Comprehensive energy system multi-dimensional resource adequacy assessment method, device, equipment and medium

Abstract

The invention provides an integrated energy system multi-dimensional resource adequacy assessment method, device and equipment and a medium, and relates to the technical field of integrated energy system optimization. The method comprises the following steps: acquiring a time sequence power curve of a target transaction resource and uncertainty data; performing morphological decomposition on the time sequence power curve, and constructing a plurality of adequacy evaluation indexes; calculating a dynamic objective weight and a basic subjective weight of each adequacy evaluation index; fusing the dynamic objective weight and the basic subjective weight through an attention mechanism to obtain an intermediate weight of each adequacy evaluation index; determining cloud model parameters according to the uncertainty data, and optimizing the intermediate weight through the cloud model to obtain the final weight of each adequacy evaluation index; and based on the final weight of each adequacy evaluation index, fusing each adequacy evaluation index to obtain a dynamic adequacy evaluation value of the target transaction resource. According to the invention, the resource adequacy of the energy system can be integrated more accurately.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system optimization technology, and in particular to a method, apparatus, equipment and medium for multi-dimensional resource adequacy assessment of integrated energy systems. Background Technology

[0002] With the large-scale application of trading resources such as diversified energy storage, virtual energy storage, and distributed power sources in integrated energy systems, multi-level energy markets (medium- and long-term markets, day-ahead markets, and real-time / ancillary service markets) are placing higher demands on the accuracy and adaptability of resource sufficiency assessment. Resource sufficiency assessment, as a core technology for ensuring the supply and demand balance of energy systems and optimizing market trading decisions, directly impacts system regulation efficiency and market operational stability. Currently, the industry has developed various assessment technologies for the resource sufficiency of integrated energy systems: single-timescale assessment techniques based on traditional signal decomposition, single-dimensional resource sufficiency modeling techniques, and solution algorithms based on a fusion of static subjective and objective factors.

[0003] Single-time-scale assessment techniques based on traditional signal decomposition often employ wavelet decomposition or Fourier transform to process the output or load curves of energy resources, and then combine this with market demand at the corresponding time scale to conduct sufficiency analysis. The advantages of this approach are its mature signal decomposition theory, wide applicability in energy resource assessment scenarios with relatively high stationarity, low computational complexity, and ease of engineering implementation. However, the output / load curves of resources such as photovoltaic, wind power, and adjustable loads in integrated energy systems exhibit significant non-stationarity and multi-time-scale fluctuations. Traditional wavelet decomposition suffers from boundary effects and insufficient noise resistance, while Fourier transform cannot effectively separate characteristic components at different time scales. Furthermore, single-time-scale assessment models can only match market demand at a certain level, failing to consider both medium- and long-term market capacity planning needs and real-time market rapid adjustment requirements, resulting in poor time-scale adaptability and difficulty in supporting cross-market collaborative decision-making.

[0004] Single-dimensional resource adequacy modeling is currently the mainstream application in the industry, with most technologies focusing on constructing evaluation indicators based on a single dimension such as "capacity adequacy" or "power adequacy." For example, some schemes assess adequacy by calculating the ratio of available resource capacity to market demand capacity, while others use the instantaneous adjustment power of resources as the core indicator. The advantage of this type of technology is that the indicator definition is simple, the physical meaning is clear, and it facilitates rapid acquisition of evaluation results, meeting basic needs in simple adjustment scenarios in the early stages of energy market development. However, with the diversification of adjustment demands in integrated energy systems, relying solely on a single-dimensional indicator is no longer sufficient to comprehensively characterize resource adjustment capabilities. Resources with ample capacity but insufficient response speed are ill-suited to the rapid dispatch demands of the real-time ancillary service market; resources with adequate power but insufficient capacity reserves cannot support continuous power supply guarantees in the medium to long term. This singularity in evaluation dimensions can easily lead to misjudgments of resource adjustment capabilities by market participants, increasing the risk of system supply-demand imbalance.

[0005] Static, subjective-objective fusion algorithms primarily calculate comprehensive adequacy by combining the entropy weight method and the analytic hierarchy process (AHP) in a fixed ratio. The advantage of this type of algorithm lies in balancing the objectivity of data-driven approaches with the subjectivity of expert experience, thus improving the reliability of the evaluation results compared to purely objective or purely subjective methods. However, the resource output and market demand of integrated energy systems exhibit significant time-varying and uncertainties, such as random fluctuations in photovoltaic output due to weather conditions and demand changes caused by load forecasting errors. Traditional algorithms, using fixed weight ratios, cannot dynamically adapt to real-time system changes. Furthermore, the algorithms do not quantify the uncertainty of the weights, only outputting a fixed value for comprehensive adequacy, which fails to reflect the probabilistic characteristics of resource regulation capabilities. This results in a low degree of alignment between the evaluation results and actual scenarios, especially in extreme fluctuation scenarios, easily leading to misjudgments of adequacy and consequently, regulation failures or resource waste.

[0006] In summary, existing technologies have significant shortcomings in terms of time scale adaptability, completeness of assessment dimensions, and dynamic integration of subjective and objective factors, and cannot meet the needs of multi-level market collaboration, multi-dimensional capability assessment, and adaptation to high-uncertainty scenarios in integrated energy systems. Summary of the Invention

[0007] This invention provides a method, apparatus, equipment, and medium for assessing the multi-dimensional resource adequacy of an integrated energy system, in order to solve the problems of poor time scale adaptability, low decomposition accuracy, single assessment dimension, and separation of subjective and objective factors in existing integrated energy system resource adequacy models.

[0008] In a first aspect, embodiments of the present invention provide a method for assessing the multi-dimensional resource adequacy of a comprehensive energy system, including: Obtain the time-series power curves of the target trading resources in the integrated energy system, as well as the uncertainty data of the integrated energy system; Based on a preset time scale, the time-series power curve is morphologically decomposed, and multiple adequacy evaluation indicators are constructed based on the decomposition results. The dynamic objective weights of each adequacy assessment index are calculated by using the sliding time window entropy, and the basic subjective weights of each adequacy assessment index are calculated by using the analytic hierarchy process. The dynamic objective weight and basic subjective weight of each adequacy assessment indicator are learned through an attention mechanism, and the dynamic objective weight and basic subjective weight of each adequacy assessment indicator are fused according to the weighting coefficient to obtain the intermediate weight of each adequacy assessment indicator. The cloud model parameters are determined based on the uncertainty data, and the intermediate weights of each adequacy assessment index are optimized through the cloud model to obtain the final weights of each adequacy assessment index. Based on the final weights of each sufficiency assessment indicator, the indicators are integrated to obtain the dynamic sufficiency assessment value of the target transaction resource.

[0009] In one possible implementation, the calculation of the dynamic objective weights of each adequacy assessment index using the sliding time window entropy includes: Based on the time-series power curve, calculate the index sequence data for each adequacy evaluation index; For each adequacy assessment index, determine the time window corresponding to the current moment, and calculate the entropy value of the index sequence data within the time window; Based on the entropy value, determine the dynamic objective weight of each adequacy assessment indicator.

[0010] In one possible implementation, the calculation of the basic subjective weights of each adequacy assessment index using the analytic hierarchy process includes: Calculate the interval judgment matrix using the analytic hierarchy process; Based on the interval judgment matrix, determine the basic subjective weight interval for each adequacy assessment indicator, and determine the basic subjective weight based on the median of the basic subjective weight interval.

[0011] In one possible implementation, the learning of the weighted coefficients of the dynamic objective weights and basic subjective weights of each sufficiency assessment indicator through an attention mechanism includes: The dynamic objective weights and basic subjective weights of each adequacy assessment index are used as a set of feature vectors and input into a single-layer perceptron to calculate the attention score. The attention scores of each sufficiency assessment indicator are normalized to obtain the dynamic objective weighting coefficients of each sufficiency assessment indicator. Based on the dynamic objective weighting coefficients of each sufficiency assessment indicator, the basic subjective weighting coefficients of each sufficiency assessment indicator are determined.

[0012] In one possible implementation, the uncertainty data includes: photovoltaic power output fluctuation data and load forecasting error data; The step of determining cloud model parameters based on the uncertainty data and optimizing the intermediate weights of each adequacy assessment indicator through the cloud model to obtain the final weight of each adequacy assessment indicator includes: Based on the photovoltaic processing fluctuation data and the load forecasting error data, the expectation, entropy, and hyperentropy of the cloud model are determined; Based on the cloud model and the intermediate weights, a preset number of cloud droplets are generated through a positive cloud generator, and all cloud droplets are fused to obtain the final weights.

[0013] In one possible implementation, the step of performing morphological decomposition on the time-series power curve of the target transaction resource according to a preset time scale includes: Adjust the length of structural elements in morphological decomposition according to the preset time scale; Based on the adjusted structural elements, the time-series power curve is subjected to opening and closing operations to obtain the trend term; The detail term is calculated based on the difference between the trend term and the time-series power curve.

[0014] In one possible implementation, the multi-dimensional resource sufficiency assessment method for integrated energy systems also includes: After the target transaction resources participate in market regulation, the actual sufficiency is obtained, and the deviation between the actual sufficiency and the dynamic sufficiency assessment value is calculated. If the deviation is greater than a preset threshold, then an adjustment feedback factor is determined based on the deviation; The parameter values ​​of the attention mechanism are adjusted based on the aforementioned adjustment feedback factor.

[0015] In one possible implementation, the multidimensional adequacy assessment metrics include: capacity adequacy metrics, power adequacy metrics, and response speed adequacy metrics. The target trading resources include: diversified energy storage, virtual energy storage, and distributed power sources; The preset time scales include: medium- and long-term scales, day-ahead scales, and real-time scales.

[0016] Secondly, embodiments of the present invention provide a multi-dimensional resource sufficiency assessment device for an integrated energy system, comprising: The acquisition module is used to acquire the time-series power curves of the target trading resources in the integrated energy system, as well as the uncertainty data of the integrated energy system; The decomposition module is used to perform morphological decomposition on the time-series power curve according to a preset time scale, and to construct multiple adequacy evaluation indicators based on the decomposition results. The calculation module is used to calculate the dynamic objective weights of each adequacy assessment index through the sliding time window entropy, and to calculate the basic subjective weights of each adequacy assessment index through the analytic hierarchy process. The fusion module is used to learn the weighting coefficients of the dynamic objective weights and basic subjective weights of each adequacy assessment indicator through an attention mechanism, and to fuse the dynamic objective weights and basic subjective weights of each adequacy assessment indicator according to the weighting coefficients to obtain the intermediate weights of each adequacy assessment indicator. The optimization module is used to determine cloud model parameters based on the uncertainty data, and optimize the intermediate weights of each adequacy assessment index through the cloud model to obtain the final weights of each adequacy assessment index. The evaluation module is used to integrate the various sufficiency evaluation indicators based on their final weights to obtain the dynamic sufficiency evaluation value of the target transaction resources.

[0017] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect or any possible implementation thereof.

[0018] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect or any possible implementation thereof.

[0019] Fifthly, embodiments of the present invention provide a computer program product, including a computer program that, when executed by a processor, implements the method described in the first aspect or any possible implementation thereof.

[0020] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows: This invention performs morphological decomposition of time-series power curves according to a preset time scale and constructs multiple-dimensional adequacy assessment indicators based on the decomposition results, overcoming the shortcomings of existing technologies that only focus on a single dimension. At the algorithm level, this invention innovatively proposes dynamic correction via an attention mechanism and quantification of uncertainty in the cloud model. Dynamic objective weights of each adequacy assessment indicator are calculated using a sliding time window entropy, while basic subjective weights are calculated using the analytic hierarchy process (AHP). The time-varying correlation characteristics of subjective and objective weights are learned through an attention mechanism, and subjective and objective weights are fused according to weighting coefficients, replacing fixed weight ratios. This allows weight allocation to adapt to the system's operating state in real time. For example, in real-time scenarios, the weight of objective data can be automatically increased; in medium- to long-term planning scenarios, the influence of expert experience weights can be strengthened, solving the problem that static weights in existing technologies cannot adapt to dynamic scenarios. Furthermore, the cloud model quantifies weight uncertainty into digital features, achieving accurate representation of weight uncertainty and effectively avoiding adequacy misjudgments caused by neglecting fluctuation risks in traditional algorithms. The embodiments of the present invention can avoid the misjudgment problem of traditional methods in extreme fluctuation scenarios, reduce the default cost caused by adjustment failure, and improve the fit between the comprehensive adequacy assessment and the actual adjustment effect, thus providing a more reliable technical guarantee for the stable operation of the system. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the implementation of the multi-dimensional resource sufficiency assessment method for integrated energy systems provided in this embodiment of the invention. Figure 2 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0022] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0023] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0024] This invention aims to address the problems of poor time scale adaptability, low decomposition accuracy, single evaluation dimension, and separation of subjective and objective factors in existing integrated energy system resource adequacy models. Specifically, it includes: (1) How to achieve accurate multi-timescale decomposition of non-stationary energy resource curves to match the adjustment needs of multi-level markets; (2) How to construct a multi-dimensional sufficiency index system covering "capacity-power-response speed" to comprehensively characterize resource regulation capabilities; (3) How to design a solution algorithm that integrates subjective and objective factors, taking into account both resource uncertainty and decision-making needs, and improve the rationality of sufficiency assessment.

[0025] The technical solution of this invention includes three parts: a resource curve decomposition method based on morphological decomposition, a multi-dimensional transaction resource sufficiency model construction, and a comprehensive sufficiency solution algorithm that integrates entropy weight-analytic hierarchy process (AHP).

[0026] See Figure 1 The document illustrates a flowchart of the implementation of a multi-dimensional resource sufficiency assessment method for integrated energy systems provided in an embodiment of the present invention, detailed below: Step S101: Obtain the time-series power curve of the target trading resource in the integrated energy system, as well as the uncertainty data of the integrated energy system.

[0027] Here, the target trading resources can be diversified energy storage, virtual energy storage, distributed power sources, etc., and the corresponding time-series power curves are output or load curves. Uncertainty data may include, but is not limited to: photovoltaic output fluctuation data and load forecast error data.

[0028] Suppose the time-series power curve of a certain type of traded resource (such as electrochemical energy storage or adjustable air conditioning load) in an integrated energy system is P(t), where The time variable is T, which is the analysis period (1 month, 1 day, etc.). The unit of P(t) is kW, which represents the actual output (load) of the resource at time t.

[0029] Step S102: Based on the preset time scale, perform morphological decomposition on the time series power curve, and construct multiple sufficiency evaluation indicators based on the decomposition results.

[0030] Here, the time scale can be set according to needs. For example, to address the adjustment needs of a multi-tiered market, three core time scales can be defined: Medium and long-term scales : Corresponding to the medium- to long-term market, take (30 days); Current Scale : In response to the current market, take ; Real-time scale : Corresponding to the real-time / auxiliary services market, take (15 minutes).

[0031] Morphological decomposition performs dilation, erosion, opening, and closing operations on signals using structural elements. It can accurately separate the trend term (reflecting long-term characteristics) and detail term (reflecting short-term fluctuations) of non-stationary signals, and its noise resistance is better than that of traditional wavelet decomposition.

[0032] This embodiment designs a morphological decomposition method based on scale-adaptive structuring elements, targeting the time scale. Design scale adaptive structural elements The structural element uses a flat structural element, and its length is... With time scale Positive correlation, the formula is:

[0033] in, Let P(t) be the data sampling interval. For example... , The discrete length (dimensionless) of the structuring element ensures that the structuring element can cover the characteristic period of the corresponding time scale.

[0034] Then, based on the structuring element, opening and closing operations are performed. Decompose P(t) to extract the trend and detail terms over the time scale: Opening operation Erosion followed by expansion is used to suppress positive spikes in signals (such as short-term load surges). Closing operation. First expand, then erode; used to fill negative valleys in signals (such as short-term power shortages).

[0035] By performing opening and closing operations on P(t) sequentially, the time scale is obtained. Trend items below :

[0036] Details The difference between the original curve and the trend term:

[0037] Trend Item This reflects the stable adjustability of resources over this timescale, such as the monthly available capacity of energy storage on a medium- to long-term timescale, and the daily adjustable power of adjustable load on a day-ahead timescale; details... It reflects the fluctuation characteristics at this scale, such as the power output fluctuation of photovoltaics at the real-time scale.

[0038] For example, this embodiment constructs a three-dimensional sufficiency index system of capacity, power, and response speed for three core trading resources: diversified energy storage, virtual energy storage, and distributed power sources. (1) First, define the core parameters of each type of transaction resource, as shown in Table 1.

[0039] Table 1

[0040] (2) Capacity adequacy index

[0041] Capacity adequacy characterizes time scale The formula for the degree to which the total regulatory capacity available from resources meets market demand capacity is as follows:

[0042]

[0043] Time scale The available adjustable capacity (kWh) of the following resources needs to be considered in conjunction with trend terms. Calculation, for example: Multi-source energy storage: (If discharging is not occurring during charging, then take...) ); Virtual energy storage: (If the load is being reduced, then take the following:) ); Distributed power sources: (Considering the uncertainty of output, take the predicted value minus 1 standard deviation); Time scale The market's regulating demand capacity (kWh) is determined by market trading rules, for example: The medium- to long-term timescale is the monthly peak load. (Base power output) Current Scale This represents the difference between the daily peak load and the predicted output of new energy sources. Real-time scale : This indicates the deviation between real-time load and predicted load. "Time": indicates that the resource capacity is fully sufficient when there is no demand, and the value is taken as an empirical safety threshold.

[0044] (3) Power adequacy index

[0045] Power adequacy characterizes time scale The degree to which the instantaneous adjustable power available from resources meets the market demand for power is given by the following formula:

[0046]

[0047] Time scale The available adjustable power (kW) of the following resources is determined by the trend term. The derivative (power change rate) constraint, for example: Multi-source energy storage: , must meet The smallest unit of time; Virtual energy storage: ; Distributed power sources: ; Time scale The formula for regulating power demand (kW) in the downstream market is: This represents the ratio of demand capacity to time scale.

[0048] (4) Response speed adequacy index

[0049] Response speed adequacy characterizes the time scale The degree to which the speed of resource adjustment meets the speed of market demand is given by the following formula:

[0050] This indicates the available response speed of a resource (1 / min), reflecting the speed at which the resource transitions from "standby state" to "full-loaded state." For example: Multi-source energy storage: (Ratio of climbing rate to available power, in units of 1 / min); Virtual energy storage: (The reciprocal of the response delay, in units of 1 / min); Distributed power sources: ; This represents the market's demand response speed (1 / min), determined by market rules, for example: Medium and long-term scales (1 day = 1440 minutes; medium- and long-term adjustments allow for a response within 1 day). Current Scale (1 hour = 60 minutes, day-ahead regulation allows for response within 1 hour); Real-time scale (The ancillary services market typically requires a response time of 5 minutes).

[0051] Step S103: Calculate the dynamic objective weights of each adequacy assessment index by using the sliding time window entropy, and calculate the basic subjective weights of each adequacy assessment index by using the analytic hierarchy process.

[0052] (1) Objective entropy weight optimization calculation

[0053] A sliding window entropy is introduced to replace the traditional static entropy weight to adapt to the time-varying characteristics of the resource curve. Let the time window length be W (e.g., W=72, corresponding to 3 days of 15-minute data). The data within the window at time t is... ,but:

[0054]

[0055]

[0056] in, The dynamic objective weight at time t is updated in real time as the window slides, solving the problem of insufficient adaptability of traditional entropy weight to time-varying data.

[0057] (2) Subjective AHP weight optimization calculation

[0058] An interval judgment matrix is ​​introduced to replace the traditional exact value matrix to reflect the ambiguity of expert judgment. Let the importance ratio of indicators i and j be the interval... , The weight interval is solved by interval eigenvalue decomposition. The midpoint is ultimately chosen as the base subjective weight.

[0059] Step S104: Learn the weighting coefficients of the dynamic objective weights and basic subjective weights of each adequacy assessment indicator through an attention mechanism, and fuse the dynamic objective weights and basic subjective weights of each adequacy assessment indicator according to the weighting coefficients to obtain the intermediate weights of each adequacy assessment indicator.

[0060] Objective entropy weight With subjective AHP weights Treating them as dual-input features, the dynamic correlation weights between the two are learned through an attention mechanism to achieve adaptive fusion in time-varying scenarios.

[0061] (1) Constructing the attention input vector Let the input feature vector be , represents the subjective and objective weight combination of the i-th type of indicator at time t.

[0062] (2) Calculate attention score Through a single-layer perceptron (including activation function) Calculate attention score This reflects the relative importance of the indicator at the current moment:

[0063] in is the perceptron weight parameter, b is the bias term, which is obtained through training with historical data (the loss function is the mean square error between the predicted adequacy and the actual adjustment effect).

[0064] (3) Normalized attention weights The attention weights are obtained by normalizing the scores using the Softmax function. :

[0065] (4) Dynamically adjust subjective and objective weights The attention weight is used as a dynamic adjustment factor to weight and fuse the basic weights to obtain the intermediate weight. : .

[0066] Step S105: Determine the cloud model parameters based on the uncertainty data, and optimize the intermediate weights of each adequacy assessment index through the cloud model to obtain the final weights of each adequacy assessment index.

[0067] (1) Set the intermediate weight Following a normal cloud distribution, its numerical characteristics are: Expected E_x_i(t) represents The mean of the weights reflects the central tendency of the weights; Entropy Eni(t) represents the degree of dispersion of the weights. (k is the adjustment coefficient); hyperentropy Represents the uncertainty of entropy, (Experience points).

[0068] (2) Cloud droplet generation and fusion N cloud droplets (N=1000, to ensure statistical significance) are generated using a forward cloud generator, with each cloud droplet being a weighted random sample. (s=1,2,...,N). All cloud droplets are fused according to the following formula to obtain the final weights. :

[0069] in, To ensure that the random numbers conform to N(0,1), an uncertainty perturbation is introduced through hyperentropy, making the weight fusion result more consistent with the actual random scenario.

[0070] Step S106: Based on the final weights of each sufficiency assessment indicator, the various sufficiency assessment indicators are integrated to obtain the dynamic sufficiency assessment value of the target transaction resources.

[0071] The formula for overall sufficiency is:

[0072] in, The time scale is t. Dynamic comprehensive adequacy After obtaining the dynamic sufficiency assessment value of the target transaction resources, a feedback factor for adjustment effect can also be introduced. When the actual adjustment effect is the same as When the deviation exceeds 5%, through Correcting attention mechanism parameters and b:

[0073]

[0074] in, The learning rate is set to 0.01. It is a gradient operator to ensure that the model continuously adapts to the dynamic changes of the system.

[0075] This embodiment, at the resource curve decomposition level, introduces a new scale-adaptive structuring element and a morphological open-closed combination decomposition process, overcoming the limitations of traditional wavelet decomposition and Fourier transform in non-stationary signal processing. This structuring element can dynamically adjust its length according to three core time scales: medium-to-long-term, day-ahead, and real-time, accurately covering the characteristic cycles of the corresponding markets. Combined with the opening operation to suppress positive peaks and the closing operation to fill negative troughs, it can efficiently separate the trend and detail terms of the resource curve. This not only provides superior noise resistance compared to traditional decomposition techniques but also achieves simultaneous and accurate decomposition of non-stationary energy resource curves across multiple time scales. This innovation allows the assessment results to directly match the adjustment needs of multi-level markets. For example, the medium-to-long-term scale trend term supports monthly capacity planning, while the real-time scale detail term adapts to the rapid response of the auxiliary service market, solving the problem that existing single-time-scale assessment techniques cannot achieve cross-market collaboration and improving the adaptability of assessments to market demands.

[0076] At the sufficiency model construction level, the pioneering three-dimensional indicator system of "capacity-power-response speed" overcomes the shortcomings of existing technologies that only focus on a single dimension. This system defines sufficiency calculation logic adapted to the characteristics of three core trading resources: diversified energy storage, virtual energy storage, and distributed power sources. The newly added "response speed sufficiency indicator" is particularly crucial—it quantifies the speed at which resources transition from standby to full regulation, and, combined with the demand response speed standards of different markets, incorporates "response timeliness" into the evaluation scope for the first time. This design allows the evaluation results to comprehensively characterize the overall regulation capability of resources, avoiding resource misjudgments caused by neglecting response speed in existing technologies. For example, in the real-time ancillary services market, it can accurately screen resources that "meet capacity requirements, have sufficient power, and respond quickly," reducing the risk of supply-demand imbalance caused by resource response lag.

[0077] At the algorithm level, this invention innovatively proposes an attention mechanism for dynamic correction and a cloud model uncertainty quantification module, which revolutionizes the limitations of the traditional static entropy weight-AHP fusion method. The attention mechanism learns the time-varying correlation characteristics of subjective and objective weights through a single-layer perceptron, replacing fixed weight ratios with dynamic correction factors. This allows weight allocation to adapt to the system's operating state in real time, automatically increasing the weight ratio of objective data in real-time scenarios and strengthening the influence of expert experience weights in medium- and long-term planning scenarios, solving the problem that static weights cannot adapt to dynamic scenarios in existing technologies. The cloud model, by quantifying weight uncertainty into three-dimensional features of expectation, entropy, and hyperentropy, and combining this with the probabilistic fusion of 1000 cloud droplets, innovatively achieves accurate representation of weight uncertainty, effectively avoiding the adequacy misjudgment caused by neglecting fluctuation risks in traditional algorithms. Simultaneously, the algorithm's built-in dynamic verification mechanism adjusts model parameters in real time by adjusting the effect feedback factor, further ensuring evaluation accuracy. In practical applications, this algorithm avoids the misjudgment problems of traditional methods in extreme fluctuation scenarios, reduces the default costs caused by adjustment failures, and improves the fit between the comprehensive adequacy assessment and the actual adjustment effect, providing a more reliable technical guarantee for stable system operation.

[0078] This method will be explained using a real-time scenario of a comprehensive energy system in a certain industrial park as an example.

[0079] I. Data Collection We collected nearly six months of 15-minute-level resource operation data from the integrated energy system (a total of 6 × 30 × 24 × 4 = 17,280 data points) for sliding window entropy calculation. We also invited five experts in the fields of energy markets and integrated energy systems to complete a questionnaire on the importance intervals of indicators at different time scales (medium- to long-term / day-ahead / real-time) to obtain the interval AHP required for calculation. Data; collect historical deviation data for 365 days of photovoltaic power output and load forecast, calculate standard deviation, fluctuation coefficient, etc., for cloud model parameter calibration.

[0080] II. Data Preprocessing Optimization Sliding window partitioning: Divide the 15-minute data into windows of length W=72 (corresponding to 3 days, 72×...). =18h) Divide the sliding window, and a total of 17280-72+1=17209 windows are generated; Expert opinion consistency verification: A consistency test was performed on the interval judgment matrix of 5 experts (interval consistency index). <0.1), after removing inconsistent questionnaires, the mean was taken as the final interval data.

[0081] III. Focusing on the real-time scale, this study examines the system within the industrial park (2MW / 8MWh electrochemical energy storage, 500kW adjustable air conditioning load, and 1MW photovoltaic system). Algorithm flow: Step 1: Definition of basic parameters and time scale Core resource parameters: Energy storage , , , Photovoltaic power output forecast Historical deviation standard deviation ; New algorithm parameters: sliding window length W=72, attention mechanism learning rate Cloud model: number of cloud droplets N=1000, adjustment coefficient k=1.638, feedback factor. threshold .

[0082] Step 2: Calculate the basic weights (sliding window entropy + interval AHP) (1) Dynamic objective entropy weight Calculation (t is the real-time point, taking the 10080th data point, corresponding to the 140th sliding window) Data extraction within the window: This window contains 72 15-minute data points from (t=10009) to (t=10080), analyzed using the capacity adequacy index. For example, the normalized value within the window The mean of s=10009,...,10080 is 0.85 and the standard deviation is 0.12; probability calculate: Take the time t=10080 ,but:

[0083] Similarly, calculate the power margin. Response speed margin ; Sliding window entropy calculate:

[0084] Similarly, we get eP(10080) = 0.485. =0.521; Dynamic objective entropy weight:

[0085] Similarly,

[0086] (2) Interval AHP subjective weights calculate Constructing an interval judgment matrix: For the real-time scale (auxiliary services market), the importance intervals of the indicators given by experts are as follows (i is the row indicator, j is the column indicator):

[0087] Where A_{C,R}=[0.3,0.5] indicates that the importance of "capacity adequacy" relative to "response speed adequacy" is 0.3-0.5, that is, response speed is more important; Interval eigenvalue decomposition: Solving for the interval of the largest eigenvalue of the judgment matrix using the eigvals function in the SciPy library. Consistency indicators This satisfies the consistency requirement; Subjective weight calculation: Extract the eigenvector interval corresponding to the largest eigenvalue, and normalize it to obtain: , , ; Take the midpoint as the basic subjective weight: =0.20, =0.27, =0.53 Step 3: Dynamic Adjustment of Attention Mechanism (1) Attention mechanism training (offline pre-training + online fine-tuning) Training data: using the sliding window entropy weights from the previous 5 months. Interval AHP weights The actual adjustment effect (such as the success rate of the ancillary services market and the adjustment deviation rate) are used as labels; Model structure: Single-layer perceptron input is The output is the attention score. The activation function used is Sigmoid. ; Training process: Using the mean squared error (MSE) between the predicted overall adequacy and the actual adjustment effect as the loss function, the Adam optimizer is used for 1000 iterations to finally obtain the perceptron parameters. (The coefficient of objective weight is higher, which meets the needs of real-time scenario data-driven requirements), b=-0.15.

[0088] (2) Real-time attention weight calculation Input vector construction: at time t=10080, , , ; Attention score calculation:

[0089] Similarly, ; ; Attention weight normalization:

[0090] Similarly, , ; Intermediate weight calculation:

[0091] Similarly, ; .

[0092] Step 4: Quantification of Uncertainty in Cloud Models (1) Calculation of digital features of cloud model Using capacity adequacy as the median weight For example: expect Take the average of the intermediate weights, E_x_C(t) = 0.243; entropy Calculate the standard deviation of the intermediate weights within this window. ,but:

[0093] hyperentropy Take 10% of the entropy.

[0094] Similarly, the power margin is: E_x_P(t) = 0.295.

[0095] Response speed margin: E_x_R(t) = 0.458, , .

[0096] (2) Cloud droplet generation and final weight fusion Forward cloud generator: Generates N=1000 cloud droplets using the CloudPy library, each cloud droplet... satisfy: Generate compliance random entropy ; Generate compliance cloud droplet value ; Taking capacity adequacy as an example, the first cloud droplet ,Pick ,have to

[0097] Weighted fusion: The 1000 cloud droplets are weighted according to their degree of uncertainty, using the following formula:

[0098] in (Random perturbation of the s-th cloud droplet); The final calculation yields: (Note: Response speed has the highest weight, which meets the demand for timely response in the ancillary services market.)

[0099] Step 5: Comprehensive sufficiency calculation and dynamic verification (1) Real-time comprehensive adequacy calculation Multi-dimensional indicator values:

[0100] Dynamic overall adequacy:

[0101] (2) Dynamic verification and parameter correction Actual adjustment effect feedback: After participating in the ancillary services market at this moment, the actual adjustment deviation rate was 2.3% (below the 5% threshold), feedback factor (No corrections required); Handling scenarios where deviation exceeds the threshold: If the actual deviation rate at a certain moment is 6.8% (exceeding 5%), then... (The larger the deviation, the stronger the correction), update the attention mechanism parameters according to the following formula:

[0102] Assuming gradient (If the objective weighting coefficient needs to be reduced, and the subjective weighting coefficient needs to be increased), then: Ensure the model dynamically adjusts with the system to continuously improve accuracy.

[0103] IV. Implementation and Verification (1) Comparison with the traditional entropy weight-AHP fusion method: Using 100 sliding windows of the real-time scale scene of the park as samples, the core performance indicators of the two fusion methods are compared, as shown in Table 2.

[0104] Table 2

[0105] (2) Scenario design: simulate extreme scenarios of photovoltaic power output fluctuation (standard deviation increases from 80kW to 120kW) and load forecast deviation (from 300kW to 500kW).

[0106] Results Comparison: The traditional method misjudged 3 times because it did not consider uncertainty (actually, the adjustment failed due to fluctuations, resulting in a default cost of 80,000 yuan); the present invention identifies uncertainty risks in advance through the cloud model, adjusts the weights, and increases the response speed weight to 0.52, avoiding all defaults and reducing losses by 80,000 yuan.

[0107] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0108] This invention provides a multi-dimensional resource sufficiency assessment device for integrated energy systems, comprising: The acquisition module is used to acquire the time-series power curves of the target trading resources in the integrated energy system, as well as the uncertainty data of the integrated energy system; The decomposition module is used to perform morphological decomposition on the time-series power curve according to a preset time scale, and to construct multiple adequacy evaluation indicators based on the decomposition results. The calculation module is used to calculate the dynamic objective weights of each adequacy assessment index through the sliding time window entropy, and to calculate the basic subjective weights of each adequacy assessment index through the analytic hierarchy process. The fusion module is used to learn the weighting coefficients of the dynamic objective weights and basic subjective weights of each adequacy assessment indicator through an attention mechanism, and to fuse the dynamic objective weights and basic subjective weights of each adequacy assessment indicator according to the weighting coefficients to obtain the intermediate weights of each adequacy assessment indicator. The optimization module is used to determine cloud model parameters based on the uncertainty data, and optimize the intermediate weights of each adequacy assessment index through the cloud model to obtain the final weights of each adequacy assessment index. The evaluation module is used to integrate the various sufficiency evaluation indicators based on their final weights to obtain the dynamic sufficiency evaluation value of the target transaction resources.

[0109] Figure 2 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. For example... Figure 2 As shown, the electronic device 2 in this embodiment includes a processor 20 and a memory 21. The memory 21 stores a computer program 22. When the processor 20 executes the computer program 22, it implements the steps in the various method embodiments described above. Alternatively, when the processor 20 executes the computer program 22, it implements the functions of each module / unit in the various device embodiments described above.

[0110] For example, computer program 22 may be divided into one or more modules / units, which are stored in memory 21 and executed by processor 20 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 22 in electronic device 2.

[0111] Electronic device 2 may include, but is not limited to, processor 20 and memory 21. Those skilled in the art will understand that... Figure 2 This is merely an example of electronic device 2 and does not constitute a limitation on electronic device 2. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 2 may also include input / output devices, network access devices, buses, etc.

[0112] The processor 20 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0113] The memory 21 can be an internal storage unit of the electronic device 2, such as a hard disk or RAM. The memory 21 can also be an external storage device of the electronic device 2, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 21 can include both internal and external storage units of the electronic device 2. The memory 21 is used to store the computer program 22 and other programs and data required by the electronic device 2. The memory 21 can also be used to temporarily store data that has been output or will be output.

[0114] For the sake of simplicity and clarity, only the above-described functional modules / units are used as examples. In practical applications, the functions described above can be assigned to different functional modules / units as needed. These modules / units can be implemented in hardware, software, or a combination of both.

[0115] This invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the methods described in the above-described method embodiments.

[0116] This invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the methods described in the above-described method embodiments.

[0117] Computer programs include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0118] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0119] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for multi-dimensional resource adequacy assessment of an integrated energy system, characterized in that, The method comprises the following steps: acquiring time sequence power curves of a target transaction resource in an integrated energy system and uncertainty data of the integrated energy system; performing morphological decomposition on the time sequence power curves according to a preset time scale, and constructing a plurality of adequacy evaluation indexes according to a decomposition result; calculating dynamic objective weights of each adequacy evaluation index through sliding time window entropy, and calculating basic subjective weights of each adequacy evaluation index through an analytic hierarchy process; learning weighting coefficients of the dynamic objective weights and the basic subjective weights of each adequacy evaluation index through an attention mechanism, and fusing the dynamic objective weights and the basic subjective weights of each adequacy evaluation index according to the weighting coefficients to obtain an intermediate weight of each adequacy evaluation index; determining cloud model parameters according to the uncertainty data, and optimizing the intermediate weight of each adequacy evaluation index through a cloud model to obtain a final weight of each adequacy evaluation index; fusing each adequacy evaluation index based on the final weight of each adequacy evaluation index to obtain a dynamic adequacy evaluation value of the target transaction resource.

2. The method of claim 1, wherein, The method of calculating the dynamic objective weights of each adequacy evaluation index through the sliding time window entropy comprises the following steps: calculating index sequence data of each adequacy evaluation index according to the time sequence power curves; for each adequacy evaluation index, determining a time window corresponding to a current time, and calculating an entropy value of the index sequence data in the time window; determining the dynamic objective weights of each adequacy evaluation index according to the entropy value.

3. The method of claim 1, wherein, The method of calculating the basic subjective weights of each adequacy evaluation index through the analytic hierarchy process comprises the following steps: calculating an interval judgment matrix through the analytic hierarchy process; determining a basic subjective weight interval of each adequacy evaluation index according to the interval judgment matrix, and determining the basic subjective weight according to a median value of the basic subjective weight interval. 4.The method of claim 1, wherein, The method of learning the weighting coefficients of the dynamic objective weights and the basic subjective weights of each adequacy evaluation index through the attention mechanism comprises the following steps: inputting the dynamic objective weights and the basic subjective weights of each adequacy evaluation index as a group of feature vectors into a single layer perception machine to calculate an attention score; normalizing the attention scores of each adequacy evaluation index to obtain a dynamic objective weight weighting coefficient of each adequacy evaluation index, and determining a basic subjective weight weighting coefficient of each adequacy evaluation index according to the dynamic objective weight weighting coefficient of each adequacy evaluation index.

5. The method of claim 1, wherein, The uncertainty data comprises photovoltaic output fluctuation data and load prediction error data. The method of determining the cloud model parameters according to the uncertainty data, and optimizing the intermediate weight of each adequacy evaluation index through the cloud model to obtain the final weight of each adequacy evaluation index comprises the following steps: determining an expectation, an entropy and a hyper entropy of the cloud model according to the photovoltaic output fluctuation data and the load prediction error data; generating a preset number of cloud droplets through a forward cloud generator according to the cloud model and the intermediate weight, fusing all the cloud droplets to obtain the final weight.

6. The method of claim 1 to 5, wherein, The method of performing morphological decomposition on the time sequence power curves of the target transaction resource according to the preset time scale comprises the following steps: According to the preset time scale, the length of a structural element in morphological decomposition is adjusted; According to the adjusted structural element, an opening operation and a closing operation are performed on the time-series power curve to obtain a trend item; A detail item is calculated according to a difference between the trend item and the time-series power curve.

7. The method of claim 1 to 5, wherein, Further comprising: After the target transaction resource participates in market regulation, an actual adequacy is obtained, and a deviation between the actual adequacy and the dynamic adequacy evaluation value is calculated; If the deviation is greater than a preset threshold, a regulation feedback factor is determined according to the deviation; Based on the regulation feedback factor, a parameter value of the attention mechanism is adjusted.

8. An integrated energy system multi-dimensional resource adequacy assessment device, characterized in that, Comprise: An acquisition module is configured to acquire a time-series power curve of a target transaction resource in a comprehensive energy system and uncertainty data of the comprehensive energy system; A decomposition module is configured to perform morphological decomposition on the time-series power curve according to a preset time scale, and construct a plurality of adequacy evaluation indexes according to a decomposition result; A calculation module is configured to calculate a dynamic objective weight of each adequacy evaluation index through sliding time window entropy, and calculate a basic subjective weight of each adequacy evaluation index through analytic hierarchy process; A fusion module is configured to learn a weighting coefficient of the dynamic objective weight and the basic subjective weight of each adequacy evaluation index through an attention mechanism, and fuse the dynamic objective weight and the basic subjective weight of each adequacy evaluation index according to the weighting coefficient to obtain an intermediate weight of each adequacy evaluation index; An optimization module is configured to determine cloud model parameters according to the uncertainty data, and optimize the intermediate weight of each adequacy evaluation index through a cloud model to obtain a final weight of each adequacy evaluation index; An evaluation module is configured to fuse each adequacy evaluation index based on the final weight of each adequacy evaluation index to obtain a dynamic adequacy evaluation value of the target transaction resource.

9. An electronic device, comprising: The computer readable storage medium stores a computer program, and the computer program is executed by the processor to realize the method in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to realize the method in any one of claims 1 to 7.

Images