Flexible load intelligent optimization scheduling method for improving photovoltaic access distribution network economy

Abstract

The application discloses a flexible load intelligent optimization scheduling method for improving the economy of photovoltaic access distribution networks. A multi-type flexible load grading and scheduling optimization model is established. On the one hand, according to the demand side response demand, the flexible load is divided into three categories, namely interruptible load, translatable load and adjustable load, a flexible load operation optimization model is constructed, the reasonable allocation of various flexible loads in different periods is realized, the peak regulation pressure and resource waste problems in the distribution network are effectively alleviated, the purchase of electricity during the power consumption peak can be effectively avoided, and the economy of the distribution network is improved. On the other hand, for the discrete and continuous mixed data distribution environment, a Latin square quantum-inspired genetic algorithm (LSQGA) is used to process the nonlinear optimization problem in the environment, so that the optimal feasible region in the macro data space can be quickly searched, and the feasibility of the optimization model under the extreme scenario is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of flexible load intelligent optimization scheduling, in particular to a flexible load intelligent optimization scheduling method for improving the economic efficiency of photovoltaic access distribution network. BACKGROUND

[0002] In recent years, with the deepening of industrial restructuring and the continuous improvement of people's living standards, the power distribution network presents the trend of large seasonal electricity demand difference, sharp peak, and increasing electricity demand. The obvious shortage of power grid peak shaving and backup resources brings great pressure to power grid dispatching. With the advancement of smart grid construction, the rapid growth of controllable resources at the user end, such as distributed generation, electric vehicles, distributed energy storage, central air conditioning, etc., greatly enriches the types of flexible load, and flexible load scheduling as a supplement to power generation scheduling can shave the peak and fill the valley, balance intermittent energy fluctuations, and provide auxiliary services. Therefore, considering multiple types of flexible load, it is necessary to consider the problem of flexible load participating in scheduling, and it is urgent to design a flexible load intelligent optimization scheduling method for improving the economic efficiency of photovoltaic access distribution network. SUMMARY

[0003] In order to overcome the technical problem that the flexible load cannot be effectively optimized and scheduled at present, the present application provides a flexible load intelligent optimization scheduling method for improving the economic efficiency of photovoltaic access distribution network, which can realize reasonable distribution of load and improve economic efficiency.

[0004] In order to achieve the above technical purpose, the technical scheme of the present application is,

[0005] A flexible load intelligent optimization scheduling method for improving the economic efficiency of photovoltaic access distribution network, comprising the following steps:

[0006] Step 1, a multi-type flexible load model including interruptible load, translatable load and adjustable load is established;

[0007] Step 2, based on demand side management, a coordinated optimization operation model considering flexible load scheduling is established with the comprehensive minimum cost of purchase cost, compensation cost and grid maintenance cost of flexible load as the target;

[0008] Step 3, the constraint conditions satisfying the objective function are set according to the minimum daily load reduction of the region where the power grid is located, the supply-demand balance of the power grid system and the working state of the energy storage system;

[0009] Step 4, the LSQGA method is used to process the nonlinear programming problem under the background of discrete-continuous mixed input data to obtain the scheduling result.

[0010] The method, in step 1, the interruptible load is represented by the following formula:

[0011] The method, in step 1, the interruptible load is represented by the following formula:

[0012] wherein W k is the interrupted load, Δt is the cumulative time of load operation, E is the cumulative power of load operation, p k is the power of W k at the time of operation;

[0013] The translatable load W p is characterized by the following formula:

[0014]

[0015] wherein p p is the operating power, [t min , t max +h] is the allowed operation time, h is the operation time of W p , t p is the start operation time of W p , and K is the Kth time period in the planned time period M;

[0016] The adjustable load P d is characterized by the following formula:

[0017] P d min ≤ P d ≤ P d max

[0018] wherein P d min is the minimum operating power of P d , and P d max is the maximum operating power.

[0019] The method, in step 2, the demand side management comprises: dividing a day into T time sequences, and indexing with t (t = 1, …, T); wherein the duration of each time sequence is fixed; each power-consuming device k has a set of non-movable devices, and is indexed with N k (t) represents the power consumption of the non-movable load of the power-consuming device k in the time sequence t, and N(t) is the total power consumption of the non-movable load in the time sequence t:

[0020]

[0021] The total energy consumption of the time sequence t is obtained from the sum of the non-movable load and the flexible load planned in the time period t, that is:

[0022]

[0023] wherein SG(t) represents the total power consumption of the time sequence t, the planned time period M is divided into K time periods, and the length of each time period is t; W i tN is the operating power of the i-th user load during time period t, and N is the total number of user loads.

[0024]

[0025] Where, p i M is the operating power of load i. i Let i be the allowable operating time for load i.

[0026] In the method described above, step 2, the coordinated optimization operation model considering flexible load scheduling is set as follows:

[0027] min P = P k +P p +P d

[0028]

[0029] Where P is the total cost, P k P p P d These represent the operating costs of three types of flexible loads; P ∈ For any type of flexible load, i.e., ∈ = k, p, d; To account for the electricity purchase cost after considering one type of flexible load; This is the compensation cost for the users under this load; It is the cost of operating and maintaining the power grid.

[0030] The method described above determines the electricity purchase cost based on both the actual load and the real-time unit price of electricity, i.e.:

[0031]

[0032] Among them, Q t S(t) represents the actual load at time t, where T is the total operating time of the load, and S(t) is the electricity purchase price at time t.

[0033] The compensation cost is expressed as:

[0034]

[0035] p(i,t)=x(i,t)Δc(θ) i )q(i,t)

[0036] L(i) = 2.67t1(i) -0.53

[0037] Where N is the number of users; q(i,t) is the load amount of user i at time t before it does not participate in interruptible load management; p(i,t) is the interrupted load amount of user i participating in peak clipping at time t. P is the compensation price of the dispatchable load unit power after shifting; shift，t P is the compensation price of the dispatchable load unit power after shifting; P is the compensation price of the dispatchable load unit power after shifting; A and B are constants greater than 0; L(i) is the unit time loss of user i participating in peak load curtailment; x(i, t) is the interrupt decision variable of day-ahead planning, taking values of 0 or 1, when x(i, t) = 1, user i participates in load interruption at t time; when x(i, t) = 0, user i does not participate in load interruption at t time; α(i, t) and β(i, t) are the shifting, adjustable decision variables of day-ahead planning; θ i is the participation rate of different types of users; Δc(θ i ) is the load shedding rate of user i; t1(i) is the peak load curtailment duration of user i;

[0038] The operation and maintenance cost of the power grid is expressed as:

[0039]

[0040] Wherein, μ is the operation and maintenance cost proportion coefficient of average load; λ is the proportion coefficient of peak load operation and maintenance cost; J is the peak load time of a day; Q j is the actual load quantity at the jth peak time.

[0041] The method, the constraint condition in the step 3 comprises:

[0042] 1) the minimum daily load shedding quantity of the region

[0043] When the power system is in the power consumption peak period, the interruptible load capacity should be greater than or equal to the load shedding requirement of the region, and the expression is as follows:

[0044]

[0045] Wherein, c is the minimum daily load shedding requirement of the region;

[0046] 2) power supply and demand balance constraint

[0047] Considering that the input power of the power system should match the output power, the following system supply and demand balance equation constraint is established, as shown in the following formula:

[0048] P cpp (t) + P gd (t) + P b (t) = P load (t)

[0049] Wherein, P cpp (t) is the output power of the self-provided power plant at time t; P b(t) is the output / input power of the energy storage system at time t; P gd (t) are the unit price of electricity sold by the grid and the electricity transmitted or consumed by the grid at t, respectively; P load (t) is the power consumed by the system load at time t;

[0050] 3) Energy storage system operating state

[0051] Considering that the energy storage system cannot charge and discharge at the same time, we have:

[0052] 0≤|λ(t)|+|υ(t)|≤1

[0053] Where λ(t) and υ(t) are the charging and discharging states of the system at time t.

[0054] The method, step 4 comprises:

[0055] Based on quantum computing, a Q-bit state represents a variable state; a Q-bit variable is represented by a string containing n groups of Q-bit states:

[0056]

[0057] Where |α i | 2 +|β i | 2 =1, i=1, 2, …, N;

[0058] There are p s groups of Q-bit variables, and initialization begins, generating an initial Q-bit population and calculating the fitness value of the population, and using the roulette method for replication operation:

[0059] Single-point crossover operation is performed on each Q-bit, and the crossover rate p c Select Q-bit individuals for crossover operation, if the i-th position is selected as a cut point, then the cut-point crossover operator is used to implement crossover on the Q-bit state system, as follows:

[0060]

[0061] Further orthogonalization test is performed by Latin square method, a new Q-bit individual is selected from the two Q-bit state individuals at each bit point, i.e. factor; at each bit point, the Q-bit is selected according to the corresponding maximum E fl value, i.e. the optimal level of each factor is determined, that is, the level that maximizes E fl value in the test region, and a better offspring is generated according to the signal-to-noise ratio η, i.e. the mean square error of the objective function; and the step is repeated until the number of loops ((1 / 4)×p s ×p c) until a termination condition is met, thereby generating a Q-bit population;

[0062] Further performing a mutation operation on the Q-bit population after the crossover operation: according to a mutation rate p m Selecting a Q-bit for mutation, for a set of Q-bit state systems, the mutation operation is performed by randomly determining a position i, and then exchanging the corresponding alpha i And beta i , the process is as follows:

[0063]

[0064] In addition to the best individual, p s Q-bit individuals are selected for Q-bit repetition operation, finally a new Q-bit population is generated, and a termination condition is checked, if the termination condition is met, the best state individual and the fitness are obtained to obtain the scheduling result, if not, the step of performing the replication operation by using the roulette method is returned to continue iterative execution.

[0065] The signal-to-noise ratio η in the under-quantity and over-quantity situations is respectively:

[0066]

[0067]

[0068] Wherein, {y1, y2, …, y n} represents a set of characteristics, η L is the under-quantity signal-to-noise ratio, and η S is the over-quantity signal-to-noise ratio.

[0069] The method, the termination condition is checked, that is, whether the iteration number reaches a preset number is checked, when the preset number is reached, the termination condition is considered to be reached, or whether the minimum fitness changes with the increase of the iteration number is checked, when the minimum fitness no longer changes, the termination condition is considered to be reached.

[0070] The technical effect of the application is that the application divides flexible load into three categories: interruptible load, translatable load and adjustable load in the scheduling mode, and makes the flexible load participate in the distribution network scheduling under certain constraints. Compared with ordinary load, flexible load has characteristics such as adaptability and flexibility, and the coordinated optimization operation of multiple flexible loads can not only realize the reasonable allocation of load distribution, relieve the peak power supply pressure and improve the smoothness, but also significantly reduce the power purchase cost and improve the economy. At the same time, the application combines the Latin square with the improved model QGA based on the traditional genetic algorithm to integrate into the LSQGA model for solving the mixed nonlinear optimization problem with discrete and continuous data. And through simulation experiments, it is verified that the proposed method has strong optimization ability and fast convergence under the background of discrete-continuous mixed input data, and can effectively improve the optimization operation efficiency and description ability. BRIEF DESCRIPTION OF DRAWINGS

[0071] Figure 1 A flowchart of the present application;

[0072] Figure 2 A flowchart of the LSQGA algorithm;

[0073] Figure 3 A schematic diagram of the power distribution network structure in the embodiment of the present application;

[0074] Figure 4 A photovoltaic output and load curve diagram on March 1 in the embodiment of the present application;

[0075] Figure 5 A schematic diagram of the total operation cost of each calculation method in the embodiment of the present application;

[0076] Figure 6 A total iteration number diagram of convergence in each calculation method in the embodiment of the present application. DETAILED DESCRIPTION

[0077] The present application will be further described below in combination with the drawings and embodiments.

[0078] The steps of the present application include:

[0079] Step 1, establishing a multi-type flexible load model;

[0080] Among them, according to the user demand and the control characteristics of each load, the flexible load is divided into three categories: interruptible load, translatable load and adjustable load.

[0081] The interruptible load is characterized by the following formula:

[0082]

[0083] Among them, W k is the interruptible load, E is the cumulative power of the load operation, p k is the power at the time of operation; k The running time of W

[0084] The translatable load W p is characterized by the following formula:

[0085]

[0086] Among them, p p is the running power, [t min , t max +h] is the allowed running time, h is the running time of W p , t p is the starting running time of W p , and K is the planned time period M divided into the Kth time period.

[0087] Adjustable load P d characterized by the following formula:

[0088] P d min ≤P d ≤P d max

[0089] where P d min is the minimum working power of P d , and P d max is the maximum working power.

[0090] Step 2, based on demand side management, taking the comprehensive minimum cost of the purchase cost of flexible load, compensation cost, and grid maintenance cost as the target, a coordinated optimization operation model considering flexible load scheduling is established.

[0091] Demand side management includes: dividing a day into T time sequences, and taking t (t=1, …, T) as the index; wherein the duration of each time sequence is fixed; each power-consuming device k has a set of non-movable devices, and is denoted as N k (t) represents the power consumption of the non-movable load of power-consuming device k in time sequence t, and N(t) is the total power consumption of the non-movable load in time sequence t:

[0092]

[0093] The total energy consumption of time sequence t is obtained from the sum of the planned non-movable load and flexible load in time period t, that is:

[0094]

[0095] Wherein, SG(t) represents the total power consumption of time sequence t, the planning time period M is divided into K time periods, and the length of each time period is t; W i t is the running power of the i-th user load in time period t, and N is the total number of user loads:

[0096]

[0097] Wherein, p i is the running power of the load i, and M i is the allowed working time of the load i.

[0098] The objective function of the load scheduling optimization model is set as:

[0099] min P=P k +P p +P d

[0100]

[0101] Where, P is the total cost, P k , P p , P d are the operation cost of three kinds of flexible load respectively; P ∈ is the purchase cost of any kind of flexible load, i.e., ∈ = k, p, d; is the purchase cost of considering a kind of flexible load (interruptible, shiftable, adjustable) load; is the compensation cost of the load user; is the operation and maintenance cost of the power grid.

[0102] The purchase cost is determined by the actual load and the real-time unit price of purchase, which can be expressed as:

[0103]

[0104] In the formula, Q t is the actual load at the t time, T is the total operation time of the load, and S(t) is the purchase unit price at the t time, unit: yuan / (kWh).

[0105] The compensation cost can be expressed as:

[0106]

[0107] p(i, t) = x(i, t)Δc(θ i )q(i, t)

[0108] L(i) = 2.67t1(i) -0.53

[0109] In the formula, N is the number of users; q(i, t) is the load amount of user i at the t time before participating in the interruptible load management; p(i, t) is the interruptible load amount of user i participating in peak shaving at the t time; is the compensation price of the unit power of the shiftable load after scheduling; P shift，t is the shiftable load at the t time; is the compensation price of the unit power of the transferable load after responding to the scheduling; A and B are constants greater than 0; L(i) is the unit time loss of user i participating in peak shaving; x(i, t) is the interruptible decision variable (0-1 integer variable) of the day-ahead plan; when x(i, t) = 1, user i participates in load interruption at the t time; when x(i, t) = 0, user i does not participate in load interruption at the t time; similarly, α(i, t) and β(i, t) are the shiftable and adjustable decision variables of the day-ahead plan; θ i is the participation rate of different types of users; Δc(θ i ) is the load reduction rate of user i; t1(i) is the peak shaving duration of user i.

[0110] The operation and maintenance cost of the power grid can be expressed as:

[0111]

[0112] where μ is the average load operation and maintenance cost proportionality coefficient; λ is the peak load operation and maintenance cost proportionality coefficient; J is the peak load time of a day; Q j is the actual load amount at the jth peak time.

[0113] Step 3, establish the constraint condition that satisfies the objective function;

[0114] 1) The minimum daily load reduction amount of the region

[0115] When the power system is in the peak period of electricity consumption, the interruptible load capacity should be greater than or equal to the load reduction requirement of the region. This constraint is also the most basic constraint of interruptible load management, and its expression is as follows:

[0116]

[0117] where c is the minimum daily load reduction requirement of the region.

[0118] 2) Power system supply-demand balance constraint

[0119] Considering that the input power of the power system should match the output power, the following system supply-demand balance equation constraint is established, as shown in the following formula:

[0120] P cpp (t) + P gd (t) + P b (t) = P load (t)

[0121] where P cpp (t) is the output power of the self-provided power plant at time t; P b (t) is the output / input power of the energy storage system at time t; P gd (t) is the unit price of the power sold by the grid and the power transmitted or consumed by the grid at t, respectively; P load (t) is the power consumed by the system load at time t.

[0122] 3) Energy storage system working state

[0123] Considering the actual situation, the energy storage system cannot charge and discharge at the same time, and therefore must satisfy the following equation.

[0124] 0 ≤ |λ(t)| + |υ(t)| ≤ 1

[0125] In the formula, λ(t) and υ(t) are the charging and discharging states of the system at time t.

[0126] Step 4: Construct the Latin square quantum-inspired genetic algorithm (LSQGA) method to address nonlinear programming problems in the context of discrete-continuous mixed input data;

[0127] In this step, based on the concept of quantum computing, a variable state can be represented by a single Q-bit state. This representation method can generate all possible solutions using linear superposition. Based on the Q-bit representation, a Q-bit variable can be represented by a string containing n sets of Q-bit states, and its mathematical expression is as follows:

[0128]

[0129] Where, |α i | 2 +|β i | 2 =1, i=1,2,…,N. Since the Q-bit state can simultaneously represent the linear superposition of different states, the Q-bit representation can ensure better population diversity compared with the representation used in other evolutionary methods.

[0130] Suppose there exists p s Set up a Q-bit variable and begin initialization to generate an initial Q-bit population. Calculate the fitness value of this population and perform the replication operation using the roulette wheel method.

[0131] Perform a single-point crossover operation on each Q bit, based on the crossover rate p. c The Q-position individuals are selected for the crossover operation. If the i-th position is selected as a tangent point, the tangent point crossover operator is used to crossover the Q-position state system, as shown below:

[0132]

[0133] Further orthogonalization experiments were conducted using the Latin square method to select a new Q-position individual from the two Q-position individuals at each locus (factor). At each locus, the individual was selected based on the largest corresponding E... fl The amplitude is selected at the Q level, which determines the optimal level for each factor, i.e., maximizing E in the experimental region. fl The value level is determined, and better offspring are generated based on the mean square error of the objective function (signal-to-noise ratio η). This step is repeated until the number of loops ((1 / 4)×p) is satisfied. s ×p c This process continues until a Q-bit population is generated. For both over- and under-supplied cases, the Taguchi model defines the signal-to-noise ratio η as follows;

[0134]

[0135]

[0136] where {y1, y2, …, y n} represents a set of features.

[0137] Further, the mutation operation is performed in the Q-bit population after the crossover operation. According to the mutation rate p m Q-bits are selected for mutation, and for a set of Q-bit state systems, the mutation operation is performed by randomly determining a position (for example, position i), and then exchanging the corresponding a i and β i , and the process is as follows:

[0138]

[0139] In addition to the best individual, p s Q-bit individuals are selected for Q-bit repetition operation, and finally a new Q-bit population is generated, and the termination condition is checked. If the termination condition is met, the best state individual and the fitness are obtained, and if the termination condition is not met, the above steps are repeated.

[0140] Embodiment

[0141] Through a 10kV 11-node small distribution network with photovoltaic access Figure 3 As an example system, simulation analysis is performed, and data of a certain day in March is taken as an example. The 24-hour photovoltaic output curve and daily load curve of the day are as shown in Figure 4 , wherein the unit is kW;

[0142] The constants and are 0.0021 and 0.65, respectively; is 0.8 yuan / (kW·h); is 0.1 yuan / (kW·h); μ is the average load operation and maintenance cost proportion coefficient, which is 0.03 yuan / (kW·h); λ is the peak load operation and maintenance cost proportion coefficient, which is 0.54 yuan / (kW·h); the grid company power purchase price S(t) is determined according to the electricity time price, the peak period is 10:00-14:00 and 17:00-22:00, the average price is 0.40 yuan / (kW·h), the valley period is 1:00-8:00, the average price is 0.15 yuan / (kW·h), and the average price of other periods is 0.30 yuan / (kW·h);

[0143] According to the LSQGA model optimization, the simulation system entries are as follows:

[0144] Table 1 Comparison of system cost optimization effects

[0145]

[0146] From Table 1, it can be seen that, under the condition of total power consumption on the same day, the flexible load is involved in the dispatching optimization, which can effectively avoid purchasing electricity during the peak period. In terms of operation cost, the comprehensive compensation cost and operation cost are reduced by 12.6% compared with the non-classification mode, which can achieve the benefits of energy saving and economy.

[0147] The effectiveness of the proposed method is verified. For the same example scenario as above, the LSQGA optimization model proposed is compared with the standard QGA model, the neural network example algorithm BPNN, and the particle swarm theoretical example algorithm PSO. The comparison is mainly for the optimization effect and optimization efficiency, and the results are shown in Figure 5 、 6 ;

[0148] As can be seen from Figure 5 , the final results of the four optimization models are not much different, demonstrating that they all have certain optimization ability in small-scale system scenarios. However, the total operation cost of the LSQGA model optimization scheme is still relatively low, and the optimization in subsequent potential large-scale application scenarios will be more significant.

[0149] In Figure 6 , compared with QGA, BPNN, and PSO models, it can be seen that the LSQGA has the smallest number of convergence iterations, and its optimization efficiency is higher, further demonstrating that it can better adapt to discrete-continuous mixed data environment, thereby forming a certain advantage in optimization efficiency.

Claims

1. A flexible load intelligent optimization scheduling method for improving the economy of photovoltaic access distribution network, characterized in that, The method comprises the following steps: Step 1, a multi-type flexible load model including three types of interruptible load, shiftable load and adjustable load is established; Step 2, based on demand side management, a coordinated optimization operation model considering flexible load scheduling is established with the minimum comprehensive cost of electricity purchasing cost, compensation cost and grid maintenance cost of the flexible load as the target; Step 3, constraint conditions satisfying the objective function are set according to the minimum daily load shedding amount of the region where the power grid is located, power system supply-demand balance and energy storage system working state; Step 4, a Latin square quantum genetic algorithm (LSQGA) is constructed by using single-cut-point crossover operation and Latin square orthogonal test to process nonlinear programming problems under the background of discrete-continuous mixed input data and obtain scheduling results; The LSQGA comprises: Based on the concept of quantum computing, a variable state is represented by a Q-bit state, and a Q-bit population is initialized by a Q-bit state initialization. Based on the concept of quantum computing, a variable state is represented by a Q-bit state, and a Q-bit population is initialized by a According to the cross rate Selecting Q-bit individuals for the single-cutpoint crossover operation; A new Q-bit individual is selected from two Q-bit state individuals at each site through the Latin square orthogonal test, and a better Q-bit individual is generated according to the signal-to-noise ratio; A mutation operation is performed on the Q-bit population after the crossover operation according to the mutation rate, and the Q-bit population is updated; The iteration is repeated until the preset termination condition is met.

2. The method of claim 1, wherein, In step 1, the interruptible load is represented by the following formula: ； wherein is the interrupted load, is the cumulative time of load operation, E is the cumulative power of load operation, is power at runtime; Translatable load By the following formula: ； wherein is the running power, is the allowed running time, h is is the running time, is the is the start running time, is the planned time period is divided into the time period; Adjustable load characterized by the formula: ； wherein is minimum operating power, is maximum operating power.

3. The method of claim 1, wherein, The step 2 includes: dividing a day into a plurality of time sequences, and indexing by ( ); wherein the duration of each time sequence is fixed; each power-consuming device has a set of non-movable loads, and represents the power consumption of the non-movable loads of the power-consuming device in the time sequence is the total power consumption of the non-movable loads in the time sequence .​​​ ； Timing The total energy consumption of the time period is obtained from the sum of the planned non-shiftable and flexible loads, i.e.: ； wherein, total power consumption of the time sequence is planned for the time period is divided into time periods, each time period has a length of ; is the operating power of the th user load in the time period , is the total number of user loads: ； wherein, the operating power of the load , the allowed operating time of the load .

4. The method of claim 2, wherein, In step 2, the coordinated optimization operation model considering flexible load scheduling is set as: ； ； wherein, is the total cost, , , are the operating costs of three kinds of flexible loads, respectively; is any one of the flexible loads, i.e. ; is the electricity purchase cost considering one of the flexible loads; is the compensation cost of the load user; is the operation and maintenance cost of the power grid.

5. The method of claim 4, wherein, The electricity purchasing cost is determined by the actual load and the real-time unit price of electricity purchasing, that is: ； wherein, is the actual load at the moment, is the total operating time of the load, is the purchase price of electricity at the moment; The compensation cost is represented as: ； ； ； in, It refers to the number of users; User Before participating in interruptible load management Load at any given moment; User In the The amount of interrupted load that participates in peak clipping at all times; The compensation price per unit power of load that can be shifted after dispatching; for The moment-to-moveable load; The compensation price per unit power of load that can be transferred after dispatch; and It is a constant greater than 0; User The unit time loss involved in peak clipping; This is the decision variable for the planned interruption, taking a value of 0 or 1. At that time, the user exist Participate in load interruptions at all times; when At that time, the user exist Never participate in load interruption; and These are the shifts and adjustable decision variables of the current plan; It is the participation rate of different types of users; ( ) is the user The load reduction rate; User Peak clipping duration; The operation and maintenance cost of the power grid is represented as: ； wherein, is the average load ratio coefficient of operation and maintenance cost; is the peak load ratio coefficient of operation and maintenance cost; is the peak load time of a day; is the actual load amount of the th peak time.

6. The method of claim 5, wherein, In step 3, the constraint conditions include: 1) The minimum daily load shedding amount of the region When the power system is in the peak power consumption period, the interruptible load capacity should be greater than or equal to the load shedding requirement of the region, expressed as: ； Where c is the minimum daily load shedding demand of the region; 2) Power system supply-demand balance constraint Considering that the input power of the power system should match the output power, the following system supply-demand balance equation constraint is established, as shown in the following formula: ； in, Is it a self-owned power plant in time? ; output power; It is time Output / input power of the energy storage system; These are the unit price of electricity sold by the power grid and the power grid's... The amount of electricity transmitted or consumed during the process; The system load in time Power consumed; 3) Energy storage system working state Considering that the energy storage system cannot charge and discharge at the same time, we have: ； wherein and is the time the state of charge and discharge of the system.

7. The method of claim 1, wherein, Step 4 comprises: Based on quantum computing, a variable state is represented by a state of a Q-bit; a variable of a Q-bit is represented by a string of Q-bits String representation of a group of Q-bit states: ； wherein , ; in the presence of The Q-bit variable is set and initialization begins, generating an initial Q-bit population and calculating the fitness values of the population. Reproduction is performed using the roulette method: Single-point crossover is performed on each Q-bit according to the crossover rate Q-bit individuals are selected for crossover. If the jth position is selected as a crossover point, then the Q-bit state system is crossed over using the single-point crossover operator as follows: Q-bit individuals are selected for crossover. If the jth position is selected as a crossover point, then the Q-bit state system is crossed over using the single-point crossover operator as follows: ； Further orthogonalization experiments were conducted using the Latin square method to select a new Q-position individual from the two Q-position individuals at each locus (i.e., the factor). At each locus, the individual was selected based on the largest corresponding value. The amplitude is selected at the Q-bit, which determines the optimal level for each factor, i.e., maximizing it within the experimental region. The level of the value, and based on the signal-to-noise ratio. That is, the mean squared error of the objective function generates better offspring; and this step is repeated until the number of loops is satisfied. This process continues until a Q-bit population is generated. Further, a mutation operation is performed on the Q-bit population after the crossover operation: according to a mutation rate Q-bits are selected for mutation, and for a set of Q-bit states, the mutation operation is performed by randomly determining a position and then swapping the corresponding and The process is as follows: ； Besides the best individual, select Each Q-bit individual performs a Q-bit repeated operation, eventually generating a new Q-bit population. The termination condition is then checked. If the termination condition is met, the optimal state individual and fitness are solved to obtain the scheduling result. If the condition is not met, the process returns to the step of using the roulette wheel method for replication and continues iterative execution.

8. The method of claim 7, wherein, Signal-to-noise ratio In both under- and over-dosing situations, respectively: ； ； wherein, represents a set of features, is an under-dose signal-to-noise ratio, is an over-dose signal-to-noise ratio.

9. The method of claim 7, wherein, The termination condition is that the number of iterations is checked whether it reaches the preset number, when it reaches the preset number, it is considered that the termination condition is reached; or the minimum fitness is checked whether it changes with the increase of the number of iterations, when it no longer changes, it is considered that the termination condition is reached.

Images