A multi-dimensional optimization dispatching method for power grid to improve wind power acceptance capacity

Abstract

A multi-dimensional optimization scheduling method for a power grid that improves the wind power acceptance capacity, step one, predicting wind power through a wind power output model, and adjusting the system's rotating reserve at the same time; step two, optimizing the output of each unit, realizing a joint optimization scheduling method in which thermal power is used as base load, nuclear power and hydrogen storage are used as auxiliary, realizing equivalent peak regulation of nuclear power, and realizing a wind-nuclear-storage-fire joint optimization scheduling process. The present application proposes a method for improving the wind power acceptance capacity by wind-fire-nuclear-storage optimization scheduling, first by establishing a wind power model with uncertain wind power output to predict wind power output, then establishing a nuclear power peak regulation depth model to determine the nuclear power peak regulation depth, establishing a joint optimization scheduling model based on the joint optimization scheduling model, and using the ant colony algorithm to analyze and determine the scheduling of thermal power, nuclear power, and energy storage, realizing energy transfer and peak shifting and valley filling, and maximizing the clean energy output by increasing the utilization and utilization rate of clean energy.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind power generation and consumption, and in particular relates to a power grid optimization dispatching method for improving wind power acceptance capacity. Technical Background

[0002] With the development of my country's social economy, the demand for energy has increased year by year. It has always relied on the support of high consumption of petrochemical energy, which has rapidly reduced the reserves of non-renewable resources such as coal, oil and natural gas. The energy crisis has intensified and a large amount of harmful pollution has also been generated. If sustainable development is to be achieved, it is necessary to find renewable energy with large reserves and low pollution. my country's power generation industry mainly uses the combustion of petrochemical energy, but with the development of science and technology, the focus is on wind power generation. my country has sufficient wind power resources, which are mainly distributed in the "Three Norths" region with rich wind power resources. Although the resources are rich, the problem of wind abandonment is serious. There are many reasons for it. Wind power generation depends on changing meteorological conditions and has randomness. The output of wind turbines is affected by wind speed and is difficult to accurately predict, and the prediction accuracy is low; there is intermittency, and the output of wind turbines will fluctuate violently between different periods, especially when the load level is low and the wind power output is high, the characteristics of anti-peak regulation are obvious, which will cause great problems. The capacity of early wind farms was small. As a distributed power source, it was decentralized and connected to the distribution network for local consumption, and the impact on power generation scheduling was negligible. As the scale and capacity of wind turbines continue to expand, the proportion of wind power in power grid generation resources has gradually increased, and the impact of large-scale wind power grid connection on the power system has become increasingly significant. The random volatility and intermittent nature of wind power make its dispatching method different from that of conventional units, which will inevitably increase the difficulty of formulating power grid generation plans and dispatching plans, and in serious cases will even affect the safety of power grid operation. Therefore, how to better utilize abandoned wind energy has become an issue that we urgently need to solve. Summary of the invention

[0003] Purpose of the invention:

[0004] The present invention provides a multi-element optimization dispatching method for power grids to improve wind power acceptance capacity, and its purpose is to solve the existing problems in the past. When the wind power output is the largest, the peak load task of thermal power is realized under the condition of base load, and nuclear power and hydrogen storage are used as auxiliary to reduce daily dispatching. The patent predicts the wind power output, and corrects the positive and negative rotating reserves of the system, and completes the day-ahead dispatching plan of the power system.

[0005] Technical solution:

[0006] A method for optimizing and dispatching a power grid to improve wind power acceptance capacity, characterized in that:

[0007] Step 1: predict wind power through a wind power output model and adjust the system's spinning reserve;

[0008] Step 2: Optimize the output of each unit to implement a joint optimization scheduling method with thermal power as the base load and nuclear power and hydrogen storage as auxiliary, achieve equivalent peak regulation of nuclear power, and realize the joint optimization scheduling process of wind-nuclear-storage-thermal.

[0009] In step one: by establishing a wind power uncertain output calculation method based on error growth, analyzing the relevant concepts of error growth, establishing a wind power output model, taking the wind power prediction error into account in the wind power output model, and considering the wind power output as an uncertain quantity in combination with the wind power output characteristics, the wind power output is predicted.

[0010] The optimization of the output of each unit described in the second step means that in the process of wind-nuclear-storage-thermal joint optimization scheduling, wind power output needs to be given priority when the load demand is met, and then the thermal power units and hydrogen storage output are arranged. Due to the uncertainty of wind power, hydrogen storage will store the wind power that cannot be accepted by the power grid in the form of pumping when wind power is generated, and generate electricity during peak load hours to alleviate the peak load pressure during peak load hours; after giving priority to the output of wind power and thermal power units, hydrogen storage is used to reduce the changes in the scheduling of thermal power. Among them, nuclear power units do not participate in conventional peak-shaving tasks. When the down-peaking capacity is insufficient during the load valley, nuclear power participates in short-term deep peak-shaving through a given judgment method.

[0011] The implementation method of predicting wind power in step 1 is as follows: (1) Determine the wind speed law through Weibull distribution and obtain the Weibull distribution of wind speed:

[0012]

[0013] In the formula, c and k represent the shape parameter and scale parameter of Weibull distribution respectively, and v is the wind speed value;

[0014]

[0015] When the wind speed value is greater than the cut-in wind speed v of the wind turbine ci When the wind speed value maintains the rated wind speed v r When the wind turbine output remains unchanged; if the wind speed value is greater than or equal to the cut-out wind speed v co When , the wind turbine stops running; the wind power output active power P that obeys the wind speed Weibull distribution can be obtained from the following formula:

[0016]

[0017] in,

[0018]

[0019] Where Pr The power required to maintain the rated wind speed for the wind speed value;

[0020]

[0021] In step 2: when the power grid is in a low period and the peak-shaving capacity is insufficient, the output of nuclear power units is reduced for a short time to provide space for wind power acceptance and achieve equivalent peak-shaving of nuclear power; equivalent peak-shaving depth model:

[0022]

[0023] Where T is the system peak surplus capacity; D is the system valley down-peak surplus capacity; P LMax and P LMin are the planned daily maximum and minimum power generation loads respectively; P GMax and P GMin are the adjustable maximum and minimum outputs of all thermal power units in the system during the peak and valley load periods on the planned day; P NMax is the rated output of all nuclear power units on the planned day; P R and P R ' are the positive and negative spinning reserve capacities of the system respectively;

[0024] If D≥0 and T≥0 are satisfied at the same time, the system meets the peak load balance and nuclear power does not need to participate in peak load regulation; if T<0, that is, the peak load capacity is insufficient during the low load period, and nuclear power is required to participate in peak load regulation. The judgment criteria are:

[0025] P LMin <P GMin +P NMax +P R ' (7)

[0026] The ant colony algorithm is used to analyze and determine the above equivalent peak-shaving depth model to achieve equivalent peak-shaving of nuclear power. The specific steps are as follows:

[0027] List the optimization objective function:

[0028]

[0029] In the formula, i is the unit number, t is the time, and f i is the power generation cost of thermal power unit i, is the power generation cost of all thermal power units in the dispatch period T, m0 is the total number of thermal power units; P NiMax is the rated output of nuclear power unit i, The actual output of nuclear power unit i, is the additional cost of nuclear power participating in peak load regulation during the dispatch period T, and m1 is the total number of nuclear power units; is the deep peak load regulation power of thermal power unit i, P′ is the additional cost of deep peak load regulation of thermal power units within the dispatch period T; GiMin is the peak load capacity of the minimum output of the thermal power unit, is the peak-shaving cost of thermal power start-up and stop, m2 is the number of thermal power units started and stopped within the dispatch period T, and the start-up and stop peak-shaving effect will act on the entire dispatch period after start-up and stop; is the capacity of hydrogen storage device i, is the cost of hydrogen production within the scheduling period T; C N , C G , C TOf , C Hy They are the peak-shaving costs per unit capacity for various peak-shaving methods.

[0030] Formula (7) needs to simultaneously satisfy the following system power balance constraints, system reserve constraints, nuclear power peak load capacity constraints, thermal power unit capacity constraints, spinning reserve constraints, conventional unit minimum start and stop time constraints, and unit ramp rate constraints. In each constraint, t is any time period within the scheduling cycle. Since the optimization target is the total operating cost within 1 day, the range of t is 1-24h;

[0031] Constraint equations

[0032] (1) System power balance equation:

[0033]

[0034] Where: The three items on the left are the total output of all thermal power units during period t, the total output of all nuclear power units during period t, and the total output of all hydrogen storage units during period t. is the power demand of the system in period t;

[0035] (2) System standby constraints:

[0036]

[0037] Where P GiMax is the maximum output of thermal power unit i during the peak period of the power grid, The actual output of thermal power unit i is is the overall reserve of the power grid during peak hours, P′ GiMin The minimum output of thermal power unit i during the off-peak period of the power grid, is the overall reserve of the power grid during off-peak hours, P rt , P′ rt They are respectively the positive and negative reserve demands of the system at time t, both of which are undertaken by thermal power units;

[0038] (3) Constraints on the ability to absorb nuclear power peak load:

[0039] P′ N ≤P N.max -P N0 (11)

[0040] Where P N0 The output of nuclear power in the pre-dispatch model;

[0041] (4) Thermal power unit capacity constraints:

[0042]

[0043]

[0044] In the formula, is the start and stop status of conventional unit i at time t, P i For conventional unit output, P Gi min , P Gi max are the upper and lower limits of the output of conventional unit i;

[0045] (5) Spinning reserve constraints:

[0046]

[0047] Where: N is the total number of units, P i t is the actual output power of the i-th generator, P i max is the maximum possible output power of the i-th generator in period t, and its value is determined by the output limit of the generator and the climbing rate of the unit; R t The spinning reserve capacity required for the system in period t is considered as 7% of the total system load;

[0048] (6) Minimum start and stop time constraints for conventional units:

[0049]

[0050]

[0051] Where: and are respectively the startup duration and shutdown duration of unit i in period t; and are the minimum continuous operating hours and minimum continuous downtime hours of unit i in the scheduling cycle, respectively;

[0052] (7) Unit climbing rate constraint:

[0053]

[0054] In the formula, P′ i.tis the output power of unit i at time t, P′ i.t-1 is the output power of unit i at time t-1, L up.i , L down.i are the upper limit of the increase and decrease of the output allowed per unit time for unit i respectively;

[0055] The ant colony algorithm used is as follows:

[0056] Introducing the concepts of state and decision

[0057] State: The combination of peak load-shaving measures taken at the current moment constitutes a state;

[0058] Decision-making: From the combination of peak-shaving means in a certain period to the determination of the peak-shaving means in the next period, that is, forming a decision from one state to the next state;

[0059] The concept of closed path to obtain the optimal solution is as follows:

[0060] Path: The set of peak-shaving methods in all time periods within the dispatching cycle. In each time period, any peak-shaving method is selected, so that the peak-shaving output of various power sources in the entire dispatching cycle, that is, the set of all decisions, constitutes a path;

[0061] The objective function is:

[0062] J=min∑(tc(s i ,s i+1 )+tc(s i ,s i-1 ))0≤i≤24 (18)

[0063] Where: tc(s i ,s i+1 ) and tc(s i ,s i-1 ) represent the decision costs from state i to state (i+1) and from state i to state (i-1) respectively; state i is selected as the point with the maximum load in 1 day, and the peak-shaving cost of this point is 0, which corresponds to the distance of 0 in the TSP problem, that is, it has not set out yet; the difference is that in the joint peak-shaving optimization problem, the combination of peak-shaving means to be adopted next time is constrained by the peak-shaving means adopted last time. The tabu table records the adjusted capacity and peak-shaving cost of the different peak-shaving means adopted, which is used to limit the combination of adjustment means and capacity for the next peak-shaving. After one cycle, the minimum peak-shaving cost in the tabu table corresponds to the shortest path of the TSP problem. Before the next cycle starts, the path information is updated and the tabu table is cleared; the converted objective function can be solved by the ant colony algorithm.

[0064] The peak-shaving output of nuclear power is arranged outside the cycle of the ant colony algorithm, and the optimization result is changed by changing the peak-shaving depth of nuclear power.

[0065] Equality constraints

[0066] The power balance constraint constrains the combination of peak-shaving means by calculating the load difference between adjacent time points. The load change at adjacent times is the sum of the adjustment capacity of the selected peak-shaving means, thereby satisfying the power balance constraint.

[0067] Inequality constraints:

[0068] Inequality constraints include upper and lower limits of output of various power sources, ramp constraints of thermal power units, spinning reserves and minimum start-up and shutdown time. The ultimate goal of various constraints is to constrain the output of the units:

[0069] The combination of peak-shaving methods that can be used at the next time point is controlled by the adjusted capacity of various peak-shaving methods recorded in the tabu table and various inequality constraints:

[0070] The calculation steps of the multi-element optimization dispatching method of the power grid to improve the wind power acceptance capacity are as follows:

[0071] ① Input the grid load data, find the time point corresponding to the maximum load, determine the state at that time point, and iterate from that point to both sides;

[0072] ② Input the nuclear power output model, the peak load depth setting is adjustable, and the adjustment amplitude and step length are set;

[0073] ③ Determine the initial state s, which is the time point corresponding to the maximum load point;

[0074] ④ACA initialization: D = 1; τ ij =0;Δτ ij =0; cost(k)=0;

[0075] ⑤Iteration on both sides:

[0076] Iterate forward - t = t-1, determine whether t is the final time period 0, if so, go to ⑧, otherwise go to ⑥;

[0077] Backward iteration - t = t + 1, determine whether t is the final time period 24, if so, go to ⑧, otherwise go to ⑥;

[0078] ⑥Form the tabu table at the current moment and calculate the peak load cost at this moment as the path length in ACA;

[0079] According to the ant transfer probability, the next state j is selected, that is, the peak load regulation means combination of the next period, and the load regulation amount of each type of unit from state i to state j and its corresponding peak load regulation cost are calculated;

[0080] ⑦Continuously accumulate the path length of the ants, i.e. the peak-shaving cost, and go to ⑤;

[0081] ⑧Record the shortest path of ⑦ and update the path information through the algorithm global update rule;

[0082] ⑨Let: D = D + 1; Δτ ij =0; cost(k)=0 clear table;

[0083] ⑩If D<D Max , and there is no stagnation, then let t return to the time point corresponding to the maximum load, go to ⑤, and start the next iteration, otherwise stop the iteration and output the optimal solution; where: cost(k) is the length of the path currently traversed by ant k, that is, the current total peak-shaving cost; D is the number of iterations, D Max is the maximum number of iterations.

[0084] Advantages and effects:

[0085] 1 Wind power output uncertainty modeling

[0086] In the process of wind-storage-thermal joint optimization dispatch, it is necessary to configure wind power first, and then configure it with thermal power units to meet the load demand. Due to the uncertainty of wind power, hydrogen storage will store wind power that cannot be accepted by the power grid in the form of pumping when wind power is in full swing; generate electricity during peak load periods to alleviate the peak load pressure during peak load periods. After giving priority to the configuration of wind power and thermal power, hydrogen storage is used to reduce the changes in the dispatch of thermal power.

[0087] Due to the randomness of wind speed, wind power output is uncertain. The patent assumes that wind speed follows Weibull distribution, and its probability density function is as follows:

[0088]

[0089] Where c and k represent the shape parameter and scale parameter of Weibull distribution respectively, and v is the wind speed value.

[0090] When the wind speed value is greater than the cut-in wind speed v of the wind turbine ci When the wind speed value maintains the rated wind speed v r When the wind turbine output remains unchanged; if the wind speed value is greater than or equal to the cut-out wind speed v co When , the wind turbine stops running. The wind power output active power P that obeys the wind speed Weibull distribution can be obtained from the following formula:

[0091]

[0092] in,

[0093]

[0094]

[0095] 2. Determination of Nuclear Power Peak Regulation

[0096] Considering only hydrogen storage, thermal and nuclear power connected to the power system, and not considering the start and stop peak load regulation of thermal power, the power balance and peak load regulation balance of the power grid are:

[0097]

[0098] Where: T is the system peak power surplus capacity; D is the system valley down-peak surplus capacity; P LMax and P LMin are the planned daily maximum and minimum power generation loads respectively; P GMax and P GMin are the adjustable maximum and minimum outputs of all thermal power units in the system during the peak and valley load periods on the planned day; P NMax is the rated output of all nuclear power units on the planned day; P R and P R ' are the positive and negative spinning reserve capacities of the system respectively.

[0099] If D≥0 and T≥0 are satisfied at the same time, the system meets the peak load balance and nuclear power does not need to participate in peak load regulation. If T<0, that is, the peak load capacity is insufficient during the low load period, nuclear power is required to participate in peak load regulation, and the judgment criteria are:

[0100] P LMin <P GMin +P NMax +P R ' (6)

[0101] 3 Joint peak load optimization modeling

[0102] The purpose of optimizing the combined peak-shaving of hydrogen storage, thermal and nuclear power in the patent is to minimize the total power generation cost of the system within the optimized scheduling cycle while ensuring the safe and stable operation of the system.

[0103] For this optimization problem, the objective function is listed as follows:

[0104]

[0105] in: is the power generation cost of all thermal power units in the dispatch period T, m0 is the total number of thermal power units; is the additional cost of nuclear power participating in peak load regulation during the dispatch period T, and m1 is the total number of nuclear power units; It is the additional cost when the thermal power units are deeply peak-shifted within the dispatch period T; is the peak-shaving cost of thermal power start-up and stop, m2 is the number of thermal power units started and stopped within the dispatch period T, and the start-up and stop peak-shaving effect will act on the entire dispatch period after start-up and stop; is the cost of hydrogen production within the scheduling period T; C N , C G , C TOf , C Hy are the peak-shaving costs per unit capacity of various peak-shaving means. Formula (7) is subject to the following system constraints, independent water system constraints, thermal power unit output constraints and nuclear power peak-shaving output constraints. In each constraint, t is any time period within the dispatch cycle. Since the optimization target is the total operating cost within 1 day, the range of t is 1-24h.

[0106] Constraint equations

[0107] (1) System power balance equation:

[0108]

[0109] In the formula, the three items on the left are the total output of all thermal power units during period t, the total output of all nuclear power units during period t, and the total output of all hydrogen storage units during period t. is the power demand of the system in period t.

[0110] (2) System backup constraints:

[0111]

[0112] Where: P rt , P′ rt They are respectively the positive and negative reserve demands of the system at time t, and both are borne by thermal power units.

[0113] (3) Constraints on the ability to absorb nuclear power peak load:

[0114] P′ N ≤P N.max -P N0 (10)

[0115] Where P N0 The output of nuclear power in the pre-scheduling model is.

[0116] (4) Thermal power unit capacity constraints:

[0117]

[0118]

[0119] Where: P Gi min , P Gi max are the upper and lower limits of the output of conventional unit i.

[0120] (5) Spinning reserve constraints:

[0121]

[0122] Where: P i max is the maximum possible output power of the i-th generator in period t, and its value is determined by the output limit of the generator and the climbing rate of the unit; R t The spinning reserve capacity required for the system during period t is considered to be 7% of the total system load.

[0123] (6) Minimum start and stop time constraints for conventional units:

[0124]

[0125]

[0126] Where: and are respectively the startup duration and shutdown duration of unit i in period t; and are the minimum continuous operating hours and the minimum continuous downtime hours of unit i during the scheduling cycle, respectively.

[0127] (7) Upper and lower limits of active output of the unit:

[0128]

[0129] In the calculation, the parameters k of thermal power units and nuclear power units are i , c v Both are 0.

[0130] (8) Unit climbing rate constraint:

[0131]

[0132] Where, L up.i , L down.i They are respectively the upper limit of the output increase and the upper limit of the output decrease allowed per unit time for unit i.

[0133] 4 Ant Colony Algorithm Optimization of Unit Joint Peak Loading

[0134] The optimization problem of multi-power combined peak load regulation requires some improvement of the model, introducing the concepts of state and decision. State: The combination of peak load regulation measures taken at the current moment constitutes a state. Decision: From the combination of peak load regulation measures in a certain period to the determination of peak load regulation measures in the next period, that is, from one state to the next state, a decision is formed.

[0135] Since in the ant colony optimization algorithm, a closed path is a feasible solution to the problem, the concept of a path is as follows: Path: A set of peak-shaving methods for all time periods in a scheduling cycle. In each time period, any peak-shaving method is selected, so that the peak-shaving output of various power sources in the entire scheduling cycle, that is, the set of all decisions, constitutes a path.

[0136] According to the above concepts and related definitions, the multi-power joint peak-shaving optimization problem can be transformed into a multi-stage dynamic search problem. The goal of peak-shaving selection optimization is to find the minimum value of the path in the decision-making at all times, which is equivalent to the optimal path found in the problem. Due to various constraints, the number of states in each time period is not the same.

[0137] (1) Ant Colony Algorithm Model for Optimization Objectives

[0138] When the ant colony algorithm solves the TSP problem, the path taken by the ants is the result of optimization. The same is true for the problem of joint peak-shaving of multiple power sources. The concept of state and decision-making can be used to solve the problem of joint peak-shaving of multiple power sources in the mode of TSP problem. The combination of peak-shaving means in each time period corresponds to each city in the TSP problem. The peak-shaving cost is the length of the path. The lowest peak-shaving cost is the optimization goal. Then the objective function of the problem of joint peak-shaving of multiple power sources can be converted into:

[0139] J=min∑(tc(s i ,s i+1 )+tc(s i ,s i-1 ))0≤i≤24 (19)

[0140] Where: tc(s i ,s i+1 ) and tc(s i ,s i-1 ) respectively represent the decision costs from state i to state (i+1) and from state to state (i-1), which corresponds to the transfer distance in the TSP problem; state i is selected as the point with the maximum load in 1 day, and the peak-shaving cost of this point is 0, corresponding to the distance in the TSP problem, that is, it has not yet set off; the difference is that in the joint peak-shaving optimization problem, the combination of peak-shaving means to be used next time is constrained by the peak-shaving means used last time, so the tabu table records the adjusted capacity and peak-shaving costs of the different peak-shaving means adopted, which is used to limit the combination of the adjustment means and capacity of the next peak-shaving. After one cycle, the minimum peak-shaving cost in the tabu table corresponds to the shortest path of the TSP problem. Before the next cycle starts, the path information is updated and the tabu table is cleared. The converted objective function can be solved by the ant colony algorithm.

[0141] (2) Transformation of Constraints

[0142] Due to the particularity of nuclear power units, they have a fixed output curve and can only change their peak-shaving depth. Therefore, the nuclear power peak-shaving output is arranged outside the ant colony algorithm cycle, and the optimization result is changed by changing the nuclear power peak-shaving depth.

[0143] 1. Equality constraints

[0144] The power balance constraint constrains the combination of peak-shaving means by calculating the load difference between adjacent time points. The load change between adjacent time points is the sum of the regulation capacity of the selected peak-shaving means, thereby satisfying the power balance constraint. The water balance constraint can be converted into an output limit for other outputs.

[0145] 2. Inequality constraints

[0146] Inequality constraints include upper and lower limit constraints on the output of various power sources, ramp constraints, spinning reserves, minimum start-up and shutdown times of thermal power units, etc. The ultimate goal of various constraints is to constrain the output of the units.

[0147] The adjusted capacity of various peak-shaving means recorded in the tabu table and various inequality constraints are used to jointly control the combination of peak-shaving means that can be used at the next time point. For example, when the unit output exceeds the limit, the maximum or minimum value is taken, and when the climbing rate is not satisfied, the maximum climbing rate of the unit is taken.

[0148] (4) Calculation steps

[0149] ① Input the grid load data, find the time point corresponding to the maximum load, determine the state at that time point (i.e. the output of each type of unit), and iterate to both sides from that point.

[0150] ② Input the nuclear power output model, the peak regulation depth setting is adjustable, and the adjustment amplitude and step length are set.

[0151] ③ Determine the initial state s, which is the time point corresponding to the maximum load point.

[0152] ④ACA initialization: D = 1; τ ij =0;Δτ ij =0; cost(k)=0.

[0153] ⑤Iteration on both sides:

[0154] Iterate forward - t = t-1, determine whether t is the final time period 0, if so, go to ⑧, otherwise go to ⑥.

[0155] Iterate backwards - t = t + 1, determine whether t is the final time period 24, if so, go to ⑧, otherwise go to ⑥.

[0156] ⑥Form the tabu table at the current moment and calculate the peak-shaving cost at this moment as the path length in ACA.

[0157] According to the ant transfer probability, the next state j is selected, that is, the combination of peak-shaving means for the next period, and the load adjustment amount of each type of unit from state i to state j and its corresponding peak-shaving cost are calculated.

[0158] ⑦Continuously accumulate the path length of the ants, i.e. the peak-shaving cost, and go to ⑤.

[0159] ⑧Record the shortest path of ⑦ and update the path information through the algorithm's global update rules.

[0160] ⑨Let: D = D + 1; Δτ ij =0; cost(k)=0 clears the table.

[0161] ⑩If D<D Max , and there is no stagnation, then set t back to the time point corresponding to the maximum load, go to ⑤, and start the next iteration, otherwise stop the iteration and output the optimal solution. Where: cost(k) is the length of the path currently traversed by ant k, that is, the current total peak-shaving cost; D is the number of iterations, D Max is the maximum number of iterations.

[0162] In summary, this application proposes a method for improving wind power acceptance capacity by optimizing the scheduling of wind-thermal-nuclear storage. First, a wind power model with uncertain wind power output is established to predict wind power output. Then a nuclear power peak-shaving depth model is established to determine the peak-shaving depth of nuclear power. A joint optimization scheduling model is established, and an ant colony algorithm is used to analyze and determine the scheduling of thermal power, nuclear power, and energy storage to achieve energy transfer and peak shifting and valley filling. The clean energy output is maximized by increasing the utilization and utilization rate of clean energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0163] Figure 1 State Space Diagram

[0164] Figure 2 Algorithm overall flow chart

[0165] Figure 3 AOA algorithm flow chart DETAILED DESCRIPTION

[0166] A method for optimizing the dispatching of a power grid to improve wind power acceptance capacity comprises the following steps:

[0167] Step 1: Predict wind power through the wind power output model and adjust the system's spinning reserve; (The spinning reserve R in the following formula is to provide space for wind power consumption)

[0168] Step 2: Optimize the output of each unit on the basis of ensuring the safety and reliability of the system through given constraints, realize the joint optimization scheduling method with thermal power as base load and nuclear power and hydrogen storage as auxiliary, realize the equivalent peak regulation of nuclear power, and realize the joint optimization scheduling process of wind-nuclear-storage-thermal.

[0169] In step one: by establishing a wind power uncertain output calculation method based on error growth, the related concepts of error growth are analyzed (that is, the related concepts of wind power prediction error are the contents described in formulas 1 to 2), a wind power (uncertain) output model is established, and the wind power prediction error is included in the wind power (uncertain) output model (this is wind power output modeling but the uncertainty of wind power output is considered in the modeling), combined with the characteristics of wind power output, wind power output is regarded as an uncertain quantity, and the output of wind power is predicted.

[0170] The optimization of the output of each unit described in the second step means that in the process of wind-nuclear-storage-thermal joint optimization scheduling, wind power output needs to be given priority when the load demand is met, and then the thermal power units and hydrogen storage output are arranged. Due to the uncertainty of wind power, hydrogen storage will store the wind power that cannot be accepted by the power grid in the form of pumping when wind power is generated, and generate electricity during peak load hours to alleviate the peak load pressure during peak load hours; after giving priority to the output of wind power and thermal power units, hydrogen storage is used to reduce the changes in the scheduling of thermal power. Among them, nuclear power units do not participate in conventional peak-shaving tasks. When the down-peaking capacity is insufficient during the low load, nuclear power participates in short-term deep peak-shaving through the given judgment method (the text description between formulas 5 and 6).

[0171] The implementation method of predicting wind power in step 1 is as follows: (1) Determine the wind speed law through Weibull distribution and obtain the Weibull distribution of wind speed:

[0172]

[0173] In the formula, c and k represent the shape parameter and scale parameter of Weibull distribution respectively, and v is the wind speed value;

[0174]

[0175] When the wind speed value is greater than the cut-in wind speed v of the wind turbine ci When the wind speed value maintains the rated wind speed v r When the wind turbine output remains unchanged; if the wind speed value is greater than or equal to the cut-out wind speed v co When , the wind turbine stops running; the wind power output active power P that obeys the wind speed Weibull distribution can be obtained from the following formula:

[0176]

[0177] in,

[0178]

[0179] Where P r The power required to maintain the rated wind speed for the wind speed value;

[0180]

[0181] In step 2: when the power grid is in a low period and the peak-shaving capacity is insufficient, the output of nuclear power units is reduced for a short time to provide space for wind power acceptance and achieve equivalent peak-shaving of nuclear power; equivalent peak-shaving depth model:

[0182]

[0183] Where: T is the system peak power surplus capacity; D is the system valley down-peak surplus capacity; P LMax and P LMin are the planned daily maximum and minimum power generation loads respectively; P GMax and P GMin are the adjustable maximum and minimum outputs of all thermal power units in the system during the peak and valley load periods on the planned day; P NMax is the rated output of all nuclear power units on the planned day; P R and P R ' are the positive and negative spinning reserve capacities of the system respectively;

[0184] If D≥0 and T≥0 are satisfied at the same time, the system meets the peak load balance and nuclear power does not need to participate in peak load regulation; if T<0, that is, the peak load capacity is insufficient during the low load period, and nuclear power is required to participate in peak load regulation. The judgment criteria are:

[0185] P LMin <P GMin +P NMax +P R '(6). (If it is determined to participate in peak load regulation, the output of nuclear power will be adjusted according to formula 7)

[0186] The ant colony algorithm is used to analyze and determine the above equivalent peak-shaving depth model to achieve equivalent peak-shaving of nuclear power. The specific steps are as follows:

[0187] List the optimization objective function:

[0188]

[0189] Where: i is the unit number, t is the time, f i is the power generation (economic) cost of thermal power unit i, is the power generation (economic) cost of all thermal power units in the dispatch period T, m0 is the total number of thermal power units; PNiMax is the rated output of nuclear power unit i, The actual output of nuclear power unit i, is the additional (economic) cost of nuclear power participating in peak load regulation during the dispatch period T, and m1 is the total number of nuclear power units; is the deep peak load regulation power of thermal power unit i, P′ is the additional (economic) cost of deep peak load regulation of thermal power units within the dispatch period T; GiMin is the peak load capacity of the minimum output of the thermal power unit, is the (economic) cost of thermal power start-stop peak regulation, m2 is the number of thermal power units started and stopped within the dispatch period T, and the start-stop peak regulation effect will act on the entire dispatch period after start-stop; is the capacity of hydrogen storage device i, is the (economic) cost of hydrogen production within the scheduling period T; C N , C G , C TOf , C Hy They are the peak-shaving costs per unit capacity for various peak-shaving methods.

[0190] Formula (7) needs to simultaneously satisfy the following system power balance constraints, system reserve constraints, nuclear power peak load capacity constraints, thermal power unit capacity constraints, spinning reserve constraints, conventional unit minimum start and stop time constraints, and unit ramp rate constraints. In each constraint, t is any time period within the scheduling cycle. Since the optimization target is the total operating cost within 1 day, the range of t is 1-24h;

[0191] Constraint equations

[0192] (1) System power balance equation:

[0193]

[0194] Where: The three items on the left are the total output of all thermal power units during period t, the total output of all nuclear power units during period t, and the total output of all hydrogen storage units during period t. is the power demand of the system in period t;

[0195] (2) System backup constraints:

[0196]

[0197] Where: P GiMax is the maximum output of thermal power unit i during the peak period of the power grid, The actual output of thermal power unit i is is the overall reserve of the power grid during peak hours, P′ GiMin The minimum output of thermal power unit i during the off-peak period of the power grid, is the overall reserve of the power grid during off-peak hours, P rt , P′ rt They are respectively the positive and negative reserve demands of the system at time t, both of which are undertaken by thermal power units;

[0198] (3) Constraints on the ability to absorb nuclear power peak load:

[0199] P′ N ≤P N.max -P N0 (10)

[0200] Where P N0 The output of nuclear power in the pre-dispatch model;

[0201] (4) Thermal power unit capacity constraints:

[0202]

[0203]

[0204] Where: is the start and stop status of conventional unit i at time t, P i For conventional unit output, P Gi min , P Gi max are the upper and lower limits of the output of conventional unit i;

[0205] (5) Spinning reserve constraints:

[0206]

[0207] Where: N is the total number of units, P i t is the actual output power of the i-th generator, P i max is the maximum possible output power of the i-th generator in period t, and its value is determined by the output limit of the generator and the climbing rate of the unit; R t The spinning reserve capacity required for the system in period t is considered as 7% of the total system load;

[0208] (6) Minimum start and stop time constraints for conventional units:

[0209]

[0210]

[0211] Where: and are respectively the startup duration and shutdown duration of unit i in period t; and are the minimum continuous operating hours and minimum continuous downtime hours of unit i in the scheduling cycle, respectively;

[0212] (7) Unit climbing rate constraint:

[0213]

[0214] In the formula, P′ i.t is the output power of unit i at time t, P′ i.t-1 is the output power of unit i at time t-1, L up.i , L down.i are the upper limit of the increase and decrease of the output allowed per unit time for unit i respectively;

[0215] The ant colony algorithm used is as follows:

[0216] Introducing the concepts of state and decision

[0217] State: The combination of peak load-shaving measures taken at the current moment constitutes a state;

[0218] Decision-making: From the combination of peak-shaving means in a certain period to the determination of the peak-shaving means in the next period, that is, forming a decision from one state to the next state;

[0219] The concept of closed path to obtain the optimal solution is as follows:

[0220] Path: The set of peak-shaving methods in all time periods within the dispatching cycle. In each time period, any peak-shaving method is selected, so that the peak-shaving output of various power sources in the entire dispatching cycle, that is, the set of all decisions, constitutes a path;

[0221] The objective function is:

[0222] J=min∑(tc(s i ,s i+1 )+tc(s i ,s i-1 ))0≤i≤24 (18)

[0223] Where: tc(s i ,s i+1 ) and tc(s i ,s i-1) represent the decision costs from state i to state (i+1) and from state i to state (i-1) respectively; state i is selected as the point with the maximum load in 1 day, and the peak-shaving cost of this point is 0, which corresponds to the distance of 0 in the TSP problem, that is, it has not set out yet; the difference is that in the joint peak-shaving optimization problem, the combination of peak-shaving means to be adopted next time is constrained by the peak-shaving means adopted last time. The tabu table records the adjusted capacity and peak-shaving cost of the different peak-shaving means adopted, which is used to limit the combination of adjustment means and capacity for the next peak-shaving. After one cycle, the minimum peak-shaving cost in the tabu table corresponds to the shortest path of the TSP (Traveling Salesman Problem). Before the next cycle starts, the path information is updated and the tabu table is cleared; the converted objective function can be solved by the ant colony algorithm.

[0224] The peak-shaving output of nuclear power is arranged outside the cycle of the ant colony algorithm, and the optimization result is changed by changing the peak-shaving depth of nuclear power.

[0225] Equality constraints

[0226] The power balance constraint constrains the combination of peak-shaving means by calculating the load difference between adjacent time points. The load change between adjacent time points (referring to the load difference between adjacent time points mentioned above) is the sum of the adjustment capacity of the selected peak-shaving means, thereby satisfying the power balance constraint.

[0227] Inequality constraints:

[0228] Inequality constraints include upper and lower limits of output of various power sources, ramp constraints of thermal power units, spinning reserves and minimum start-up and shutdown time. The ultimate goal of various constraints is to constrain the output of the units:

[0229] The adjusted capacity of various peak-shaving means recorded in the tabu table and various inequality constraints are used to jointly control the combination of peak-shaving means that can be used at the next time point. For example, when the unit output exceeds the limit, the maximum or minimum value is taken, and when the climbing rate is not satisfied, the maximum climbing rate of the unit is taken:

[0230] The calculation steps of the multi-element optimization dispatching method of the power grid to improve the wind power acceptance capacity are as follows:

[0231] ① Input the grid load data, find the time point corresponding to the maximum load, determine the state at that time point (i.e. the output of each type of unit), and iterate from that point to both sides;

[0232] ② Input the nuclear power output model, the peak load depth setting is adjustable, and the adjustment amplitude and step length are set;

[0233] ③ Determine the initial state s, which is the time point corresponding to the maximum load point;

[0234] ④ACA initialization: D = 1; τ ij =0;Δτ ij =0; cost(k)=0;

[0235] ⑤Iteration on both sides:

[0236] Iterate forward - t = t-1, determine whether t is the final time period 0, if so, go to ⑧, otherwise go to ⑥;

[0237] Backward iteration - t = t + 1, determine whether t is the final time period 24, if so, go to ⑧, otherwise go to ⑥;

[0238] ⑥Form the tabu table at the current moment and calculate the peak load cost at this moment as the path length in ACA;

[0239] According to the ant transfer probability, the next state j is selected, that is, the peak load regulation means combination of the next period, and the load regulation amount of each type of unit from state i to state j and its corresponding peak load regulation cost are calculated;

[0240] ⑦Continuously accumulate the path length of the ants, i.e. the peak-shaving cost, and go to ⑤;

[0241] ⑧Record the shortest path of ⑦ and update the path information through the algorithm global update rule;

[0242] ⑨Let: D = D + 1; Δτ ij =0; cost(k)=0 clear table;

[0243] ⑩If D<D Max , and there is no stagnation, then let t return to the time point corresponding to the maximum load, go to ⑤, and start the next iteration, otherwise stop the iteration and output the optimal solution; where: cost(k) is the length of the path currently traversed by ant k, that is, the current total peak-shaving cost; D is the number of iterations, D Max is the maximum number of iterations.

[0244] Step three uses the nuclear-thermal-wind-storage joint scheduling model to reasonably determine the output forecasts of nuclear power and thermal power, and use the energy transfer of energy storage to achieve equivalent peak regulation of nuclear power, thereby increasing the utilization rate of clean energy.

[0245] The present invention is described in further detail below:

[0246] 1 Wind power output uncertainty modeling

[0247] In the process of wind-storage-thermal joint optimization dispatch, it is necessary to configure wind power first, and then configure it with thermal power units to meet the load demand. Due to the uncertainty of wind power, hydrogen storage will store wind power that cannot be accepted by the power grid in the form of pumping when wind power is in full swing; generate electricity during peak load periods to alleviate the peak load pressure during peak load periods. After giving priority to the configuration of wind power and thermal power, hydrogen storage is used to reduce the changes in the dispatch of thermal power.

[0248] Due to the randomness of wind speed, wind power output is uncertain. The patent assumes that wind speed follows Weibull distribution, and its probability density function is as follows:

[0249]

[0250] Where c and k represent the shape parameter and scale parameter of Weibull distribution respectively, and v is the wind speed value.

[0251] When the wind speed value is greater than the cut-in wind speed v of the wind turbine ci When the wind speed value maintains the rated wind speed v r When the wind turbine output remains unchanged; if the wind speed value is greater than or equal to the cut-out wind speed v co When , the wind turbine stops running. The wind power output active power P that obeys the wind speed Weibull distribution can be obtained from the following formula:

[0252]

[0253] in,

[0254]

[0255]

[0256] 2. Determination of Nuclear Power Peak Regulation

[0257] Considering only hydrogen storage, thermal and nuclear power connected to the power system, and not considering the start and stop peak load regulation of thermal power, the power balance and peak load regulation balance of the power grid are:

[0258]

[0259] Where: T is the system peak power surplus capacity; D is the system valley down-peak surplus capacity; P LMax and P LMin are the planned daily maximum and minimum power generation loads respectively; P GMax and P GMin are the adjustable maximum and minimum outputs of all thermal power units in the system during the peak and valley load periods on the planned day; P NMax is the rated output of all nuclear power units on the planned day; P Rand P R ' are the positive and negative spinning reserve capacities of the system respectively.

[0260] If D≥0 and T≥0 are satisfied at the same time, the system meets the peak load balance and nuclear power does not need to participate in peak load regulation. If T<0, that is, the peak load capacity is insufficient during the low load period, nuclear power is required to participate in peak load regulation, and the judgment criteria are:

[0261] P LMin <P GMin +P NMax +P R ' (6)

[0262] 3 Joint peak load optimization modeling

[0263] The purpose of optimizing the combined peak-shaving of hydrogen storage, thermal and nuclear power in the patent is to minimize the total power generation cost of the system within the optimized scheduling cycle while ensuring the safe and stable operation of the system.

[0264] For this optimization problem, the objective function is listed as follows:

[0265]

[0266] in: is the power generation cost of all thermal power units in the dispatch period T, m0 is the total number of thermal power units; is the additional cost of nuclear power participating in peak load regulation during the dispatch period T, and m1 is the total number of nuclear power units; It is the additional cost when the thermal power units are deeply peak-shifted within the dispatch period T; is the peak-shaving cost of thermal power start-up and stop, m2 is the number of thermal power units started and stopped within the dispatch period T, and the start-up and stop peak-shaving effect will act on the entire dispatch period after start-up and stop; is the cost of hydrogen production within the scheduling period T; C N , C G , C TOf , C Hy are the peak-shaving costs per unit capacity of various peak-shaving means. Formula (7) is subject to the following system constraints, independent water system constraints, thermal power unit output constraints and nuclear power peak-shaving output constraints. In each constraint, t is any time period within the dispatch cycle. Since the optimization target is the total operating cost within 1 day, the range of t is 1-24h.

[0267] Constraint equations

[0268] (1) System power balance equation:

[0269]

[0270] In the formula, the three items on the left are the total output of all thermal power units during period t, the total output of all nuclear power units during period t, and the total output of all hydrogen storage units during period t. is the power demand of the system in period t.

[0271] (2) System backup constraints:

[0272]

[0273] Where: P rt , P′ rt They are respectively the positive and negative reserve demands of the system at time t, and both are borne by thermal power units.

[0274] (3) Constraints on the ability to absorb nuclear power peak load:

[0275] P′ N ≤P N.max -P N0 (10)

[0276] Where P N0 The output of nuclear power in the pre-scheduling model is.

[0277] (4) Thermal power unit capacity constraints:

[0278]

[0279]

[0280] Where: P Gi min , P Gi max are the upper and lower limits of the output of conventional unit i.

[0281] (5) Spinning reserve constraints:

[0282]

[0283] Where: P i max is the maximum possible output power of the i-th generator in period t, and its value is determined by the output limit of the generator and the climbing rate of the unit; R t The spinning reserve capacity required for the system during period t is considered to be 7% of the total system load.

[0284] (6) Minimum start and stop time constraints for conventional units:

[0285]

[0286]

[0287] Where: and are respectively the startup duration and shutdown duration of unit i in period t; and are the minimum continuous operating hours and the minimum continuous downtime hours of unit i during the scheduling cycle, respectively.

[0288] (7) Upper and lower limits of active output of the unit:

[0289]

[0290] In the calculation, the parameters k of thermal power units and nuclear power units are i , c v Both are 0.

[0291] (8) Unit climbing rate constraint:

[0292]

[0293] Where, L up.i , L down.i They are respectively the upper limit of the output increase and the upper limit of the output decrease allowed per unit time for unit i.

[0294] 4 Ant Colony Algorithm Optimization of Unit Joint Peak Loading

[0295] The optimization problem of multi-power combined peak load regulation requires some improvement of the model, introducing the concepts of state and decision. State: The combination of peak load regulation measures taken at the current moment constitutes a state. Decision: From the combination of peak load regulation measures in a certain period to the determination of peak load regulation measures in the next period, that is, from one state to the next state, a decision is formed.

[0296] Since in the ant colony optimization algorithm, a closed path is a feasible solution to the problem, the concept of a path is as follows: Path: A set of peak-shaving methods for all time periods in a scheduling cycle. In each time period, any peak-shaving method is selected, so that the peak-shaving output of various power sources in the entire scheduling cycle, that is, the set of all decisions, constitutes a path.

[0297] According to the above concepts and related definitions, the multi-power joint peak-shaving optimization problem can be transformed into a multi-stage dynamic search problem. The goal of peak-shaving selection optimization is to find the minimum value of the path in the decision-making at all times, which is equivalent to the optimal path found in the problem. Due to various constraints, the number of states in each time period is not the same.

[0298] (1) Ant Colony Algorithm Model for Optimization Objectives

[0299] When the ant colony algorithm solves the TSP problem, the path taken by the ants is the result of optimization. The same is true for the problem of joint peak-shaving of multiple power sources. The concept of state and decision-making can be used to solve the problem of joint peak-shaving of multiple power sources in the mode of TSP problem. The combination of peak-shaving means in each time period corresponds to each city in the TSP problem. The peak-shaving cost is the length of the path. The lowest peak-shaving cost is the optimization goal. Then the objective function of the problem of joint peak-shaving of multiple power sources can be converted into:

[0300] J=min∑(tc(s i ,s i+1 )+tc(s i ,s i-1 ))0≤i≤24 (18)

[0301] Where: tc(s i ,s i+1 ) and tc(s i ,s i-1 ) respectively represent the decision costs from state i to state (i+1) and from state to state (i-1), which corresponds to the transfer distance in the TSP problem; state i is selected as the point with the maximum load in 1 day, and the peak-shaving cost of this point is 0, corresponding to the distance in the TSP problem, that is, it has not yet set off; the difference is that in the joint peak-shaving optimization problem, the combination of peak-shaving means to be used next time is constrained by the peak-shaving means used last time, so the tabu table records the adjusted capacity and peak-shaving costs of the different peak-shaving means adopted, which is used to limit the combination of the adjustment means and capacity of the next peak-shaving. After one cycle, the minimum peak-shaving cost in the tabu table corresponds to the shortest path of the TSP problem. Before the next cycle starts, the path information is updated and the tabu table is cleared. The converted objective function can be solved by the ant colony algorithm.

[0302] (2) Transformation of Constraints

[0303] Due to the particularity of nuclear power units, they have a fixed output curve and can only change their peak-shaving depth. Therefore, the nuclear power peak-shaving output is arranged outside the ant colony algorithm cycle, and the optimization result is changed by changing the nuclear power peak-shaving depth.

[0304] 1. Equality constraints

[0305] The power balance constraint constrains the combination of peak-shaving means by calculating the load difference between adjacent time points. The load change between adjacent time points is the sum of the regulation capacity of the selected peak-shaving means, thereby satisfying the power balance constraint. The water balance constraint can be converted into an output limit for other outputs.

[0306] 2. Inequality constraints

[0307] Inequality constraints include upper and lower limit constraints on the output of various power sources, ramp constraints, spinning reserves, minimum start-up and shutdown times of thermal power units, etc. The ultimate goal of various constraints is to constrain the output of the units.

[0308] The adjusted capacity of various peak-shaving means recorded in the tabu table and various inequality constraints are used to jointly control the combination of peak-shaving means that can be used at the next time point. For example, when the unit output exceeds the limit, the maximum or minimum value is taken, and when the climbing rate is not satisfied, the maximum climbing rate of the unit is taken.

[0309] (4) Calculation steps

[0310] ① Input the grid load data, find the time point corresponding to the maximum load, determine the state at that time point (i.e. the output of each type of unit), and iterate to both sides from that point.

[0311] ② Input the nuclear power output model, the peak regulation depth setting is adjustable, and the adjustment amplitude and step length are set.

[0312] ③ Determine the initial state s, which is the time point corresponding to the maximum load point.

[0313] ④ACA initialization: D = 1; τ ij =0;Δτ ij =0; cost(k)=0.

[0314] ⑤Iteration on both sides:

[0315] Iterate forward - t = t-1, determine whether t is the final time period 0, if so, go to ⑧, otherwise go to ⑥.

[0316] Iterate backwards - t = t + 1, determine whether t is the final time period 24, if so, go to ⑧, otherwise go to ⑥.

[0317] ⑥Form the tabu table at the current moment and calculate the peak-shaving cost at this moment as the path length in ACA.

[0318] According to the ant transfer probability, the next state j is selected, that is, the combination of peak-shaving means for the next period, and the load adjustment amount of each type of unit from state i to state j and its corresponding peak-shaving cost are calculated.

[0319] ⑦Continuously accumulate the path length of the ants, i.e. the peak-shaving cost, and go to ⑤.

[0320] ⑧Record the shortest path of ⑦ and update the path information through the algorithm's global update rules.

[0321] ⑨Let: D = D + 1; Δτ ij =0; cost(k)=0 clears the table.

[0322] ⑩If D<DMax , and there is no stagnation, then set t back to the time point corresponding to the maximum load, go to ⑤, and start the next iteration, otherwise stop the iteration and output the optimal solution. Where: cost(k) is the length of the path currently traversed by ant k, that is, the current total peak-shaving cost; D is the number of iterations, D Max is the maximum number of iterations.

Claims

1. A multi-element optimization dispatching method for a power grid to improve wind power acceptance capacity, characterized in that: Step 1: predict wind power through a wind power output model and adjust the system's spinning reserve; In step 1: by establishing a wind power uncertain output calculation method based on error growth, analyzing the relevant concepts of error growth, establishing a wind power output model, taking the wind power prediction error into the wind power output model, and considering the wind power output as an uncertain quantity in combination with the wind power output characteristics, the wind power output is predicted; Step 2: Optimize the output of each unit, realize the joint optimization scheduling method with thermal power as base load and nuclear power and hydrogen storage as auxiliary, realize the equivalent peak regulation of nuclear power, and realize the joint optimization scheduling process of wind-nuclear-storage-thermal; The optimization of the output of each unit described in step 2 means that in the process of wind-nuclear-storage-thermal joint optimization scheduling, when the load demand is met, the wind power output needs to be given priority, and then the thermal power units and hydrogen storage output are arranged; due to the uncertainty of wind power, hydrogen storage stores the wind power that cannot be accepted by the power grid in the form of pumping when wind power is in high demand, and generates electricity during peak load hours to alleviate the peak load pressure during peak load hours; after giving priority to the output of wind power and thermal power units, hydrogen storage is used to reduce the changes in the scheduling of thermal power, among which nuclear power units do not participate in conventional peak-shaving tasks. When the down-peaking capacity is insufficient during the load valley, nuclear power participates in short-term deep peak-shaving through a given determination method; In step 2: when the power grid is in a low period and the peak-shaving capacity is insufficient, the output of nuclear power units is reduced for a short time to provide space for wind power acceptance and achieve equivalent peak-shaving of nuclear power; equivalent peak-shaving depth model: Where: T is the system peak power surplus capacity; D is the system valley down-peak surplus capacity; P LMax and P LMin are the planned daily maximum and minimum power generation loads respectively; P GMax and P GMin are the adjustable maximum and minimum outputs of all thermal power units in the system during the peak and valley load periods on the planned day; P NMax is the rated output of all nuclear power units on the planned day; P R and P R ' are the positive and negative spinning reserve capacities of the system respectively; If D≥0 and T≥0 are satisfied at the same time, the system meets the peak load balance and nuclear power does not need to participate in peak load regulation; if T<0, that is, the peak load capacity is insufficient during the low load period, and nuclear power is required to participate in peak load regulation. The judgment criteria are: P LMin <P GMin +P NMax +P R ' (2); The ant colony algorithm is used to analyze and determine the above equivalent peak-shaving depth model to achieve equivalent peak-shaving of nuclear power. The specific steps are as follows: List the optimization objective function: Where: i is the unit number, t is any time period within the scheduling cycle, f i is the power generation cost of thermal power unit i, is the power generation cost of all thermal power units in the dispatch period T, m0 is the total number of thermal power units; P NiMax is the rated output of nuclear power unit i, The actual output of nuclear power unit i, is the additional cost of nuclear power participating in peak load regulation during the dispatch period T, and m1 is the total number of nuclear power units; is the deep peak load regulation power of thermal power unit i, P' is the additional cost of deep peak load regulation of thermal power units within the dispatch period T; GiMin is the peak load capacity of the minimum output of the thermal power unit, is the peak-shaving cost of thermal power start-up and stop, m2 is the number of thermal power units started and stopped within the dispatch period T, and the start-up and stop peak-shaving effect will act on the entire dispatch period after start-up and stop; is the capacity of hydrogen storage device i, is the cost of hydrogen production within the scheduling period T; C N , C G , C TOf , C Hy They are the peak-shaving costs per unit capacity for various peak-shaving methods.

2. A multi-element optimization dispatching method for a power grid to improve wind power acceptance capacity according to claim 1, characterized in that: The implementation method of predicting wind power in step 1 is as follows: (1) Determine the wind speed law through Weibull distribution and obtain the Weibull distribution of wind speed: In the formula, c and k represent the shape parameter and scale parameter of Weibull distribution respectively, v is the wind speed value; the probability density function is: .

3. A multi-element optimization dispatching method for a power grid to improve wind power acceptance capacity according to claim 2, characterized in that: When the wind speed value is greater than the cut-in wind speed v of the wind turbine ci When the wind speed value maintains the rated wind speed v r When , the wind turbine output remains unchanged; If the wind speed value is greater than or equal to the cut-out wind speed v co When , the wind turbine stops running; the wind power output active power P that obeys the wind speed Weibull distribution can be obtained from the following formula: in, Where P r The power required to maintain the rated wind speed for the wind speed value; 。 4. A multi-element optimization dispatching method for power grid to improve wind power acceptance capacity according to claim 1, characterized in that: Formula (7) needs to simultaneously satisfy the following system power balance constraints, system reserve constraints, nuclear power peak load capacity constraints, thermal power unit capacity constraints, spinning reserve constraints, conventional unit minimum start and stop time constraints, and unit ramp rate constraints. Among the constraints, since the optimization target is the total operating cost within 1 day, the range of t is 1-24h; Constraint equations (1) System power balance equation: Where: The three items on the left are the total output of all thermal power units during period t, the total output of all nuclear power units during period t, and the total output of all hydrogen storage units during period t. is the power demand of the system in period t; (2) System backup constraints: Where: P GiMax is the maximum output of thermal power unit i during the peak period of the power grid, is the actual output of thermal power unit i, It is the overall reserve of the power grid during peak hours. is the overall reserve of the power grid during off-peak hours, P rt , P' rt They are respectively the positive and negative reserve demands of the system at time t, both of which are undertaken by thermal power units; (3) Constraints on the ability to absorb nuclear power peak load: P′ N ≤P NMax -P N0 (11) Where P N0 The output of nuclear power in the pre-dispatch model; (4) Thermal power unit capacity constraints: Where: is the start and stop status of conventional unit i at time t, P i For conventional unit output, P Gimin , P Gimax are the upper and lower limits of the output of conventional unit i; (5) Spinning reserve constraints: Where: N is the total number of units, is the actual output power of the i-th generator, P imax is the maximum possible output power of the i-th generator in period t, and its value is determined by the output limit of the generator and the climbing rate of the unit; R t The spinning reserve capacity required for the system in period t is considered as 7% of the total system load; (6) Minimum start and stop time constraints for conventional units: Where: and are respectively the startup duration and shutdown duration of unit i in period t; and are the minimum continuous operating hours and minimum continuous downtime hours of unit i in the scheduling cycle, respectively; (7) Unit climbing rate constraint: Where P' i,t is the output power of unit i at time t, P' it-1 is the output power of unit i at time t-1, L up.i , L down.i They are respectively the upper limit of the output increase and the upper limit of the output decrease allowed per unit time for unit i.

5. A multi-element optimization dispatching method for power grid to improve wind power acceptance capacity according to claim 4, characterized in that: The ant colony algorithm used is as follows: Introducing the concepts of state and decision State: The combination of peak load-shaving measures taken at the current moment constitutes a state; Decision-making: From the combination of peak-shaving means in a certain period to the determination of the peak-shaving means in the next period, that is, forming a decision from one state to the next state; The concept of closed path to obtain the optimal solution is as follows: Path: The set of peak-shaving methods in all time periods within the dispatching cycle. In each time period, any peak-shaving method is selected, so that the peak-shaving output of various power sources in the entire dispatching cycle, that is, the set of all decisions, constitutes a path; The objective function is: J=minΣ(tc(s i ,s i+1 )+tc(s i ,s i-1 ))0≤i≤24 (18) Where: tc(s i ,s i+1 ) and tc(s i ,s i-1 ) represent the decision costs from state i to state (i+1) and from state i to state (i-1) respectively; state i is selected as the point with the maximum load in 1 day, and the peak-shaving cost of this point is 0, which corresponds to the distance of 0 in the TSP problem, that is, it has not set out yet; the difference is that in the joint peak-shaving optimization problem, the combination of peak-shaving means to be adopted next time is constrained by the peak-shaving means adopted last time. The tabu table records the adjusted capacity and peak-shaving cost of the different peak-shaving means adopted, which is used to limit the combination of adjustment means and capacity for the next peak-shaving. After one cycle, the minimum peak-shaving cost in the tabu table corresponds to the shortest path of the TSP problem. Before the next cycle starts, the path information is updated and the tabu table is cleared; the converted objective function can be solved by the ant colony algorithm.

6. A multi-element optimization dispatching method for power grid to improve wind power acceptance capacity according to claim 5, characterized in that: The calculation steps of the multi-element optimization dispatching method of the power grid to improve the wind power acceptance capacity are as follows: ① Input the grid load data, find the time point corresponding to the maximum load, determine the state at that time point, and iterate to both sides from that time point; ② Input the nuclear power output model, the peak load depth setting is adjustable, and the adjustment amplitude and step length are set; ③ Determine the initial state s, which is the time point corresponding to the maximum load point; ④ACA initialization:D=1;t ij =0;Δτ ij =0;cost(k)=0; ⑤Iteration on both sides: Iterate forward - t = t-1, determine whether t is the final time period 0, if so, go to ⑧, otherwise go to ⑥; Backward iteration - t = t + 1, determine whether t is the final time period 24, if so, go to ⑧, otherwise go to ⑥; ⑥Form the tabu table at the current moment and calculate the peak load cost at this moment as the path length in ACA; According to the ant transfer probability, the next state j is selected, that is, the peak load regulation combination of the next period, and the calculation from state i is The load regulation amount of each type of unit from state j to state j and its corresponding peak regulation cost; ⑦Continuously accumulate the path length of the ants, i.e. the peak-shaving cost, and go to ⑤; ⑧Record the shortest path of ⑦ and update the path information through the algorithm global update rule; ⑨Let: D = D + 1; Δτ ij =0; cost(k)=0 clear table; ⑩If D<D Max , and there is no stagnation, then let t return to the time point corresponding to the maximum load, go to ⑤, and start the next iteration, otherwise stop the iteration and output the optimal solution; where: cost(k) is the length of the path currently traversed by ant k, that is, the current total peak-shaving cost; D is the number of iterations, D Max is the maximum number of iterations.

Images