Staged power generation resource credible capacity evaluation method and device

Abstract

The invention discloses a staged power generation resource credible capacity evaluation method and device, and belongs to the field of power market capacity market mechanism design. Aiming at the problem of insufficient precision caused by neglecting system state time-varying characteristics in long-time scale evaluation in the existing method, a penalty type segmented dynamic planning model is established to perform optimal stage division on the whole year by constructing a multi-dimensional daily characteristic sequence reflecting system net load and renewable energy output, so that the system characteristics in the stages are highly consistent; then, a staged marginal effective load capacity evaluation model is constructed based on the division scheme, and unit operation and energy storage constraints are depicted finely; and finally, determining the reliable contribution proportion by comparing the improvement degrees of the introduced ideal resources and target resources on the system energy shortage. According to the method, the accuracy of credible capacity evaluation is improved, and support is provided for scientific planning and resource adequacy analysis of a high-proportion new energy power system.

Description

Technical Field

[0001] This invention relates to the field of electricity market capacity market mechanism design, and in particular to a method, apparatus, equipment and storage medium for phased assessment of reliable capacity of power generation resources. Background Technology

[0002] As my country accelerates the construction of its new power system, the installed capacity and power generation of renewable energy sources such as wind and solar power continue to rise, while new loads such as electric vehicles and electric heating are growing rapidly. This has led to stronger seasonal and periodic fluctuations in both supply and demand. The intermittency and correlation of renewable energy significantly amplify the strain on the system during certain periods, placing higher demands on the assessment and allocation of capacity adequacy while ensuring a safe and reliable power supply.

[0003] As the most widely used mechanism for adequacy of capacity, capacity markets in practice generally organize capacity procurement and settlement using an "annual average" or a single reliable capacity coefficient, implicitly assuming that "the capacity value of resources is roughly balanced throughout the year." However, under conditions of high-penetration renewable and diverse flexible resources, the frequency, duration, and seasonality of system shortage periods vary significantly: high-temperature loads and low-wind nights in summer, cold waves and rigid heating output constraints in winter, and high proportions of wind and solar output with limited portability in spring and autumn lead to significant differences in the marginal capacity value of resources across different seasons and periods. Continuing to use annual averages will produce two prominent problems: first, capacity value mismatch—underestimating capacity demand during high-risk periods and overestimating capacity demand during low-risk periods; second, incentive distortion—resources cannot obtain commensurate capacity returns based on their reliability contributions during "truly scarce periods."

[0004] To address the aforementioned mismatch and incentive issues, domestic and international practices have gradually explored organizing capacity procurement by season or phase. However, this approach faces three key technical challenges: First, how to objectively identify time periods with homogeneous risks from annual data without relying on subjective calendar segmentation, while maintaining reliability objectives. Second, how to determine the specific number of phases after introducing phased approaches, balancing statistical robustness with market implementation complexity. Third, how to ensure the consistency between phased divisions and reliability indicators such as ELCC (Elastic Computing Capacity Control Center) and capacity market settlement rules. Therefore, a phased, reliable capacity calculation technology approach that addresses both capacity adequacy and market implementation is urgently needed. Summary of the Invention

[0005] The present invention aims to at least partially solve one of the technical problems in the related art.

[0006] To address this, the present invention proposes a phased method for assessing the reliable capacity of power generation resources, comprising: constructing a multi-dimensional daily characteristic sequence based on the annual load and renewable energy output; establishing a penalized segmented dynamic programming model to solve for the optimal phase division for the whole year that satisfies the shortest phase constraint; constructing a phased marginal effective load-carrying capacity assessment model based on the division results, and calculating the reliable capacity of each phase in a coordinated manner with unit operation and energy storage constraints; and finally determining the reliable capacity coefficient of the target resource by comparing the degree of improvement of the unused energy of the system by introducing ideal resources and target resources.

[0007] Another objective of this invention is to provide a staged power generation resource reliability capacity assessment device.

[0008] The third objective of this invention is to provide a computer device.

[0009] A fourth objective of this invention is to provide a non-transitory computer-readable storage medium.

[0010] To achieve the above objectives, this invention proposes a phased method for assessing the reliable capacity of power generation resources, comprising: S1, based on the hourly load and renewable energy output data of the target capacity delivery year, constructs a natural daily feature vector sequence including peak net load, ramping demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean and load shedding amount; S2. Establish a penalized piecewise least squares dynamic programming model. Under the condition of satisfying the shortest stage length constraint, take the weighted sum of the eigenvector dispersion within the stage and the stage number penalty term as the optimization objective to solve the optimal stage division scheme for the whole year and the calendar continuity. S3. Construct a phased marginal effective load-carrying capacity assessment model based on the phase division scheme, comprehensively consider the load loss value, unit start-up and ramp-up constraints, heating rigid constraints, and energy storage charging and discharging and SOC constraints, and calculate the reliable capacity of power generation resources in each phase. S4. By comparing the expected improvement in unused energy of the system after adding a unit of installed capacity of perfect power generation resources and target power generation resources to the initial power structure, the reliable capacity coefficient of the target power generation resources is determined.

[0011] The phased power generation resource reliability capacity assessment method of this invention may also have the following additional technical features: In one embodiment of the present invention, the construction of a daily feature vector sequence based on the hourly load and renewable energy output data of the target capacity delivery year, including peak net load, ramp-up demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, and load shedding, includes: S11 uses median and interquartile range standardization to process each feature component, and normalizes the feature components using a formula. ; in, Indicates the first Year The first of the natural days The original values ​​of the characteristic components, j=1,2,3,4,5, correspond to the five characteristic values: peak net load, ramp demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, and load shedding, respectively. Represents characteristic components The median of all natural days. Represents characteristic components Interquartile range; S12 assigns weights to the conditional mean of the wind power capacity factor. Weights are assigned to the conditional mean of the photovoltaic capacity factor. The weighted feature vector is constructed using the following formula: ; in, arrive The five characteristic components are: peak net load, ramp-up demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, and load shedding. arrive This represents the weight assigned to each corresponding feature component.

[0012] In one embodiment of the present invention, the step of establishing a penalized piecewise least squares dynamic programming model, under the condition of satisfying the shortest stage length constraint, and using the weighted sum of the eigenvector dispersion within a stage and the stage number penalty term as the optimization objective, to solve for the optimal stage division scheme for the whole year and with a continuous calendar, includes: S21 calculates the squared error within a multi-year period using the cross-year weighted prefix, and calculates the weighted sum of any interval [i,j] based on the prefix. vectors and , sum of squares : ; ; ; in, This represents the weight of the data in year y. Indicates the year y, the year The weighted feature vector of days, where d represents the d-th natural day of the year; S22, the optimal piecewise strategy is solved recursively using dynamic programming state transition equations: ; in, This represents the optimal objective function value from day 1 to day i-1. This represents the eigenvector dispersion within stage [i, j]. This represents the penalty coefficient for the stage number.

[0013] In one embodiment of the present invention, the step of constructing a phased marginal effective load-carrying capacity assessment model according to the phase division scheme, comprehensively considering the load shedding value, unit start-up and ramp-up constraints, heating rigid constraints, and energy storage charging and discharging and SOC constraints, to calculate the reliable capacity of power generation resources in each phase includes: S31 constrains the state of charge of the energy storage unit, limiting the upper and lower limits of the energy storage capacity through formulas: ; in, Indicates time energy storage capacity, Representing time The state of charge, and These are the minimum and maximum allowable limits for the state of charge; S32 introduces transmission line constraints, using a formula to ensure that the system power transmission does not exceed the maximum line capacity: ; ; in, It is a line Maximum transmission capacity It is the power transfer distribution factor of the system.

[0014] In one embodiment of the present invention, it further includes: S5, When calculating the squared error within a multi-year period, define the cross-year weighted prefix quantity. , and And based on the formula, optimize the solution efficiency of the dynamic programming model: ; in, This represents the sum of squares of the weighted eigenvectors for all days and all years within the time period [i, j]. This represents the total weight of all days and all years within the time period [i, j]. This represents the vector sum of the eigenvectors of all days and all years within the time period [i, j].

[0015] To achieve the above objectives, another aspect of the present invention provides a phased power generation resource reliability capacity assessment device, comprising: The feature vector construction module is used to construct a natural daily feature vector sequence based on the hourly load and renewable energy output data throughout the year of the target capacity delivery year. This sequence includes peak net load, ramp-up demand, the conditional mean of wind power capacity factor, the conditional mean of photovoltaic capacity factor, and the amount of unloaded load. The dynamic programming model building module is used to build a penalized piecewise least squares dynamic programming model. Under the condition of satisfying the shortest stage length constraint, the module uses the weighted sum of the eigenvector dispersion within the stage and the stage number penalty term as the optimization objective to solve for the optimal stage division scheme that is calendar-continuous throughout the year. The reliable capacity calculation module is used to construct a phased marginal effective load-carrying capacity assessment model according to the phase division scheme, and comprehensively consider the load loss value, unit start-up and ramp-up constraints, heating rigid constraints, and energy storage charging and discharging and SOC constraints to calculate the reliable capacity of power generation resources in each phase. The reliable capacity coefficient determination module is used to determine the reliable capacity coefficient of the target power generation resource by comparing the expected improvement value of the unused energy of the system after adding a unit of installed capacity of perfect power generation resource to the initial power structure with that of the target power generation resource.

[0016] In one embodiment of the present invention, it further includes: The cross-year weighted prefix calculation module is used to define the cross-year weighted prefix when calculating the squared error over multiple years. , and And based on the formula, optimize the solution efficiency of the dynamic programming model: ; in, This represents the sum of squares of the weighted eigenvectors for all days and all years within the time period [i, j]. This represents the total weight of all days and all years within the time period [i, j]. This represents the vector sum of the eigenvectors of all days and all years within the time period [i, j].

[0017] This invention discloses a method and apparatus for assessing the reliable capacity of power generation resources in stages. By constructing a multi-dimensional daily characteristic sequence reflecting the system's net load and renewable energy output, and utilizing penalized dynamic programming to achieve optimal stage division throughout the year, it effectively solves the problems of insufficient accuracy and poor adaptability caused by traditional assessment methods neglecting the time-varying characteristics of system states. It achieves full-process optimization from feature standardization, weighted fusion, efficient segmentation to constraint-integrated assessment, significantly improving the accuracy and computational efficiency of reliable capacity calculation. While ensuring the reliability of power system planning, it enhances the pertinence of resource adequacy analysis, providing effective support for the scientific decision-making and stable operation of high-proportion renewable energy power systems.

[0018] To achieve the above objectives, a third aspect of this application provides a computer device, including a processor and a memory; wherein the processor reads executable program code stored in the memory to run a program corresponding to the executable program code, for implementing a phased power generation resource reliability capacity assessment method as described in the first aspect embodiment.

[0019] To achieve the above objectives, a fourth aspect of this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a phased power generation resource reliability capacity assessment method as described in the first aspect embodiment.

[0020] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0021] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a phased power generation resource reliability capacity assessment method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a phased power generation resource reliability capacity assessment device according to an embodiment of the present invention; Figure 3 It is a computer device according to an embodiment of the present invention. Detailed Implementation

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0024] The following description, with reference to the accompanying drawings, describes a method, apparatus, equipment, and storage medium for assessing the reliable capacity of power generation resources in stages according to embodiments of the present invention.

[0025] The core idea of ​​this invention is to construct a multi-dimensional daily characteristic system integrating power load and renewable energy output, mapping peak net load, ramp demand, wind and solar capacity factor conditional mean, and load shedding into a time-series vector in a unified feature space, thereby achieving a multi-dimensional quantitative representation of the power system's operating state. Based on this feature sequence, a penalized piecewise dynamic programming model is further established. Under the condition of satisfying the stage length constraint, the model aims to minimize the weighted sum of feature dispersion within a stage and the stage number penalty, achieving an optimal and calendar-continuous stage division throughout the year. On this basis, a staged marginal effective load-carrying capacity assessment model is constructed, comprehensively considering the value of load shedding, unit operation constraints, and energy storage dynamic characteristics to accurately calculate the reliable capacity of power generation resources within each stage. Finally, by comparing the improvement effects of introducing ideal resources and actual resources on the system's unused energy, the reliable capacity contribution of target resources in ensuring system adequacy is scientifically determined. This transforms the traditional static assessment into a dynamic reliable capacity assessment system that adapts to renewable energy fluctuations, multi-stage coordination, and fine-grained constraint modeling, significantly improving the accuracy, interpretability, and engineering applicability of the assessment results.

[0026] Example 1 To achieve the above invention, embodiments of the present invention provide a method for assessing the reliable capacity of power generation resources in stages, such as... Figure 1 As shown, it includes: S1, based on the hourly load and renewable energy output data of the target capacity delivery year, constructs a natural daily feature vector sequence including peak net load, ramp-up demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, and load shedding.

[0027] Specifically, the sequence of this step includes five key characteristic components: peak net load, 3-hour ramp-up demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, and daily load shedding, which are used for subsequent stage division and reliable capacity assessment.

[0028] Specifically, the smallest unit of division is first defined as the calendar day. And for each natural day Extract its feature vector Among them, peak net load By hourly net load Calculate and take the maximum value of the day as the characteristic component. 3-hour uphill requirement. The focus is then on the evening period (17:00 to 21:00), calculating the maximum 3-hour net load change during this period, i.e. It is used to reflect the system's adjustment pressure during the rapid load increase phase.

[0029] Furthermore, the mean values ​​of wind and solar capacity factors. and The construction is based on the risk hour set This set is configured by setting a scarcity threshold. We set q = 95% to filter out periods where the net load is higher than the threshold. If no such periods exist, we select the period with the highest net load. Each time period is designated as a risk hour. Wind power capacity factor. With photovoltaic capacity factor These represent the ratios of actual output to installed capacity of wind power and solar power during risk hours, respectively. Their conditional mean is used to assess the reliability of renewable energy during critical periods.

[0030] Furthermore, natural daily load loss By accumulating the expected load loss within the risk hour To obtain, that is This reflects the system's power supply reliability level on that day. To eliminate the influence of extreme values, all characteristic components are standardized using the median and interquartile range (IQR), as shown in the formula: ,in .

[0031] Specifically, weights are assigned to each feature component. Construct weighted feature vectors This vector sequence serves as input in the subsequent dynamic programming model to identify consecutive time periods with similar capacity shortage characteristics, thereby enabling phased reliable capacity assessment.

[0032] Specifically, this step provides a data foundation for subsequent stage division by quantifying the key operational characteristics of the system in the natural day dimension, effectively improving the accuracy and reliability assessment capabilities of resource allocation in the capacity market at different risk stages.

[0033] Furthermore, S1 includes: S11 uses median and interquartile range standardization to process each feature component, and normalizes the feature components using a formula. ; in, Indicates the first Year The first of the natural days The original values ​​of the characteristic components, j=1,2,3,4,5, correspond to the five characteristic values: peak net load, ramp demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, and load shedding, respectively. Represents characteristic components The median of all natural days. Represents characteristic components The interquartile range.

[0034] Specifically, in this invention, the median and interquartile range are used to standardize each feature component to eliminate dimensional differences and the influence of extreme values ​​between different feature dimensions, thereby improving the robustness and accuracy of the subsequent dynamic programming model for stage division. This step is based on robust standardization methods in statistics, using the formula: .

[0035] Furthermore, for each natural day In the year The first in each feature component Standardization processing is performed, among which This represents the median of the characteristic component. This indicates its interquartile range, i.e., the upper quartile. and lower quartile difference, This method is more suitable for power system data with outliers than the traditional Z-score standardization (mean and standard deviation) because it is not sensitive to extreme values ​​and can more realistically reflect the distribution characteristics of the data.

[0036] Specifically, the standardization process first applies each feature component... (Equivalents include peak net load, ramp-up demand, mean capacity factor for wind and solar power, and daily unload.) Statistical calculations are performed year-by-year and day-by-day to obtain the historical median and interquartile range. Then, the median is subtracted from the original eigenvalues, and the result is divided by the interquartile range to obtain the standardized eigenvalues. This process can process multiple years and multiple feature dimensions in parallel, making it suitable for the preprocessing of large-scale power system data.

[0037] Furthermore, the standardization process does not require the introduction of additional hyperparameters, relying solely on the statistical properties of historical data. In practical applications, this step is typically used to construct the natural day feature vector. This was completed previously to ensure that all feature components are weighted and combined on a uniform scale, and its form is as follows: ; in, The weights of each feature component can be set based on actual system operation experience or optimization algorithms.

[0038] Specifically, through robust standardization, the impact of data distribution skewness or outliers on the phase division model can be effectively reduced, the accuracy and stability of the dynamic programming model in identifying homogeneous time periods of capacity shortage risk can be improved, and high-quality input feature vectors can be provided for subsequent phased credible capacity assessment, thereby enhancing the resource allocation efficiency and system reliability assurance capabilities of the capacity market at different stages.

[0039] S12 assigns weights to the conditional mean of the wind power capacity factor. Weights are assigned to the conditional mean of the photovoltaic capacity factor. The weighted feature vector is constructed using the following formula: ; in, arrive The five characteristic components are: peak net load, ramp-up demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, and load shedding. arrive This represents the weight assigned to each corresponding feature component.

[0040] Specifically, this step is technically implemented based on a weighted synthesis mechanism of multi-dimensional feature vectors, aiming to comprehensively reflect the actual contribution capacity of different renewable energy sources during system risk periods. By introducing weight parameters, the average capacity factors of wind power and photovoltaic power during risk hours are weighted, thereby constructing a representative daily feature vector. This is used for subsequent phase division and reliable capacity assessment.

[0041] Specifically, the mean capacity factor conditions for wind power and solar power are defined as follows: and ,in Represents a calendar day In the year The set of risk hours in the data is defined based on a preset scarcity threshold. Typically, q = 95%. If a certain natural day exists... If the time period is within a certain range, then it is defined as the risk hour set; otherwise, the period with the highest net load on that day is selected. One time period (reference value is 3) is used as an alternative.

[0042] Furthermore, to avoid the influence of extreme values ​​on the feature vector, it is necessary to... and Standardization is performed, and the formula is: ,in IQR is the interquartile range. The standardized eigenvalues ​​and preset weights... and Multiplying them together, the final combination forms the feature vector. This is used to characterize the capacity shortage risk of a natural day.

[0043] Specifically, this step is typically used in power system planning and capacity market design, especially in the context of high-proportion renewable energy integration. By quantifying the availability of wind and solar power during risky periods, it provides input for subsequent dynamic planning phase divisions. Weighting and The settings need to be combined with system operation experience and historical data, and are usually determined through sensitivity analysis or expert weighting methods to ensure that the model has a reasonable degree of differentiation in the reliability assessment of wind power and photovoltaic resources.

[0044] Specifically, by introducing the average capacity factor and its weight for wind and solar power during risky periods, the actual supporting capacity of renewable energy in system operation can be reflected more accurately, thereby improving the rationality of phase division and the accuracy of credible capacity assessment.

[0045] S2. Establish a penalized piecewise least squares dynamic programming model. Under the condition of satisfying the shortest stage length constraint, take the weighted sum of the eigenvector dispersion within the stage and the stage number penalty term as the optimization objective, and solve for the optimal stage division scheme that is calendar-continuous throughout the year.

[0046] Specifically, the model obtains the optimal and calendar-continuous stage division scheme for the whole year by minimizing the weighted sum of the dispersion of feature vectors within a stage and the stage number penalty term while satisfying the constraint of the shortest stage length.

[0047] Specifically, the model is first based on the constructed natural day feature vector. Define the annual natural day sequence And set a minimum stage length constraint. To ensure sufficient statistical representativeness at each stage, the model is solved using dynamic programming, defining state DP[j] as the minimum objective function value from day 1 to day j. The state transition equation is: ; in, This represents the squared error within the interval [i, j], and its calculation depends on a predefined weighted prefix. , and The specific expression is: ; in, This represents the weighted number of natural days within the interval. For the weighted eigenvector sum, This is a weighted sum of squares. Penalty term. Used to control the number of stages To avoid over-segmentation, This is the penalty coefficient, which is usually set through cross-validation or empirically, and is typically on the order of 10^3 to 10^5.

[0048] Furthermore, the shortest stage length The duration is typically set between 7 and 30 days to ensure that the phase has seasonal characteristics. The value of needs to balance the model's fitting accuracy with the economy of the number of stages; too large a value will result in too few stages, while too small a value may introduce redundant stages. Furthermore, the model supports weighted processing of data spanning multiple years, with weights... It can be set according to the reliability of historical years or market performance, such as using equal weight or weighting based on historical EUE.

[0049] Specifically, in practical applications, this step is applicable to scenarios such as power system capacity market design, resource optimization and allocation, and reliability assessment. Through data-driven dynamic programming methods, consecutive time periods with similar capacity shortage characteristics throughout the year can be automatically identified, avoiding biases caused by subjective segmentation and improving the capacity market's guarantee capacity during high-risk phases and resource allocation efficiency during low-risk phases.

[0050] Specifically, this model effectively reduces the subjectivity of stage division and improves the accuracy and robustness of staged reliable capacity assessment. Its innovation lies in combining the piecewise least squares method from statistics with the operating characteristics of power systems. By introducing a stage number penalty term, it achieves optimized control over stage division, providing structured input data for subsequent ELCC calculations.

[0051] Furthermore, S2 includes: S21 calculates the squared error within a multi-year period using the cross-year weighted prefix, and calculates the weighted sum of any interval [i,j] based on the prefix. vectors and , sum of squares : ; ; ; in, This represents the weight of the data in year y. Indicates the year y, the year The weighted feature vector of days, where d represents the d-th natural day of the year.

[0052] Specifically, the technical implementation of this step is based on the idea of ​​combining statistical modeling and dynamic programming, aiming to quantify the degree of dispersion of the feature vector of natural days in different time periods, thereby providing a mathematical basis for subsequent stage division.

[0053] Specifically, this step first defines the cross-year weighted prefix quantity. , and These are used to accumulate the weighted sum, weighted sum of squares, and weighted number of days of the feature vectors for each natural day, respectively. Specifically, Indicates from day 1 to day 2. Natural Day Feature Vector for All Years The weighted cumulative value is mathematically expressed as: ; in, For the year The weights are used to reflect the relative importance of data from different years in the model. Similarly, The weighted sum of squares of the eigenvectors: ; and This is the cumulative value of the weighted days: .

[0054] Furthermore, based on the aforementioned prefix quantities, we define a weighted feature vector sum within any interval [i, j]. Weights and and the sum of squares Their expressions are as follows: .

[0055] Specifically, by calculating the squared error over multiple years. This is used to measure the degree of dispersion of the feature vectors within the interval [i,j]. .

[0056] Furthermore, this formula embodies the least squares principle, obtaining an estimate of the variance of the eigenvector within the given interval by subtracting the sum of squares of the weighted means. In practical applications, this step is typically performed after preprocessing historical load and renewable energy output data, and is suitable for joint modeling of multi-year data, particularly important in capacity market assessments that consider interannual variations. Its technical advantage lies in providing an efficient and accurate basis for stage division in dynamic programming models, thereby improving the spatiotemporal resolution and market responsiveness of reliable capacity assessments.

[0057] S22, the optimal piecewise strategy is solved recursively using dynamic programming state transition equations: ; in, This represents the optimal objective function value from day 1 to day i-1. This represents the eigenvector dispersion within stage [i, j]. This represents the penalty coefficient for the stage number.

[0058] Specifically, this step is the core link in the entire technical solution to achieve data-driven stage identification. Its technical principle is based on the combination of piecewise least squares model and dynamic programming algorithm. By minimizing the dispersion of feature vectors within a stage and introducing a segment number penalty term, it achieves adaptive division of the annual load and renewable energy output fluctuation characteristics.

[0059] Specifically, the dynamic programming model uses the natural day as the smallest unit of division, defining the state DP[j] as the optimal objective function value from day 1 to day j, i.e., minimizing the dispersion of the feature components within the stage. To avoid the penalty term in the first stage interfering with the initial state, an initial state is set... The state transition equation is: ; in, This represents the multi-year segment-inclusive squared error within the interval [i, j], and its calculation depends on a predefined weighted prefix quantity. , and It is calculated by the difference between the vector sum and the sum of squares. To constrain the shortest stage length, it is typically set to 7 to 30 days to ensure that the stage division is statistically robust over time. This is the segment penalty coefficient, used to balance the relationship between the number of segments and the dispersion within a segment. Its value range is usually between [0.1, 10], and it needs to be calibrated according to the volatility of historical data.

[0060] Furthermore, The calculation formula is: ; in, Let be the sum of squares of the eigenvectors within the interval. For vector sum, For the weighted sum, all three are preprocessed weighted prefix values. , and Fast computation significantly improves the efficiency of solving dynamic programming problems.

[0061] Specifically, this step is applicable to power system capacity market design, especially in the context of high renewable energy penetration. By identifying time periods throughout the year with similar capacity shortage risks, it provides structured input for subsequent phased credible capacity assessments. The dynamic programming model can be deployed on power market simulation platforms or dispatch systems, combining historical load and wind / solar output data to achieve an objective division of annual operating characteristics.

[0062] Specifically, by introducing a dynamic programming state transition mechanism, the capacity value mismatch problem caused by the annual average assessment method in the traditional capacity market is effectively solved. Its innovation lies in: 1) Dynamic programming modeling based on multi-dimensional feature vectors improves the accuracy of stage identification; 2) Through a segment number penalty term... With the shortest stage length The joint optimization ensures that the phase division achieves a balance between statistical robustness and market implementation complexity; 3) it provides continuous and calendar-consistent phase boundaries for subsequent phased ELCC assessments, enhancing the scientific nature and operability of the capacity market mechanism.

[0063] S3. Based on the stage division scheme, construct a staged marginal effective load-carrying capacity assessment model, comprehensively consider the load loss value, unit start-up and ramp-up constraints, heating rigid constraints, and energy storage charging and discharging and SOC constraints, and calculate the reliable capacity of power generation resources in each stage.

[0064] Specifically, based on the optimal stage division result obtained from dynamic programming, the model divides the year into several consecutive calendar stages. In each stage, an energy market clearing model that considers multiple system operation constraints is established to assess the marginal effective load-carrying capacity (ELCC) of power generation resources.

[0065] Furthermore, the model uses minimizing the total power generation cost as its objective function, comprehensively considering key economic indicators such as the variable operating cost, start-up and shutdown cost, and loss load cost of the generating units. The objective function takes the form: ; in, This represents a collection of coal-fired power units. It represents a collection of energy storage units. Represents the set of load nodes. For the value of loss of load, Represents a node During the period The expected unmet load is assessed by simulating the economically optimal solution of the system operation at different stages, evaluating the marginal contribution of power generation resources in meeting load demand.

[0066] Specifically, regarding constraints, the model comprehensively considers operational limitations of power generation resources, including unit start-up and shutdown and ramp-up constraints, rigid heating constraints, renewable energy availability constraints, energy storage charging and discharging and SOC constraints, and system power balance and transmission line constraints. For example, the ramp-up constraint for coal-fired units is: ; in, , The units The uphill and downhill climbing rates, For the unit During the period The operating status (0-1 variable). Energy storage operation constraints include upper and lower limits of charging and discharging power, upper and lower limits of SOC, and mutually exclusive charging and discharging states, to ensure the rational operation of energy storage resources at both physical and economic levels.

[0067] Furthermore, the model introduces Value at Loss of Load (VOLL) as a quantitative indicator of reliability cost, typically set to a value of [value missing]. This reflects the economic cost of system load shedding. Furthermore, the model calculates the reliable capacity ratio by introducing a comparison mechanism between "perfect" power generation resources and target power generation resources. Its definition is: ; in, This represents the unmet energy expectation in the initial scenario. and These represent the improvement values ​​after adding the "perfect" resource and the target resource, respectively. This ratio reflects the marginal contribution of the target resource to system reliability at a specific stage and is a core indicator for trusted capacity assessment.

[0068] Specifically, by combining system operating characteristics with the physical and economic constraints of power generation resources, this model enables a refined assessment of the reliability value of power generation resources at different stages. Compared to traditional year-round unified assessment methods, this model can more accurately identify resource scarcity during high-risk phases, thereby providing a scientific basis for phased pricing and resource allocation in the capacity market and improving the economy and security of system operation.

[0069] Furthermore, S3 includes: S31 constrains the state of charge of the energy storage unit, limiting the upper and lower limits of the energy storage capacity through formulas: ; in, Indicates time Energy storage capacity; Representing time The state of charge; and These are the minimum and maximum allowable limits for the state of charge.

[0070] Specifically, in constructing a phased generation resource reliability capacity assessment model based on marginal effective load capacity (ELCC), constraining the state of charge (SOC) of energy storage units is a crucial step in ensuring the safety and economy of system operation. This step involves defining the state of charge (SOC) of energy storage units at any given time. The SOC range ensures that the system stability and equipment lifespan will not be affected by over-discharge or over-charge during operation.

[0071] Furthermore, the SOC constraint of the energy storage unit is modeled using the following formula: ; in, Indicates energy storage unit In time The state of charge, For time energy storage capacity, For energy storage units Rated capacity, and These are the minimum and maximum permissible limits for SOC, respectively. Typically, The value is 0.2. The value is set to 0.9 to avoid battery life degradation caused by deep discharge and safety risks caused by overcharging.

[0072] Furthermore, energy storage capacity The dynamic evolution is described by the following equation: ; In the formula, and Energy storage units In time The discharge power and charging power, and These are discharge efficiency and charging efficiency, respectively. The time interval is typically 1 hour. This formula reflects the energy conservation relationship of the energy storage unit during the charging and discharging process, and incorporates efficiency parameters to correct for actual charging and discharging behavior.

[0073] Specifically, this constraint is embedded in the optimization framework of the phased reliable capacity assessment model, working in conjunction with constraints such as generator start-up and shutdown, ramp-up, and output limits to ensure that the operating status of energy storage resources remains within a safe range at different stages. Especially in high-risk stages (such as extreme weather or peak load periods), the SOC constraint helps prevent energy storage resources from losing their regulatory capacity due to overuse, thereby improving the reliability of the system during critical periods.

[0074] Specifically, by precisely controlling the SOC range of energy storage units, the abuse or idleness of energy storage resources is avoided, and their effective load-carrying capacity in the phased capacity market is improved. At the same time, combined with the optimal stage divided by the dynamic programming model, the SOC constraint can adapt to the load and renewable energy output characteristics of different stages, realize the optimal allocation of energy storage resources in the time dimension, thereby improving the operating efficiency of the entire system and the scientific nature of the capacity market mechanism.

[0075] S32 introduces transmission line constraints, using a formula to ensure that the system power transmission does not exceed the maximum line capacity: ; ; in, It is a line Maximum transmission capacity It is the power transfer distribution factor of the system.

[0076] Specifically, this step uses the power transfer distribution factor (PTDF) to model the impact of each generator unit, new energy unit, energy storage unit, and load shedding on the power flow of transmission lines in the power system, thereby avoiding line overload and ensuring stable system operation during the optimization process.

[0077] Furthermore, transmission line constraints are modeled using the following formula: .

[0078] Furthermore, this formula describes the time at any given moment. and any line The power flow should meet its maximum transmission capacity. Upper and lower bound constraints. Among them, Indicates the line For coal-fired power units The power transfer distribution factor, For the unit At any moment The output power; similarly, and Corresponding to new energy units Power of the line The impact; With energy storage units charging and discharging power , Related; Used for quantization nodes Loss of load With load demand For the line The influence of trends.

[0079] Specifically, The maximum transmission capacity is typically set based on power grid planning standards or actual operating data, such as determining the line thermal stability limit or voltage stability limit. The PTDF value is obtained through power flow sensitivity analysis, and its calculation requires a basis on the system topology and baseline power flow distribution, usually using a DC power flow model for approximate calculation.

[0080] Specifically, this constraint is embedded in the phased power market clearing model to ensure that the system power flow remains within a safe range during capacity allocation and operation scheduling at each stage. Especially during high-risk phases (such as peak summer loads or winter cold waves), this constraint helps identify and mitigate line overload risks caused by uneven resource distribution, thereby improving system reliability.

[0081] Specifically, by quantifying the power flow constraints of transmission lines, the physical constraints of the power grid are closely integrated with the capacity market mechanism, avoiding system operation risks caused by ignoring line transmission capacity, and enhancing the engineering feasibility and market guidance significance of credible capacity assessment results.

[0082] S4. By comparing the expected improvement in unused energy of the system after adding a unit of installed capacity of perfect power generation resources and target power generation resources to the initial power structure, the reliable capacity coefficient of the target power generation resources is determined.

[0083] Specifically, this step is based on the principle of Marginal Effective Load Carrying Capability (ELCC), which quantifies the marginal contribution of different resources to the system reliability index, thereby enabling a refined assessment of the capacity value of power generation resources at different stages.

[0084] Specifically, the first step is to calculate each stage based on the initial power structure operation results. Expected Unsupplied Energy (EUE) within the range is denoted as This value reflects the system's reliability gap without any additional resources. Subsequently, during this phase, 1MW of "perfect" power generation resources—a virtual generator group—is added to the system. This virtual generator group is fully controllable, has no ramp-up limitations, no fuel constraints, and can operate at full capacity around the clock. The EUE value after adding the "perfect" resources is obtained by re-running the market clearing model. And calculate its improvement value. This value represents the maximum potential of ideal resources to improve system reliability.

[0085] Furthermore, by adding 1MW of target power generation resources (such as coal-fired units, wind power, photovoltaics, etc.) to the same initial power structure, and re-running the market model, a new EUE value is obtained. And calculate its improvement value. . Reliable capacity coefficient of target power generation resources Defined as the ratio of this improvement value to the "perfect" resource improvement value, that is: ; Furthermore, this coefficient reflects the relative contribution of the target resource to the improvement of system reliability in a specific stage. The closer the value is to 1, the higher its reliability and the closer it is to the performance of a "perfect" resource.

[0086] Specifically, the calculation of EUE needs to consider multiple factors such as load demand, renewable energy output, energy storage operation status, unit start-up and shutdown constraints, and ramp-up constraints. Its numerical accuracy is typically required to be within 0.1 MW·h. The calculation of the reliability capacity coefficient needs to be performed independently in each stage to ensure that it reflects the stage-specific capacity value differences. Furthermore, this step must be performed based on the optimal stage boundaries determined by dynamic programming to ensure consistency between stage division and reliability assessment.

[0087] Specifically, this step is applicable to scenarios such as capacity market settlement, resource planning, and scheduling optimization. By calculating the staged reliable capacity coefficient, differentiated capacity pricing can be provided for different seasons and periods with different load characteristics, thereby improving the incentive efficiency of the market mechanism for resource reliability.

[0088] Specifically, by quantifying the marginal reliability contribution of resources, the capacity value mismatch problem caused by the traditional annual average reliable capacity coefficient is avoided, providing a scientific basis for the refined management of the capacity market and enhancing the system's power supply guarantee capability during high-risk phases.

[0089] S5, When calculating the squared error within a multi-year period, define the cross-year weighted prefix quantity. , and And based on the formula, optimize the solution efficiency of the dynamic programming model: ; in, This represents the sum of squares of the weighted eigenvectors for all days and all years within the time period [i, j]. This represents the total weight of all days and all years within the time period [i, j]. This represents the vector sum of the eigenvectors of all days and all years within the time period [i, j].

[0090] Specifically, this step involves defining the cross-year weighted prefix quantity. , and This enables the rapid calculation of the weighted squared error of the eigenvectors within any time interval [i, j], thereby improving the efficiency of solving dynamic programming problems.

[0091] Furthermore, the definition of the cross-year weighted prefix is ​​based on historical years. Natural day feature vector ,in This represents the weighted sum of eigenvectors for all years from day 1 to day d. This represents the weighted sum of squares of the eigenvectors within the same time period. This represents the weighted total number of days. The specific formula is as follows: ; ; ; in, For the year The weights are used to reflect the importance of different years in historical data. By pre-compiling these prefixes, the weighted mean of any interval [i, j] can be quickly calculated. With weighted squared error Its expression is: ; in, This represents the weighted sum of days within the interval. This represents the weighted sum of eigenvectors within the interval. This represents the weighted sum of squares within the interval. Based on the principle of variance calculation in statistics, this formula quantifies the dispersion of the eigenvectors within the interval by subtracting the squared mean term.

[0092] Specifically, weight The minimum stage length is typically set based on the representativeness of the year or the quality of the data, such as using equal weighting or a weighting method based on historical load fluctuations. The timeframe is typically set between 7 and 30 days to ensure that the phase divisions are meaningful in terms of time and to avoid excessive segmentation that could increase model complexity.

[0093] Specifically, this step is mainly used in practical applications for the recursive solution of dynamic programming models. By quickly calculating the squared error of any time period, it supports finding the optimal stage division that minimizes the objective function while satisfying the minimum stage length constraint. Its technical effect is to significantly reduce computational complexity, improve model solution efficiency, and ensure that the stage division is statistically robust, providing a clear and risk-homogeneous time interval division basis for subsequent phased reliable capacity assessment.

[0094] This invention presents a phased reliable capacity assessment method for power generation resources. Based on dynamic phase division and multi-constraint integrated modeling, it effectively solves the problems of capacity value mismatch and insufficient accuracy caused by neglecting the time-varying characteristics of system state in traditional year-round unified assessments. This method achieves fully automated optimization of the entire process, from constructing multi-dimensional daily feature sequences and penalized dynamic programming for optimal phase division to refined assessment of marginal effective load-carrying capacity, significantly improving the accuracy, computational efficiency, and model robustness of reliable capacity calculation. While ensuring the reliability of power supply, it enhances the targeting and economy of capacity market resource allocation, providing scientific decision support for system planning, dispatching operation, and market mechanism design under high-proportion renewable energy access, and comprehensively improving the efficiency of risk assessment and capacity management in complex operating scenarios.

[0095] Example 2 To achieve the above invention, embodiments of the present invention also provide specific steps for a phased power generation resource reliability capacity assessment method, including: S101, Construct the feature vector of natural day: First, build year by year The smallest unit of division within an area is the natural day. Define its eigenvectors .

[0096] S1011, Peak Net Load .

[0097] Specifically, define hourly net load. :

[0098] The maximum peak net load of the day is taken as the characteristic component. .

[0099] S1012, 3-hour uphill climb requirement .

[0100] Specifically, the characteristic component is the maximum 3-hour uphill climb demand during the evening of that day. .

[0101] S1013, Average wind power conditions during risk hours .

[0102] First, based on historical data from year to year, day to day, and hour to hour, a certain quantile is taken. The reference value is 95%, and the scarcity threshold is defined as follows:

[0103]

[0104] That is, when there is a net load greater than the scarcity threshold on a natural day, the corresponding time period is defined as the risk hour set of that natural day; otherwise, the m time periods with the maximum net load on that day are selected, with a reference value of 3.

[0105] Furthermore, the historical wind power generation capacity factor is defined. :

[0106] In the formula, These represent wind power generation capacity and installed capacity, respectively.

[0107] Specifically, the average wind power conditions during risk hours. .

[0108] S1014, Average value of photovoltaic capacity factor under risk hour conditions. .

[0109] Specifically, the mean value of the photovoltaic capacity factor under risk hour conditions. .

[0110] S1015, daily load loss .

[0111] Specifically, calculations are based on expected unserved energy (EUE). .

[0112] To avoid the influence of extreme values, each characteristic value is standardized using the following formula:

[0113] Assign weights to each feature component Then, construct the natural day feature vector. :

[0114] S102, Construct a dynamic programming model: Specifically, on a unified natural day Choose the stage boundary above, and assume the whole year is divided into... In stages, then: S1021, Calculation of the mean within a segment: Specifically, breakpoints include ,stage The range of natural days covered is Shortest stage length: The mean within the segment is defined as follows:

[0115] in, Weighting is based on annual changes.

[0116] S1022, The objective function of the dynamic programming model is to minimize the dispersion of the feature components within each stage:

[0117]

[0118] S1023, Calculate the squared error over multiple years: Specifically, define the cross-year weighted prefix:

[0119]

[0120]

[0121] Furthermore, let For any interval We have: The weighted sum is: The vector sum is: The sum of squares is: .

[0122] Furthermore, the squared error within a multi-year period is:

[0123] S103, determine the optimal continuous segment for reliable capacity assessment.

[0124] S1031, Determine the state: First, define the state. Let be the minimum value of the objective function from day 1 to day j, i.e., minimize the dispersion of the feature components within the stage. To avoid additional penalties for the first state transition, let .

[0125] S1032, Determine the state transition equation: This describes the evolutionary relationship between states, i.e., how to determine the state at the next moment based on the current state. That is, starting from the candidate starting point... The state changes that come with the addition of new stages.

[0126] ; S1033, Constructing a dynamic programming table: Based on the state and state transition equations, construct a dynamic programming table.

[0127] S1034, Recursively Solving for the Optimal Solution: The dynamic programming table is filled in recursively, starting from the initial state and updating the values ​​in the table step by step according to the state transition equation until the target state is reached.

[0128] S1035, Backtracking to obtain the optimal strategy: Based on the dynamic programming table, the optimal decision sequence is obtained through backtracking, that is, the number of stages K and the optimal continuous segmentation strategy are obtained. .

[0129] S104, Construct a phased reliable capacity assessment model based on ELCC.

[0130] S1041, Objective function: Specifically, the proposed model minimizes the total generation cost, including the variable operating cost of the generator set, start-up and shutdown costs, and loss load costs:

[0131] in, It is a collection of coal-fired power units. It is a collection of ESS. It is a set of net nodes. It is a generator set In time 'output power' It is a generator set In time The cost of bidding and It indicates the generator set In time The power on / off action, and The units Start-up and shutdown costs and Energy storage units In time The charging and discharging power, Represents a node exist The expected energy not provided at the time Value at the loss of load.

[0132] S1042, Constraints: S10421, Power generation resource output constraints: Specifically, each unit in the first tThe power output during a given time period should meet the upper and lower limits of the output for that time period.

[0133] in, It is a generator set In time The output, It is a 0-1 variable used to represent the unit. In time Are you online? and It is a generator set In time The minimum and maximum outputs.

[0134] S10422, Power generation resource ramping constraint: Specifically, generator sets, especially coal-fired units, face upward and downward power constraints in terms of output:

[0135]

[0136] in, , Generator sets The uphill and downhill climbing rates.

[0137] S10423, Forced Downtime Rate Constraint:

[0138] in, For generator sets The installed capacity, For generator sets Forced shutdown rate.

[0139] S10424, Special Unit Constraints:

[0140] in, For heating and the collection of units that must be in operation, For the unit exist The operational status at any given moment.

[0141] S10425, Constraints on New Energy Units:

[0142] in, For new energy units The installed capacity, For new energy units At any moment Available coefficients.

[0143] S10426, Energy Storage Operation Constraints: Specifically, the energy storage power source is modeled as a price taker, and its charging and discharging behavior is influenced by market prices. The specific operational constraints are as follows:

[0144]

[0145]

[0146]

[0147]

[0148]

[0149]

[0150] in, and Representing energy storage units In time Charging power and discharging power; and These are the lower and upper limits of the charging power; and The lower and upper limits of discharge power; binary variables. and Used to indicate the energy storage unit at time Is it in a charging / discharging state? Indicates time Energy storage capacity; This represents the energy storage value from the previous time period; and These are discharge efficiency and charging efficiency, respectively. It is a time interval; Representing time The state of charge (SOC); and These are the minimum and maximum allowable limits for the state of charge.

[0151] S10426, Power balance constraint:

[0152] in, It is a collection of new energy generating units. It is a moment The load demand.

[0153] S10427, Transmission line constraints:

[0154] in, It is a line Maximum transmission capacity It is the power transfer distribution factor of the system.

[0155] S1043, phased reliable capacity solution.

[0156] Specifically, based on the market solution results, each stage is calculated separately. Within this context, the expected power unload (EUE) of the system is denoted as the initial test scenario. If the system does not meet the specified reliability standards, the system load requirements can be adjusted proportionally.

[0157] Furthermore, a "perfect" power generation resource per unit installed capacity (1MW) is added to the initial power structure. This "perfect" power generation resource refers to a perfectly controllable virtual machine group that is not subject to fuel limitations or ramp-up / downward constraints. Subsequently, the expected unused energy of the system under the new power structure scenario is calculated. Calculate the expected improvement in unused energy resulting from "perfect" power generation resources, i.e. .

[0158] Furthermore, a target generating resource with a unit installed capacity (1MW) is added to the initial power structure scenario. This target generating resource refers to the generating units whose credible capacity values ​​are to be calculated. Subsequently, the expected unused energy of the system under the new power structure scenario is calculated. Calculate the expected improvement in unused energy brought about by the target power generation resources, i.e. .

[0159] Specifically, the credible capacity value of the target power generation resource is calculated as the ratio of the expected improvement in unused energy brought about by the target power generation resource to the expected improvement in unused energy brought about by the "perfect" power generation resource, as shown in the following formula:

[0160] This invention discloses a phased reliable capacity assessment method for power generation resources. Based on multi-dimensional daily feature sequences and penalized dynamic programming, it achieves optimal phase division throughout the year, effectively overcoming the inherent limitations of traditional assessment methods, such as capacity value estimation bias and insufficient adaptability caused by neglecting the time-varying characteristics of system states. It realizes a systematic assessment throughout the entire process, from key feature extraction, robust standardization, adaptive phase clustering to multi-constraint refined modeling, significantly improving the accuracy, computational efficiency, and engineering practicality of reliable capacity calculation. While ensuring the power supply reliability of high-proportion renewable energy power systems, it provides accurate decision-making basis for scientific pricing, resource optimization, and long-term planning in the capacity market, comprehensively enhancing the efficiency and robustness of system adequacy management under complex operating environments.

[0161] Example 3 To achieve the above invention, such as Figure 2 As shown, this embodiment also provides a phased power generation resource reliability capacity assessment device 10, which includes: The feature vector construction module 100 is used to construct a natural daily feature vector sequence containing peak net load, ramping demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, and load shedding amount based on the hourly load and renewable energy output data of the whole year of the target capacity delivery year. The dynamic programming model building module 200 is used to build a penalized piecewise least squares dynamic programming model. Under the condition of satisfying the shortest stage length constraint, the optimal stage division scheme with the weighted sum of the eigenvector dispersion within the stage and the stage number penalty term is used as the optimization objective. The reliable capacity calculation module 300 is used to construct a phased marginal effective load-carrying capacity assessment model according to the phase division scheme, and comprehensively consider the load loss value, unit start-up and ramp-up constraints, heating rigid constraints, and energy storage charging and discharging and SOC constraints to calculate the reliable capacity of power generation resources in each phase. The credible capacity coefficient determination module 400 is used to determine the credible capacity coefficient of the target power generation resource by comparing the expected unused energy improvement value of the system after adding a unit of installed capacity of perfect power generation resource and target power generation resource to the initial power structure.

[0162] In one embodiment of the present invention, it further includes: a cross-year weighted prefix calculation module, used to define the cross-year weighted prefix when calculating the squared error within a multi-year period. , and And based on the formula, optimize the solution efficiency of the dynamic programming model: ;in, This represents the sum of squares of the weighted eigenvectors for all days and all years within the time period [i, j]. This represents the total weight of all days and all years within the time period [i, j]. This represents the vector sum of the eigenvectors of all days and all years within the time period [i, j].

[0163] This invention provides a phased power generation resource reliability capacity assessment device. By integrating feature vector construction, dynamic programming model solving, and refined reliability capacity calculation modules, it effectively solves the problems of capacity value mismatch and coarse assessment results caused by traditional reliability capacity assessment methods neglecting the time-varying characteristics of system states. This device achieves fully automated assessment from multi-dimensional feature extraction and adaptive optimal stage division to multi-constraint operation simulation and marginal contribution quantification, significantly improving the accuracy, computational efficiency, and engineering applicability of reliability capacity assessment. While ensuring the power supply reliability of the power system under high-proportion renewable energy access, it provides a powerful decision support tool for accurate pricing in the capacity market, scientific resource allocation, and long-term system planning, comprehensively enhancing the intelligence and robustness of system adequacy management under complex operating environments.

[0164] To implement the methods of the above embodiments, the present invention also provides a computer device, such as... Figure 3 As shown, the computer device 600 includes a memory 601 and a processor 602; wherein, the processor 602 reads the executable program code stored in the memory 601 to run a program corresponding to the executable program code, so as to implement the various steps of the phased power generation resource reliable capacity assessment method described above.

[0165] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a phased power generation resource reliability capacity assessment method as described in the foregoing embodiments.

[0166] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0167] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for phased generation resource credible capacity assessment, characterized in that, The method comprises the following steps: S1, based on the hourly load and renewable energy output data of the target capacity delivery year, a natural day feature vector sequence is constructed, which comprises peak net load, ramping demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean and loss of load; S2, a penalty piecewise least squares dynamic programming model is established, and under the condition of meeting the shortest stage length constraint, the weighted sum of the feature vector dispersion in the stage and the stage number penalty term is taken as the optimization objective to solve the optimal and calendar continuous stage division scheme for the whole year; S3, a staged marginal effective load carrying capacity evaluation model is constructed according to the stage division scheme, and the loss of load value, unit start-stop and ramping constraint, heating rigidity constraint and energy storage charging and discharging and SOC constraint are comprehensively considered to calculate the reliable capacity of each stage of the power generation resource; S4, the reliable capacity coefficient of the target power generation resource is determined by comparing the expected unused energy improvement value of the system after adding unit installed capacity of perfect power generation resource and target power generation resource to the initial power supply structure.

2. The method of claim 1, wherein, The method comprises the following steps: S11, the median and quartile range standardization processing is adopted to process each feature component, and the feature component is normalized by formula; ； wherein, represents the day of the week of the year, and represents the day of the year, and represents the original numerical value of the j-th feature component, j = 1, 2, 3, 4, 5, respectively corresponding to the peak load, the ramping demand, the wind power capacity factor conditional mean, the photovoltaic capacity factor conditional mean, and the loss of load amount, represents the feature component in the median of all natural days, represents the feature component interquartile range. S12, assign weights to the wind capacity factor conditional mean S14, assign weights to the PV capacity factor conditional mean S15, construct a weighted feature vector by the formula ； wherein, to corresponding peak load, ramping demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean, loss of load amount five characteristic components, to represent the weight assigned to each corresponding characteristic component.

3. The method of claim 1, wherein, The method comprises the following steps: S21, calculate the square error in the multi-year section by cross-year weighting prefix amount, and calculate the weight of any interval [i,j] based on the prefix amount , vector sum , square sum : ； ； ； in, This represents the weight of the data in year y. Indicates the year y, the year The weighted feature vector of days, where d represents the d-th natural day of the year; S22, the optimal segmentation strategy is recursively solved by using the dynamic programming state transition equation: ； wherein, represents the optimal objective function value from day 1 to day i-1, represents the feature vector dispersion within phase [i,j], represents the phase number penalty coefficient.

4. The method of claim 1, wherein, The method comprises the following steps: S31, the state of charge of the energy storage unit is constrained, and the upper and lower limits of the energy storage capacity are limited by formula: ； wherein represents the energy storage capacity, of time represents the state of charge, of time and are the allowed minimum and maximum limits of the state of charge; S32, the transmission line constraint is introduced, and the system power transmission is ensured not to exceed the maximum capacity of the line by formula: ； ； wherein is the maximum transmission capacity of the line , is the power transfer distribution factor of the system.

5. The method of claim 1, wherein, The method comprises the following steps: S5, define the cross-year weighting prefix quantity when calculating the squared error over multiple years , and and optimize the solution efficiency of the dynamic programming model based on the formula: ； wherein, denotes the sum of the squared weighted feature vectors for all days, all years, within the time period [i, j], denotes the sum of the total weights for all days, all years, within the time period [i, j], denotes the vector sum of the feature vectors for all days, all years, within the time period [i, j].

6. An apparatus for staged generation resource credible capacity assessment, characterized in that, The feature vector construction module is configured to construct a natural day feature vector sequence based on the hourly load and renewable energy output data of the target capacity delivery year, which comprises peak net load, ramping demand, wind power capacity factor conditional mean, photovoltaic capacity factor conditional mean and loss of load; The dynamic programming model establishment module is configured to establish a penalty piecewise least squares dynamic programming model, and under the condition of meeting the shortest stage length constraint, the weighted sum of the feature vector dispersion in the stage and the stage number penalty term is taken as the optimization objective to solve the optimal and calendar continuous stage division scheme for the whole year; ​ A credible capacity calculation module is configured to construct a staged marginal effective band capacity evaluation model according to the stage division scheme, comprehensively consider the lost load value, unit start-stop and climbing constraints, heat supply rigidity constraints, and energy storage charging and discharging and SOC constraints, and calculate the credible capacity of the power generation resources in each stage. A credible capacity coefficient determination module is configured to determine the credible capacity coefficient of the target power generation resources by comparing the expected unused energy improvement values of the system after respectively adding unit installed capacity of perfect power generation resources and target power generation resources to the initial power source structure.

7. The apparatus of claim 6, wherein, Further comprising: The cross-year weighting prefix quantity calculation module is configured to define the cross-year weighting prefix quantity when calculating the square error in the multi-year section , and and optimize the solving efficiency of the dynamic programming model based on the formula: ； wherein, denotes the sum of the squares of the weighted feature vectors for all days, all years, within the time period [i, j], denotes the sum of the total weights for all days, all years, within the time period [i, j], denotes the vector sum of the feature vectors for all days, all years, within the time period [i, j].

8. An electronic device comprising: a processor; a memory storing executable instructions; the processor executes the instructions to implement the method of any one of claims 1-5.

9. A computer readable storage medium storing a computer program, the computer program being executed by a processor to implement the method of any one of claims 1-5.

Images