Electric power spot market robust clearing efficient solving method considering uncertainty

Abstract

The invention discloses an electric power spot market robust clearing efficient solving method considering uncertainty, and relates to the technical field of electric power market optimization and scheduling, and the method comprises the steps: constructing a security constraint unit combination model, and constructing a multi-scene new energy power prediction model; constructing a robust economic dispatching model by embedding random scene constraints; and performing feasibility verification of the power generation plan on the robust economic dispatching model through a parallel solving method, performing optimization calculation of the robust economic dispatching model according to a feasibility verification result, and outputting a clearing result of each unit. According to the method, the fluctuation of new energy output is simulated through a kernel density estimation method and a Markov chain technology, random input data which widely covers different power fluctuation conditions is provided for a robust economic dispatching model, and through parallel solving and dynamic screening of infeasible scenes, the power fluctuation of the new energy output is greatly reduced. The overall solving speed and the real-time scheduling capability of the system are improved, and the scheduling effect of considering economy and reliability is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of power market optimization and dispatching, and in particular to an efficient solution method for robust clearing of a power spot market taking uncertainty into consideration. Background Art

[0002] As the environmental pollution caused by fossil fuel consumption becomes more and more serious, the global power industry is actively turning to low-carbon, low-pollution renewable energy such as wind power and photovoltaics. However, these new energy power generation has significant volatility and randomness, which brings huge challenges to the stability of the power grid. In the current power market system, due to the imperfect technology and mechanism, the process of power marketization is still relatively slow, and it is difficult to effectively promote the development of renewable energy. The participation of market players and the scope of transactions are limited, making it difficult for transactions in the new energy market to grow steadily. In addition, due to the imperfect pricing mechanism and cross-regional transmission mechanism, cross-provincial power transmission and transactions face bottlenecks, especially the lack of auxiliary services such as frequency regulation and peak regulation, which leads to unstable power grid operation and further aggravates the dilemma of new energy consumption. Traditional frequency regulation and standby services often adopt more conservative strategies to ensure the stability of the system, but also sacrifice the economy of the system. At the same time, due to the increasing capacity of renewable energy grid connection, it occupies the space of traditional units, making the problem of insufficient peak regulation and frequency regulation capacity more prominent, especially in the case of not entering the auxiliary service market, which further increases the difficulty and uncertainty of system operation.

[0003] In the existing technology, in order to cope with the uncertainty of renewable energy output, methods such as opportunity constraints, robust optimization and scenario methods are widely used to embed uncertainty constraints into the electricity market clearing model, which has improved the new energy absorption capacity to a certain extent. However, due to the complexity of the constraints, including a large number of nonlinearities and non-convexities, the model calculation efficiency is low and the solution time is long, which is difficult to meet the high efficiency requirements of the spot market. Especially when dealing with large-scale multi-scenario problems, the lack of computing performance has become a bottleneck restricting the application of the electricity market clearing model. In addition, traditional models are difficult to dynamically update and cope with the rapid volatility of renewable energy power, and it is difficult to truly meet the needs of widespread access to renewable energy. Therefore, how to effectively consider complex uncertainty constraints in the electricity spot market clearing model and ensure the reliability and economy of the solution while improving the computing efficiency is a technical problem that needs to be solved urgently. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: the existing robust optimization method of the electricity spot market has the problems of low computational efficiency, insufficient scenario feasibility, poor scheduling flexibility, and how to efficiently achieve robust power clearing under uncertain conditions.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: an efficient solution method for robust clearing of the electricity spot market taking into account uncertainty, comprising constructing a safety-constrained unit combination model and simultaneously constructing a multi-scenario new energy power prediction model; constructing a robust economic dispatch model by embedding random scenario constraints; and performing a feasibility check of the power generation plan of the robust economic dispatch model through a parallel solution method, performing optimization calculation of the robust economic dispatch model based on the feasibility check results, and outputting the clearing results of each unit.

[0007] As a preferred solution for the efficient solution method of robust clearing of the electricity spot market considering uncertainty described in the present invention, wherein: the construction of the safety constraint unit combination model includes constructing a safety constraint unit combination model under the prediction scenario based on the predicted boundary information, and obtaining the start-up and shutdown plans of each unit.

[0008] As a preferred solution for the method for efficiently solving the robust clearing of the electricity spot market taking into account uncertainty described in the present invention, the multi-scenario new energy power prediction model includes generating at least 1,000 random scenarios of new energy power sequences.

[0009] As a preferred solution to the method for efficiently solving the robust clearing of the electricity spot market taking into account uncertainty described in the present invention, wherein: the construction of the robust economic dispatch model includes fixing the start-up and shutdown plans of each unit before constructing the robust economic dispatch model.

[0010] As a preferred solution to the efficient solution method for robust clearing of the electricity spot market considering uncertainty described in the present invention, wherein: the embedded random scenario constraints include the number of constructed robust economic dispatch models being the same as the number of random scenarios of new energy power sequences.

[0011] As an optimal solution for the efficient solution method for robust clearing of the electricity spot market considering uncertainty described in the present invention, wherein: the parallel solution method includes performing a feasibility check on the power generation plan for each robust economic dispatch model. If the robust economic dispatch model has a solution, the feasibility check is passed under the random scenario of the new energy power sequence, and the economic dispatch model optimization calculation is performed on the prediction scenario model, and the clearing results of each unit are output; if the robust economic dispatch model has no solution, an infeasible random scenario of the new energy power sequence is output to form an infeasible scenario set, and the scenarios with the most serious cross-section flow over-limit and the largest amount of wind and solar power abandonment are selected to construct a robust dispatch optimization model.

[0012] As a preferred solution of the method for efficiently solving the robust clearing of the electricity spot market considering uncertainty described in the present invention, wherein: the feasibility verification result includes screening out infeasible scenarios if the feasibility verification fails, and further selecting the most constrained scenarios, constructing a robust scheduling optimization model, obtaining new start-up and shutdown plans of each unit based on the robust scheduling optimization model and fixing them, reconstructing the robust economic scheduling model and performing a feasibility verification of the power generation plan, when the number of infeasible scenario sets is equal to 0, the feasibility verification passes, and the economic scheduling model is optimized and calculated for the new start-up and shutdown plans of each unit, and the clearing results of each unit that meet the uncertainty of intra-day new energy are output; when the number of infeasible scenario sets is greater than 0, repeatedly constructing the robust scheduling optimization model and the robust economic scheduling model, and performing a feasibility verification of the power generation plan, until the number of infeasible scenario sets is equal to 0, the feasibility verification of the power generation plan is completed.

[0013] Another object of the present invention is to provide an efficient solution system for robust clearing of the electricity spot market taking into account uncertainty, which can use a parallel solution method to verify the feasibility of the power generation plan of the robust economic dispatch model, perform optimization calculations on the robust economic dispatch model based on the feasibility verification results, and output the clearing results of each unit, thereby solving the problem of low computational efficiency of the current robust optimization technology in the electricity spot market.

[0014] As a preferred solution of the robust clearing and efficient solution system of the power spot market considering uncertainty described in the present invention, it includes: a multi-scenario construction module, a robust scheduling construction module, and a parallel solution and verification module; the multi-scenario construction module is used to construct a safety constraint unit combination model, and at the same time construct a multi-scenario new energy power prediction model; the robust scheduling construction module is used to embed random scenario constraints, construct a robust economic scheduling model, and generate a corresponding scheduling model for each new energy power sequence by fixing the unit start-up and shutdown plan; the parallel solution and verification module is used to solve the robust economic scheduling model in parallel to realize the feasibility verification of the power generation plan, optimize the scheduling model according to the verification results to form an infeasible scenario set, and further screen out the most constrained scenario to update the scheduling plan, until all scenarios pass the feasibility verification, and finally output the clearing results of each unit that meets the uncertainty of intra-day new energy.

[0015] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a step of an efficient solution method for robust clearing of a power spot market taking uncertainty into consideration.

[0016] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of an efficient solution method for robust clearing of a power spot market taking uncertainty into consideration.

[0017] Beneficial effects of the present invention: The efficient solution method for robust clearing of the power spot market considering uncertainty provided by the present invention realizes the optimization of the start-up and shutdown schemes of different units in the power system by constructing a safety constraint unit combination model combined with a multi-scenario new energy power prediction model. The safety constraint unit combination model generates a unit start-up and shutdown plan that meets the system safety constraints by predicting boundary information, and further ensures the stability of the system under the condition of large volatility of new energy output. The multi-scenario new energy power prediction model generates at least 1,000 random scenarios, simulates the fluctuation of new energy output through kernel density estimation and Markov chain technology, can ensure the rigid demand of safe scheduling, and provides random input data that widely covers different power fluctuations for the subsequent robust economic scheduling model. In the process of constructing the prediction model, it provides comprehensive data support for different power scenarios, thereby effectively coping with the randomness and volatility of new energy, and laying a data and technical foundation for the safe operation of the power grid and large-scale access of new energy. On the basis of constructing the safety constraint unit combination model, the random scenario constraints are further embedded in the economic scheduling model to form a robust economic scheduling model. By fixing the unit in each random scenario The start-up and shutdown states are determined, and the corresponding random scenario constraints are generated. The robust economic dispatch model realizes the economic dispatch requirements of the power grid under different renewable energy output scenarios. By embedding the random scenario constraints, the system has higher adaptability in the power fluctuations between different scenarios, ensuring the load balance and system stability under different scenarios, reducing the adverse effects that may be caused by renewable energy fluctuations, ensuring the optimization adaptability of the power dispatching scheme and the stability of system operation, and improving the responsiveness of the system in different scenarios. It provides a reliable technical guarantee for the dual requirements of power grid economy and safety. The robust economic dispatching model is efficiently solved by a parallel solution method, and the feasibility of the power generation plan is verified, realizing the screening and optimization of infeasible scenarios. Through multiple rounds of infeasible scenario feedback and verification, the system finally outputs the optimal start-up and shutdown plan that meets the requirements of all random scenarios. Through parallel solution and dynamic screening of infeasible scenarios, the problem of wind and solar abandonment caused by renewable energy volatility in traditional methods is reduced, the overall solution speed and real-time dispatching capability of the system are improved, and the dispatching effect that takes into account both economy and reliability is achieved. The present invention achieves better results in terms of robustness, reliability and economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1An overall flow chart of an efficient solution method for robust clearing of the electricity spot market taking uncertainty into consideration provided in the first embodiment of the present invention.

[0020] Figure 2 A technical detail diagram of an efficient solution method for robust clearing of the electricity spot market taking into account uncertainty provided in the first embodiment of the present invention.

[0021] Figure 3 A schematic diagram of modules of an efficient solution system for robust clearing of the electricity spot market taking uncertainty into consideration provided in the third embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0023] Example 1, reference Figure 1-Figure 2 , which is an embodiment of the present invention, provides an efficient solution method for robust clearing of power spot market considering uncertainty, including:

[0024] S100: Build a safety constraint unit combination model and a multi-scenario new energy power prediction model.

[0025] Furthermore, building a safety constrained unit commitment model includes constructing a safety constrained unit commitment (SCUC) model under the prediction scenario based on the predicted boundary information, using CPLEX and Gurobi solvers to efficiently solve the SCUC model, and obtaining the start-up and shutdown plans of each unit, including the following specific steps:

[0026] S101-1, construct the objective function of the SCUC model under the prediction scenario, assuming that the number of thermal power units in a certain power grid is N (n = 1, 2, ..., N), the number of hydropower stations is M (m = 1, 2, ..., M), the number of gas power units is R (r = 1, 2, ..., R), the number of nuclear power units is D (d = 1, 2, ..., D), the number of pumped storage power stations is C (c = 1, 2, ..., C), the number of wind farms is J (j = 1, 2, ..., J), the number of photovoltaic power stations is K (k = 1, 2, ..., K), except for wind power and photovoltaic, other units are collectively referred to as conventional units, that is, the number of conventional units is G = N + M + R + D (g = 1, 2, ..., G), and the scheduling period is T (t = 1, 2, ..., T).

[0027] The solution of robust scheduling is the optimal solution under the prediction scenario and can adapt to all random scenarios. Therefore, the goal of the scheduling model is the total cost under the prediction scenario. Since the power generation cost of new energy units is 0, the objective function includes the power generation costs of thermal power, gas power, and nuclear power, as well as the environmental costs of thermal power and gas power, which can be expressed as:

[0028]

[0029] Among them, Min:F is the sum of the power generation costs of thermal power, gas power and nuclear power under the prediction scenario adopted by the objective function, and F g (n,t),F q (r,t),F e (d, t) are the power generation costs of thermal power unit n, gas power unit r, and nuclear power unit d in time period t. Their cost functions are quadratic functions with the actual output of the unit as the decision variable. Their variable coefficients are obtained through actual operation or experiments. en The environmental costs of thermal power and gas power.

[0030] S101-2, construct the constraint function of the SCUC model under the prediction scenario, including system power balance constraints, pumped storage unit constraints, unit spinning reserve constraints, output upper and lower limits and ramp constraints of conventional units, total power constraints of hydropower stations and gas power plants, storage capacity constraints and conversion efficiency constraints of pumped storage power stations, line flow constraints, etc., expressed as:

[0031]

[0032] P x (c,t)=P x,pm (c,t)+P x,gen (c,t) (3)

[0033]

[0034] P min (g)≤P(g,t)≤P max (g) (5)

[0035] -R d (g)×T 15 ≤P(g,t)-P(g,t-1)≤R u (g)×T 15 (6)

[0036]

[0037] W min (c)≤W0(c)-L(c)×ΔP x (c,t)≤W max (c) (9)

[0038]

[0039] Wherein, formula (2) is the system power balance constraint, P g (n,t),P v (m,t),P q (r,t),P e (d,t),P x (c,t),P w (j,t),P f (k, t) are the outputs of thermal power unit n, hydropower station m, gas power unit r, nuclear power unit d, pumped storage power station c, wind farm j, and photovoltaic power station k in time period t, respectively. L (t) is the load in time period t.

[0040] Formula (3) is the constraint of the pumped storage unit. The pumped storage unit includes two states: pumping and power generation. When in the pumping state, it is equivalent to a running virtual motor with an output power of P x,pm (c,t),P x,pm (c, t)<0; when in power generation state, it is equivalent to a running virtual generator with an output power of P x,gen (c,t),P x,gen (c,t)>0, obviously, at most one virtual machine can be running at any time.

[0041] Formula (4) is the unit spinning reserve constraint, L% is the load demand for spinning reserve, P max (g) is the output upper limit of conventional unit g, and P(g,t) is the output of conventional unit g in time period t.

[0042] Formula (5) is the upper and lower limit constraints of the output of conventional units, P min (g) is the lower limit of the output of conventional unit g.

[0043] Formula (6) is the ramp speed constraint of conventional units, R d (g), R u (g) are the downward and upward climbing speeds of the conventional unit g, respectively, and the scheduling period T 15 =15min, usually the pumped storage unit pumps water at full power, and the output is allowed to be adjusted only in the power generation state. At this time, the upper and lower limit constraints must be met. The hydropower unit and pumped storage unit have a fast adjustment speed, and the climbing speed constraint can be ignored.

[0044] Formula (7) and Formula (8) are the total power constraints of hydropower stations and gas power plants, respectively. Due to the limitation of water and gas volume, the allowable power generation of hydropower station m and gas power plant r in the dispatching period is W respectively. v (m), W q (r).

[0045] Formula (9) requires that the upper reservoir of the pumped storage power station should meet the storage capacity constraint in each period, W0(c), W min (c) W max (c) are the initial storage capacity and upper and lower limits of the storage capacity of the pumped storage power station c, ΔP x (c,t) is the cumulative power generation of the pumped storage power station c from the initial time to time t, ΔP x (c,t) multiplied by the storage capacity to electricity conversion coefficient L(c) represents the cumulative water consumption at time t.

[0046] Formula (10) represents ΔP x (c, t) is the equivalent cumulative power generation that combines the two processes of pumping and power generation. η is the conversion efficiency of the pumped power station, which is generally 80%.

[0047] Formula (11) indicates that the equivalent cumulative power generation of the pumped storage power station during the entire dispatching period is zero, which means that the upper reservoir is required to return to the initial water level after the dispatching is completed.

[0048] Formula (12) is the line power flow constraint, γ(g,l), γ w (j,l),γ f (k,l) are the power distribution factors of conventional unit g, wind farm j and photovoltaic power station k on line l, respectively, and L(l) is the line flow limit.

[0049] In S101-3, CPLEX and Gurobi solvers are used to efficiently solve the SCUC model with objective functions and constraints added in S101 and S102 to obtain the start-up and shutdown plans of each unit.

[0050] It should be noted that a multi-scenario new energy power prediction model includes generating at least 1,000 random scenarios of new energy power sequences. There are many ways to construct a multi-scenario new energy power prediction model, such as kernel density estimation combined with Markov chain or Bayesian optimization combined with quantile regression.

[0051] As a preferred embodiment, in this embodiment, a multi-scenario new energy power prediction model is constructed to use the kernel density estimation method to perform interval estimation on the new energy prediction error, and a Markov chain is used to perform time series modeling on the prediction error to generate a simulation sequence of the prediction error, and the prediction error simulation sequence is superimposed with the predicted output sequence to generate at least 1000 new energy power sequence random scenarios. The principle steps are as follows:

[0052] S102-1, first, given the number of prediction boxes, the prediction boxes are constructed through the Bayesian optimization method. Then, the probability distribution function at each moment is sampled by combining kernel density estimation and Markov chain to generate a prediction error simulation sequence that conforms to the state transfer matrix in a temporal manner. Finally, the prediction error simulation sequence is superimposed with the predicted output sequence to obtain the new energy power sequence scenario.

[0053] Among them, the prediction box means that the predicted output and actual output data at the same time in the historical data are first formed into data pairs, the prediction error corresponding to each data pair is calculated, and then the data pairs are arranged from small to large according to the value of the predicted output and divided into B groups. Each group is a prediction box. The prediction error data in a single prediction box is fitted, and the probability distribution of the prediction error when the predicted output is within a certain interval can be obtained. The prediction error calculation is expressed as:

[0054]

[0055] Among them, E i is the prediction error at time i, P Pi is the actual output at time i, P Mi A is the prediction contribution for time i. i is the startup capacity at time i.

[0056] S102-2, based on the historical new energy predicted output-actual output data, the number of prediction boxes is given, and the predicted output interval corresponding to each prediction box is determined through Bayesian optimization to construct the prediction box.

[0057] Bayesian optimization is a model-based sequential optimization method. Its main idea is to infer the area where the most likely optimal value of the function is located by establishing a posterior probability model of the objective function in each iteration, and then sample the area to obtain new function values ​​and update the posterior probability model. Based on the prediction box partitioning method of Bayesian optimization, the objective function is the average value of the determination coefficient of all prediction boxes. The hyperparameter that needs to be adjusted is the prediction output range covered by each prediction box, which is expressed as:

[0058] z * =argmaxg(z) (14)

[0059] Among them, z is the hyperparameter vector, g(z) is the objective function, and z * is the optimization result of the hyperparameters.

[0060] The main components of the Bayesian optimization framework include a probabilistic proxy model and a collection function. The probabilistic proxy model is an estimation model of the objective function. It predicts the value of the objective function at unknown points based on existing sampling points, and gradually approaches the true objective function through dynamic updates during the optimization process. Commonly used probabilistic proxy models include Gaussian processes and random forests. Since the random forest method has better computational efficiency, this embodiment selects random forests for model proxy.

[0061] The acquisition function determines which points are selected for sampling in each iteration to update the proxy model and continue to optimize the objective function. Commonly used methods include probability improvement, expected improvement, and confidence boundary strategy. This embodiment selects expected improvement as the acquisition function, that is, based on the existing proxy model, calculates the expected E[I(x)] of the objective function improvement I(x) at each possible sampling point x, and then selects the point with the highest E[I(x)] as the next sampling point.

[0062] S102-3, using kernel density estimation to fit the probability distribution of the prediction error in each prediction box one by one.

[0063] When fitting the prediction errors in the prediction box, since the probability distributions of the prediction errors corresponding to different prediction outputs are different, the same distribution function cannot be used to fit all the prediction errors. In this embodiment, the kernel density estimation method is used to fit the prediction errors in different prediction boxes respectively. Assuming that e1, e2, ..., e n is a random sample of prediction errors in a prediction box, let the probability density function of the random variable E be f(e), then the kernel density estimate of f(e) can be expressed as:

[0064]

[0065] Where n is the sample size, h is the bandwidth, is the kernel function, e k is the kth error sample.

[0066] In kernel density estimation, kernel function and bandwidth are hyperparameters that need to be selected. Different kernel functions have little effect on the accuracy of the model, but the selection of bandwidth will greatly affect the final fitting result. This embodiment selects the commonly used Gaussian kernel function as the probability density estimation model for the prediction error in the prediction box, and rewrites the kernel density estimation of f(e) as follows:

[0067]

[0068] Calculate the optimal bandwidth h best , expressed as:

[0069] h best =1.06σh-1 / 5 (17)

[0070] Among them, σ is the standard deviation of the sample variable.

[0071] S102-4, according to the new energy output prediction sequence, obtain the prediction box corresponding to each moment, and then obtain the probability distribution of the prediction error at each moment.

[0072] S102-5, define the discrete state of the prediction error, and calculate the state transfer matrix P of the prediction error and the cumulative state transfer matrix P according to the historical new energy prediction error data cum ;

[0073] If the maximum value in the historical data of the forecast error is E max , the number of states is M, then the definition falls within the range The state of the prediction error between is j (j = 1, 2, ..., M). The Markov chain assumes that the probability of state transition at a certain moment depends only on its previous state. If the prediction error sequence state is represented as X1, X2, ..., X i-1 ,X i ,..., then the state X at time i i The conditional probability of the state depends only on the state X at time i-1 i-1 , which is independent of the state before time i-1, can be expressed as:

[0074] P(X i |X1,X2,...,X i-1 )=P(X i |X i-1 ) (18)

[0075] Among them, P(·|·) is the conditional transition probability.

[0076] The Markov chain describes the changing law of prediction errors between different moments through the state transfer matrix and the cumulative state transfer matrix. The state transfer matrix P describes the probability of prediction error state transitions at adjacent moments. All prediction error state transition probabilities can be obtained through historical data statistics. The matrix P is expressed as:

[0077]

[0078] Among them, the element p ab Represents the probability of the prediction error transferring from state a to state b.

[0079] Cumulative state transfer matrix P cum The elements in are accumulated by the elements in the state transfer matrix P, expressed as:

[0080]

[0081] S102-6, at time i=1, the generation of the prediction error simulation sequence begins, and the probability distribution function of the prediction error at the initial time is sampled to obtain the value of the prediction error simulation sequence at the initial time.

[0082] S102-7, find the prediction error E at this moment i The corresponding state j is based on the cumulative state transfer matrix P of the prediction error cum , sample and obtain the state of the prediction error simulation sequence at the next moment.

[0083] S102-8, sample the probability distribution of the prediction error at the next moment, and determine whether the state corresponding to the sampling result is consistent with the state obtained in step S102-6. If it is consistent, the sampling result is used as the value of the prediction error simulation sequence at the next moment. If it is not consistent, re-sample.

[0084] S102-9, if i reaches the preset time period number T, the prediction error simulation sequence generation ends, otherwise go to step S102-7.

[0085] S102-10, superimposing the prediction error simulation sequence with the predicted output sequence to obtain a new energy power sequence scenario.

[0086] It should be noted that in another feasible embodiment, a multi-scenario new energy power prediction model is constructed by using Bayesian optimization combined with quantile regression method, and its principle steps are as follows:

[0087] First, the prediction error is divided into multiple intervals using the quantile regression method. The confidence level of each interval is set to 95% to ensure that the error value in each interval falls within the corresponding confidence interval. For the error distribution in each interval, the boundary of the prediction box is dynamically adjusted in each iteration through Bayesian optimization to reduce the mean square error of the error in the interval. In the specific optimization process, random forest is used as a proxy model for Bayesian optimization, which can effectively approximate the true characteristics of the error distribution. In order to maximize the accuracy of the optimization, the acquisition function uses the expected improvement, and each time selects the sampling point that can maximize the improvement of the mean square value of the error. The accuracy of the optimized prediction box on the error The error and distribution characteristics are adjusted so that the error is evenly distributed in each interval. When generating the error sequence, the optimized prediction boxes are sequentially sampled to ensure that the generated error sequence not only conforms to the continuity characteristics of the time series, but also accurately reflects the interval distribution characteristics of the power forecast. After the error sequence is generated, at least 1,000 random scenarios of new energy power are formed by superimposing the error sequence with the predicted output data of new energy. These scenarios respond to power fluctuations more smoothly and accurately at a certain confidence level, which effectively meets the input requirements of the robust scheduling model and ensures the stability and adaptability under new energy power fluctuations.

[0088] S200: Constructing a robust economic dispatch model by embedding random scenario constraints.

[0089] Furthermore, constructing a robust economic dispatch model includes fixing the start and shutdown plans of each unit before constructing the robust economic dispatch model, fixing the start and shutdown plans of the units obtained in step S101, and constructing a robust economic dispatch (RSCED) model embedded with random scenario constraints for each new energy power sequence scenario generated in step S102, and the number of models is the same as the number of random scenarios of the new energy power sequence.

[0090] It should be noted that constructing a robust economic dispatch model specifically includes the following steps:

[0091] S201, in the SCUC model generated in step S101, the start and stop plans of the fixed units are formed into an economic dispatch (SCED) model.

[0092] S202, adding constraint equations under the current random scenario to the SCED model. The constraints under the random scenario are similar to those under the prediction scenario. The difference is that in the constraint equations of equations (2) to (12), the outputs of all types of units are all the outputs under the random scenario. Taking the system power balance constraint as an example, the random scenario s i The lower power balance constraint is expressed as:

[0093]

[0094] Among them, P g (s i ,n,t) is a random scene s i The output of thermal power unit n in the next time period t, other physical quantities are similar, and the remaining constraint equations under the current random scenario can be obtained in the same way.

[0095] S203, adding the scene transition constraint equation to the SCED model to form the RSCED model, which is expressed as:

[0096]

[0097]

[0098] Among them, R d (g) is the power reduction rate of conventional unit g, R u (g) is the power increase rate of conventional unit g, T C is the scene transition time, and formula (22) requires the system to adjust the time T C The transition from the operating point P(g,t) in the prediction scenario to the operating point P(s i,g,t), formula (23) requires the system to adjust the time T C The transition between any two running points in random scenarios can be completed.

[0099] S204, performing steps S201-S203 for each new energy random scenario until the same number of RSCED models as the new energy power sequence random scenario is formed.

[0100] S300: A feasibility check of the power generation plan of the robust economic dispatch model is performed through a parallel solution method. Based on the feasibility check result, the robust economic dispatch model is optimized and the clearing results of each unit are output.

[0101] Furthermore, a parallel solution method is used to verify the feasibility of the power generation plan for each robust economic dispatch model. If the robust economic dispatch model has a solution, the feasibility check is passed under the random scenario of the new energy power sequence, and the economic dispatch model optimization calculation is performed on the prediction scenario model to output the clearing results of each unit; if the robust economic dispatch model has no solution, an infeasible random scenario of the new energy power sequence is output to form an infeasible scenario set, and the scenarios with the most serious cross-section flow over-limit and the largest amount of wind and solar power abandonment are selected to construct a robust dispatch optimization model.

[0102] A parallel solution method may be a variety of methods, such as RSCED model solution based on CPLEX or Gurobi or a parallel verification framework based on distributed computing.

[0103] As a preferred embodiment, in this embodiment, a parallel solution method adopts CPLEX and Gurobi solvers to solve the RSCED model formed in step S200 in parallel to perform feasibility verification of the power generation plan, and its principle steps are as follows:

[0104] S301-1, for each RSCED model, solve and obtain the output of each unit.

[0105] S301-2, if the RSCED model has a solution and the output of each unit is reasonable, it is considered that the feasibility check of the new energy power sequence under the random scenario has passed, and the passed scenario is discarded to release computer memory and speed up the solution.

[0106] S301-3, if the RSCED model has no solution or the output of each unit is unreasonable, the infeasible new energy power sequence random scenario is output to the infeasible scenario set.

[0107] S301 - 4 , repeating step S301 - 2 and step S301 - 3 until all RSCED models are solved and a total set of infeasible scenarios is obtained.

[0108] In another feasible embodiment, a parallel solution method adopts a parallel verification framework based on distributed computing, and its principle steps are as follows:

[0109] In the parallel verification framework based on distributed computing, each RSCED model task is first assigned to different computing nodes in a distributed computing environment, so that each node independently processes a RSCED model solution task, thereby making full use of the parallel computing capabilities of multiple nodes and speeding up the overall solution speed.

[0110] During the solution process, once a node completes its task and its output results pass the feasibility check, the resources of the node are immediately released and reallocated to other unfinished scenario tasks to avoid idle resources and achieve dynamic balancing of task loads between nodes. For example, when the solution of a scenario takes a short time, the node will be automatically allocated to a more complex scenario to maintain high utilization of computing resources. Finally, the solution results of all computing nodes are summarized, and scenarios with no solutions or unreasonable output are recorded in the infeasible scenario set to form systematic feedback. The distributed parallel computing framework not only improves the overall solution efficiency through dynamic scheduling of nodes, but also uses the system's horizontal scalability to meet the efficient parallel verification requirements of multi-scenario power generation plans in ultra-large-scale computing tasks, thereby improving the practicality and reliability of the model in complex scenarios.

[0111] It should be noted that from the infeasible scenario set formed in step S301-4, the scenario with the most severe cross-section flow over-limit and the largest amount of wind and solar power abandonment is selected and recorded as s non , add the random scenario constraint equation and random scenario transition constraint in this scenario to the SCUC model of the prediction scenario to form a robust scheduling optimization (RSCUC) model, which specifically includes the following steps:

[0112] S302-1, in the SCUC model generated in step S101, add scene s non The constraint equations, processes and formulas below are consistent with those in step S202.

[0113] S302-2, adding scene s based on step S302-1 non The scene transition constraint equation, process and formula are consistent with step S203 to form the RSCUC model.

[0114] It should also be noted that according to the feasibility check results, if the feasibility check fails, infeasible scenarios are screened out, and the most constrained scenarios are further selected to construct a robust scheduling optimization model. Based on the robust scheduling optimization model, new start-up and shutdown plans for each unit are obtained and fixed, and the robust economic scheduling model is reconstructed to perform a feasibility check on the power generation plan. When the number of infeasible scenario sets is equal to 0, the feasibility check passes, and the economic scheduling model is optimized and calculated for the new start-up and shutdown plans for each unit, and the clearing results of each unit that meet the uncertainty of intraday new energy are output; when the number of infeasible scenario sets is greater than 0, the robust scheduling optimization model and the robust economic scheduling model are repeatedly constructed to perform a feasibility check on the power generation plan until the number of infeasible scenario sets is equal to 0, and the feasibility check of the power generation plan is completed.

[0115] Re-check the feasibility of the robust economic dispatch model, which includes the following steps:

[0116] S303: Use a parallel solution method to efficiently solve the updated RSCUC model to determine new start-up and shutdown plans for each unit.

[0117] S304, fix the new unit start-up and shutdown plan generated in step S303, such as Figure 2 As shown, execute step S200, step S301-step S303 to perform feasibility verification of the power generation plan under the random scenario of renewable energy, until all random scenarios of renewable energy power sequences can pass the feasibility verification under the latest start-up and shutdown plans of each unit, that is, the number of infeasible scenario sets under the latest start-up and shutdown plans of each unit is 0.

[0118] S305, fixing the latest start-up and shutdown plans of each unit formed in step S304, performing SCED optimization calculations, and obtaining unit output results that meet the uncertainty of intra-day new energy, wherein the SCED model is generated according to step S201.

[0119] Example 2 is an embodiment of the present invention, which provides an efficient solution method for robust clearing of the electricity spot market taking into account uncertainty. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0120] This embodiment aims to verify the feasibility and optimization effect of the robust economic dispatch model of the present invention in dealing with renewable energy power fluctuations. The main steps of the experiment are to construct a safety constrained unit combination model (SCUC) and a multi-scenario renewable energy power prediction model, to construct a robust economic dispatch model (RSCED), and to perform feasibility verification and optimization calculation of the power generation plan through a parallel solution method. The experimental data mainly includes the number and related parameter settings of thermal power units, hydropower stations, gas-fired power units, nuclear power units, pumped storage power stations, wind farms and photovoltaic power stations to simulate the operation of the power grid system under the scenario of renewable energy power fluctuations.

[0121] In the first stage of the experiment, by predicting the boundary information of renewable energy output, the safety constrained unit commitment model (SCUC) was constructed using CPLEX and Gurobi solvers respectively. The model included 15 thermal power units, 5 hydropower stations, 8 gas-fired power units, 3 nuclear power units, 4 pumped storage power stations, 10 wind farms and 12 photovoltaic power stations in the simulation environment. The experiment assumed that the scheduling cycle was 24 hours, with one scheduling period per hour. Based on this model, a multi-scenario renewable energy power prediction model was constructed, and more than 1,000 random scenarios of renewable energy power were generated by combining the kernel density estimation method with the Markov chain.

[0122] In the second stage of the experiment, the start-up and shutdown plan of the SCUC model was fixed, and a robust economic dispatch model (RSCED) embedded with random scenario constraints was further constructed. Under each random scenario, a corresponding RSCED model was generated based on the fixed start-up and shutdown plan to ensure that the model under different renewable energy output scenarios can adapt to the corresponding power fluctuations. Then, each RSCED model was solved in parallel, and the actual output of each unit was calculated by the CPLEX solver. If the model has a solution and the output is reasonable, the random scenario is considered to have passed the feasibility check, otherwise the scenario is added to the infeasible scenario set. For the scenarios in the infeasible scenario set, the experiment selected the 10 scenarios with the most serious cross-section flow over-limit and the 10 scenarios with the largest amount of wind and solar power abandonment, and rebuilt the robust scheduling optimization model, updated the start-up and shutdown plans of the units, until all random scenarios passed the feasibility check, and finally output the clearing results of each unit that met the uncertainty of intraday renewable energy.

[0123] Refer to Table 1 for a comparative analysis of the experimental data.

[0124] Table 1 Experimental data record table

[0125]

[0126]

[0127] By comparing the output and environmental cost data of each unit in Table 1, this experiment found that the robust economic dispatch model adopted by the present invention can effectively adjust the start-up and shutdown plans of each unit under the situation of large fluctuations in renewable energy power to ensure the stability and economy of power grid operation. Especially when dealing with high-frequency fluctuations in renewable energy, the experimental model realizes dynamic adjustment of power generation plans by embedding random scenario constraints. The data in Table 1 show that the environmental costs of thermal power units and nuclear power units are relatively high, but by allocating more output from wind power and photovoltaic power stations, the overall power generation cost is reduced and the economy is significantly improved. In addition, the pumped storage power station adjusts the storage capacity according to the power demand in each time period, providing flexible backup support for the power grid, thereby meeting the demand for rotating reserve.

[0128] Compared with the existing dispatching methods, the present invention is technically innovative. Traditional dispatching methods are usually difficult to adapt to complex and multi-scenario power fluctuations. The present invention incorporates the uncertainty of new energy into the dispatching system by constructing a robust economic dispatching model and a multi-scenario prediction model, thereby achieving high adaptability and robustness of the power generation plan. Experimental data show that the present invention effectively reduces the environmental cost and dispatching risk of power generation, and meets the dual needs of new energy consumption and output regulation. Therefore, compared with traditional dispatching methods, the present invention not only improves the model solution efficiency and the economy of the power grid, but also improves the dispatching reliability in dealing with complex and multi-scenario power fluctuations, providing a new technical solution for the safety and sustainable development of the power system.

[0129] Example 3, reference Figure 3 , which is an embodiment of the present invention, provides an efficient solution system for robust clearing of the power spot market taking into account uncertainty, including a multi-scenario construction module, a robust scheduling construction module, and a parallel solution and verification module.

[0130] The multi-scenario construction module is used to construct a safety-constrained unit combination model and a multi-scenario new energy power prediction model; the robust scheduling construction module is used to embed random scenario constraints and construct a robust economic scheduling model. By fixing the unit start-up and shutdown plan, a corresponding scheduling model is generated for each new energy power sequence; the parallel solution and verification module is used to solve the robust economic scheduling model in parallel to realize the feasibility verification of the power generation plan. According to the verification results, the scheduling model is optimized to form an infeasible scenario set, and the most constrained scenarios are further selected to update the scheduling plan until all scenarios pass the feasibility verification, and finally the output of each unit clearing result that meets the intraday new energy uncertainty is output.

[0131] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0132] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0133] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0134] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0135] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An efficient solution method for robust clearing of electricity spot market considering uncertainty, characterized by: include: Construct a safety constraint unit combination model and a multi-scenario new energy power prediction model; By embedding random scenario constraints, a robust economic dispatch model is constructed; Through a parallel solution method, the feasibility of the power generation plan of the robust economic dispatch model is verified. According to the feasibility verification results, the robust economic dispatch model is optimized and the clearing results of each unit are output.

2. The method for solving the robust clearing of the power spot market considering uncertainty according to claim 1 is characterized by: The construction of the safety constraint unit combination model includes constructing the safety constraint unit combination model under the prediction scenario based on the predicted boundary information, and obtaining the start-up and shutdown plan of each unit.

3. The method for solving the robust clearing of the power spot market considering uncertainty according to claim 2, characterized in that: The multi-scenario new energy power prediction model includes generating at least 1,000 random scenarios of new energy power sequences.

4. The method for solving the robust clearing of the power spot market considering uncertainty according to claim 3 is characterized by: The constructing of the robust economic dispatch model includes fixing the start and stop plans of each unit before constructing the robust economic dispatch model.

5. The method for solving the robust clearing of the power spot market considering uncertainty according to claim 4 is characterized by: The embedded random scenario constraints include that the number of constructed robust economic dispatch models is the same as the number of random scenarios of new energy power sequences.

6. The method for solving the robust clearing of the power spot market considering uncertainty according to claim 5, characterized in that: The parallel solution method includes performing a feasibility check on the power generation plan for each robust economic dispatch model. If the robust economic dispatch model has a solution, the feasibility check is passed under the random scenario of the new energy power sequence, and the economic dispatch model optimization calculation is performed on the prediction scenario model to output the clearing results of each unit. If the robust economic dispatch model has no solution, then an infeasible random scenario of the new energy power sequence is output to form an infeasible scenario set. The scenario with the most serious cross-section power flow exceeding the limit and the largest amount of wind and solar power abandonment is selected to construct a robust dispatch optimization model.

7. The method for solving the robust clearing of the power spot market considering uncertainty according to claim 6, characterized in that: The feasibility check result includes screening out infeasible scenarios if the feasibility check fails, and further selecting the most constrained scenarios, constructing a robust scheduling optimization model, obtaining new start-up and shutdown plans for each unit based on the robust scheduling optimization model and fixing them, reconstructing the robust economic scheduling model and performing a feasibility check on the power generation plan. When the number of infeasible scenario sets is equal to 0, the feasibility check passes, and the new start-up and shutdown plans for each unit are optimized and calculated by the economic scheduling model, and the clearing results of each unit that meet the uncertainty of intraday new energy are output; When the number of infeasible scenario sets is greater than 0, the robust scheduling optimization model and the robust economic scheduling model are repeatedly constructed to perform feasibility verification of the power generation plan until the number of infeasible scenario sets is equal to 0, completing the feasibility verification of the power generation plan.

8. A system using the method for solving the power spot market robust clearing efficiently considering uncertainty as claimed in any one of claims 1 to 7, characterized in that: It includes multi-scenario construction module, robust scheduling construction module, and parallel solution and verification module; The multi-scenario construction module is used to construct a safety constraint unit combination model and a multi-scenario new energy power prediction model; The robust scheduling construction module is used to embed random scenario constraints, construct a robust economic scheduling model, and generate a corresponding scheduling model for each new energy power sequence by fixing the start and stop plan of the unit; The parallel solution and verification module is used to solve the robust economic dispatch model in parallel to realize the feasibility verification of the power generation plan. According to the verification results, the dispatch model is optimized to form a set of infeasible scenarios, and the most constrained scenarios are further screened out to update the dispatch plan until all scenarios pass the feasibility verification, and finally the clearing results of each unit that meet the uncertainty of intraday new energy are output.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the efficient solution method for robust clearing of the electricity spot market considering uncertainty described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for efficiently solving the robust clearing of the electricity spot market considering uncertainty described in any one of claims 1 to 7 are implemented.

Images