Day-ahead scheduling plan making method based on quotient gradient system assistance and considering frequency security constraint

Abstract

The invention relates to a quotient gradient system assisted day-ahead scheduling plan making method considering frequency safety. The method comprises the following steps: collecting and processing power grid data; constructing an objective function and a constraint set of a day-ahead economic dispatching optimization model according to the processed power grid data; and calculating the constructed previous economic dispatching optimization model based on a quotient gradient system auxiliary interior point method to obtain a day-ahead dispatching plan. According to the invention, the scheduling result generated by model optimization solution can consider both economic efficiency and safety margin, and the applicability of actual power grid production planning is effectively improved. Besides, by adopting a mode of combining quotient gradient system integration and interior point method accurate solution, reliable convergence of the algorithm is ensured on the premise that the problem is feasible, and an accurate diagnosis result is provided for an operator for reference when the problem is not feasible.

Description

Technical Field

[0001] This invention belongs to the field of power system dispatching technology, and in particular to a method for formulating day-ahead dispatching plans based on frequency security constraints and assisted by a quotient gradient system. Background Technology

[0002] In formulating day-ahead unit allocation and economic dispatch plans, the power industry widely adopts linear power flow equations as approximations of nonlinear power flow equations to ensure reliable convergence of the algorithms. Among these, the Direct Current Optimal Power Flow (DCOPF) method is the most widely used. This method replaces the original nonlinear power flow equations with linear mainstream power flow equations to achieve a linear approximation of the optimal power flow model. However, DC power flow equations assume a linear relationship between active power and phase angle, resulting in a certain loss of accuracy. Calculations indicate that DC power flow approximations cause the power industry losses of hundreds of billions of yuan annually. Furthermore, because reactive power and voltage information are not considered, the resulting unit dispatch results cannot reflect active power transmission bottlenecks caused by reactive power and voltage constraints in the power grid, and cannot guarantee the safety of their application in actual production operations. Therefore, the power grid urgently needs effective optimal power flow modeling and solution methods.

[0003] In day-ahead scheduling models based on AC optimal power flow, the nonlinearity of the power flow equations leads to a nonconvex optimization problem. Solving nonconvex optimization problems presents the following bottlenecks: 1) The algorithm cannot guarantee reliable convergence; 2) Even if the algorithm converges, it cannot guarantee the global optimality of the solution. In fact, solving the optimal power flow problem is extremely difficult. When using numerical methods such as the interior point method to solve the optimal power flow problem, if the algorithm diverges, it is difficult to determine whether it is because the optimal solution cannot be found due to numerical reasons or because the original problem itself does not have an optimal solution.

[0004] In terms of day-ahead scheduling with security constraints, a large number of studies have been conducted on operational optimization problems and some results have been achieved. However, security constraints and network transmission capacity are not adequately considered, and frequency constraints are rarely taken into account. In particular, with the large-scale integration of renewable energy, the impact of fluctuations on system frequency security from both the source and load sides is becoming increasingly prominent. Therefore, it is more necessary to consider frequency constraints when formulating day-ahead scheduling plans. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and propose a day-ahead scheduling plan formulation method based on the quotient gradient system and taking into account frequency security constraints. This method covers power balance constraints, frequency security constraints, voltage security constraints, transmission limit constraints, unit output constraints, unit ramp-up constraints, spinning reserve constraints, energy storage operation constraints, and new energy output constraints. It can reliably solve scheduling plans that are both economically efficient and have a safety margin when the problem is feasible.

[0006] The technical problem solved by this invention is achieved through the following technical solution: A method for formulating day-ahead scheduling plans based on frequency security constraints and assisted by a quotient gradient system includes the following steps: Step 1: Collect and process power grid data.

[0007] It is necessary to obtain network topology and line parameters, 24 / 96 time period load and renewable energy output forecast data, and system operating conditions from the power grid data platform, and use methods such as deletion, mean replacement and filling to handle missing and outlier values ​​in the data.

[0008] Step 2: Construct the objective function and constraint set of the day-ahead economic dispatch optimization model based on the processed power grid data.

[0009] Operating costs include fuel costs for conventional units and regulation costs for units with automatic generation control (AGC) functions. (1) in, For time index set, and These are the index sets for conventional units (including AGC units and non-AGC units) and AGC units, respectively. For the collection of new energy units, For conventional units i At any moment t Those who have made contributions and AGC units j At any moment t The power setpoint and power allocation factor, and For new energy units k At any moment t The power forecast and the actual dispatch value, For the cost function of conventional generating units, The adjustment cost function for the AGC unit. The cost per unit of renewable energy curtailment.

[0010] The day-ahead AC optimal power flow model used in this invention considers the following power balance constraints under a single regulation. (2) in, For the system node index set, For nodes i At any moment t voltage amplitude, For nodes i and j At any moment t The phase angle difference, and branch road i - j Conductivity and susceptance, superscripts of each variable p This represents the value under one adjustment.

[0011] Adjust the lower node once i At any moment t The net active and reactive power are (3) in , , and They are nodes i A set of indexes for connecting conventional generating units, loads, renewable energy generating units, and energy storage. For the unit g The active power output setpoint, For the unit g The reactive power output under one adjustment To adjust the entire network frequency in one go, The system's rated frequency, and Energy storage units e The charging and discharging power, and They are conventional units g and load d The unit adjustment power under one adjustment.

[0012] Under one adjustment, the unit output changes with the frequency under the gearbox adjustment, and the load power will also change due to its static frequency characteristics.

[0013] Under secondary adjustment (4) Among them, the superscripts of each variable s This represents the value under secondary adjustment.

[0014] Secondary adjustment of the lower node i At any moment tThe net active and reactive power are (5) Under secondary regulation, the system frequency should recover to or be very close to the rated power. The system's unbalanced power... It is not entirely borne by the balancing unit, but rather according to the allocation coefficient. Assigned to each AGC unit, for non-AGC units .

[0015] The day-ahead scheduling model used in this invention needs to satisfy the following frequency constraint: under a single adjustment, the system frequency must be guaranteed not to exceed the allowable deviation range. (6) This invention considers the following voltage safety constraints (7) in, and They are nodes i The lower and upper limits of the voltage amplitude.

[0016] The system circuit set is defined as follows: For any branch l ( ), whose beginning and end nodes are marked as i and j Under one adjustment, the transmission limit constraint that this branch must satisfy is: (8) in, and Each is a single adjustment of the lower meridian line l From node i Transport to node j Active power and reactive power, and Each is a single adjustment of the lower meridian line l From node j Reverse transport to node i Active power and reactive power, For the line l The power transmission limit.

[0017] Specifically, by line l Taking the forward transmission power as an example, the specific expression is as follows: (9) Similarly, under secondary adjustment, the transmission limit constraint satisfied by this branch is: (10) in, , , and The circuit under secondary adjustment l Active and reactive power transmitted in two power directions.

[0018] Regarding the unit g ( Regarding operational constraints, this invention considers active and reactive power output constraints and ramp-up constraints. Under primary and secondary regulation, the output constraints are respectively expressed as follows: (11) (12) At adjacent time points, the ramp-up constraints of this unit are as follows: (13) in, and For the unit g The lower and upper limits of ramp power.

[0019] To ensure power balance and frequency stability in the power grid under conditions of power deficit or surplus caused by source-load uncertainty, this invention considers the following spinning reserve constraints. (14) in, and For the unit g At any moment t The up / down rotation provided below is available for use. and These represent the minimum required up / down spinning reserve of the power grid, typically set at 2% to 5% of the total load.

[0020] This invention considers the following energy storage operation constraints (15) in For energy storage units e exist t The electrical energy value at that moment, and Energy storage units e The charge / discharge efficiency, The time interval for each period, and Energy storage units e Minimum / maximum charge, and Energy storage units e Maximum / minimum charge / discharge power.

[0021] Furthermore, to ensure reliable dispatch of energy storage in subsequent periods, it is necessary to ensure that the state of charge of each energy storage unit is consistent at the beginning and end of the period. (16) This invention considers the following renewable energy output constraints to ensure that a certain degree of wind / solar curtailment is allowed when the system cannot fully absorb renewable energy. (17) Step 3: Calculate the pre-economic scheduling optimization model constructed in Step 2 based on the quotient gradient system-assisted interior point method strategy to generate the day-ahead scheduling plan.

[0022] Construct a constraint set for the above problems (18) Each sub-constraint set involves a power balance constraint set. (19) For conventional units under secondary regulation i At any moment t Those who have made contributions For secondary adjustment of the lower node i At any moment t voltage amplitude, For secondary adjustment of the lower node j At any moment t voltage amplitude, Conventional units under secondary regulation i At any moment t Unproductive efforts; Frequency security constraint set (20) in, , The upper and lower limits of the system frequency are, in order. Voltage safety constraint set (twenty one) Line thermal constraint set (twenty two) Unit output constraint set (twenty three) in, and These represent the active and reactive power output dispatch values ​​of unit g at time t under a single adjustment. and These represent the active and reactive power output dispatch values ​​of unit g at time t under the secondary adjustment. He is a conventional unitg Unit adjustment power under one adjustment This refers to the adjusted network frequency. The system's rated frequency, and They are conventional units g The upper and lower limits of effort and contribution. and They are conventional units g The upper and lower limits of no-efficiency output. For AGC units g The power allocation factor, For the entire network t Imbalanced power caused by forecasting errors and real-time fluctuations on both the supply and demand sides; Unit ramping constraint set (twenty four) in, For the unit g exist t The active power output during the -1 time period; Rotational spare constraint set (25) Energy storage operation constraint set (26) in, and Energy storage units e exist t The charging and discharging power at any given time, and Energy storage units at the initial and final moments, respectively. e electrical energy value New energy output constraint set (27) in, , For new energy units r exist t The output scheduling values ​​and predicted values ​​at each moment, where the original problem decision variables are... , s These are auxiliary variables used to transform inequality constraints into equality constraints.

[0023] Using constraint set vectors and its Jacobi matrix The following quotient gradient system can be constructed. (28) in, Decision variablesx The rate of change with respect to time. Integrating this dynamical system using ode15s, it eventually converges to a conventional stable equilibrium point, which has a one-to-one correspondence with the feasible solution. This invention uses this feasible solution as the initial solution and employs IPOPT to solve the problem precisely.

[0024] The advantages and positive effects of this invention are: This invention collects and processes power grid data; constructs an objective function and its constraints for a pre-economic dispatch optimization model based on the processed power grid data; and calculates the pre-economic dispatch optimization model using the quotient gradient system-assisted interior-point method to obtain the day-ahead dispatch plan. This invention ensures that the dispatch results optimized by the model can balance economic efficiency and safety margin, effectively improving the applicability for actual power grid production planning. Furthermore, by combining quotient gradient system integration and exact solution using the interior-point method, it guarantees reliable convergence of the algorithm when the problem is feasible, and provides accurate diagnostic results for operators when the problem is infeasible. Attached Figure Description

[0025] Figure 1 This is an overall flowchart of a day-ahead scheduling plan formulation method based on a quotient gradient system assisted by the present invention.

[0026] Figure 2 This is a 96-hour day-ahead dispatch plan curve for Unit 10, formulated by a provincial power grid in a certain region of southern China on a certain day in September 2025, according to Embodiment 2 of the present invention.

[0027] Figure 3 This is a graph showing the charging and discharging plan curve of the No. 3 energy storage unit, which was formulated by a provincial power grid in a certain region of southern China on a certain day in September 2025, according to Embodiment 2 of the present invention.

[0028] Figure 4 This is a graph showing the frequency change of the entire provincial power grid over 96 hours after the frequency adjustment following the formulation of the day-ahead dispatch plan on a certain day in September 2025, according to Embodiment 2 of the present invention. Detailed Implementation

[0029] The present invention will be further described in detail below with reference to the accompanying drawings.

[0030] A method for day-ahead scheduling planning that takes into account frequency security constraints and is assisted by a quotient gradient system, such as... Figure 1 As shown, it includes the following steps: S1: Preprocess the predicted data from both the system network structure and the source / load sides to establish a day-ahead coordinated scheduling optimization model considering optimal AC power flow and various security constraints. Specifically, obtain network topology and line parameters, and collect day-ahead load forecast data and renewable energy forecast data for the power system over 24 / 96 time periods. Use methods such as deletion, mean replacement, and filling to handle missing and outlier values ​​in the data. Furthermore, the established day-ahead scheduling optimization model includes constructing an objective function based on minimizing the operating costs of conventional generating units and the cost of renewable energy curtailment penalties. f , represented as (29) The constraints include power balance constraints under primary / secondary regulation. (30) (31) The net power injection amount for each node in the primary / secondary regulation is as follows: (32) (33) Frequency security constraints are expressed as follows: (34) Voltage safety constraints are expressed as follows: (35) The transmission limit constraint of the line is expressed as: (36) The unit output constraints under primary and secondary regulation are expressed as follows: (37) (38) The unit ramp-up constraint is expressed as: (39) The spin-off reserve constraint is expressed as: (40) Energy storage operation constraints are expressed as follows: (41) (42) The power output constraint of new energy sources is expressed as follows: (43) S2: Construct a constraint set based on known problem parameters This constraint set involves the power balance constraint set. (44) Frequency security constraint set (45) Voltage safety constraint set (46) Transmission Limit Constraint Set (47) Unit output constraint set (48) Unit ramping constraint set (49) Rotational spare constraint set (50) Energy storage operation constraint set (51) New energy output constraint set (52) Constructing constraint set vectors for decision variables x Jacobi matrix And construct the following quotient gradient system: (53) S3: Starting from any initial point, integrate the constructed quotient gradient system. If the trajectory converges to a conventional stable equilibrium point, proceed to S4; otherwise, proceed to S6. It should be noted that the conventional stable equilibrium point... There is a one-to-one correspondence between these points and the feasible points of the original problem, thus satisfying the feasibility conditions of the original problem. . S4: Starting from the conventional stable equilibrium point / feasible point obtained by integration, apply the interior point method IPOPT to solve the original problem accurately, and formulate a day-ahead scheduling scheme based on the solution results.

[0031] S5: Users verify the feasibility and superiority of the developed plan based on actual operating experience. Only after verification can it be released for operators' reference.

[0032] S6: Based on the obtained degenerate stable equilibrium point Evaluate its constraint set The more restrictive the constraint, the more relaxed the upper and lower bound constraints of the variables at the boundary are, and the integration is repeated. The relaxation amount is 5% of the average of the upper and lower bounds of this variable. For example, the node voltage magnitude constraint is [0.95, 1.05], and If the value of this variable is 0.95, then the constraint is relaxed to the range of [0.9, 1.1]. Record the degree to which the maximum constraint exceeds the limit after relaxing the constraint for each critical variable to form a sensitivity vector.

[0033] S7: Sort the sensitivity vectors to identify the key factors that restrict the feasibility of the original problem, and submit them to the user as diagnostic results along with infeasibility warning information.

[0034] Example 2, an embodiment of the present invention, is used in conjunction with the day-ahead dispatching plan of a provincial power grid on a certain day in September 2025 to verify the beneficial effects of the present invention. The results are demonstrated in detail through economic benefit calculations and simulation experiments.

[0035] First, system network information, along with predicted curves for each load and renewable energy power, are obtained from the system data platform. Missing and outlier values ​​in the data are handled using methods such as deletion, mean replacement, and filling. Further, a day-ahead dispatch optimization model is established, aiming to minimize the operating costs of conventional generating units and the penalty costs of renewable energy curtailment. This model encompasses power balance constraints, frequency security constraints, voltage security constraints, transmission limit constraints, unit output constraints, unit ramp-up constraints, spinning reserve constraints, energy storage operation constraints, and renewable energy output constraints.

[0036] After the model is established, a quotient gradient system is constructed based on the constraint set and integrated to a conventional stable equilibrium / feasible point, corresponding to a set of feasible day-ahead scheduling schemes. Starting from this feasible point, the interior-point method IPOPT is used to solve the optimal solution of this problem in detail, outputting a set of day-ahead scheduling schemes that balance operational economic efficiency and safety margins. After verification, the schemes are published to the scheduling platform for operators' reference. Note that directly solving this case using IPOPT may result in algorithm non-convergence. For details of some unit scheduling plans, please refer to the appendix. Figure 2 It can be pointed out that the proposed solution can significantly improve the absorption of new energy sources such as wind power and photovoltaics, reduce the overall operating cost of the system, and effectively ensure the frequency security of the system.

[0037] It should be emphasized that the embodiments described in this invention are illustrative rather than limiting. Therefore, this invention includes, but is not limited to, the embodiments described in the specific implementation. Any other implementations derived by those skilled in the art based on the technical solutions of this invention are also within the scope of protection of this invention.

Claims

1. A method for formulating day-ahead scheduling plans based on frequency security constraints and assisted by a quotient gradient system, characterized in that: Includes the following steps: Step 1: Collect and process power grid data; Step 2: Construct the objective function and constraint set of the day-ahead economic dispatch optimization model based on the processed power grid data; Step 3: Calculate the day-ahead economic dispatch optimization model constructed in Step 2 based on the quotient gradient system-assisted interior point method strategy to generate the day-ahead dispatch plan, including the active and reactive power dispatch plans of various units within the time period, the energy storage charging and discharging plan, and the new energy output plan.

2. The day-ahead scheduling plan formulation method based on quotient gradient system assistance and considering frequency security constraints as described in claim 1, characterized in that: In step 1, the power grid data includes network topology and line parameters, time-period load and new energy output forecast data, and system operating conditions. Missing and outlier values ​​in the data are processed by deletion, mean replacement, and filling.

3. The day-ahead scheduling plan formulation method based on quotient gradient system assistance and considering frequency security constraints according to claim 1, characterized in that: In step 2, the objective function is to minimize the operating cost of conventional generating units and the cost of renewable energy curtailment penalties. The operating cost includes the fuel cost of conventional generating units and the regulation cost of AGC (Automatic Generation Control) units. ； in, For time index set, and These are index sets for conventional units and AGC units, respectively. For the collection of new energy units, For conventional units g At any moment t Those who have made contributions and AGC units j At any moment t The power setpoint and power allocation factor, and For new energy units k At any moment t The power forecast and the actual dispatch value, For the cost function of conventional generating units, The adjustment cost function for the AGC unit. The cost per unit of renewable energy curtailment.

4. The day-ahead scheduling plan formulation method based on quotient gradient system assistance and considering frequency security constraints according to claim 1, characterized in that: The constraints of the optimization model in step 2 include: power balance constraints, frequency constraints, voltage safety constraints, line thermal constraints, unit output constraints, ramping constraints, spinning reserve constraints, energy storage operation constraints, and new energy output constraints.

5. The day-ahead scheduling plan formulation method based on quotient gradient system assistance and considering frequency security constraints according to claim 1, characterized in that: The specific implementation method of step 3 is as follows: construct a constraint set based on the constraints in step 2: ； Each subset of constraints includes a power balance constraint set: ； in, For a single adjustment of a conventional unit i At any moment t Those who have made contributions For adjusting the lower node i At any moment t voltage amplitude, For adjusting the lower node j At any moment t voltage amplitude, and branch road i - j The conductivity and susceptance, For nodes i and j At any moment t The phase angle difference, For a single adjustment of a conventional unit i At any moment t Unproductive efforts For conventional units under secondary regulation i At any moment t Those who have made contributions For secondary adjustment of the lower node i At any moment t voltage amplitude, For secondary adjustment of the lower node j At any moment t voltage amplitude, Conventional units under secondary regulation i At any moment t Unproductive efforts; Frequency security constraint set: ； in, To adjust the entire network frequency in one go, The system's rated frequency, , These are the upper and lower limits of the system frequency, respectively. Voltage safety constraint set: ； in, and They are nodes i The lower and upper limits of the voltage amplitude; Line thermal constraint set: ； in, For the system's circuit set, and Each is a single adjustment of the lower meridian line l From node i Transport to node j Active power and reactive power, and Each is a single adjustment of the lower meridian line l From node j Reverse transport to node i Active power and reactive power, For the line l The power transmission capacity; Unit output constraint set: ； in, and These represent the active and reactive power output dispatch values ​​of unit g at time t under a single adjustment. and These represent the active and reactive power output dispatch values ​​of unit g at time t under the secondary adjustment. He is a conventional unit g The unit adjustment power under one adjustment, This refers to the adjusted network frequency. The system's rated frequency, and They are conventional units g The upper and lower limits of effort and contribution. and They are conventional units g The upper and lower limits of no-efficiency output. For AGC units g The power allocation factor, For the entire network t Imbalanced power caused by forecasting errors and real-time fluctuations on both the supply and demand sides; Unit ramp-up constraint set: ； in, and For the unit g The lower and upper limits of ramp power. For the unit g The active power output setting value, For the unit g exist t The active power output during the -1 time period; Rotational spare constraint set: ； in, and For the unit g At any moment t The up / down rotation provided below is available for use. and These represent the minimum required spin-up / spin-down reserves for the power grid, respectively. Energy storage operation constraint set: ； in, For energy storage units e exist t The electrical energy value at that moment. For energy storage units e exist t The electrical energy value at that moment. and Energy storage units e The charge / discharge efficiency, The time interval for each period, and Energy storage units e Minimum / maximum charge, and Energy storage units e Maximum / minimum charge / discharge power and Energy storage units e exist t The charging and discharging power at any given time, and Energy storage units at the initial and final moments, respectively. e The electrical energy value; New energy output constraints set: ； in, , For new energy units r exist t Output scheduling values ​​and predicted values ​​at any given time; Among them, the decision variables of the original problem Using constraint set vectors and its Jacobi matrix Construct a quotient gradient system: ； in, Decision variables x The rate of change over time is integrated using ode15s for this dynamical system, eventually converging to a conventional stable equilibrium point. There is a one-to-one correspondence between this equilibrium point and the feasible solution. This feasible solution is used as the initial solution, and IPOPT is used to solve the problem exactly.

Images