Industrial park load aggregation optimization control method based on improved star-sparrow optimization algorithm

Abstract

The invention provides an industrial park load aggregation optimization control method based on an improved star-sparrow optimization algorithm. According to the method, common benefits of a load aggregator and a high-energy-consumption factory are considered, an optimized objective function is constructed, and meanwhile, an improved star-sparrow optimization algorithm is used for solving. The optimization algorithm is improved by adopting Logistic chaotic mapping, Gaussian perturbation and sine and cosine optimization algorithms, and the search capability of the algorithm and the escape capability of a local optimal solution are improved. The method introduces solving steps in detail, and comprises the steps of population initialization, disturbance updating and the like. The method is suitable for a large-scale industrial park, electric energy optimal configuration can be achieved, and the economic benefits of the industrial park are improved.

Description

Technical Field

[0001] The present invention belongs to the field of power grids, and more specifically, relates to a method for optimizing the load aggregation control in industrial parks based on an improved starling optimization algorithm. Background Art

[0002] The load optimization control in industrial parks is an important link in energy management. Its goal is to reduce energy costs, improve energy utilization efficiency, and reduce the burden on the power grid through scientific and reasonable load regulation. Load aggregation refers to integrating multiple dispersed load resources into a whole to more effectively conduct energy scheduling and management. Load optimization control refers to, according to various factors such as real-time electricity prices, electricity demand, and the operating conditions of energy equipment, realizing the dynamic regulation and distribution of loads through optimization algorithms to achieve the purpose of reducing energy consumption costs and improving the operating stability of the system.

[0003] Existing load optimization control methods mainly include mathematical optimization models, machine learning algorithms, and heuristic optimization algorithms, etc. Among them, the method based on a mathematical optimization model optimizes the load control by establishing an accurate load model and objective function and using algorithms such as linear programming and dynamic programming; the method based on machine learning generates load prediction and control strategies through a data-driven approach using a large amount of load historical data; while the method based on heuristic algorithms, such as particle swarm optimization, genetic algorithms, and ant colony algorithms, etc., utilize their global optimization ability and flexibility and play an important role in solving complex load optimization problems.

[0004] However, due to the large number of industries in industrial parks, the load characteristics of each production process are different, and the load has characteristics such as non-linearity, high dimension, and dynamic change, there are still certain limitations in the solution efficiency and optimization accuracy of existing technologies. Therefore, further research on load aggregation optimization control methods and improved optimization algorithms for improving the effect of load aggregation optimization in industrial parks has important practical significance and application value. Summary of the Invention

[0005] To improve the efficiency and accuracy of load optimization control in industrial parks, the present invention proposes a method for optimizing the load aggregation control in industrial parks based on an improved starling optimization algorithm to solve the technical problems existing in the current load management in industrial parks. The present invention can effectively respond to the dynamic changes in the load demand of industrial parks, improve the accuracy of load scheduling, reduce energy consumption costs, and at the same time maximize the energy utilization efficiency. Further promote the intelligent development of energy management in industrial parks, and promote the optimal allocation of regional energy resources and the development of green and low-carbon industries.

[0006] To achieve the above object, the present invention is implemented by adopting the following technical solutions: The method includes:

[0007] Considering the interests of the load aggregator side, comprehensively considering factors such as time-of-use electricity price, day-ahead winning bid volume, and demand response resource income, construct the objective function on the LA side;

[0008] Considering the interests of the high-energy-consuming factory side, obtain the objective function on the factory side, and construct the load aggregation scheduling optimization model for the industrial park;

[0009] Solve the load aggregation scheduling optimization model for the industrial park.

[0010] In one solution, the construction of the objective function on the LA side includes:

[0011] Firstly, define the characteristic vectors of three types of industrial loads for the process;

[0012] Determine the time-of-use electricity price in the park according to user requirements;

[0013] Calculate the load response resource income of the LA;

[0014] Calculate the demand response subsidy paid by the LA to the factory;

[0015] Calculate the cost on the LA side, which consists of the system load shedding penalty cost and the demand response invocation cost.

[0016] In one solution, the construction of the load aggregation scheduling optimization model for the industrial park includes:

[0017] Calculate the total economic benefit of the industrial park;

[0018] Calculate the cost of the industrial park's electricity interaction with the power grid;

[0019] Obtain the objective function on the high-energy-consuming factory side;

[0020] Introduce two weights to weighted combine the two objective functions into one objective function, and obtain the total objective function of the load aggregation scheduling optimization model for the industrial park.

[0021] In one solution, use the starling optimization algorithm improved by Logistic chaos mapping, Gaussian perturbation, and sine-cosine optimization algorithm for solving.

[0022] In one solution, the solving steps are as follows:

[0023] Initialize the algorithm parameters, and obtain the initial population based on the Logistic chaos mapping mechanism;

[0024] Generate a random number between 0 and 1 and;

[0025] Set two reference points;

[0026] Set a random number between [0,1];

[0027] The sine-cosine algorithm is fused to perturb and update individuals;

[0028] The position of individuals is perturbed and updated based on Gaussian perturbation;

[0029] Check whether the new position is within the legal search space range;

[0030] Output the optimal solution and the optimal value, and the algorithm ends.

[0031] In a solution, the industrial loads are divided into three categories: shiftable, transferable, and curtailable. Their characteristic vectors are defined respectively, including the start and end times before and after shifting, the start and end times and load amounts before and after transfer, and the start and end times and load amounts before and after curtailment.

[0032] In a solution, the improved starling optimization algorithm perturbs and updates 30% of the individuals with the lowest fitness through the Gaussian perturbation mechanism.

[0033] In a solution, the legality of the new position is determined by comparing the new position with the upper and lower bounds of the search space. If it exceeds the boundary, it is reflected back into the boundary.

[0034] Advantages of the present invention:

[0035] 1) The industrial park load aggregation optimization control method based on the improved starling optimization algorithm of the present invention can take into account the interests of load aggregators and high-energy-consuming factories, making up for the deficiency of traditional methods that only consider the energy-saving effect of the power grid and ignore the energy-saving benefits of users, and improving the economic benefits of industrial parks.

[0036] 2) The present invention innovatively introduces the Logistic chaotic mapping, Gaussian perturbation, and sine-cosine optimization algorithm to improve the starling optimization algorithm, enhancing the global search ability of the algorithm and the ability to jump out of local optima, and improving the solution effect.

[0037] 3) The present invention provides a new load aggregation scheduling control method, which can not only guide users to participate in load response, relieve the load pressure of the power grid, realize the optimal allocation of electric energy, but also maximize the benefits of users' energy use. Description of the Drawings

[0038] Figure 1 It is the flow chart of the method of the present invention;

[0039] Figure 2 It is the flow chart of the improved starling optimization algorithm of the present invention. Detailed Embodiment

[0040] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Typical embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0041] Unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Typical embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0042] The present invention provides an optimized control method for industrial park load aggregation based on an improved starling optimization algorithm, which includes:

[0043] As Figure 1 shown, Step 1: First, consider the interests of the Load Aggregator (LA) side, and comprehensively consider factors such as time-of-use electricity price, day-ahead winning bid volume, demand response resource income, etc., to construct the LA-side objective function.

[0044] Step 1.1: Considering the diverse types of industries in the industrial park and the different production processes of different industries, first define the characteristic vectors of three types of industrial loads for the processes:

[0045]

[0046] Among them, S Shift 、S Div and S Cut represent the characteristic vectors of shiftable, transferable, and curtailable load types respectively, and m, q, and r represent the time sets of shiftable, transferable, and curtailable load types respectively. Among them and represent the start and end times of the load before shifting respectively, and represent those after shifting. P t Shift represents the load at each time, which remains unchanged before and after shifting. Among them and represent the start and end times of the transferable load respectively, P t DivO and P tDiv respectively represent the load amounts at each time before and after the transfer, which represents the total load amount. Among them, and respectively represent the start and end times of the load that can be reduced, represents the total load amount before the reduction, represents the total load amount after the reduction, P t CutO and P t Cut respectively represent the initial load value and the load value after the reduction.

[0047] Step 1.2: Determine the time-of-use electricity price Price in the park according to the user's needs t , as shown in Equation (4).

[0048]

[0049] where N m is the number of users of the mth type of load, is the electricity price in the electricity spot market at time t, is the basic incentive price of the mth type of load at time t; represents the load reduction amount of the high-energy-consuming factory i at time t.

[0050] Step 1.3: Calculate the revenue of the load response resources of LA, as shown in Equation (7).

[0051]

[0052] where P t Contract represents the day-ahead winning bid electricity quantity, Price re represents the declared electricity price during the power shortage period, which is agreed upon by the contract between LA and the electricity market.

[0053] Step 1.4: Calculate the demand response subsidy paid by LA to the factory, as shown in Equation (8).

[0054]

[0055] It is assumed that the adjustment period of the shiftable load is 3 hours. Therefore, 24 hours a day is equally divided into 8 segments of unit shiftable load, G = {1, 2,..., 8}. Among them represents the start time after the shift of the jth segment of the shiftable load of factory i, then represents the time before the shift. ψ Shift represents the subsidy standard coefficient of the shiftable load. and respectively represent the load amounts of factory i before and after the transfer at time t. ψ DivThe subsidy standard coefficient representing the transferable load. and respectively represent the load amounts before and after curtailment of factory i at time t. ψ Cut The subsidy standard coefficient representing the curtailable load.

[0056] Step 1.5: Calculate the cost on the LA side, which consists of the system load shedding penalty cost and the demand response invocation cost, as shown in Equation (12).

[0057]

[0058] where N s is the number of scenarios; is the unit penalty coefficient for load shedding; is the load shedding amount of the system at time t in scenario s. C s is the incentive-based response capacity cost; D d,s is the invocation capacity of the incentive-based response in scenario s; N e is the number of user economic compensations; C d,e is the unit cost coefficient for load reduction of the aggregator on segment e. q d,e,t,s is the load reduction power of aggregator d on segment e at the corresponding time in the scenario, p t,s is the probability of different scenarios.

[0059] Step 1.6: Obtain the LA objective function, as shown in Equation (13).

[0060] maxF LA = F Income - F Sup - F Cost (13)

[0061] Step Two: Further consider the interests of the high-energy-consuming factories, obtain the factory-side objective function, and construct the load aggregation and dispatching optimization model for the industrial park.

[0062] Step 2.1: Calculate the total economic benefit of the industrial park, as shown in Equation (14).

[0063]

[0064] where Q i represents the total electricity consumption of factory i. KHV i represents the per-kWh output value of the products produced by factory i.

[0065] Step 2.2: Calculate the cost of the industrial park's electricity interaction with the power grid, as shown in Equation (16).

[0066]

[0067] where Pt is the power purchased from the power grid at time t; C pv is the feed-in tariff for PV power; P togrid,t is the power of PV power fed into the grid at time t.

[0068] Step 2.3. Thus, the objective function on the side of the high-energy-consuming factory is obtained, as shown in Equation (17).

[0069] maxF Factory = F Benefit + F Sup - F Grid (17)

[0070] Step 2.4. By introducing two weights ω1 and ω2, the two objective functions are weighted and combined into one objective function to facilitate the solution of the model, and thus the total objective function of the industrial park load aggregation scheduling optimization model is obtained:

[0071] maxf = ω1F LA + ω2F Factory (18)

[0072] As Figure 2 shown, in Step 3, for the industrial park load aggregation scheduling optimization model, a starling optimization algorithm improved based on Logistic chaotic mapping, Gaussian perturbation, and sine-cosine optimization algorithm is designed for solution. The specific steps are as follows:

[0073] Step 3.1. Initialize the algorithm parameters and obtain the initial population based on the Logistic chaotic mapping mechanism.

[0074] The Logistic chaotic mapping is a chaotic mapping technology with advantages such as strong randomness, good ergodicity, and sensitivity to initial values. It can effectively improve the diversity of the initial solutions in the optimization algorithm, thereby improving the optimization performance of the algorithm. The steps for initializing the population based on the Logistic chaotic mapping are as follows:

[0075] For each dimension j (j = 1, 2,..., D), select an initial value x0 between (0, 1). Generate a series of chaotic values through the following formula.

[0076] x i+1 = μx i (1 - x i ), i = 1, 2,..., N (19)

[0077] Continue to initialize each individual as follows:

[0078] X i,j = (ub j - lb j )·x i + lbj , where \(i = 1, 2, \ldots, N\) and \(j = 1, 2, \ldots, D\) (20)

[0079] where \(N\) represents the population size and \(D\) represents the individual dimension. \(X\) i,j represents the \(j\)-th dimensional vector of individual \(i\), \(ub\) j and \(lb\) j represent the upper and lower bounds of the \(j\)-th dimensional vector respectively.

[0080] Step 3.2: Generate random numbers \(\sigma\) and \(\sigma_1\) between 0 and 1. If \(\sigma < \sigma_1\), go to Step 3.3; otherwise, go to Step 3.4.

[0081] Step 3.3: Set is a random number between \([0, 1]\), \(P\) a1 linearly decreases from 1 to 0 as the number of iterations increases. If is greater than \(P\) a1 then apply formula (21); otherwise, apply formula (23).

[0082]

[0083] where \(\gamma\) is a random number generated according to the Levy flight function, is the optimal individual of the current population, \(A\), \(B\), \(C\) are three different individuals randomly selected from the population, is the average position of all individuals, \(\tau_1\), \(\tau_2\), \(\tau_3\), \(r\), \(r_1\) are all random numbers between \([0, 1]\), \(\tau_4\) is a random number subject to a normal distribution, \(\tau_5\) is a random number generated by Levy flight, \(t\) is the current number of iterations, and \(\delta\) is set to 0.05. Go to Step 3.6.

[0084] Step 3.4: Set two reference points. The calculation methods of the two reference points are as follows:

[0085]

[0086]

[0087] where \(\theta\) is a random radian between \([0, \pi]\), \(A\), \(B\) are two different individuals randomly selected from the population, \(r_1\), \(r_2\), \(\tau_3\) are random vectors between \([0, 1]\), and \(T\) is the maximum number of iterations.

[0088] Step 3.5: Set \(\varphi\) as a random number between \([0, 1]\), \(P\) a2 = 0.4. If \(\varphi\) is greater than \(P\) a2 then apply formula (30); otherwise, apply formula (33).

[0089]

[0090] In formulas (27) to (29), C is an individual randomly selected from the population, and r1, r2, τ i (i = 3, 4, 5, 6, 7, 8) are all random numbers between [0, 1].

[0091]

[0092] Step 3.6: Use the sine cosine algorithm (SCA) to perturb and update the individuals.

[0093] There is a problem that the local search of the starling optimization algorithm is not fine enough during the search process. As a result, when the algorithm is in the region close to the local optimal solution, it cannot quickly and accurately locate the optimal solution. In response to this, the sine cosine algorithm (SCA) is introduced to perform a secondary update on the individuals to improve the local optimization ability of the algorithm. The update formula based on SCA is shown in formulas (34 - 35).

[0094]

[0095] where a and b are algorithm constants. r3 is a random number between [0, 1]. Let the current individual position be X i , and the optimal individual position be X best , X j and X k are two positions randomly selected from the population.

[0096] Step 3.7: Perturb and update the individual positions based on Gaussian perturbation.

[0097] When the starling optimization algorithm deals with large-scale problems, the convergence speed is slow. Gaussian perturbation can accelerate the convergence speed of the algorithm and find a better solution faster by searching near the current solution. Therefore, this model introduces a Gaussian perturbation operator to perform a secondary update on the individuals to further improve the global exploration ability of the algorithm at this stage.

[0098] For the 30% of individuals with the lowest fitness, set the Gaussian walk probability p gw , for each individual, generate a random number r gw between [0, 1]. If r gw < p gw , then the current position is a D-dimensional vector (x1, x2,..., x D ). Independently draw a random number from a Gaussian distribution with a mean of 0 and a standard deviation of 10 as the perturbation value. The drawn perturbation vector is (δ1, δ2,..., δ D ), where δ i ~N(0, σ). Update the individual position to (x1 + δ1, x2 + δ2,..., x D+δ D )。

[0099] Step 3.8. Check whether the new position is within the legal search space range.

[0100] For each dimension j, compare the j-th dimensional component of the new position with the upper bound U j and the lower bound L j to determine their magnitude relationship: If x j +δ j > U j , reflect the position beyond the boundary back inside the boundary, and adjust the j-th dimensional component of the new position to 2U j -(x j +δ j ); If x j +δ j < L j , adjust the j-th dimensional component of the new position to 2L j -(x j +δ j ).

[0101] Step 3.9. Set the current iteration number t = t + 1.

[0102] Step 3.10. Determine whether t is greater than the maximum iteration parameter T. If so, go to Step 3.11; otherwise, go to Step 3.2.

[0103] Step 3.11. Output the optimal solution and the optimal value, and the algorithm ends.

[0104] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0105] It should be understood that the detailed description of the technical solutions of the present invention with the aid of the preferred embodiments above is illustrative rather than restrictive. Those of ordinary skill in the art can modify the technical solutions recorded in each embodiment on the basis of reading the specification of the present invention, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. An industrial park load aggregation optimization control method based on an improved star-sparrow optimization algorithm, characterized in that: The method includes: Considering the interests of the load aggregator, the objective function of the LA side is constructed by comprehensively considering the time-of-use electricity price, the day-ahead bidding volume, and the demand response resource revenue factors; Considering the interests of high-energy-consuming factories, the objective function of the factories is obtained, and the load aggregation scheduling optimization model of industrial parks is constructed; The optimization model of load aggregation scheduling in industrial parks is solved.

2. According to claim 1, an industrial park load aggregation optimization control method based on an improved star-sparrow optimization algorithm is characterized by: The construction of the LA side objective function includes: firstly, defining the characteristic vectors of three types of industrial loads in the process; Determine the time-of-use electricity price within the park based on user demand; Calculate LA's load response resource revenue; Calculate the demand response subsidy paid by LA to the plant; Calculate the LA side cost, which consists of the system load shedding penalty cost and the demand response call cost.

3. The industrial park load aggregation optimization control method based on the improved star-sparrow optimization algorithm according to claim 1 is characterized by: The construction of the industrial park load aggregation scheduling optimization model includes: Calculate the total economic benefits of the industrial park; Calculate the cost of industrial parks interacting with the power grid; Obtain the objective function of the high energy consumption factory side; Two weights are introduced to weight the two objective functions into one objective function, and the total objective function of the load aggregation scheduling optimization model of the industrial park is obtained.

4. The method for optimizing the load aggregation of industrial parks based on the improved star-sparrow optimization algorithm according to claim 1 is characterized in that: The solution is solved by using the improved starfinch optimization algorithm based on Logistic chaotic mapping, Gaussian perturbation and sine-cosine optimization algorithm.

5. The method for optimizing the control of industrial park load aggregation based on the improved star-sparrow optimization algorithm according to claim 4 is characterized by: The steps of solving are as follows: Initialize the algorithm parameters and obtain the initialized population based on the Logistic chaotic mapping mechanism; Generate a random number between 0 and 1; Set two reference points; Set a random number between [0,1]; The sine-cosine algorithm is integrated to perform disturbance updates on individuals; Perform disturbance updates on individual positions based on Gaussian perturbations; Check whether the new position is within the legal search space; Output the optimal solution and optimal value, and the algorithm ends.

6. The method for optimizing the control of industrial park load aggregation based on the improved star-sparrow optimization algorithm according to claim 2 is characterized by: The industrial loads are divided into three categories: movable, transferable and reducible, and their characteristic vectors are defined respectively. It includes the start and end time before and after translation, the start and end time and load before and after transfer, and the start and end time and load before and after reduction.

7. The method for optimizing the control of industrial park load aggregation based on the improved star-sparrow optimization algorithm according to claim 4 is characterized by: The improved starfinch optimization algorithm performs disturbance updates on 30% of individuals with the lowest fitness through a Gaussian perturbation mechanism.

8. The method for optimizing the control of industrial park load aggregation based on the improved star-sparrow optimization algorithm according to claim 5 is characterized by: The legitimacy of the new position is determined by comparing the new position with the upper and lower bounds of the search space, and if it exceeds the bounds, it is reflected back into the bounds.

Images