Spot clearing electricity price prediction system fusing new energy output fluctuation characteristics

Abstract

The invention relates to the technical field of electricity market electricity price prediction, and discloses a spot clearing electricity price prediction system fusing new energy output fluctuation characteristics, and the system comprises a fluctuation characteristic extraction module which carries out the variational mode decomposition of a new energy output prediction sequence, and generates fluctuation impact momentum and duration; the climbing scarcity calculation module is used for calculating a reference equivalent climbing scarcity in combination with the polymerization residual climbing capability and the fluctuation impact momentum; the hysteresis state correction module is used for correcting and generating a scarcity degree state index by using a state attenuation parameter based on fluctuation duration; the supply curve reconstruction module is used for generating a dynamic effective supply curve by carrying out linear deformation on the reference residual supply curve by utilizing the physical capacity cut-off quantity mapped by the index and a quotation slope correction coefficient; and the clearing electricity price solving module is used for outputting a predicted value based on the curve. The new energy fluctuation impact and hysteresis effect are effectively quantified, and the prediction precision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power market electricity price prediction, in particular to a spot market clearing electricity price prediction system fusing new energy output fluctuation characteristics. BACKGROUND

[0002] At present, with the promotion of energy transformation, the penetration rate of new energy represented by wind power and photovoltaic in the power system is rising. The randomness and volatility of new energy output have profoundly changed the operation logic of the power spot market, making the electricity price signal show strong nonlinear oscillation characteristics. For market participants such as power generators and power sellers, accurately capturing the spot price trend is the key to formulating an optimized bidding strategy and avoiding market risks.

[0003] For spot electricity price prediction, existing technologies mainly adopt two paths of data-driven or structured simulation. Data-driven methods mostly use algorithms such as long short-term memory network (LSTM) and support vector machine (SVM) to establish an input-output mapping model by taking historical load, weather data, and historical electricity prices as inputs. Structured simulation is to build a unit commitment and optimal power flow model to simulate the clearing process of the dispatching institution, and to solve the electricity price by calculating the intersection point of the supply curve and the demand curve. These methods can provide relatively reliable prediction results in conventional scenarios with stable system operation and small changes in boundary conditions.

[0004] However, existing technologies have limitations in dealing with extreme fluctuation scenarios with high proportion of new energy access. Traditional structured models often focus on power balance at a time point, easily ignoring the dynamic physical constraint of unit ramp rate. When new energy output experiences a high-frequency sudden drop for a very short time, the system has surplus capacity but is blocked by the ramp rate and cannot provide immediate service. Existing models are difficult to convert this physical scarcity of little amount into a price signal, resulting in an underestimation of the peak electricity price. In addition, market transactions have psychological inertia, and prices often do not fall instantly after the crisis is resolved. Existing algorithms are usually based on instantaneous supply-demand matching and lack a description mechanism for this price hysteresis effect, causing the predicted value to drop too quickly during the fluctuation recovery period. At the same time, the supply curve in existing models is mostly static or fixed in shape based on historical statistics, and cannot dynamically adjust the intercept and slope of the curve according to real-time system shocks, making it difficult to accurately map the physical layer fluctuation risk to the economic layer bidding behavior, affecting the physical consistency of the prediction.

[0005] Therefore, the present application provides a spot market clearing electricity price prediction system fusing new energy output fluctuation characteristics to solve the deficiencies in the prior art. SUMMARY

[0006] In view of the deficiencies of the prior art, the present application provides a spot market clearing price prediction system fusing new energy output fluctuation characteristics, which solves the problem of insufficient spot market peak price prediction and large deviation of price falling trend in the prior art under the scene of severe fluctuation of new energy output due to lack of quantitative description of physical climbing scarcity and market hysteresis effect.

[0007] To achieve the above object, the present application is implemented by the following technical scheme: a spot market clearing price prediction system fusing new energy output fluctuation characteristics, comprising: a fluctuation feature extraction module for variational mode decomposition of a new energy output prediction sequence to generate new energy output fluctuation characteristics containing fluctuation impact momentum and fluctuation duration; a climbing scarcity calculation module for estimating aggregated residual climbing capacity based on unit commitment state, superimposing the fluctuation impact momentum to the change rate of system net load, and calculating the benchmark equivalent climbing scarcity relative to the aggregated residual climbing capacity; a hysteresis state correction module for time correction of the benchmark equivalent climbing scarcity by using a state decay parameter based on the fluctuation duration to generate a corrected scarcity state index; a supply curve reconstruction module for establishing a mapping relationship between the corrected scarcity state index and physical capacity truncation amount and offer slope correction coefficient, and performing geometric deformation on a benchmark residual supply curve by using the physical capacity truncation amount and the offer slope correction coefficient to generate a dynamic effective supply curve; a clearing price solving module for projecting the system net load at the time to be predicted to the dynamic effective supply curve, reading the corresponding price coordinates, and outputting a spot market clearing price prediction value.

[0008] By adopting the above technical scheme, the variational mode decomposition technology is used to quantify the instantaneous impact of new energy high-frequency components on the system, the hysteresis state correction mechanism is introduced to simulate the stress inertia of the power system in response to severe fluctuations, and the scarcity index is used to drive the geometric deformation of the supply curve to represent the shortage of physical resources and the risk premium of market offer. Therefore, the prediction effect of accurately capturing the nonlinear jump characteristics of spot market price under the scene of sudden change of new energy output is obtained, the problems of insufficient prediction of peak price and too fast prediction of price falling process of traditional models in the period of climbing resource shortage are effectively solved, and the accuracy and robustness of spot market clearing price prediction in high-proportion new energy power systems are improved.

[0009] Preferably, the process by which the fluctuation feature extraction module performs variational mode decomposition on the new energy output prediction sequence to generate fluctuation impact momentum and fluctuation duration includes: decomposing the new energy output prediction sequence into several intrinsic mode functions using a variational mode decomposition algorithm, reconstructing it using a frequency segmentation threshold to obtain high-frequency fluctuation components, differentiating the high-frequency fluctuation components with respect to time to obtain the real-time rate of change; comparing the absolute value of the real-time rate of change with the average ramp rate threshold of the system's marginal units, calculating the integral of the difference between the portion of the real-time rate of change exceeding the average ramp rate threshold within a sliding time window to obtain the fluctuation impact momentum; and calculating the duration for which the absolute value of the real-time rate of change continuously exceeds the average ramp rate threshold to obtain the fluctuation duration.

[0010] By adopting the above technical solution, the trend term and random fluctuation term in the output of new energy can be effectively separated. By integrating and timing the part of the high-frequency component whose rate of change exceeds the limit, the extreme fluctuation energy and duration that exceed the system's conventional adjustment capability are accurately quantified, providing a precise physical quantitative input for subsequent evaluation of the system's adjustment pressure.

[0011] Preferably, the process of the ramp scarcity calculation module performing the estimation of aggregated remaining ramp capacity based on unit combination status includes: acquiring the real-time unit combination status of the power system, identifying the set of online units currently in operation; reading the rated maximum ramp rate and the remaining capacity space from the maximum technical output limit for each unit in the set of online units; selecting the smaller value between the rated maximum ramp rate and the remaining capacity space as the effective upward ramp capacity of a single unit, and summing the effective upward ramp capacities of all the single units to obtain the aggregated remaining ramp capacity.

[0012] By adopting the above technical solution, the physical constraints of the unit's operating point are fully considered, avoiding the misjudgment of units with capacity but no ramp rate as effective regulatory resources, and ensuring that the aggregated remaining ramp capacity index truly reflects the maximum physical potential that the system can call upon at the current moment.

[0013] Preferably, the process by which the ramp scarcity calculation module superimposes the fluctuation impact momentum onto the rate of change of the system net load and calculates the benchmark equivalent ramp scarcity relative to the aggregated remaining ramp capacity includes: calculating the derivative of the system net load with respect to time to obtain the rate of change of the net load; adding the absolute value of the rate of change of the net load to the weighted fluctuation impact momentum to form the total system ramp demand reflecting the total system regulation demand; and dividing the total system ramp demand by the aggregated remaining ramp capacity to obtain the benchmark equivalent ramp scarcity used to characterize the degree of tension in the supply and demand of physical regulation resources.

[0014] By adopting the above technical solution, the conventional net load change demand is superimposed with the random fluctuation impact demand of new energy sources to construct a total system regulation demand that covers both determinism and randomness. By dynamically comparing it with the supply-side ramp-up capacity, a dimensionless scarcity benchmark that can reflect the physical supply and demand tension of the system in real time is established.

[0015] Preferably, the process by which the hysteresis state correction module obtains the state decay parameter based on the fluctuation duration includes: receiving the fluctuation duration from the fluctuation feature extraction module, wherein the fluctuation duration reflects the persistence characteristics of the system under high-frequency fluctuation impact; and mapping the fluctuation duration to the state decay parameter using a linear mapping relationship, using a preset basic hysteresis constant as the intercept, and a preset coupling coefficient as the slope, wherein the state decay parameter is used to quantify the rate at which the system recovers from the stress state to the normal state.

[0016] By adopting the above technical solution, a quantitative relationship between the duration of fluctuations and the system recovery rate was established, reflecting the physical characteristic that the longer the duration of fluctuations, the longer the time required for the system to return to normal, providing a key time parameter for simulating the hysteresis effect of prices.

[0017] Preferably, the process by which the hysteresis state correction module performs time correction on the benchmark equivalent ramp scarcity using a state decay parameter based on the duration of fluctuation to generate a corrected scarcity state index includes: comparing the benchmark equivalent ramp scarcity input at the current moment with the corrected scarcity state index stored at the previous moment; when the benchmark equivalent ramp scarcity is greater than or equal to the corrected scarcity state index stored at the previous moment, the system is determined to be in stress lock mode, and the benchmark equivalent ramp scarcity at the current moment is directly assigned to the corrected scarcity state index at the current moment to maintain an immediate response to stress; when the benchmark equivalent ramp scarcity is less than the corrected scarcity state index stored at the previous moment, the system is determined to be in hysteresis recovery mode, an exponential decay function is constructed using the state decay parameter, and the exponential decay function is used to approximate the benchmark equivalent ramp scarcity at the current moment based on the corrected scarcity state index value stored at the previous moment, to calculate the corrected scarcity state index at the current moment.

[0018] By adopting the above technical solution, a state correction mechanism with asymmetric characteristics of rapid rise and slow fall was constructed. It can simulate the rapid rise of market prices when a scarcity event occurs and the slow fall after the crisis is resolved, effectively correcting the problem of the traditional model's low prediction value during the fluctuation recovery period.

[0019] Preferably, before performing the establishment of the mapping relationship and geometric deformation, the process of obtaining the benchmark residual supply curve by the supply curve reconstruction module includes: reading the time attributes and total system load level of the period to be predicted; using the weighted Euclidean distance algorithm to retrieve the historical trading day in the historical operating database that is closest to the time attributes and total system load level of the period to be predicted; extracting the declaration data of the historical trading day; sorting the declaration prices of all market-based units monotonically in ascending order of capacity; and generating a sequence composed of discrete price and capacity coordinate points as the benchmark residual supply curve to be reconstructed.

[0020] By adopting the above technical solution and using actual declaration data from similar historical days as a basis, the basic pricing strategies and behavioral habits of market participants can be preserved, providing an initial benchmark that conforms to the actual market environment for subsequent deformation adjustments.

[0021] Preferably, the process by which the supply curve reconstruction module establishes the mapping relationship between the corrected scarcity status index and the physical capacity cutoff and the price slope correction coefficient includes: constructing an S-shaped function that monotonically increases with the corrected scarcity status index based on the total installed capacity of the system and the cutoff sensitivity coefficient; calculating the physical capacity cutoff representing the capacity of units that cannot provide services due to ramp-up obstruction; and constructing a regression model that maps the corrected scarcity status index to a value greater than or equal to 1 based on a preset slope sensitivity coefficient; and calculating the price slope correction coefficient representing the risk premium of market participants.

[0022] By adopting the above technical solution, the abstract scarcity state index is transformed into specific supply curve deformation parameters. The S-shaped function accurately describes the nonlinear characteristic of the sharp decrease in available capacity after the scarcity reaches a certain threshold, while the regression model quantifies the game behavior of market participants raising their bid levels when supply and demand are expected to be tight.

[0023] Preferably, the process by which the supply curve reconstruction module performs geometrical deformation on the benchmark residual supply curve using the physical capacity cutoff and the price slope correction coefficient to generate a dynamic effective supply curve includes: traversing each coordinate point on the benchmark residual supply curve using the physical capacity cutoff and the price slope correction coefficient; shifting the coordinate point to the left along the capacity axis by a shift value equal to the physical capacity cutoff, reflecting the unavailability of low-priced resources; and multiplying the price value of the coordinate point by the price slope correction coefficient along the price axis, reflecting the price increase brought about by the risk premium, thereby generating the dynamic effective supply curve.

[0024] By adopting the above technical solution, the supply curve is dynamically reconstructed through coordinate transformation. The leftward shift of the capacity axis directly reflects that some low-priced unit capacity cannot be called up by the system due to ramp-up restrictions, while the correction of the price axis reflects the overall price level rising during periods of scarcity, thus constructing an effective supply boundary that truly determines the market clearing price.

[0025] Preferably, the process by which the clearing price solution module projects the system net load at the time to be predicted onto the dynamic effective supply curve, reads the corresponding price coordinates, and outputs the predicted spot clearing price includes: calculating the difference between the predicted total system load at the time to be predicted and the predicted output of new energy sources and the fixed switching power of the tie line, to obtain the system net load; projecting the system net load as the abscissa onto the capacity axis of the dynamic effective supply curve, and retrieving the supply function segment whose capacity covers the system net load; reading the price ordinate value corresponding to the supply function segment, correcting it in conjunction with the key section constraint information of the power grid, and outputting the final predicted spot clearing price.

[0026] By adopting the above technical solution, the actual clearing logic of the spot market was simulated, and the supply and demand balance point was found in the reconstructed dynamic effective supply curve. This ensured that the forecast results not only took into account the load balance, but also fully integrated the ramp-up scarcity risk and market lag effect brought about by the fluctuation of new energy sources.

[0027] This invention provides a spot clearing electricity price prediction system that integrates the fluctuation characteristics of new energy output. It has the following beneficial effects: 1. This invention quantifies the instantaneous impact of high-frequency fluctuations in renewable energy on the system's regulatory capacity by introducing a fluctuation feature extraction and ramp scarcity calculation mechanism. Based on this, it uses physical capacity cutoff and price slope correction coefficients to geometrically deform the benchmark supply curve. This mechanism can accurately simulate the supply-side contraction caused by the inability of low-priced generating units to output power when ramp resources are scarce. This enables accurate prediction of nonlinear jumps in spot electricity prices under scenarios of sudden drops in renewable energy output, solving the problem that traditional models struggle to capture extreme price characteristics caused by physical ramp constraints.

[0028] 2. This invention constructs a hysteresis state correction module based on fluctuation duration. Utilizing asymmetric stress lock-in and exponential decay logic, it simulates the recovery inertia of the power system and market participants after experiencing severe fluctuations. By mapping the fluctuation duration to a state decay parameter, the system accurately reflects the long-tail effect of the slow decline in market risk premium after the crisis subsides. This effectively avoids the shortcomings of existing technologies, such as the excessively rapid decline in predicted values ​​during the fluctuation recovery period and deviations from actual trading results, thus improving the robustness of the prediction results in the time-series dimension.

[0029] 3. This invention establishes a hybrid forecasting architecture that integrates physical constraints and market game behavior, tightly coupling unit operating status with supply curve reconstruction through a ramp-up scarcity index. This scheme not only reconstructs the basic market declaration structure using historical transaction data, but also, through a dynamic effective supply curve generation mechanism, concretizes the impact path of physical supply and demand tension on market clearing points, ensuring that the forecast results have clear physical interpretability and high engineering application value even under complex operating conditions involving a high proportion of renewable energy access. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the architecture of the spot clearing electricity price prediction system that integrates the fluctuation characteristics of new energy output according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the spot clearing electricity price prediction method that integrates the fluctuation characteristics of new energy output according to an embodiment of the present invention.

[0031] Among them, 10 is the fluctuation feature extraction module; 20 is the ramp scarcity calculation module; 30 is the hysteresis state correction module; 40 is the supply curve reconstruction module; and 50 is the clearing price solution module. Detailed Implementation

[0032] The technical solutions in 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, 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 are within the scope of protection of the present invention.

[0033] See attached document Figure 1 The present invention provides a spot clearing electricity price prediction system that integrates the fluctuation characteristics of new energy output. The system includes a fluctuation feature extraction module 10, a ramp scarcity calculation module 20, a hysteresis state correction module 30, a supply curve reconstruction module 40, and a clearing electricity price solution module 50.

[0034] The fluctuation feature extraction module 10 is configured to receive the new energy output prediction sequence for the period to be predicted. The fluctuation feature extraction module 10 performs variational mode decomposition on the new energy output prediction sequence to separate low-frequency trend components and high-frequency fluctuation components. The fluctuation feature extraction module 10 further monitors the rate of change of the high-frequency fluctuation components and calculates the integral of the rate of change exceeding a preset physical threshold to obtain the fluctuation impact momentum. The fluctuation feature extraction module 10 also calculates the duration for which the high-frequency fluctuation components continuously exceed a preset physical threshold to obtain the fluctuation duration. The fluctuation feature extraction module 10 finally outputs a fluctuation feature tensor containing power amplitude, fluctuation impact momentum, and fluctuation duration.

[0035] The ramp scarcity calculation module 20 is communicatively connected to the fluctuation feature extraction module 10. The ramp scarcity calculation module 20 acquires the real-time unit combination status and operating parameters of the power system, and estimates the current aggregated remaining ramp capacity based on the unit combination status. The ramp scarcity calculation module 20 also calculates the rate of change of the system's net load. Based on the rate of change of the system's net load, the fluctuation impact momentum in the fluctuation feature tensor, and the aggregated remaining ramp capacity, the ramp scarcity calculation module 20 calculates the baseline equivalent ramp scarcity. The baseline equivalent ramp scarcity characterizes the degree of strain on the system's physical regulation resources at the current moment.

[0036] The hysteresis state correction module 30 is communicatively connected to both the fluctuation feature extraction module 10 and the ramp scarcity calculation module 20. The hysteresis state correction module 30 receives the fluctuation duration and the baseline equivalent ramp scarcity from the fluctuation feature tensor. It dynamically calculates a state decay parameter based on the fluctuation duration, reflecting the rate at which the system recovers from a stress state to a normal state. The hysteresis state correction module 30 uses the state decay parameter to perform asymmetric time-dimensional correction on the baseline equivalent ramp scarcity, generating a corrected scarcity state index. This corrected scarcity state index characterizes the overall system state, including unit start-up / shutdown inertia and market behavior inertia.

[0037] The supply curve reconstruction module 40 is communicatively connected to the hysteresis correction module 30. The supply curve reconstruction module 40 accesses a historical database to match a baseline residual supply curve corresponding to the current load level. It calculates a physical capacity cutoff based on a corrected scarcity index, representing the unit capacity unable to provide service due to ramp-up constraints. The module also calculates a bid slope correction coefficient based on the corrected scarcity index, representing the bid premium generated by market participants due to risk expectations. By applying the physical capacity cutoff and the bid slope correction coefficient, the supply curve reconstruction module 40 geometrically deforms the baseline residual supply curve to generate a dynamic effective supply curve.

[0038] The clearing price calculation module 50 is communicatively connected to the supply curve reconstruction module 40. The clearing price calculation module 50 calculates the difference between the total system load and the predicted renewable energy output at the time to be predicted to obtain the system net load. The clearing price calculation module 50 projects the system net load as the abscissa onto the dynamic effective supply curve and reads the corresponding ordinate value as the base clearing price. The clearing price calculation module 50 also corrects the base clearing price by incorporating key cross-sectional constraint information of the power grid topology, ultimately outputting the predicted spot clearing price.

[0039] In the spot clearing electricity price prediction system that integrates the fluctuation characteristics of new energy output provided by this invention, the core modules achieve cascaded data processing and transmission through well-defined data interfaces and communication buses.

[0040] The fluctuation feature extraction module 10 is located at the beginning of the data processing flow and is equipped with an input interface for receiving external new energy output prediction sequences. The output of the fluctuation feature extraction module 10 establishes communication connections with the inputs of the ramp scarcity calculation module 20 and the hysteresis state correction module 30, respectively. The fluctuation feature extraction module 10 splits and transmits different dimensional components of the processed fluctuation feature tensor. Specifically, the fluctuation feature extraction module 10 transmits the fluctuation impact momentum data, representing the integral of the rate of change, from the fluctuation feature tensor to the ramp scarcity calculation module 20, and simultaneously transmits the fluctuation duration data, representing the duration of high-frequency fluctuations, from the fluctuation feature tensor to the hysteresis state correction module 30.

[0041] The output of the ramp scarcity calculation module 20 is connected to another input of the hysteresis state correction module 30. The ramp scarcity calculation module 20 combines the wave impact momentum received from the wave characteristic extraction module 10 with the unit combination state parameters obtained from the external system to calculate and generate a baseline equivalent ramp scarcity. The ramp scarcity calculation module 20 uses this baseline equivalent ramp scarcity as a fundamental signal characterizing the supply and demand tension at the current physical moment and sends it to the hysteresis state correction module 30.

[0042] The hysteresis state correction module 30, acting as an intermediate node connecting the physical layer and the state layer, has its output connected to the supply curve reconstruction module 40. The hysteresis state correction module 30 simultaneously receives fluctuation duration data from the fluctuation feature extraction module 10 and baseline equivalent ramp scarcity data from the ramp scarcity calculation module 20. The hysteresis state correction module 30 uses the fluctuation duration to determine the state decay parameter and accordingly performs asymmetric time-dimensional correction on the baseline equivalent ramp scarcity, generating a corrected scarcity state index that includes a memory effect. The hysteresis state correction module 30 then transmits this corrected scarcity state index to the supply curve reconstruction module 40.

[0043] The output of the supply curve reconstruction module 40 is connected to the clearing price solution module 50. Based on the received corrected scarcity status index, the supply curve reconstruction module 40 determines the physical capacity cutoff and the price slope correction coefficient, and applies these two parameters to geometrically deform the retrieved benchmark residual supply curve to generate a dynamic effective supply curve. The supply curve reconstruction module 40 sends the dynamic effective supply curve data, which includes the price-capacity mapping relationship, to the clearing price solution module 50. Based on this dynamic effective supply curve and the calculated system net load, the clearing price solution module 50 solves for and outputs the final spot clearing price forecast.

[0044] See attached document Figure 2 In one specific embodiment of the present invention, the fluctuation feature extraction module 10 first performs a signal mode decomposition step, which aims to decompose the original new energy prediction data into frequency components with different physical characteristics, so as to isolate the fluctuation components that have a substantial impact on the power system ramping constraints.

[0045] The fluctuation feature extraction module 10 is equipped with a data receiving interface for acquiring the new energy output prediction sequence for the target time period. The fluctuation feature extraction module 10 processes this new energy output prediction sequence using a variational mode decomposition algorithm. The variational mode decomposition algorithm iteratively searches for the extreme points of the variational model, decomposing the input new energy output prediction sequence into several intrinsic mode functions (IMFs). Each IMF has an independent center frequency and a finite bandwidth. The fluctuation feature extraction module 10 presets a frequency segmentation threshold to divide the decomposed IMFs into a low-frequency trend set and a high-frequency fluctuation set. The low-frequency trend set represents the stable change trend of new energy output, while the high-frequency fluctuation set represents the random disturbance of new energy output. The fluctuation feature extraction module 10 linearly superimposes all IMFs in the high-frequency fluctuation set to reconstruct the high-frequency fluctuation component sequence.

[0046] After obtaining the high-frequency fluctuation component sequence, the fluctuation feature extraction module 10 performs an impact momentum calculation step. The purpose of this step is to quantify the instantaneous crowding-out effect of the new energy fluctuation rate on system regulation resources. The fluctuation feature extraction module 10 first calculates the first derivative of the high-frequency fluctuation component sequence with respect to time, thereby obtaining the real-time rate of change of the high-frequency fluctuation components. The fluctuation feature extraction module 10 internally stores the average ramp-up rate threshold of the system's marginal units, which represents the maximum regulation rate boundary of the system without mobilizing expensive emergency resources.

[0047] The fluctuation feature extraction module 10 compares the absolute value of the real-time rate of change of the high-frequency fluctuation component with the average ramp rate threshold. When the absolute value of the real-time rate of change is less than or equal to the average ramp rate threshold, the fluctuation feature extraction module 10 determines that the current fluctuation is within the system's self-adjusting range and does not generate impact momentum. When the absolute value of the real-time rate of change is greater than the average ramp rate threshold, the fluctuation feature extraction module 10 calculates the difference between the two, which represents the overflow rate exceeding the system's normal adjustment capability. The fluctuation feature extraction module 10 integrates this overflow rate over a preset sliding time window backward from the current moment to obtain the fluctuation impact momentum.

[0048] The wave feature extraction module 10 calculates the wave impact momentum using the following formula: ; in, This represents the momentum of the fluctuation at the current moment; Indicates the length of the preset sliding time window; Indicates the integral variable High-frequency fluctuation components at any given moment; This represents the derivative of the high-frequency fluctuation component with respect to time. This represents the average ramp rate threshold for marginal units in the system. This represents the maximum value function, used to eliminate the integral contribution of the part that has not exceeded the limit.

[0049] While completing the impact momentum calculation, the wave feature extraction module 10 further performs a feature tensor construction step to generate a multi-dimensional feature vector for use by subsequent modules. The wave feature extraction module 10 continuously monitors the real-time rate of change of the high-frequency wave components. The wave feature extraction module 10 calculates the duration for which the absolute value of the real-time rate of change continuously exceeds the average ramp rate threshold, defining this duration as the wave duration. This wave duration reflects the time span characteristics of a single wave event.

[0050] The fluctuation feature extraction module 10 acquires the numerical value of the current time's new energy output prediction sequence as the power amplitude. The fluctuation feature extraction module 10 then arranges and combines the power amplitude, the calculated fluctuation impact momentum, and the fluctuation duration in a predetermined order to construct a three-dimensional fluctuation feature tensor. This fluctuation feature tensor fully describes the physical characteristics of new energy fluctuations in three dimensions: energy magnitude, regulation rate impact, and time span.

[0051] The fluctuation feature extraction module 10 constructs the fluctuation feature tensor according to the following formula: ; in, Indicates at time The output fluctuation feature tensor; This represents the power amplitude of the predicted output of new energy sources at the current moment. This represents the momentum of the fluctuation at the current moment; This represents the duration of the fluctuation as statistically obtained at the current moment. The fluctuation feature extraction module 10 sends this fluctuation feature tensor as output data to the subsequent processing module of the system.

[0052] In an embodiment of the present invention, the ramp scarcity calculation module 20 first performs a remaining ramp capacity assessment step, which aims to accurately quantify the maximum regulation rate resources that the power system can call upon at the current moment from the physical supply side.

[0053] The ramp scarcity calculation module 20 is equipped with a communication interface with the power dispatch control center or historical operation database to obtain the unit combination status and operating parameters of the power system in real time. The ramp scarcity calculation module 20 first identifies the set of online units currently in operation based on the unit start-up / shutdown status identifiers. For each generator unit in the online unit set, the ramp scarcity calculation module 20 reads its rated maximum ramp rate parameter, maximum technical output limit, and current real-time power output value. The ramp scarcity calculation module 20 calculates the increase in output that each unit can achieve per unit time. This calculation process follows the principle of minimizing physical constraints, meaning that the actual adjustable capacity of the unit is limited by its maximum ramp rate of mechanical characteristics, and also by the remaining capacity space between its current output point and the maximum technical output limit. The ramp scarcity calculation module 20 selects the smaller value between the maximum ramp rate and the remaining capacity space as the effective upward ramp capacity of the single unit.

[0054] The ramp scarcity calculation module 20 sums the effective ramp capabilities of all individual units in the online unit set to obtain the aggregated remaining ramp capability at the system level. This indicator represents the maximum positive power regulation rate that the entire power system can provide solely based on the currently available online spinning reserve resources, without starting any new cold standby units.

[0055] The climbing scarcity calculation module 20 calculates the aggregated remaining climbing capacity based on the following formula: ; in, This indicates the system's remaining upward climbing capacity at time t. This represents the set of online generating units that are currently powered on. Indicates the first The rated maximum gradeability of the unit; Indicates the first The maximum technical output limit of the Taiwanese unit; Indicates the first Taiwanese crew at all times Real-time power output value; This represents the function that takes the minimum value.

[0056] After completing the supply-side capacity assessment, the ramp-up scarcity calculation module 20 then executes the benchmark scarcity index generation step. This aims to construct a dimensionless physical index reflecting the system's supply-demand tension by combining demand-side changes with supply-side capacity. The ramp-up scarcity calculation module 20 first obtains the predicted total system load and the predicted renewable energy output at the time to be predicted, and calculates the system net load sequence by subtracting the two. The ramp-up scarcity calculation module 20 then differentiates the system net load sequence with respect to time to obtain the net load change rate. The net load change rate reflects the benchmark adjustment rate required by the system to maintain power balance under normal conditions, without considering the impact of drastic fluctuations in renewable energy.

[0057] The ramp scarcity calculation module 20 receives fluctuation impact momentum data from the fluctuation feature extraction module 10. Fluctuation impact momentum, as a non-linear penalty term, characterizes the additional regulation demand brought about by the high-frequency and drastic fluctuations of new energy sources. The ramp scarcity calculation module 20 adds the absolute value of the net load change rate to the weighted fluctuation impact momentum to form the total ramp demand of the system. The ramp scarcity calculation module 20 divides the total ramp demand of the system by the aggregated remaining ramp capacity calculated in step one, thereby generating the benchmark equivalent ramp scarcity. The larger the value of this index, the more scarce the current regulation resources of the system are relative to the regulation demand, and the higher the physical probability of a price peak.

[0058] The slope scarcity calculation module 20 calculates the benchmark equivalent slope scarcity according to the following formula: ; in, Indicates at time The calculated baseline equivalent ramp scarcity; Indicates time The system net load; Indicates the rate of change of net load; This represents the momentum of the fluctuation at the current moment; This represents the preset impact weighting coefficient, used to adjust the proportion of contribution of fluctuation impact to scarcity. This indicates the system's aggregate remaining upward ramping capability; This represents a very small positive constant to prevent numerical calculation errors where the denominator is zero. The slope scarcity calculation module 20 outputs the calculated baseline equivalent slope scarcity to the hysteresis state correction module 30.

[0059] In an embodiment of the present invention, the hysteresis state correction module 30 first performs a state memory and hysteresis mechanism construction step. This step aims to establish a state variable with time-dependent characteristics to correct the bias of linear prediction based solely on physical parameters at the current moment.

[0060] The hysteresis state correction module 30 is equipped with a bidirectional data input channel, used to receive fluctuation duration data from the fluctuation feature extraction module 10 and baseline equivalent ramp scarcity data from the ramp scarcity calculation module 20. The hysteresis state correction module 30 internally has a preset state register to store the corrected scarcity state index calculated at the previous moment. Based on the physical operating characteristics of the power system, the hysteresis state correction module 30 defines two state update modes. The first mode is the stress-locked-off mode, corresponding to scenarios where the physical supply and demand tension in the system intensifies. In this scenario, the ramp constraint of the generator units is triggered, and the market price responds quickly to the scarcity signal; therefore, the system state should immediately follow the rise of physical indicators. The second mode is the hysteresis recovery mode, corresponding to scenarios where renewable energy output recovers and the physical supply and demand tension eases. In this scenario, due to the minimum operating time constraint, minimum downtime constraint of thermal power units, and the risk-averse inertia of market participants, the recovery speed of the system state lags behind the recovery speed of physical parameters. The hysteresis state correction module 30 simulates this dual physical and economic inertia by introducing a time decay mechanism.

[0061] The hysteresis state correction module 30 then executes the dynamic attenuation parameter calculation step. The core of this step is to dynamically adjust the speed of system state recovery based on the severity of the renewable energy fluctuation event. The hysteresis state correction module 30 reads the fluctuation duration, which reflects the length of the high-frequency fluctuation event the system has just experienced. The longer the fluctuation duration, the more deeply the system's frequency regulation resources are consumed, and the higher the level of safety warning for dispatchers and market participants in the short term. Therefore, the time constant required for the system to recover to a normal state should be larger.

[0062] The hysteresis state correction module 30 utilizes a linear mapping relationship to transform the fluctuation duration into state decay parameters used in the control state update equation. The hysteresis state correction module 30 calculates the state decay parameters according to the following formula: ; in, This represents the state decay parameter at the current moment; This represents the system's preset basic hysteresis constant, and signifies the natural recovery rate under conditions of no severe fluctuations or shocks. This represents the preset coupling coefficient, used to quantify the degree to which the duration of fluctuation affects the recovery rate; This indicates the duration of the fluctuation as statistically obtained at the current moment.

[0063] After determining the current decay characteristics, the hysteresis state correction module 30 performs an asymmetric state update step. The hysteresis state correction module 30 compares the current input baseline equivalent ramp scarcity with the corrected scarcity state index stored in the previous time step.

[0064] When the current baseline equivalent ramp scarcity is greater than or equal to the revised scarcity state index from the previous moment, the hysteresis state correction module 30 determines that the system is in the stress lock-in phase. At this time, the hysteresis state correction module 30 directly assigns the current baseline equivalent ramp scarcity to the revised scarcity state index, achieving zero-delay capture of risk signals.

[0065] When the current equivalent ramp scarcity is less than the corrected scarcity state index from the previous moment, the hysteresis state correction module 30 determines that the system is in the hysteresis recovery phase. At this time, the hysteresis state correction module 30 uses the state decay parameters calculated above to construct an exponential decay function or a weighted smoothing function. Based on the state value from the previous moment, the hysteresis state correction module 30 slowly approximates the current physical baseline value, thereby calculating the corrected scarcity state index for the current moment. This process ensures that even if wind and solar power output suddenly increases at the physical level, the predicted system's output state index will remain high for a period of time, accurately reflecting the market phenomenon that wind power supply has stopped but prices have not fallen.

[0066] Hysteresis state correction module 30 calculates the corrected scarcity state index according to the following formula: ; in, Indicates at time The output is a corrected scarcity status index; This indicates the state index at the previous moment; This indicates the baseline equivalent ramp scarcity at the current moment; Indicates the time step of the forecast; This represents the state decay parameter at the current moment. The hysteresis state correction module 30 will calculate the... Output to the supply curve reconstruction module 40.

[0067] In an embodiment of the present invention, the supply curve reconstruction module 40 performs a series of operations based on the physical and economic coupling mechanism, aiming to transform static historical data into a dynamic model that can reflect the current instantaneous supply and demand tension.

[0068] The supply curve reconstruction module 40 first performs a baseline curve matching step. The purpose of this step is to obtain an undisturbed initial market quote structure as the basic framework for subsequent deformation processing. The supply curve reconstruction module 40 is configured with access to a historical operating database, which stores spot market declaration data and clearing results for historical periods. The supply curve reconstruction module 40 reads the time attributes (including season, weekday / holiday type, and period index) of the period to be predicted, as well as the total system load level provided by an external load forecasting system. The supply curve reconstruction module 40 uses a weighted Euclidean distance algorithm to retrieve, from the historical database, the historical trading day that is closest to the period to be predicted in terms of time attributes and load level.

[0069] The supply curve reconstruction module 40 extracts the benchmark residual supply curve from the selected historical trading day data. The benchmark residual supply curve is generated by sorting the bid prices of all market-based generating units monotonically in ascending order of capacity using a step function or piecewise linear function. This curve describes the trajectory of the marginal clearing price as the system's net load demand increases under standard operating conditions. The supply curve reconstruction module 40 stores the extracted benchmark residual supply curve as a sequence of discrete coordinate points representing price and capacity.

[0070] After obtaining the baseline curve, the supply curve reconstruction module 40 performs an effective capacity truncation step. This step aims to simulate the physical phenomenon where low-priced units have capacity but cannot generate power due to ramp-up rate limitations. The supply curve reconstruction module 40 receives a corrected scarcity status index from the hysteresis correction module 30. The higher the index value, the more urgent the system's need for adjustment rate, which leads to some baseload units with lower ramp-up rates being in the low-price zone in the bidding sequence but unable to respond to load changes in a timely manner within the current scheduling interval due to physical ramp-up constraints. This unavailable capacity is mathematically equivalent to being temporarily removed from the low-price zone on the left side of the supply curve.

[0071] The supply curve reconstruction module 40 calculates the physical capacity cutoff using the modified scarcity status index. The supply curve reconstruction module 40 has preset parameters for the total installed capacity of the system and a cutoff sensitivity coefficient. The physical capacity cutoff is defined as a function that monotonically increases with the scarcity status index.

[0072] The supply curve reconstruction module 40 calculates the physical capacity cutoff based on the following formula: ; in, Indicates at time Physical capacity cutoff; This indicates the total installed capacity of generating units participating in the market; This represents the preset cutoff sensitivity coefficient, used to control the response rate of the cutoff amount as the state changes; This indicates the corrected scarcity status index from the hysteresis state correction module 30. This indicates the midpoint of the preset state threshold.

[0073] Subsequently, the supply curve reconstruction module 40 performs a price slope drift step. This step aims to simulate the economic behavior of market participants who, when perceiving system supply and demand tensions and volatility risks, tend to raise their price levels to obtain a risk premium. The supply curve reconstruction module 40 believes that as the revised scarcity state index increases, market participants will not only reduce effective supply by withholding sales, but also steepen the overall slope of the supply curve by adjusting their pricing strategies.

[0074] The supply curve reconstruction module 40 calculates the price slope correction coefficient based on the revised scarcity state index. This coefficient is a dimensionless value greater than or equal to 1, used to multiplicatively expand the price axis of the benchmark curve. The supply curve reconstruction module 40 maps the scarcity state to the price slope correction coefficient using a linear or polynomial regression model.

[0075] Finally, the supply curve reconstruction module 40 performs a curve synthesis step, superimposing the aforementioned physical cutoff effect and economic drift effect onto the baseline residual supply curve to generate the final dynamic effective supply curve. The supply curve reconstruction module 40 performs coordinate transformation on each coordinate point on the baseline residual supply curve. On the capacity axis, the supply curve reconstruction module 40 shifts the coordinate point to the left by an amount equal to the physical capacity cutoff, reflecting the unavailability of low-priced resources. On the price axis, the supply curve reconstruction module 40 multiplies the price value of the coordinate point by a price slope correction coefficient to reflect the price increase resulting from the risk premium.

[0076] The supply curve reconstruction module 40 constructs the functional relationship of the dynamic effective supply curve based on the following formula: ; in, This indicates the restructured dynamic effective supply curve at effective capacity. The corresponding price at that location; The price function representing the baseline residual supply curve; This represents the horizontal axis offset introduced when reading the benchmark price, which is determined by the physical capacity truncation. This represents the preset slope sensitivity coefficient; This indicates the revised scarcity status index; This is the price quote slope correction coefficient. The supply curve reconstruction module 40 transmits the generated dynamic effective supply curve data to the clearing price solution module 50.

[0077] In an embodiment of the present invention, the clearing price solution module 50 serves as the system's terminal processing unit, responsible for performing simulation calculations of the market clearing logic. This module couples the dynamic cost characteristics of the supply side with the real-time load level of the demand side to solve for the final clearing price prediction.

[0078] The clearing price solution module 50 first performs the net load calculation step. This step aims to determine the actual power gap that needs to be met through competitive bidding in the spot market. The clearing price solution module 50 receives the system total load forecast sequence and the renewable energy output forecast sequence for the forecast period through a data interface. The system total load forecast sequence reflects the total electricity demand of all electricity users during that period, while the renewable energy output forecast sequence reflects the expected power generation of uncontrollable power sources such as wind power and photovoltaics.

[0079] The clearing price calculation module 50, based on the principle of priority consumption in the electricity market, treats new energy sources as mandatory generating units or priority dispatch resources with zero marginal cost. The module processes data for each forecast time segment, subtracting the corresponding new energy output forecast from the total system load forecast; the result is the system net load. This system net load represents the surplus electricity demand that must be met by dispatchable units such as thermal and hydropower through market bidding. In the presence of tie-line exchange plans or fixed bilateral contracts, the clearing price calculation module 50 further deducts these non-market fixed power components in the above calculation, thereby accurately locking in the target electricity volume for spot bidding.

[0080] After determining the market demand-side parameters, the clearing price solution module 50 performs the projection solution and output steps. This step simulates the process of a power trading center matching transactions and determining marginally clearing units. The clearing price solution module 50 receives the dynamic effective supply curve generated by the supply curve reconstruction module 40. Mathematically, this curve represents a function of price versus capacity, and already includes capacity truncation and price slope correction features caused by ramp scarcity.

[0081] The clearing price solution module 50 uses the calculated system net load as the independent variable and projects it onto the capacity axis of the dynamic effective supply curve. The clearing price solution module 50 then retrieves the price coordinate point on the dynamic effective supply curve corresponding to the system net load value. Specifically, the clearing price solution module 50 finds the supply function segment whose capacity covers the system net load and calculates its corresponding function value. This function value represents the marginal cost that must be paid at the current moment to meet the system's last megawatt of electricity demand, which is the predicted clearing price value in the spot market.

[0082] The clearing price calculation module 50 calculates the final spot clearing price forecast based on the following formula: ; in, Indicates at time Output of the spot clearing electricity price forecast; This represents the dynamic effective supply curve function output by the supply curve reconstruction module 40; Indicates time The predicted total system load; Indicates time Forecast values ​​of new energy power output; Indicates time The tie lines and fixed contract power (0 if none). The clearing price solution module 50 will calculate the... The output is formatted according to the time series to complete the entire prediction process.

[0083] This invention provides a computer device configured to perform the functions of a spot clearing electricity price prediction system that integrates the fluctuation characteristics of new energy output. The computer device mainly includes a processor, memory, communication interface, and communication bus at the hardware level.

[0084] The processor, memory, and communication interface are physically connected and transmit signals to each other via a communication bus. The communication bus, as the data transmission channel, is responsible for transmitting control signals, address signals, and data signals between these components. The communication interface is the physical port through which the computer device interacts with external networks or data sources. In this embodiment, the communication interface is used to receive real-time output prediction sequence data from the new energy power prediction system, receive historical operating data and real-time unit status parameters from the power grid dispatch center, and send the calculated clearing price prediction results to the power market trading auxiliary decision-making terminal.

[0085] The memory is used to store computer programs and temporary and persistent data required for system operation. Physically, the memory includes at least one type of computer-readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory, random access memory, or read-only memory. The memory stores an operating system for managing the hardware resources of the computer device. The memory also stores application code specifically for implementing the methods of the embodiments of the present invention. This application code contains sequences of computer instructions corresponding to the fluctuation feature extraction module 10, the ramp scarcity calculation module 20, the hysteresis state correction module 30, the supply curve reconstruction module 40, and the clearing price solution module 50.

[0086] The processor is the core of a computer device's computation and control center. It processes data and makes logical judgments by reading and executing computer program instructions stored in memory. Specifically, when the processor executes application code, it is configured to perform the following operations: first, it calls the instructions of the fluctuation feature extraction module 10 to perform variational mode decomposition on the input new energy output prediction sequence, calculates the impact momentum of high-frequency components, and constructs a three-dimensional feature tensor containing power amplitude, impact momentum, and fluctuation duration.

[0087] Subsequently, the processor invokes instructions from the ramp scarcity calculation module 20 to assess the system's aggregate remaining ramp capacity based on the physical parameters of the online units, and calculates the equivalent ramp scarcity by combining the net load change rate and the fluctuation impact momentum. Next, the processor invokes instructions from the hysteresis state correction module 30 to calculate dynamic decay parameters based on the fluctuation duration, and uses asymmetric state update logic to calculate the corrected scarcity state index to simulate the hysteresis recovery characteristics of market prices.

[0088] Based on this, the processor calls the instructions of the supply curve reconstruction module 40 to retrieve the historical benchmark residual supply curve, and uses the corrected scarcity status index to perform physical capacity truncation and price slope drift correction on the curve to generate a dynamic effective supply curve. Finally, the processor calls the instructions of the clearing price solution module 50 to calculate the system net load, and projects the net load onto the dynamic effective supply curve to solve for the final spot clearing price forecast.

[0089] This invention further provides a computer-readable storage medium. The computer-readable storage medium can be non-volatile or volatile. A computer program is stored on the computer-readable storage medium. The computer program includes a series of computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the processors perform all the steps of the aforementioned spot-clearing electricity price prediction method that incorporates the fluctuation characteristics of new energy output.

[0090] Specific implementations of the computer-readable storage medium include, but are not limited to, phase-change memory, static random access memory, dynamic random access memory, electrically erasable programmable read-only memory, optical disc read-only memory or other optical disc storage or disk storage media. By carrying computer programs, this storage medium enables the aforementioned technical solutions for fluctuation feature extraction, ramp-up scarcity calculation, hysteresis state correction, supply curve reconstruction, and clearing price solution to be distributed and deployed independently of specific hardware terminals, thereby achieving the technical effects of this invention on any device with general-purpose computing capabilities.

Claims

1. A spot clearing electricity price prediction system that integrates the fluctuation characteristics of new energy output, characterized in that, include: The fluctuation feature extraction module (10) is used to perform variational mode decomposition on the new energy output prediction sequence to generate fluctuation impact momentum and fluctuation duration. The ramp scarcity calculation module (20) is used to estimate the aggregated remaining ramp capacity based on the unit combination state, add the fluctuation impact momentum to the rate of change of the system net load, and calculate the benchmark equivalent ramp scarcity relative to the aggregated remaining ramp capacity. Hysteresis state correction module (30) is used to perform time correction on the benchmark equivalent climbing scarcity using the state decay parameter based on the duration of the fluctuation, and generate the corrected scarcity state index. The supply curve reconstruction module (40) is used to geometrically deform the benchmark residual supply curve by using the physical capacity cutoff amount and the price slope correction coefficient obtained by mapping the modified scarcity state index to generate a dynamic effective supply curve. The clearing price solution module (50) is used to project the system net load at the time to be predicted onto the dynamic effective supply curve, read the corresponding price coordinates, and output the spot clearing price prediction value.

2. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, The process by which the fluctuation feature extraction module (10) performs variational mode decomposition on the new energy output prediction sequence to generate fluctuation impact momentum and fluctuation duration includes: The new energy output prediction sequence is decomposed into several intrinsic mode functions using the variational mode decomposition algorithm. High-frequency fluctuation components are obtained by reconstructing the sequence using frequency segmentation thresholds. The real-time rate of change is obtained by differentiating the high-frequency fluctuation components with respect to time. The absolute value of the real-time rate of change is compared with the average ramp rate threshold of the system's marginal units, and the integral of the difference between the real-time rate of change and the average ramp rate threshold within the sliding time window is calculated to obtain the fluctuation impact momentum. The duration of the fluctuation is obtained by calculating the length of time during which the absolute value of the real-time rate of change continuously exceeds the average ramp rate threshold.

3. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, The process of the ramp scarcity calculation module (20) performing the aggregated ramp capacity estimation based on unit combination status includes: Obtain the real-time unit combination status of the power system and identify the set of online units currently in operation; Read the rated maximum ramp rate and remaining capacity space from the maximum technical output limit for each unit in the online unit set; The smaller of the rated maximum ramp rate and the remaining capacity space is selected as the effective upward ramp capacity of a single unit. The effective upward ramp capacities of all the single units are summed to obtain the aggregated remaining ramp capacity.

4. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, The process by which the ramp scarcity calculation module (20) performs the addition of the rate of change of the fluctuation impact momentum to the system net load and calculates the benchmark equivalent ramp scarcity relative to the aggregated remaining ramp capacity includes: Calculate the derivative of the system net load with respect to time to obtain the net load change rate. Add the absolute value of the net load change rate to the weighted fluctuation momentum to form the total system ramp-up demand that reflects the total system regulation demand. Dividing the total system ramp demand by the aggregated remaining ramp capacity yields the benchmark equivalent ramp scarcity, which characterizes the degree of tension in the supply and demand of physical regulation resources.

5. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, The process by which the hysteresis state correction module (30) obtains the state decay parameter based on the fluctuation duration includes: The fluctuation duration is received from the fluctuation feature extraction module (10), the fluctuation duration reflecting the persistence of the system under the impact of high-frequency fluctuations; Using a linear mapping relationship, the system's preset basic hysteresis constant is used as the intercept, and the preset coupling coefficient is used as the slope to map the fluctuation duration to the state decay parameter. The state decay parameter is used to quantify the rate at which the system recovers from the stress state to the normal state.

6. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, The process by which the hysteresis state correction module (30) performs time correction on the benchmark equivalent ramp scarcity using a state decay parameter based on the duration of fluctuation, and generates a corrected scarcity state index, includes: Compare the baseline equivalent climbing scarcity input at the current moment with the corrected scarcity state index stored at the previous moment; When the baseline equivalent ramp scarcity is greater than or equal to the modified scarcity state index stored at the previous moment, the system is determined to be in stress lock mode, and the baseline equivalent ramp scarcity at the current moment is directly assigned to the modified scarcity state index at the current moment to maintain an immediate response to stress. When the baseline equivalent ramp scarcity is less than the corrected scarcity state index stored at the previous time, the system is determined to be in hysteresis recovery mode. An exponential decay function is constructed using the state decay parameter. The exponential decay function is used to approximate the baseline equivalent ramp scarcity at the current time based on the corrected scarcity state index value stored at the previous time, and the corrected scarcity state index at the current time is calculated.

7. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, Before performing the geometric deformation of the reference residual supply curve, the process of obtaining the reference residual supply curve by the supply curve reconstruction module (40) includes: Read the time attributes and total system load level of the period to be predicted, and use the weighted Euclidean distance algorithm to search the historical operation database for the historical trading day that is closest to the time attributes and total system load level of the period to be predicted. Extract the application data from the historical trading days and sort all market-based generating units by capacity in a monotonically increasing order. A sequence of discrete price and capacity coordinate points is generated as the baseline residual supply curve to be reconstructed.

8. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, The supply curve reconstruction module (40) performs the process of using the physical capacity cutoff and the price slope correction coefficient obtained by mapping the modified scarcity status index, including: Based on the total installed capacity of the system and the cutoff sensitivity coefficient, an S-shaped function that monotonically increases with the modified scarcity state index is constructed to calculate the physical capacity cutoff amount representing the unit capacity that cannot provide services due to ramp-up obstruction. Based on the preset slope sensitivity coefficient, a regression model is constructed that maps the corrected scarcity state index to a value greater than or equal to 1, and the price slope correction coefficient representing the risk premium of market participants is calculated.

9. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, The supply curve reconstruction module (40) performs the following process: using the physical capacity cutoff amount and the price slope correction coefficient obtained by mapping the modified scarcity state index to geometrically deform the benchmark residual supply curve and generate a dynamic effective supply curve: Using the physical capacity cutoff and the price slope correction coefficient, traverse every coordinate point on the benchmark residual supply curve; The coordinate point is shifted to the left along the capacity axis, and the shift value is equal to the physical capacity cutoff, reflecting the unavailability of low-cost resources; Multiply the price value at the coordinate point along the price axis by the price slope correction coefficient to reflect the price increase caused by the risk premium, and generate the dynamic effective supply curve.

10. The spot clearing electricity price prediction system integrating the fluctuation characteristics of new energy output according to claim 1, characterized in that, The clearing price solution module (50) performs the process of projecting the system net load at the time to be predicted onto the dynamic effective supply curve, reading the corresponding price coordinates, and outputting the predicted spot clearing price value, including: The net system load is obtained by subtracting the predicted output of the new energy source and the fixed switching power of the tie line from the predicted total system load at the time to be predicted. Project the system net load as the abscissa onto the capacity axis of the dynamic effective supply curve, and retrieve the supply function segment whose capacity covers the system net load; Read the price ordinate value corresponding to the supply function segment, correct it in combination with the key section constraint information of the power grid, and output the final spot clearing electricity price prediction value.

Images