updated confidence level threshold. After calculation, the color distribution of the heatmap in the electronic map is updated synchronously, ensuring a positive correlation between the heatmap's color intensity and the updated reliable aggregation capability value. This reflects the real-time distribution of the reliable adjustment capability of schedulable resources in each area under the current confidence level. Simultaneously with updating the heatmap's color distribution, the reliable aggregation capability value displayed in the information panel is also updated. When an operator clicks on the geographical area corresponding to any virtual aggregate, the pop-up information panel directly displays the latest reliable aggregation capability data matching the current confidence level.

[0079] Through the aforementioned synchronous update mechanism, operators can intuitively compare the distribution, capacity, and changing trends of schedulable resources under different risk tolerance levels, providing dynamic support for flexibly adjusting scheduling strategies and adapting to scheduling needs of different security levels.

[0080] In summary, the embodiments of this application have at least the following technical effects: This invention firstly constructs a multi-state finite automaton model to achieve a unified and standardized description of the operating states and constraints of flexible resources on different types of load sides, breaking through the limitations of traditional assessment methods that only target a single resource type and use simplified models. Secondly, by generating multi-dimensional capability envelopes for individual resources that include power regulation capability, duration, and ramp rate, this invention overcomes the limitation of traditional assessments that only provide static power indicators, achieving multi-dimensional dynamic quantification of resource adjustability and providing comprehensive resource capability data support for the scheduling needs of power systems at different time scales. Thirdly, this invention relies on a multi-dimensional value tagging system to cluster and group resources and construct virtual aggregates. Simultaneously, it employs robust quantification methods to account for uncertainties such as user behavior and communication reliability, generating a multi-dimensional vector of the virtual aggregate's regulation capability at a given confidence level. This avoids the scheduling risks brought about by deterministic aggregation in traditional resource aggregation, achieving an upgrade from technology-driven aggregation to value-oriented aggregation, adapting to the multi-objective scheduling needs of new power systems.

[0081] Finally, this invention generates a system-level capability map by processing the spatiotemporal visualization of multidimensional vectors, transforming multi-source heterogeneous quantitative data into intuitive visual decision-making information, improving the efficiency of dispatchers' understanding of the overall resource regulation capability, and providing efficient decision support for precise power grid dispatch and virtual power plant operation.

[0082] Example 2, as Figure 2As shown, based on the same inventive concept as the flexible resource adjustment capability assessment method based on tagging and risk analysis provided in Embodiment 1, this embodiment of the invention also provides a flexible resource adjustment capability assessment system based on tagging and risk analysis, including: Individual resource capability analysis module 11 is used to collect real-time operating data of load-side flexibility resources, construct a multi-state finite automaton model to describe the operating state of resources, predict the adjustable capability of each flexibility resource in the future scheduling cycle based on the multi-state finite automaton model, and generate a multi-dimensional capability envelope of individual resources. The virtual aggregate construction module 12 is used to construct virtual aggregates by clustering and grouping each flexible resource based on the multi-dimensional value tags of each flexible resource and the multi-dimensional capability envelope of the individual resources, and to robustly quantify the aggregation and adjustment capabilities of each virtual aggregate to generate a multi-dimensional vector characterizing the overall adjustment capability of the virtual aggregate. The visualization decision support module 13 is used to visualize the multi-dimensional vector and generate a system-level capability map to support scheduling decisions.

[0083] The individual resource capability analysis module 11 is specifically used for: Specifically, real-time operational data of load-side flexibility resources are collected, a multi-state finite automaton model is constructed to describe the operational state of the resources, and the adjustable capabilities of each flexibility resource in future scheduling cycles are predicted based on the multi-state finite automaton model, generating a multi-dimensional capability envelope for each individual resource, including: The collected load-side flexibility resources include at least one of industrial and commercial adjustable loads, electric vehicles, and energy storage systems. By extracting the state switching nodes and parameter change thresholds of each flexibility resource during operation, a multi-state finite automaton model is constructed. The multi-state finite automaton model contains multiple dynamic operating states, and the transition conditions between each dynamic operating state include at least one of time threshold, power change value, or external excitation signal. After instantiating the multi-state finite automaton model for each of the aforementioned flexible resources, the physical constraint parameters, dynamic operating state, and response constraint parameters of each of the aforementioned flexible resources are identified and stored. The physical constraint parameters include at least the rated power, the upper limit of power adjustment, and the lower limit of power adjustment. The dynamic operating state includes at least the operating mode and instantaneous power at the current moment. The response constraint parameters include at least the maximum ramp rate, the minimum stable operating time, and the minimum downtime. The multi-state finite automaton model of each of the aforementioned flexible resources is set to autonomous operation mode, and the predicted external driving data sequence is input. The power output sequence from the current time to the end of the scheduling cycle is obtained through simulation, and the baseline load curve of each of the aforementioned flexible resources is generated. Based on the baseline load curve, for each discrete moment in the future scheduling cycle, the maximum upward adjustment potential and the maximum downward adjustment potential of each flexibility resource at that discrete moment are calculated using a constrained instantaneous power boundary search algorithm. Forward simulation is performed based on the multi-state finite automaton model to calculate the maximum duration for which the adjustment potential can be continuously invoked. At the same time, based on the static parameters in the multi-state finite automaton model and the adjustment space of the current running point, the maximum available ramp rate at that discrete moment is calculated. For each discrete moment discretized at a set time resolution within the future scheduling period, iterative calculations are repeatedly performed to generate the multidimensional capability envelope of each flexibility resource. The multidimensional capability envelope includes at least a power regulation capability value, a duration value, and a ramp rate value.

[0084] The virtual aggregate construction module 12 is specifically used for: First, based on the multi-dimensional value tags of each flexibility resource, and combined with the multi-dimensional capability envelope of each individual resource, the flexibility resources are clustered and grouped to construct a virtual aggregate, including: At least one label dimension is selected as a component of the multi-dimensional value label system, wherein the label dimension includes at least one of social importance label, spatial attribute label, technological attribute label and environmental attribute label, and a quantifiable label level is set for each label dimension; All the aforementioned flexibility resources are initially grouped according to the priority of the aforementioned social importance tags to ensure that the flexibility resources within the same priority group have similar levels of social importance. The multidimensional capability envelope and multidimensional value label of each individual resource of the flexibility resource are used together as the input features for clustering and grouping. The label information and capability information of each flexibility resource after the initial grouping at any time are represented as a multidimensional feature vector. Using the multidimensional feature vector as input, a clustering algorithm is used to perform secondary clustering on the flexibility resources within the same priority group. Based on the comprehensive similarity of resources in the multidimensional feature space, resources with similar features are assigned to the same virtual aggregate, thus completing the construction of the virtual aggregate.

[0085] The social importance label is used to characterize the social function of the industry to which the resource belongs and the severity of the consequences of power outages, and the ordinal value reflecting the priority of scheduling outages is allocated according to the social function importance of the resource; the spatial attribute label is used to characterize the geographical area where the resource is located, and the spatial distance between resources is calculated using a method based on geographical coordinates; the technical attribute label is used to characterize the response characteristics and regulation accuracy of the resource, and the resource is classified into levels based on the response time and regulation accuracy; the environmental attribute label is used to characterize the carbon emission impact when the resource participates in regulation, and the resource is classified into levels based on the equivalent carbon emissions when providing regulation services.

[0086] Furthermore, the aggregation and regulation capabilities of each virtual aggregate are robustly quantized to generate a multi-dimensional vector characterizing the overall regulation capability of the virtual aggregate, including: Obtain the multidimensional capability envelope of the individual resources of all the flexibility resources in the virtual aggregate, and calculate the nominal aggregation capability of each virtual aggregate. The nominal aggregation capability is obtained by summing the power adjustment capability values ​​of all the flexibility resources in the virtual aggregate at each time in the future scheduling period, and the nominal upward aggregation capability and the nominal downward aggregation capability are obtained respectively. A robust aggregation method based on chance constraints is used to calculate the reliable aggregation capability of each virtual aggregate, wherein the reliable aggregation capability characterizes the power value that the virtual aggregate can ensure the call at a given confidence level after comprehensively considering the uncertainty risk, and obtains the reliable up-aggregation capability and the reliable down-aggregation capability respectively. Calculate the duration aggregation value for each of the virtual aggregates, wherein the duration aggregation value is calculated by aggregating the duration values ​​in the multidimensional capability envelope of all the flexibility resources in the virtual aggregate using a power-weighted average method, and obtain the up-adjusted duration aggregation value and the down-adjusted duration aggregation value respectively. Calculate the ramp rate aggregation value for each of the virtual aggregates, wherein the ramp rate aggregation value is obtained by summing the ramp rate values ​​in the multidimensional capability envelope of all the flexibility resources in the virtual aggregate, and obtain the ramp rate aggregation value for upward adjustment and the ramp rate aggregation value for downward adjustment respectively. The nominal upward aggregation capability, nominal downward aggregation capability, reliable upward aggregation capability, reliable downward aggregation capability, upward sustainable aggregation value, downward sustainable aggregation value, upward climbing rate aggregation value, and downward climbing rate aggregation value are combined to generate a multi-dimensional vector of adjustment capability for each virtual aggregate.

[0087] Specifically, a robust aggregation method based on chance constraints is used to calculate the reliable aggregation capability of each virtual aggregate, including: The actual callable capacity of each of the aforementioned flexibility resources is regarded as a random variable that follows a specific probability distribution, wherein the distribution parameters of the random variable are estimated based on historical response data and real-time operational status prediction deviations. For each virtual aggregate, its actual callable capacity after aggregation is the sum of random variables of all the flexibility resources in the virtual aggregate. According to the central limit theorem in probability theory, when the number of resources in the virtual aggregate is large enough, the actual callable capacity after aggregation approximately follows a normal distribution. A confidence level threshold is set, wherein the confidence level threshold is used to characterize the risk level acceptable to the scheduling agency, and the value of the confidence level threshold is greater than zero and less than or equal to one; Based on the probability distribution function of the actual callable capability after aggregation, the quantiles at the confidence level threshold are calculated, where the power value corresponding to the quantile is the reliable aggregation capability that can be guaranteed to be called at the confidence level. The reliable upward aggregation capability and the reliable downward aggregation capability are calculated separately for the upward and downward adjustment directions.

[0088] Specifically, calculating the duration aggregation value for each of the virtual aggregates includes: Extract the up-adjustment duration value and down-adjustment duration value from the multi-dimensional capability envelope of each flexibility resource in the virtual aggregate, and extract the corresponding power adjustment capability value as a weighting factor. For the upward adjustment direction, the upward adjustment duration value of each of the aforementioned flexibility resources is multiplied by the corresponding power regulation capability value to obtain the weighted upward adjustment duration of each of the aforementioned flexibility resources. The weighted upward adjustment durations of all the aforementioned flexibility resources are summed and then divided by the sum of the power regulation capability values ​​of all the aforementioned flexibility resources to obtain the aggregated value of the upward adjustment duration. For the downward adjustment direction, the downward adjustment duration value of each flexibility resource is multiplied by the corresponding power regulation capability value to obtain the weighted downward adjustment duration of each flexibility resource. The weighted downward adjustment durations of all flexibility resources are summed and then divided by the sum of the power regulation capability values ​​of all flexibility resources to obtain the aggregated downward adjustment duration value.

[0089] The visualization decision support module 13 is specifically used for: The multidimensional vectors are visualized to generate a system-level capability map to support scheduling decisions, including: The adjustment capability multidimensional vector of all the virtual aggregates is fused with geographic information system data to associate corresponding spatial location information for each virtual aggregate, wherein the spatial location information includes at least latitude and longitude coordinates or administrative division codes; In the time dimension, the multidimensional vector of adjustment capability is segmented according to the time resolution within the future scheduling period to form the time-series data of the adjustment capability of each virtual aggregate in different time periods; Using an electronic map as the visualization base, a heat map is used to display the distribution of schedulable resources in different geographical areas within a specific time period. The color intensity of the heat map is proportional to the magnitude of reliable aggregation capability. A timeline control is integrated on the electronic map, and the heat map of different time periods can be dynamically switched by sliding the timeline control, so as to realize the dynamic evolution of the system-level capability map in the time dimension. An interactive click event is set on the electronic map. When an operator clicks on any geographical area corresponding to the virtual aggregate, an information panel pops up to display a detailed multi-dimensional vector of the virtual aggregate's adjustment capabilities.

[0090] Specifically, using an electronic map as the visualization base, a heat map is used to display the distribution of schedulable resources in different geographical areas over a specific time period, followed by: A confidence level adjustment control is set in the visual interface, wherein the confidence level adjustment control is used to receive confidence level adjustment instructions input by the operator; When the value of the confidence level adjustment control changes, the reliable aggregation capability of each virtual aggregate is recalculated in real time, and the color distribution of the heatmap is updated synchronously. At the same time, the reliable aggregation capability value displayed in the information panel is updated, enabling operators to intuitively compare the distribution of schedulable resources under different risk tolerance levels.

Claims

1. A method for assessing flexible resource adjustment capabilities based on labeling and risk analysis, characterized in that, include: Collect real-time operational data of load-side flexibility resources, construct a multi-state finite automaton model to describe the operational status of the resources, predict the adjustable capabilities of each flexibility resource in future scheduling cycles based on the multi-state finite automaton model, and generate a multi-dimensional capability envelope for individual resources. Based on the multi-dimensional value tags of each flexible resource, and combined with the multi-dimensional capability envelope of the individual resources, the flexible resources are clustered and grouped to construct virtual aggregates. The aggregation and adjustment capabilities of each virtual aggregate are robustly quantified to generate a multi-dimensional vector characterizing the overall adjustment capability of the virtual aggregate. The multidimensional vectors are visualized to generate a system-level capability map to support scheduling decisions.

2. The method for assessing flexible resource adjustment capabilities based on tagging and risk analysis according to claim 1, characterized in that, Real-time operational data of load-side flexibility resources are collected, and a multi-state finite automaton model is constructed to describe the operational state of the resources. Based on the multi-state finite automaton model, the adjustable capabilities of each flexibility resource in future scheduling cycles are predicted, and a multi-dimensional capability envelope of individual resources is generated, including: The collected load-side flexibility resources include at least one of industrial and commercial adjustable loads, electric vehicles, and energy storage systems. By extracting the state switching nodes and parameter change thresholds of each flexibility resource during operation, a multi-state finite automaton model is constructed. The multi-state finite automaton model contains multiple dynamic operating states, and the transition conditions between each dynamic operating state include at least one of time threshold, power change value, or external excitation signal. After instantiating the multi-state finite automaton model for each of the aforementioned flexible resources, the physical constraint parameters, dynamic operating state, and response constraint parameters of each of the aforementioned flexible resources are identified and stored. The physical constraint parameters include at least the rated power, the upper limit of power adjustment, and the lower limit of power adjustment. The dynamic operating state includes at least the operating mode and instantaneous power at the current moment. The response constraint parameters include at least the maximum ramp rate, the minimum stable operating time, and the minimum downtime. The multi-state finite automaton model of each of the aforementioned flexible resources is set to autonomous operation mode, and the predicted external driving data sequence is input. The power output sequence from the current time to the end of the scheduling cycle is obtained through simulation, and the baseline load curve of each of the aforementioned flexible resources is generated. Based on the baseline load curve, for each discrete moment in the future scheduling cycle, the maximum upward adjustment potential and the maximum downward adjustment potential of each flexibility resource at that discrete moment are calculated using a constrained instantaneous power boundary search algorithm. Forward simulation is performed based on the multi-state finite automaton model to calculate the maximum duration for which the adjustment potential can be continuously invoked. At the same time, based on the static parameters in the multi-state finite automaton model and the adjustment space of the current running point, the maximum available ramp rate at that discrete moment is calculated. For each discrete moment discretized at a set time resolution within the future scheduling period, iterative calculations are repeatedly performed to generate the multidimensional capability envelope of each flexibility resource. The multidimensional capability envelope includes at least a power regulation capability value, a duration value, and a ramp rate value.

3. The method for assessing flexible resource adjustment capabilities based on tagging and risk analysis according to claim 1, characterized in that, Based on the multi-dimensional value tags of each flexibility resource, and combined with the multi-dimensional capability envelope of each individual resource, the flexibility resources are clustered and grouped to construct a virtual aggregate, including: At least one label dimension is selected as a component of the multi-dimensional value label system, wherein the label dimension includes at least one of social importance label, spatial attribute label, technological attribute label and environmental attribute label, and a quantifiable label level is set for each label dimension; All the aforementioned flexibility resources are initially grouped according to the priority of the aforementioned social importance tags to ensure that the flexibility resources within the same priority group have similar levels of social importance. The multidimensional capability envelope and multidimensional value label of each individual resource of the flexibility resource are used together as the input features for clustering and grouping. The label information and capability information of each flexibility resource after the initial grouping at any time are represented as a multidimensional feature vector. Using the multidimensional feature vector as input, a clustering algorithm is used to perform secondary clustering on the flexibility resources within the same priority group. Based on the comprehensive similarity of resources in the multidimensional feature space, resources with similar features are assigned to the same virtual aggregate, thus completing the construction of the virtual aggregate.

4. The method for assessing flexible resource adjustment capabilities based on tagging and risk analysis according to claim 3, characterized in that, The social importance label is used to characterize the social function of the industry to which the resource belongs and the severity of the consequences of power outages. The ordinal value reflecting the priority of scheduling outages is assigned according to the social function importance of the resource. The spatial attribute label is used to characterize the geographical area where the resource is located. The spatial distance between resources is calculated using a method based on geographical coordinates. The technical attribute label is used to characterize the response characteristics and regulation accuracy of the resource. The resource is classified into levels based on its response time and regulation accuracy. The environmental attribute labels are used to characterize the carbon emission impact when resources participate in regulation, and are classified based on the equivalent carbon emissions when resources provide regulation services.

5. The method for assessing flexible resource adjustment capabilities based on tagging and risk analysis according to claim 1, characterized in that, Robust quantization is performed on the aggregation and regulation capabilities of each virtual aggregate to generate a multi-dimensional vector characterizing the overall regulation capability of the virtual aggregates, including: Obtain the multidimensional capability envelope of the individual resources of all the flexibility resources in the virtual aggregate, and calculate the nominal aggregation capability of each virtual aggregate. The nominal aggregation capability is obtained by summing the power adjustment capability values ​​of all the flexibility resources in the virtual aggregate at each time in the future scheduling period, and the nominal upward aggregation capability and the nominal downward aggregation capability are obtained respectively. A robust aggregation method based on chance constraints is used to calculate the reliable aggregation capability of each virtual aggregate, wherein the reliable aggregation capability characterizes the power value that the virtual aggregate can ensure the call at a given confidence level after comprehensively considering the uncertainty risk, and obtains the reliable up-aggregation capability and the reliable down-aggregation capability respectively. Calculate the duration aggregation value for each of the virtual aggregates, wherein the duration aggregation value is calculated by aggregating the duration values ​​in the multidimensional capability envelope of all the flexibility resources in the virtual aggregate using a power-weighted average method, and obtain the up-adjusted duration aggregation value and the down-adjusted duration aggregation value respectively. Calculate the ramp rate aggregation value for each of the virtual aggregates, wherein the ramp rate aggregation value is obtained by summing the ramp rate values ​​in the multidimensional capability envelope of all the flexibility resources in the virtual aggregate, and obtain the ramp rate aggregation value for upward adjustment and the ramp rate aggregation value for downward adjustment respectively. The nominal upward aggregation capability, nominal downward aggregation capability, reliable upward aggregation capability, reliable downward aggregation capability, upward sustainable aggregation value, downward sustainable aggregation value, upward climbing rate aggregation value, and downward climbing rate aggregation value are combined to generate a multi-dimensional vector of adjustment capability for each virtual aggregate.

6. The method for assessing flexible resource adjustment capabilities based on tagging and risk analysis according to claim 5, characterized in that, The reliable aggregation capability of each virtual aggregate is calculated using a chance-constrained robust aggregation method, including: The actual callable capacity of each of the aforementioned flexibility resources is regarded as a random variable that follows a specific probability distribution, wherein the distribution parameters of the random variable are estimated based on historical response data and real-time operational status prediction deviations. For each virtual aggregate, its actual callable capacity after aggregation is the sum of random variables of all the flexibility resources in the virtual aggregate. According to the central limit theorem in probability theory, when the number of resources in the virtual aggregate is large enough, the actual callable capacity after aggregation approximately follows a normal distribution. A confidence level threshold is set, wherein the confidence level threshold is used to characterize the risk level acceptable to the scheduling agency, and the value of the confidence level threshold is greater than zero and less than or equal to one; Based on the probability distribution function of the actual callable capability after aggregation, the quantiles at the confidence level threshold are calculated, where the power value corresponding to the quantile is the reliable aggregation capability that can be guaranteed to be called at the confidence level. The reliable upward aggregation capability and the reliable downward aggregation capability are calculated separately for the upward and downward adjustment directions.

7. The method for assessing flexible resource adjustment capabilities based on tagging and risk analysis according to claim 5, characterized in that, Calculating the duration aggregation value for each of the virtual aggregates includes: Extract the up-adjustment duration value and down-adjustment duration value from the multi-dimensional capability envelope of each flexibility resource in the virtual aggregate, and extract the corresponding power adjustment capability value as a weighting factor. For the upward adjustment direction, the upward adjustment duration value of each of the aforementioned flexibility resources is multiplied by the corresponding power regulation capability value to obtain the weighted upward adjustment duration of each of the aforementioned flexibility resources. The weighted upward adjustment durations of all the aforementioned flexibility resources are summed and then divided by the sum of the power regulation capability values ​​of all the aforementioned flexibility resources to obtain the aggregated value of the upward adjustment duration. For the downward adjustment direction, the downward adjustment duration value of each flexibility resource is multiplied by the corresponding power regulation capability value to obtain the weighted downward adjustment duration of each flexibility resource. The weighted downward adjustment durations of all flexibility resources are summed and then divided by the sum of the power regulation capability values ​​of all flexibility resources to obtain the aggregated downward adjustment duration value.

8. The method for assessing flexible resource adjustment capabilities based on tagging and risk analysis according to claim 1, characterized in that, The multidimensional vectors are visualized to generate a system-level capability map to support scheduling decisions, including: The adjustment capability multidimensional vector of all the virtual aggregates is fused with geographic information system data to associate corresponding spatial location information for each virtual aggregate, wherein the spatial location information includes at least latitude and longitude coordinates or administrative division codes; In the time dimension, the multidimensional vector of adjustment capability is segmented according to the time resolution within the future scheduling period to form the time-series data of the adjustment capability of each virtual aggregate in different time periods; Using an electronic map as the visualization base, a heat map is used to display the distribution of schedulable resources in different geographical areas within a specific time period. The color intensity of the heat map is proportional to the magnitude of the reliable aggregation capability. A timeline control is integrated on the electronic map, and the heat map of different time periods can be dynamically switched by sliding the timeline control, so as to realize the dynamic evolution of the system-level capability map in the time dimension. An interactive click event is set on the electronic map. When an operator clicks on any geographical area corresponding to the virtual aggregate, an information panel pops up to display a detailed multi-dimensional vector of the virtual aggregate's adjustment capabilities.

9. The method for assessing flexible resource adjustment capabilities based on tagging and risk analysis according to claim 8, characterized in that, Using an electronic map as the visualization base, a heat map is used to display the distribution of schedulable resources in different geographical areas over a specific time period. This is followed by: A confidence level adjustment control is set in the visual interface, wherein the confidence level adjustment control is used to receive confidence level adjustment instructions input by the operator; When the value of the confidence level adjustment control changes, the reliable aggregation capability of each virtual aggregate is recalculated in real time, and the color distribution of the heatmap is updated synchronously. At the same time, the reliable aggregation capability value displayed in the information panel is updated, enabling operators to intuitively compare the distribution of schedulable resources under different risk tolerance levels.

10. A flexible resource adjustment capability assessment system based on tagging and risk analysis, characterized in that, The method for evaluating flexible resource adjustment capabilities based on labeling and risk analysis as described in any one of claims 1-9 includes: The individual resource capability analysis module is used to collect real-time operating data of load-side flexibility resources, construct a multi-state finite automaton model to describe the operating state of resources, predict the adjustable capability of each flexibility resource in the future scheduling cycle based on the multi-state finite automaton model, and generate a multi-dimensional capability envelope of individual resources. The virtual aggregate construction module is used to cluster and group the flexible resources based on the multi-dimensional value tags of each flexible resource and the multi-dimensional capability envelope of the individual resources to construct virtual aggregates, and to robustly quantify the aggregation and adjustment capabilities of each virtual aggregate to generate a multi-dimensional vector characterizing the overall adjustment capability of the virtual aggregate. The visualization decision support module is used to visualize the multi-dimensional vectors and generate a system-level capability map to support scheduling decisions.

Images