Method for realizing coordinated evolution of power market price mechanism

Abstract

The application discloses a power market electricity price mechanism coordinated evolution implementation method, and particularly relates to the technical field of power market operation and regulation; based on a multi-dimensional power market scene set, evolution track characteristic parameter sets of the electricity price mechanism are extracted; a similarity clustering algorithm is used to divide the electricity price mechanism into coupled mechanism clusters, and a coupling degree coefficient matrix is constructed; a coordinated evolution function group is constructed based on the coupling degree to represent the linkage response relationship between the mechanisms; a disturbance factor is introduced to optimize the robustness of the model, forming a robust evolution model; the optimized model is applied to the current operating state of the target area, and the electricity price mechanism state prediction result of the next period is output; the electricity price tearing point and the mechanism conflict point in the prediction result are further identified, and the high-risk regulation interval is marked; the application can realize dynamic linkage modeling and stability early warning between multiple mechanisms, and improve the adaptability and regulation precision of the electricity price mechanism under high uncertainty scenarios.

Description

Technical Field

[0001] This invention relates to the field of power market operation and regulation technology, specifically to a method for the coordinated evolution of power market pricing mechanisms. Background Technology

[0002] With the continuous increase in the proportion of renewable energy and the increasing participation of users, the structural uncertainty of the electricity market has increased significantly. Frequent fluctuations in electricity price signals have become the norm, and in severe cases, "abrupt" jumps can occur in a short period of time, disrupting user trading behavior and generation-side dispatch plans. Current electricity pricing mechanisms are mostly based on marginal price models, but they suffer from problems such as signal feedback delays and insufficient coordination when dealing with scenarios involving large-scale distributed resource integration and rapid changes in grid power flow.

[0003] Especially in regional electricity spot market pilots, when rare scenarios occur such as drastic cross-regional load migration and concentrated high-proportion renewable energy output, the independent operation of regional electricity pricing mechanisms often leads to price fragmentation (discontinuous pricing), a surge in ancillary service costs, and a sharp increase in grid regulation pressure, ultimately causing an imbalance in the linkage between "price-dispatch-load". Traditional methods lack the ability to construct a coordinated evolution relationship between various sub-mechanisms (such as spot market, capacitor, power forecasting, and user-side response) at the mechanism level, and cannot adapt to the rapidly evolving operational landscape.

[0004] Therefore, there is an urgent need for a method that can dynamically construct the collaborative relationship between electricity pricing mechanisms and support the coordinated evolution of multiple mechanisms in the electricity market. In particular, it is necessary to address the mechanism integration problem under extremely high uncertainty scenarios, so as to improve the collaborative and adaptive capabilities of electricity pricing mechanisms and avoid systemic disturbances during operation. Summary of the Invention

[0005] The purpose of this invention is to provide a method for the coordinated evolution of electricity market pricing mechanisms to address the shortcomings in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for the coordinated evolution of electricity market pricing mechanisms, comprising: S100: Collect historical operational data of the target area and construct a multi-dimensional set of electricity market scenarios that includes time, region, load level, and energy type; S200. Based on a multi-dimensional electricity market scenario set, extract the evolution trajectory characteristic parameter set M of each electricity price mechanism under typical scenarios. M includes the rate of change of the spot electricity price slope, the periodicity of the peak-valley price difference, and the ancillary service cost coefficient. S300. Based on the set of evolution trajectory feature parameters M, the electricity pricing mechanism is divided into coupling mechanism clusters G using a similarity clustering algorithm, and the coupling coefficient matrix C between different mechanism clusters is quantified. S400. Based on the coupling coefficient matrix C, construct a mechanism co-evolution function group F to characterize the linkage response relationship between different mechanisms. Each function in F takes the dynamic features in M ​​as input variables. S500 introduces a perturbation factor δ to simulate the mechanism perturbation response under abnormally high uncertainty scenarios, and adjusts the function parameters in F to form a robust evolution model F′; S600. Apply the robust evolution model F′ to the current operating state of the target area, apply F′ to the current electricity price mechanism state vectors for linkage adjustment, and output a new electricity price mechanism state prediction vector P. S700 identifies potential electricity price tear points or mechanism conflict points based on the electricity price mechanism state prediction vector P and marks them as high-risk control intervals.

[0007] Preferably, the extraction of the rate of change of the spot electricity price slope includes the following steps: Based on a multi-dimensional electricity market scenario set, the spot market clearing price in each time frame is reconstructed to generate a continuous electricity price time series. The sliding window method was used to perform local linear regression on the electricity price time series, and the slope value of the electricity price within each window was calculated. Wavelet transform was used to perform multi-scale analysis on the electricity price slope sequence to extract the rate of change of local extreme slopes in different frequency bands. Based on preset sensitivity thresholds, abnormal slope jump points are screened to form the rate of change of spot electricity price slope in high uncertainty scenarios.

[0008] Preferably, the periodic extraction of the peak-valley price difference includes the following steps: Based on the multi-dimensional electricity market scenario-based intraday load level dimension, a typical daily template set for the daily electricity price curve is constructed. Fourier transform is used to map the daily spot electricity price curve to the frequency domain to identify the dominant frequency components; The peak-valley price difference cyclical index is calculated based on the dominant frequency and amplitude.

[0009] Preferably, the extraction of the ancillary service cost coefficient includes the following steps: Based on the multi-dimensional electricity market scenario, extract the ancillary service call records and response costs within the corresponding time period. A support vector regression model is used to model the relationship between ancillary service costs and regional load levels, forming a cost prediction function; The marginal cost of ancillary services corresponding to a unit change in load is calculated and defined as the ancillary service cost coefficient.

[0010] Preferably, the step of using a similarity clustering algorithm to divide the electricity pricing mechanism into coupling mechanism clusters G, and quantifying the coupling coefficient matrix C between different mechanism clusters includes the following process: Based on the set of evolutionary trajectory feature parameters M, a multi-dimensional feature vector space is constructed to uniformly map the feature parameters of various electricity pricing mechanisms under different typical scenarios into a high-dimensional vector representation. A density-based similarity clustering algorithm is used to cluster high-dimensional feature vectors to identify a set of electricity pricing mechanisms that are highly consistent in their electricity price response behavior, which is defined as the coupling mechanism cluster G. Between the formed coupling mechanism clusters, the coupling strength is calculated based on the cosine similarity between the center points of the feature vectors, and the coupling coefficient between the mechanism clusters is obtained. The coupling coefficients are organized into cluster pairs to construct the coupling coefficient matrix C.

[0011] Preferably, the step of constructing a mechanism co-evolution function set F based on the coupling coefficient matrix C to characterize the linkage response relationship between different mechanisms includes: Using the coupling coefficients between clusters in the coupling coefficient matrix C as weight parameters, the dynamic features of the corresponding mechanism clusters in the evolution trajectory feature parameter set M are weighted and correlated to establish a feature transfer relationship across mechanism clusters; For each target mechanism cluster, a co-evolution function is constructed with its own dynamic characteristics and the weighted characteristics of the associated mechanism clusters as independent variables. The function adopts a nonlinear mapping form to characterize the amplification or suppression effect of the linkage response. The co-evolution functions corresponding to each mechanism cluster are combined to form a mechanism co-evolution function group F.

[0012] Preferably, the step of adjusting the function parameters in F to form a robust evolutionary model F′ includes: The perturbation factor δ is injected into the input of the co-evolution function group F of the mechanism to perform perturbation enhancement processing on the input variables in the set of evolution trajectory feature parameters M, forming a perturbation feature sample set; Perform parameter sensitivity analysis on each co-evolution function to identify highly sensitive parameters that cause output shift under perturbation samples; All perturbation-adjusted evolution functions are recombined to form a robust evolution model F′.

[0013] Preferably, the step of applying the robust evolution model F′ to the current operating state of the target area and applying F′ linkage adjustment to the current electricity price mechanism state vectors includes: The electricity pricing mechanism state vector of the target area is collected within the current time period. The electricity pricing mechanism state vector is composed of the evolution trajectory feature parameters of each mechanism cluster. The current state vector is used as an input variable and input into the robust evolution model F′ after perturbation optimization to perform nonlinear mapping calculations on the linkage relationship between each mechanism cluster; Based on the calculation results of the robust evolution model F′, the output is the prediction vector P of the electricity price mechanism state of the target region in the next time period, where each element corresponds to the predicted evolution value of the mechanism cluster.

[0014] Preferably, the step of identifying potential electricity price tear points or mechanism conflict points based on the electricity price mechanism state prediction vector P includes: The difference between the state value of each mechanism cluster in the electricity price mechanism state prediction vector P and the current state vector is calculated to construct an offset vector; Based on the degree of abrupt change in the rate of change of the slope of the spot electricity price in the offset vector, it is determined whether there is a price tear point. If the degree of abrupt change exceeds a preset threshold, it is marked as a tear risk. Calculate the collaborative bias of predicted feature values ​​among different mechanism clusters. When the bias directions are opposite and the coupling coefficient is greater than the threshold, it is marked as a mechanism conflict point. The time period in which the tear or conflict occurs will be output as the high-risk control range.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention constructs a set of characteristic parameters for the evolution trajectory of electricity pricing mechanisms based on multi-dimensional scenarios, and combines a coupling coefficient matrix and a co-evolution function group to achieve dynamic modeling of the linkage relationship of multiple mechanism clusters in the electricity market. It can accurately capture the evolution trend and mutual influence between different mechanisms, improve the fine-grained expression capability of electricity pricing mechanism response modeling, and overcome the limitations of existing methods that model individual electricity pricing mechanisms and lack coupling analysis.

[0016] 2. This invention further introduces disturbance factors for robustness modeling, and achieves accurate positioning of electricity price tear points and mechanism conflict points through evolution prediction output and risk identification algorithm, so that the electricity price mechanism prediction results have strong anti-disturbance and forward-looking early warning capabilities, significantly enhancing the regulation accuracy and safety margin of the power market in high uncertainty operation scenarios, and has good engineering applicability. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] For examples, please refer to Figure 1 As shown in this embodiment, the method for the coordinated evolution of the electricity market pricing mechanism includes: S100: Collect historical operational data of the target area and construct a multi-dimensional set of electricity market scenarios that includes time, region, load level, and energy type.

[0021] In this embodiment, the initial basis for the coordinated evolution of the electricity pricing mechanism is high-precision electricity market scenario modeling. Specifically, the historical operating data of the target area is first collected and structured. The target area is a typical pilot area of ​​the electricity spot market, such as a coastal power grid section, covering an area of ​​approximately 8,000 square kilometers, with a total load of approximately 15 million kilowatts during peak hours, and a renewable energy installation penetration rate of over 35%.

[0022] The data collected comes from, but is not limited to, the following four categories: Dispatch operation logs of provincial dispatch centers (5-minute intervals); settlement data of the spot market of the power trading center; real-time power flow monitoring data of the energy management system (EMS); and data transmitted back from distributed power sources and user-side smart terminals.

[0023] Based on the above data, the following four-dimensional set of electricity market scenarios is constructed: Time dimension (T): Divided by granularity into: annual seasons (spring, summer, autumn, winter); monthly load levels (high, medium, low); intraday operating characteristics (peak, off-peak, valley); time resolution is set to one frame every 15 minutes, covering a total of approximately 1.05 million frames of data over three consecutive years.

[0024] Regional Dimension (R): The regional division adopts a partitioned coding method, dividing the target area into 6 grid sub-regions (R1~R6). Each sub-region establishes a coding weight based on its dominant load type (industrial, residential, commercial) and the concentration of new energy sources. For example, R1 is the industrial load-dominated area with strong load fluctuations; R3 and R4 are new energy-rich areas, with photovoltaic and wind power installed capacity accounting for more than 60% in total.

[0025] Load level dimension (L): The total load in the target area is normalized according to the operating curve and divided into five levels: L1: load is less than 30% of rated capacity; L2: 30% to 50%; L3: 50% to 70%; L4: 70% to 90%; L5: above 90% (peak load section); This division is used to quantify the pressure of load response to the electricity pricing mechanism.

[0026] Energy type dimension (E): Consider the following five types of energy supply structure characteristics: E1: thermal power dominance; E2: wind power dominance; E3: photovoltaic dominance; E4: energy storage + new energy; E5: multi-energy integration (including gas power, hydropower, etc.); each frame time node is accompanied by the corresponding energy supply structure ratio vector, for example [E1=0.35, E2=0.25, E3=0.20, E4=0.15, E5=0.05].

[0027] Finally, based on the above four dimensions, a scenario matrix S is constructed, defined as: S={(T_i,R_j,L_k,E_m)}, where i∈[1,N_T], j∈[1,N_R], k∈[1,N_L], m∈[1,N_E], and N_T, N_R, N_L, and N_E are the number of discrete segments for time, region, load level, and energy type, respectively.

[0028] In this embodiment, the total number of multi-dimensional scenario samples generated exceeds 86,400. Each scenario point is bound to corresponding historical electricity price mechanism parameters (such as day-ahead clearing price, real-time market price, ancillary service fees, etc.), providing high-precision input support for extracting mechanism evolution features in subsequent steps. This scenario set is not only used for mechanism correlation modeling, but also supports mechanism coordination simulation and robustness verification under extreme operating conditions (such as high proportion of wind and solar curtailment, rapid load fluctuations), and is the core data foundation for realizing the coordinated evolution of electricity price mechanisms.

[0029] S200. Based on a multi-dimensional electricity market scenario set, extract the set of characteristic parameters M of the evolution trajectory of each electricity price mechanism in typical scenarios. M includes the rate of change of the slope of the spot electricity price, the periodicity of the peak-valley price difference, and the ancillary service cost coefficient.

[0030] After constructing the multi-dimensional electricity market scenario set, this step is used to extract the set of characteristic parameters M of the evolution trajectory of the electricity price mechanism under typical operating scenarios. The set of characteristic parameters M consists of three types of parameters: the rate of change of the spot electricity price slope, the periodicity of the peak-valley price difference, and the ancillary service cost coefficient. These three types of parameters characterize the dynamic evolution behavior of the electricity price mechanism from the perspectives of the speed of electricity price change, the rhythmic characteristics of electricity price, and the sensitivity of regulation costs.

[0031] The extraction process for the rate of change of the spot electricity price slope is as follows: First, based on the time frame sequence of the concentrated time dimension of the multi-dimensional electricity market scenario, the time series reconstruction of the corresponding spot market clearing price under each scenario is performed. Specifically, the spot clearing prices within continuous time frames are arranged in chronological order to form an uninterrupted electricity price time series, wherein the time interval between adjacent electricity price data remains consistent to ensure the continuity and comparability of the time series.

[0032] Subsequently, a sliding window method was used to perform local linear regression on the electricity price time series. The window length of the sliding window was preset according to the scene's temporal resolution, for example, using eight consecutive time frames as a window, and the window slid along the time axis frame by frame. Within each window, a local linear function was constructed using least squares fitting with time as the independent variable and electricity price as the dependent variable. The slope value of this function was defined as the electricity price slope value corresponding to that window, used to characterize the rate of change of electricity price within that time period.

[0033] After obtaining the electricity price slope sequence corresponding to the continuous window, the wavelet transform method is used to perform multi-scale analysis on the electricity price slope sequence. Specifically, a continuous wavelet function with good time-frequency localization characteristics is selected as the basis function to decompose the electricity price slope sequence scale by scale, thereby splitting the original slope sequence into multiple subsequences corresponding to different frequency bands to reflect the differences in electricity price changes at the short-period and long-period levels.

[0034] Within the slope subsequences at each scale, local extreme points of slope change are identified, and the magnitude of slope change between adjacent extreme points is calculated. The magnitude of slope change is filtered using a pre-set sensitivity threshold, determined by the statistical distribution of historical slope change magnitudes, for example, by taking a multiple of the average historical slope change magnitude as the criterion. When the slope change magnitude exceeds this threshold, the corresponding time point is identified as an abnormal slope jump point. This forms a spot electricity price slope change rate index under high uncertainty scenarios, which serves as a component of the evolution trajectory feature parameter set M.

[0035] The process for extracting the periodicity of peak-valley price differences is as follows: Firstly, based on the periodicity parameter of the peak-valley price difference, historical operating days are classified according to the intraday load level dimension of a multi-dimensional electricity market scenario. Operating days with the same load level distribution characteristics are grouped into the same category, and representative operating days are selected from each category to construct a corresponding typical day template set for the electricity price daily curve. The typical day template is stored in the form of a complete intraday spot clearing electricity price curve.

[0036] Frequency domain analysis was performed on each typical daily electricity price curve. Specifically, the Fourier transform method was used to map the intraday spot electricity price curve from the time domain to the frequency domain. By sorting the spectral amplitudes, the dominant frequency components that contribute the most to electricity price fluctuations were identified. The dominant frequency reflects the recurring cycle of electricity price changes on an intraday scale.

[0037] After identifying the dominant frequency components, the peak-valley price difference periodicity index is calculated by combining the amplitude of the corresponding frequencies. This index characterizes the stability and repetition intensity of peak and valley electricity prices during periodic changes. The larger the amplitude and the more stable the dominant frequency, the more significant the periodicity of the peak-valley price difference. The calculated peak-valley price difference periodicity index is incorporated into the evolutionary trajectory characteristic parameter set M.

[0038] The process of extracting ancillary service cost coefficients involves first identifying ancillary service call records under different energy structure conditions based on the energy type dimension of a multi-dimensional electricity market scenario. For each time frame, the number of corresponding ancillary service calls (such as frequency regulation and reserve) and their corresponding actual response costs are collected to form an ancillary service cost time series.

[0039] Subsequently, a support vector regression model was constructed with ancillary service costs as the dependent variable and regional load levels as the independent variables. This support vector regression model establishes a nonlinear mapping relationship by selecting a radial basis function kernel and determines the model parameters through training with historical samples. This enables the model to accurately fit the correspondence between load changes and ancillary service costs, thereby forming an ancillary service cost prediction function.

[0040] After obtaining the cost prediction function, the marginal cost of ancillary services corresponding to a unit load change is calculated by analyzing the magnitude of the change in predicted cost when the load level changes by a unit. This marginal cost is defined as the ancillary service cost coefficient, which measures the sensitivity of the control costs required to maintain the stable operation of the power system under different energy structures and load levels. The ancillary service cost coefficient, as an important component of the evolution trajectory characteristic parameter set M, provides a quantitative basis for subsequent collaborative analysis of electricity pricing mechanisms.

[0041] S300. Based on the set of evolutionary trajectory feature parameters M, the electricity pricing mechanism is divided into coupling mechanism clusters G using a similarity clustering algorithm, and the coupling coefficient matrix C between different mechanism clusters is quantified.

[0042] After extracting the set of evolutionary trajectory feature parameters M, in order to further reveal the differences in response patterns of different electricity pricing mechanisms under various typical operating scenarios, it is necessary to perform similarity clustering on the electricity pricing mechanisms and construct a coupling relationship description structure.

[0043] First, for each electricity pricing mechanism, under multiple typical operating scenarios, the corresponding set of evolution trajectory characteristic parameters M is extracted, including three parameters: the rate of change of the spot electricity price slope, the peak-valley price difference periodic index, and the ancillary service cost coefficient.

[0044] The parameters were standardized by uniformizing their dimensions. For each type of parameter, a maximum-minimum normalization method was used to map the parameter values ​​to the [0,1] interval, thus eliminating the influence of differences in dimensions and numerical levels between different parameters on the subsequent clustering results.

[0045] The normalized parameters are arranged sequentially to form a feature vector. Specifically, the M value of each electricity pricing mechanism in a typical scenario is represented as a three-dimensional feature vector. If N typical scenarios are considered, the high-dimensional vector corresponding to the electricity pricing mechanism is represented as a vector of length 3×N. Ultimately, all electricity pricing mechanisms are mapped to vector representations located in a unified multidimensional space.

[0046] In the aforementioned high-dimensional feature vector space, a density-based spatial clustering algorithm is used for cluster analysis. Specifically, the density clustering algorithm DBSCAN is employed, where the parameter ε represents the maximum neighbor distance between points, and MinPts represents the minimum number of samples required to form the density core.

[0047] In this algorithm, if a certain electricity pricing mechanism vector contains at least MinPts other mechanism vectors within a radius of ε, then that point is considered a density core point and is grouped into the same cluster as the vectors in its neighborhood. This method can effectively identify a set of mechanisms that are closely distributed in the feature space and have similar response behaviors.

[0048] Ultimately, the set of all electricity pricing mechanisms classified into the same cluster is defined as a coupled mechanism cluster G, with each cluster representing a set of electricity pricing mechanisms that exhibit a high degree of consistency in their operational characteristics.

[0049] For each identified cluster of coupling mechanisms, its cluster center point is calculated. The cluster center point is defined as the mean vector of the feature vectors of all member mechanisms within the cluster, which serves as the representative response feature of the cluster.

[0050] The coupling strength between any two clusters of coupling mechanisms is determined by calculating the cosine similarity between their cluster centers. Specifically, for two cluster center vectors A and B, the cosine similarity S is expressed as the dot product of vectors A and B divided by the product of their respective magnitudes, with a value ranging from 0 to 1. This cosine similarity value measures the directional similarity between the two clusters in the evolutionary trajectory feature space; the closer the value is to 1, the tighter the coupling relationship, reflecting a higher degree of consistency in their operational behavior.

[0051] The coupling coefficients between all mechanism clusters are organized into cluster pairs, forming a symmetric matrix structure. The rows and columns of this matrix correspond to the respective coupling mechanism clusters, and the element in the i-th row and j-th column represents the coupling coefficient between the i-th and j-th clusters. The constructed coupling coefficient matrix C serves as the basis for subsequent mechanism co-modeling and evolution function group construction, used to quantify the response linkages between various electricity price mechanisms and provide accurate parameter support for evolution direction prediction.

[0052] S400. Based on the coupling coefficient matrix C, construct a mechanism co-evolution function group F to characterize the linkage response relationship between different mechanisms. Each function in F takes the dynamic features in M ​​as input variables.

[0053] After constructing the coupling mechanism cluster G and the coupling coefficient matrix C, in order to further describe and predict the linkage response behavior among the various electricity price mechanism clusters, it is necessary to construct a mechanism co-evolution function group F to characterize the co-evolution trend of multiple electricity price mechanism clusters under the dynamic operating characteristics.

[0054] First, the cluster pair coupling information in the coupling coefficient matrix C is used as a weight parameter for cross-mechanism cluster feature transfer. Each element C(i,j) of matrix C represents the coupling strength between the i-th mechanism cluster and the j-th mechanism cluster, and its value ranges from 0 to 1. The closer it is to 1, the higher the similarity of the evolutionary behavior between the two clusters.

[0055] For each target mechanism cluster Gi, its evolutionary trajectory features consist of its own feature vector Vi and the weighted feature vectors of other mechanism clusters Gj. The weighting rule is as follows: multiply the feature vector Vj of mechanism cluster Gj with the corresponding coupling coefficient C(i,j) to obtain the weighted feature contribution value. The weighted features of all related mechanism clusters are accumulated and used as the external feature input to cluster Gi. This operation realizes cross-mechanism cluster feature transfer based on coupling relationships.

[0056] For each target mechanism cluster Gi, a co-evolution function fi is constructed, with its own dynamic feature vector Vi and the weighted feature vector set of associated mechanism clusters as independent variables. This function is used to predict the feature change trend of mechanism cluster Gi in the next time step, and its form is: output variable Yi = fi(Vi, ΣC(i,j)×Vj), where j≠i. To enhance the function's ability to characterize nonlinear response relationships, the co-evolution function fi is constructed using a nonlinear mapping structure. Specifically, a feedforward artificial neural network is selected as the implementation framework for the mapping function, with one input layer, one hidden layer, and one output layer. The hyperbolic tangent function is used as the activation function to enhance the ability to distinguish between strong and weak linkage effects.

[0057] Each co-evolutionary function fi constructed in the above steps corresponds to a mechanism cluster Gi, characterizing the linkage evolution trend of that cluster. The co-evolutionary functions {f1, f2, ..., fn} of all mechanism clusters are combined to form a mechanism co-evolutionary function group F. In each running cycle, function group F uses the set of evolutionary trajectory feature parameters M as input and dynamically updates the response paths between each mechanism cluster using the weight relationships provided by the coupling coefficient matrix C, ensuring that the function group can adapt to the dynamic changes in the linkage relationships of electricity pricing mechanisms under multiple scenarios.

[0058] The final constructed mechanism co-evolution function group F will serve as the core logical model for dynamic adjustment of the electricity price mechanism, predictive response, and identification of high-risk linkages in subsequent steps. It has clear input-output relationships and structural mapping characteristics, and can support multi-mechanism linkage modeling and deduction in the process of electricity price mechanism co-evolution.

[0059] S500 introduces a perturbation factor δ to simulate the mechanism perturbation response under abnormally high uncertainty scenarios, and adjusts the function parameters in F to form a robust evolution model F′.

[0060] After completing the construction of the mechanism co-evolution function group F, in order to improve the stability and generalization ability of the model in high uncertainty operation scenarios, it is necessary to introduce a perturbation factor δ to enhance the perturbation processing of the function group input, thereby identifying sensitive parameters and optimizing the function structure to form a robust evolution model F′, so as to enhance the adaptability and fault tolerance of mechanism co-modeling.

[0061] The disturbance factor δ is designed to simulate three types of high uncertainty sources in electricity market operation: wind power output fluctuations, electricity price jumps, and load forecasting errors. Specifically, these include: Δ1 represents the standard deviation of wind power output; Δ2 represents the frequency of electricity price fluctuations exceeding a preset threshold within a unit of time. Δ3 represents the average deviation rate between the actual load and the predicted load.

[0062] The above-mentioned disturbance factors are derived from historical operating data and are standardized in terms of units, with values ​​ranging from 0 to 1.

[0063] The perturbation factor δ is used as the input to the co-evolution function set F of the input perturbation term injection mechanism. Specifically, each input variable Xi in the evolution trajectory feature parameter set M is replaced with Xi plus δ multiplied by the perturbation ratio of Xi, i.e. This operation generates a perturbation-enhanced feature sample set, which is used to characterize the response behavior of the mechanism cluster under extreme conditions.

[0064] Based on the perturbation-enhanced sample set, parameter sensitivity analysis was performed on each function fi in the mechanism co-evolution function group. The analysis method was the local parameter perturbation method, that is, the weight parameters and threshold parameters inside each function were changed one by one, and the degree of deviation of the model output result Yi was observed.

[0065] The specific operation is as follows: For each parameter θj to be estimated in the function fi, a perturbation range of ±5% is set. Forward propagation is performed on the perturbation feature samples, and the change ΔYi in the function output value is recorded. The perturbation sensitivity index Sj = ΔYi divided by Δθj is calculated to measure the strength of the output response to parameter changes.

[0066] The parameter θj corresponding to the sensitivity index Sj being higher than the preset threshold (e.g., 1.5 times the historical average) is identified as a highly sensitive parameter and marked as the focus of subsequent optimization and adjustment.

[0067] For the identified highly sensitive parameters, a multi-objective optimization algorithm is used for parameter reconstruction. This optimization algorithm has two objectives: minimizing output offset and improving output stability. It employs a particle swarm optimization algorithm for iterative search to determine the optimal parameter configuration combination under perturbation conditions.

[0068] All the co-evolutionary functions fi′, after perturbation analysis and optimization, are recombined to form a new function group F′, i.e., the robust evolutionary model. While maintaining the linkage structure of the original function group F, the function group F′ enhances its tolerance to input perturbations, ensuring that it can still output stable and controllable mechanism response prediction results under high uncertainty operating scenarios.

[0069] S600. Apply the robust evolution model F′ to the current operating state of the target area, apply F′ to the current electricity price mechanism state vectors for linkage adjustment, and output a new electricity price mechanism state prediction vector P.

[0070] After constructing the robust evolution model F′, in order to dynamically predict the future evolution trend of the electricity price mechanism in the target area, the model needs to be applied to actual operating scenarios. By mapping the input of the current electricity price mechanism state vector with the model, the predicted state of the mechanism for the next cycle can be obtained, thus providing a priori response basis for electricity market regulation.

[0071] First, data on the electricity pricing mechanism status of the target area within the current time period is collected. The time period is the prediction step size set by the model, such as 15 minutes or 60 minutes.

[0072] The operational states of each coupling mechanism cluster within the current time period are represented as feature parameters to construct the electricity pricing mechanism state vector S. This state vector consists of three core dimensions from the set of evolution trajectory feature parameters: Current rate of change in the slope of spot electricity prices; Current peak-to-trough price spread cyclical index; Current ancillary service cost coefficient.

[0073] Each mechanism cluster corresponds to a set of three-dimensional feature values. If there are N mechanism clusters, the state vector S is a vector of length 3×N, used to fully describe the current mechanism evolution state.

[0074] The constructed state vector S is used as the input variable and fed into the robust evolutionary model F′ after perturbation enhancement and sensitivity parameter optimization. Model F′ consists of several nonlinear cooperative evolution functions fi′, each function corresponding to a mechanism cluster. The function input includes the features of the current mechanism cluster and the weighted features of its coupled clusters. The input structure is consistent with the aforementioned model structure.

[0075] During the model operation, each function fi′ performs nonlinear mapping calculations. The function structure is a feedforward neural network, and a hyperbolic tangent activation function is used to enhance the ability to approximate nonlinear boundaries. The parameter weights have been optimized through perturbation training in the previous stage.

[0076] The model output is the predicted feature value for the next time period of the mechanism cluster, representing the evolution trend of the electricity price mechanism in the current state. The output value is driven by the linkage between current state variables and the model's adaptive adjustment capability to abnormal disturbances.

[0077] The outputs of each function fi′ of model F′ are summarized to form the electricity price mechanism state prediction vector P for the target region in the next time period. The prediction vector P has the same dimension as the input state vector S, which is still a 3×N vector structure. Each three-dimensional sub-vector represents the predicted state of a mechanism cluster, including the predicted rate of change of electricity price slope, the predicted peak-valley periodicity index, and the predicted ancillary service cost coefficient.

[0078] Vector P can serve as the basic input for subsequent mechanism risk analysis and control strategy generation. The difference between its numerical change trend and the current state vector S can be used to identify risk signs of mechanism jumps, abnormal linkages, or coordinated instability, which has strong practicality and foresight.

[0079] S700 identifies potential electricity price tear points or mechanism conflict points based on the electricity price mechanism state prediction vector P and marks them as high-risk control intervals.

[0080] To achieve proactive response in electricity market regulation, after obtaining the electricity price mechanism state prediction vector P, it is necessary to further conduct risk analysis on the prediction results to identify potential abnormal changes in the mechanism, including electricity price tear points and mechanism conflict points, thereby dynamically marking high-risk regulation intervals and providing a basis for operational intervention and optimization.

[0081] First, the difference between the electricity price mechanism state vector S and the prediction vector P within the current time period is calculated to obtain the mechanism evolution offset vector ΔS. The offset value of each mechanism cluster Gi consists of the difference between the predicted value and the current value of three types of features, which are as follows: Difference in the rate of change of the slope of spot electricity prices; Peak-valley price difference cyclical index difference; Difference in cost coefficients for ancillary services.

[0082] Right now It is used to quantify the future evolution direction and magnitude of each mechanism cluster.

[0083] For the component of the rate of change of the spot electricity price slope in the evolutionary offset vector ΔS, it is determined whether its change constitutes a "tear" risk. The electricity price tear point is defined as: within a unit prediction period, the growth rate of the rate of change of the electricity price slope for a certain mechanism cluster exceeds the critical jump threshold η1. The threshold η1 is set by: statistically analyzing the mean and standard deviation of the rate of change of the electricity price slope based on historical operating data, and taking η1 as the mean plus twice the standard deviation. When the rate of change of the electricity price slope in ΔS(i) > η1, it is determined that the mechanism cluster Gi has an electricity price tear risk, and its prediction period is marked as the electricity price tear point.

[0084] For the predicted feature values ​​between different mechanism clusters, the evolution direction and cooperative relationship are analyzed. Specifically, the cooperative deviation σ(i,j) of any two mechanism clusters Gi and Gj is calculated, which is defined as a measure of the directional consistency of similar features in the predicted state.

[0085] The method for calculating the coordination deviation σ(i,j) is as follows: if the similar characteristic values ​​of ΔS(i) and ΔS(j) have opposite signs (i.e., one increases and the other decreases), and the coupling coefficient C(i,j) between them is greater than a preset coupling threshold η2 (e.g., η2=0.75), then it is determined that there is a linkage conflict between them. This conflict indicates that highly coupled mechanism clusters may exhibit behavioral deviations in the next cycle, leading to increased difficulty in regulation. Such conflict points are marked as mechanism conflict points.

[0086] The time periods corresponding to the identified electricity price tear points and mechanism conflict points are marked as high-risk control intervals, and a unified set of high-risk intervals H is output.

[0087] The set H can be provided as input to the market regulation strategy formulation process, assisting in triggering response mechanisms such as electricity price smoothing, ancillary service compensation, or constraint adjustment, thereby improving the overall stability and responsiveness of the electricity price mechanism.

[0088] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for the coordinated evolution of electricity market pricing mechanisms, characterized by: include: S100: Collect historical operational data of the target area and construct a multi-dimensional set of electricity market scenarios that includes time, region, load level, and energy type; S200. Based on a multi-dimensional electricity market scenario set, extract the evolution trajectory characteristic parameter set M of each electricity price mechanism under typical scenarios. M includes the rate of change of the spot electricity price slope, the periodicity of the peak-valley price difference, and the ancillary service cost coefficient. S300. Based on the set of evolution trajectory feature parameters M, the electricity pricing mechanism is divided into coupling mechanism clusters G using a similarity clustering algorithm, and the coupling coefficient matrix C between different mechanism clusters is quantified. S400. Based on the coupling coefficient matrix C, construct a mechanism co-evolution function group F to characterize the linkage response relationship between different mechanisms. Each function in F takes the dynamic features in M ​​as input variables. S500 introduces a perturbation factor δ to simulate the mechanism perturbation response under abnormally high uncertainty scenarios, and adjusts the function parameters in F to form a robust evolution model F′; S600. Apply the robust evolution model F′ to the current operating state of the target area, apply F′ to the current electricity price mechanism state vectors for linkage adjustment, and output a new electricity price mechanism state prediction vector P. S700 identifies potential electricity price tear points or mechanism conflict points based on the electricity price mechanism state prediction vector P and marks them as high-risk control intervals.

2. The method for realizing the coordinated evolution of electricity market pricing mechanisms according to claim 1, characterized in that: The extraction of the rate of change of the spot electricity price slope includes the following steps: Based on a multi-dimensional electricity market scenario set, the spot market clearing price in each time frame is reconstructed to generate a continuous electricity price time series. The sliding window method was used to perform local linear regression on the electricity price time series, and the slope value of the electricity price within each window was calculated. Wavelet transform was used to perform multi-scale analysis on the electricity price slope sequence to extract the rate of change of local extreme slopes in different frequency bands. Based on preset sensitivity thresholds, abnormal slope jump points are screened to form the rate of change of spot electricity price slope in high uncertainty scenarios.

3. The method for realizing the coordinated evolution of electricity market pricing mechanisms according to claim 1, characterized in that: The periodic extraction of the peak-valley price difference includes the following steps: Based on the multi-dimensional electricity market scenario-based intraday load level dimension, a typical daily template set for the daily electricity price curve is constructed. Fourier transform is used to map the daily spot electricity price curve to the frequency domain to identify the dominant frequency components; The peak-valley price difference cyclical index is calculated based on the dominant frequency and amplitude.

4. The method for realizing the coordinated evolution of electricity market pricing mechanisms according to claim 1, characterized in that: The extraction of the ancillary service cost coefficient includes the following steps: Based on the multi-dimensional electricity market scenario, extract the ancillary service call records and response costs within the corresponding time period. A support vector regression model is used to model the relationship between ancillary service costs and regional load levels, forming a cost prediction function; The marginal cost of ancillary services corresponding to a unit change in load is calculated and defined as the ancillary service cost coefficient.

5. The method for realizing the coordinated evolution of electricity market pricing mechanisms according to claim 1, characterized in that: The steps involved in using a similarity clustering algorithm to divide the electricity pricing mechanism into coupling mechanism clusters G, and quantifying the coupling coefficient matrix C between different mechanism clusters, include the following processes: Based on the set of evolutionary trajectory feature parameters M, a multi-dimensional feature vector space is constructed to uniformly map the feature parameters of various electricity pricing mechanisms under different typical scenarios into a high-dimensional vector representation. A density-based similarity clustering algorithm is used to cluster high-dimensional feature vectors to identify a set of electricity pricing mechanisms that are highly consistent in their electricity price response behavior, which is defined as the coupling mechanism cluster G. Between the formed coupling mechanism clusters, the coupling strength is calculated based on the cosine similarity between the center points of the feature vectors, and the coupling coefficient between the mechanism clusters is obtained. The coupling coefficients are organized into cluster pairs to construct the coupling coefficient matrix C.

6. The method for realizing the coordinated evolution of electricity market pricing mechanisms according to claim 1, characterized in that: The step of constructing a mechanism co-evolution function set F based on the coupling coefficient matrix C to characterize the linkage response relationship between different mechanisms includes: Using the coupling coefficients between clusters in the coupling coefficient matrix C as weight parameters, the dynamic features of the corresponding mechanism clusters in the evolution trajectory feature parameter set M are weighted and correlated to establish a feature transfer relationship across mechanism clusters; For each target mechanism cluster, a co-evolution function is constructed with its own dynamic characteristics and the weighted characteristics of the associated mechanism clusters as independent variables. The function adopts a nonlinear mapping form to characterize the amplification or suppression effect of the linkage response. The co-evolution functions corresponding to each mechanism cluster are combined to form a mechanism co-evolution function group F.

7. The method for realizing the coordinated evolution of electricity market pricing mechanisms according to claim 1, characterized in that: The step of adjusting the function parameters in F to form a robust evolutionary model F′ includes: The perturbation factor δ is injected into the input of the co-evolution function group F of the mechanism to perform perturbation enhancement processing on the input variables in the set of evolution trajectory feature parameters M, forming a perturbation feature sample set; Perform parameter sensitivity analysis on each co-evolution function to identify highly sensitive parameters that cause output shift under perturbation samples; All perturbation-adjusted evolution functions are recombined to form a robust evolution model F′.

8. The method for realizing the coordinated evolution of electricity market pricing mechanisms according to claim 1, characterized in that: The step of applying the robust evolution model F′ to the current operating state of the target area and applying F′ to the current electricity price mechanism state vectors includes: The electricity pricing mechanism state vector of the target area is collected within the current time period. The electricity pricing mechanism state vector is composed of the evolution trajectory feature parameters of each mechanism cluster. The current state vector is used as an input variable and input into the robust evolution model F′ after perturbation optimization to perform nonlinear mapping calculations on the linkage relationship between each mechanism cluster; Based on the calculation results of the robust evolution model F′, the output is the prediction vector P of the electricity price mechanism state of the target region in the next time period, where each element corresponds to the predicted evolution value of the mechanism cluster.

9. The method for realizing the coordinated evolution of electricity market pricing mechanisms according to claim 1, characterized in that: The step of identifying potential electricity price tear points or mechanism conflict points based on the electricity price mechanism state prediction vector P includes: The difference between the state value of each mechanism cluster in the electricity price mechanism state prediction vector P and the current state vector is calculated to construct an offset vector; Based on the degree of abrupt change in the rate of change of the slope of the spot electricity price in the offset vector, it is determined whether there is a price tear point. If the degree of abrupt change exceeds a preset threshold, it is marked as a tear risk. Calculate the collaborative bias of predicted feature values ​​among different mechanism clusters. When the bias directions are opposite and the coupling coefficient is greater than the threshold, it is marked as a mechanism conflict point. The time period in which the tear or conflict occurs will be output as the high-risk control range.

Images