Multi-objective safety and economic optimization dispatch method for multi-energy system considering demand response
Through the multi-objective evolution algorithm based on decomposition and particle swarm optimization algorithm, combined with the Chebishev method and adaptive weight adjustment, the problems of power generation cost, carbon emissions and power transmission safety optimization under demand response in multi-energy systems are solved, and the safety optimization of power generation costs, carbon emissions and power transmission in multi-energy systems is achieved. The optimal configuration of power generation costs and carbon emission reductions and power transmission safety.
Patent Information
- Application Number
- CN202210685305.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-17
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2042-06-17
AI Technical Summary
When considering demand response, how to achieve optimal configuration of power generation costs, carbon emissions and power transmission safety in multi-energy systems to solve the challenges of peak-to-valley pressure regulation and power transmission safety.
Using a multi-objective evolution algorithm based on decomposition, the multi-objective model is decomposed into a single-objective subproblem through particle swarm optimization algorithm and Chebischev method, and an adaptive weight adjustment and random trajectory correction of Cauchy-Lorentz distribution are introduced to ensure that The algorithm balances the global and local searches, avoids premature convergence, and obtains the Pareto optimal solution set.
It realizes the optimal configuration to reduce power generation costs, low carbon emissions and transmission line safety, provides a decision-making solution with lower computing complexity, and enhances the flexibility and controllability of the system.
Smart Images

Figure FDA0004940099260000011 
Figure FDA0004940099260000012 
Figure FDA0004940099260000013
Abstract
Description
Technical Field
[0001] The present invention relates to the field of power system automation, and more specifically to a multi-objective safety and economic optimization scheduling method for a multi-energy system taking demand response into consideration. Background Art
[0002] The greenhouse effect limits the sustainable development of mankind, so carbon emissions have received widespread attention around the world.
[0003] Wind energy has the advantage of being pollution-free and has become one of the most prominent renewable energy sources for solving emission problems. However, it should be noted that the daily output of wind power generation is uncertain due to the influence of wind profiles and may sometimes have anti-peak characteristics. Therefore, high wind penetration in the power system may cause pressure and adaptation problems for peak-valley regulation, and also pose a great challenge to the safety of power transmission. In addition, smart grids give the demand side greater flexibility, which is enough for demand response to play an important role in the operation of the power system.
[0004] ZL2021113544051 discloses a multi-objective optimization scheduling method for a multi-energy system considering the reliability of power transmission. This method adopts a multi-population co-evolution strategy to solve complex problems with multiple objectives and multiple constraints. At the same time, it adopts a non-inferior solution screening monitoring mechanism and a simulated binary crossover method to increase the diversity of the Pareto solution set, thereby providing more reasonable decision-making options for decision makers in power dispatching.
[0005] ZL 2021101769548 discloses a multi-objective optimization scheduling method that considers the uncertainty and demand of wind power. This method establishes an optimization scheduling model, uses the weighted Chebyshev decomposition method to decompose it into multiple sub-problems for easy solution, and uses the MOEA / D algorithm to solve the decomposed single-objective optimization model obtained in the previous steps. The integration of wind power into the system is safer and more reliable. At the same time, demand response is taken into account, so that the flexibility, controllability and economy of the system operation are improved. Summary of the invention
[0006] In order to solve the above technical problems, the present invention provides a multi-objective safe and economic optimization scheduling method for a multi-energy system taking demand response into consideration, which achieves the optimal configuration of power generation cost, carbon emissions and power transmission security, and solves the problem of optimal scheduling of a multi-energy system that achieves reduced power generation cost, low carbon emissions and transmission line safety while considering demand response.
[0007] In order to achieve the above object, the present invention is achieved through the following technical solutions:
[0008] The present invention is a multi-objective safety and economic optimization scheduling method for a multi-energy system considering demand response, comprising the following steps:
[0009] A. Establish a multi-objective, multi-constraint safety and economic optimization dispatch model for a multi-energy system with demand response;
[0010] B. According to the decomposition-based multi-objective evolutionary algorithm, the multi-objective model is decomposed into a set of scalar sub-problems by presetting a set of weight vectors uniformly distributed in the target space, and then a number of neighbors are assigned to each sub-problem. Then, each particle and a number of neighboring particles constitute a sub-population of the particle. The optimization of each sub-problem is optimized in its sub-population through the replacement strategy of the particle swarm optimization algorithm based on the Tchebycheff approach.
[0011] C. Adopt an adaptive weight adjustment strategy to balance the global and local exploration capabilities of the algorithm. And add a random trajectory correction part based on Cauchy-Lorentz distribution to avoid premature convergence of the particle swarm optimization algorithm and ensure that the algorithm task is dynamically transformed from the early global search to the later local search. In addition, a non-dominated solution set is added to record and update the non-dominated solutions obtained in each iteration of the algorithm;
[0012] D. Use the improved decomposition-based multi-objective evolutionary algorithm to solve the multi-objective overall optimization model and obtain the Pareto optimal solution set.
[0013] As a multi-objective safety and economic optimization dispatch of a multi-energy system with demand response, which targets economy, environmental protection and power transmission safety, step A takes power generation cost, carbon emission and transmission line safety as objective functions, and establishes a multi-objective safety and economic optimization model under multi-energy access with demand response based on power balance constraint, rotation reserve constraint, generator unit operating power constraint, unit ramp rate constraint, transmission line power constraint, generator start and stop time constraint and demand response capacity constraint:
[0014] Multiple Targets:
[0015] Power balance constraints:
[0016]
[0017] Ascending rotation reserve constraint:
[0018]
[0019] R Uit =min(P imax -P i,t , R.U. i )
[0020] Descending Spin Reserve Constraint:
[0021]
[0022] R Dit =min(P i,t -P imin , R.D. i )
[0023] Operating power constraints of generator sets:
[0024] P imax ≥P i,t I i,t ≥P imin i∈N g , t∈T
[0025]
[0026] Generator set ramp rate constraints:
[0027] RU i ≥P i,t -P i,(t-1) ≥-RD i i∈N g , t∈T
[0028] RU wj ≥P j,t -P j,(t-1) ≥-RD wj j∈M,t∈T
[0029] Transmission line power constraints:
[0030]
[0031] Start and stop time limits for thermal power units:
[0032]
[0033] Demand response capacity constraints:
[0034] |P DRdmax |≥λ d,t |P DRd,t |≥0 d∈N DR , t∈T
[0035] Among them, f 1 、f 2 and f 3 are the objective functions of power generation cost, carbon emission and transmission line safety respectively. T is a scheduling cycle, N g 、N DR and M are the number of thermal power units, the number of demand response units and the number of wind power units respectively; P i,t and P DRd,tare the power corresponding to thermal power unit i and demand response d at time t respectively; P wjt is the predicted active power of wind turbine j at time t; and are the minimum output power and maximum output power of the j-th wind turbine respectively; I i,t and λ d,t They are the switch operation status of the thermal power unit and the in-and-out scheduling of demand response, and their values switch between 0 and 1; a i 、b i and c i is the coefficient, and the power generation cost of the unit can be obtained by taking appropriate values; c d is the fixed coefficient of demand response cost; S i,t and D i,t is the switch operation cost; C DR0 is the capacity cost of demand response, and its value changes with the change of demand response capacity; α i , β i and γ i is the carbon emission coefficient; P LNt is the net load of the system at time t; N s is the number of feeder lines; is the safety distance of the lth feeder; P imax and P imin is the maximum output power and minimum output power of thermal power unit i; RU i and RD i are the maximum and minimum ramp limits of thermal power unit i; RU wj ,RD wj is the maximum and minimum climbing limit of the j-th wind turbine; u s % and d s % is the demand coefficient of wind power forecast error on positive and negative reserves; and are the start-up and shutdown time of thermal power unit i at time t-1; P l,t is the line transmission loss at time t; G l2 and G l1 are the upper and lower limits of the power flow of transmission line l respectively; F l,i and F l,k is the matrix associated with the corresponding transmission line; and are the minimum start-up and shutdown time of thermal power unit i; P DRdmax is the capacity cap for demand response.
[0036] A further improvement of the present invention is that: Step B firstly generates a set of uniformly distributed weight vectors {ω 1 ,…,ωN}(N is the number of weight vectors, i.e., the population size). The weight vectors are generated based on the unit hyperplane, and the total number is Among them, m is the number of objective functions, H is the number of intervals of the coordinate axis of each dimension in the multi-objective problem, and is called are uniformly spaced values. Specifically, the unit hyperplane can be expressed as In this way, each weight vector ω can be expressed as and are the components of the weight vector ω.
[0037] After the weight vector is generated, the nearest T weight vectors are assigned to each weight vector according to the Euclidean distance between the weight vectors, that is, T neighbors are assigned to each subproblem. Then each particle and its T neighboring particles constitute a subpopulation of the particle. The optimization of each subproblem is optimized in its subpopulation by the particle swarm optimization algorithm based on the replacement strategy of the Tchebycheff approach. The speed and position update equations of the particle swarm optimization algorithm are:
[0038]
[0039] The position of the i-th particle is represented by x i ={x i1 , x i2 , …, x iD}, and x id ∈[x min,d, x max,d ], d = 1, 2, ..., D, x min,d and x max,d are the upper and lower bounds of the dimension d respectively; p best ={p i1 , p i2 ,…,p iD} is the best position of the i-th particle in the previous iteration; v i = {v i1 , v i2 , …, v iD} is the current speed, and v id ∈(v min , v max ), v min and v max are the upper and lower limits of speed respectively; p gbest ={p g1 , p g2 ,…,p gD}, represents the optimal particle of all particles in the sub-population; ω(k)∈[0,1] is the adaptive inertia weight, which determines the influence of historical speed on current speed; k is the number of iterations of the algorithm, rand1d and rand 2d is a random number uniformly distributed between [0, 1], c 1 and c 2 is the learning factor.
[0040] The Chebyshev aggregation method is used to compare the pros and cons of two solutions. The fitness value (Fitness Value), that is, the aggregation function value, is calculated based on the same weight vector. In this way, it can be determined which solution is better for the weight vector. That is, based on this aggregation function value, the particle swarm optimization algorithm is used to find a better aggregation function value. For a multi-objective optimization problem, the Chebyshev method can be used to convert the problem into a scalar optimization problem, as follows:
[0041]
[0042] subject to x∈Ω
[0043] in is the reference point and
[0044] Furthermore, in the multi-objective safe and economic optimization dispatch of multi-energy systems with demand response, which targets economy, environmental protection and power transmission safety, step C completes the improvement of the particle swarm optimization algorithm in the previous step, adopts an adaptive weight adjustment strategy to balance the global and local exploration capabilities of the algorithm, and introduces the random trajectory correction part of the Cauchy-Lorenz distribution to avoid premature convergence of the particle swarm optimization algorithm and ensure that the algorithm task is dynamically transformed from the early global search to the later local search. In addition, the introduction of the non-dominated solution set is used to record and update the non-dominated solution obtained in each iteration of the algorithm.
[0045] The inertia weight ω is used to control the impact of historical speed on the current speed. Therefore, the parameter ω adjusts the trade-off between the global and local exploration capabilities of the population. A larger inertia weight is conducive to global search, and a smaller inertia weight is conducive to local search. A suitable inertia weight ω value can usually balance the global and local exploration capabilities, thereby reducing the number of iterations required to find the optimal solution. The proposal of adaptive inertia weight is based on the diversity of particles in the population. The inertia weight is changed nonlinearly and adaptively according to the maximum Euclidean distance between good and bad individuals. Good and bad individuals are distinguished according to the value of the aggregation function in the Chebyshev method. The adjustment of the inertia weight is based on the maximum Euclidean distance between the best individual and the worst individual in this iteration and the maximum Euclidean distance between the best individual and the worst individual at initialization. If it is not greater than, the inertia weight is reduced, otherwise, the inertia weight is increased. The nonlinear method is used to adaptively adjust the inertia weight ω to ensure that the algorithm has a larger global search capability in the early stage of search and continuous optimization capability in the later stage of search. That is, the inertia weight of the k+1th iteration is as follows,
[0046]
[0047] Where k is the number of iterations; x best (k) is the optimal particle of the kth iteration; x worst (k) the worst particle in the kth iteration; x best (0) is the optimal particle at initialization; x worst (0) is the worst particle at initialization.
[0048] The addition of the random trajectory correction based on the Cauchy-Lorentz distribution is to avoid premature convergence of the particle swarm optimization algorithm and to ensure that the algorithm task is dynamically transformed from the early global search to the later local search. There are two parameters in the probability density function of the Cauchy distribution, namely the position parameter x of the peak position of the distribution and 0 and the scale parameter γ of the general width at half the maximum value, whose probability density function is as follows,
[0049]
[0050] Let the position parameter x of the peak position be 0 is the global optimal particle among all particles. When the remaining particles are closer to the optimal particle, the larger the trajectory correction is. Relatively speaking, the particles far away from the optimal particle receive smaller corrections. Because when particles are close to the optimal particle, their optimization ability will weaken. Therefore, these particles that converge too early will be subject to larger position corrections and then diffused into the solution space to prevent the particle swarm algorithm from converging too early. The additional position update formula is as follows, where P g (k) is the global optimal particle, β∈[-1, 1], indicating that the particle position is mutated in the positive or negative direction.
[0051] x j (k+1)=x j (k)+β·L j (k)·||P g (k)-x j (k)|| 2
[0052] Among them, L j (k) is the density function of the introduced Cauchy distribution. According to the different positions of the particles, its update formula is as follows:
[0053]
[0054] In order to ensure the development capability of the algorithm in the later stages of iteration, the variation of particles should gradually decrease with the increase of the number of iterations. γ(k) can well meet this requirement. Its form is as follows. Because γ(k) determines the half width at half of the maximum value, where γ 0 and τ are parameters, k is the number of iterations. When the number of iterations increases, γ(k) decreases, making the Cauchy distribution density curve steeper and steeper. This makes the variation range at the same position smaller and smaller, ensuring the development capability of the algorithm in the later stage of iteration.
[0055]
[0056] Furthermore, in the multi-objective safe and economical optimal dispatch of the multi-energy system with demand response that targets economy, environmental protection and power transmission safety, step D uses an improved decomposition-based multi-objective evolutionary algorithm to solve the multi-objective overall optimization model to obtain the Pareto optimal solution set. The particles in each population are updated and iterated 200 times, and the non-dominated solutions they searched are stored in the non-dominated solution set, and they are updated in each subsequent update until the iteration is completed. In this way, the multi-objective safe and economical optimal dispatch model under multi-energy access is solved to obtain a non-dominated solution set that meets complex constraints. The solutions in the solution set can all simultaneously meet the safe and economical optimal dispatch of economy, environmental protection and power transmission safety, thereby providing guidance for relevant dispatching personnel.
[0057] The beneficial effects of the present invention are as follows: the present invention proposes a multi-objective particle swarm optimization algorithm based on decomposition in view of the multi-objective, multi-constraint and nonlinear characteristics of the multi-energy system, and according to the target requirements of the economy, environmental protection and power transmission safety of the multi-energy system, a multi-objective safety and economic optimization model under multi-energy access with demand response is established in combination with the unit's own output, ramp rate constraints and demand response constraints, and the multi-objective evolutionary algorithm based on decomposition is used to convert the multi-objective model into a series of single-objective optimization sub-problems;
[0058] Compared with traditional optimization algorithms, it has lower computational complexity, adaptive weight search strategy and particle trajectory correction, which ensures that the algorithm has continuous global search capabilities while reducing the number of iterations to reach the optimal solution. The improved particle swarm optimization algorithm is used to solve each subproblem to obtain the optimal solution for each subsystem, and finally achieves the optimal Pareto solution set of the entire system with uniform distribution at a lower computational complexity, providing reliable decision support for decision makers.
[0059] The multi-objective safe and economical optimization scheduling method for a multi-energy system considering demand response proposed in the present invention achieves the optimal configuration of reducing power generation costs, low carbon emissions and ensuring the safety of transmission lines, solves the problem of safe optimization scheduling of a multi-energy system that achieves low carbon emissions and power transmission reliability while considering demand response, and provides more solution guidance for relevant dispatching personnel. BRIEF DESCRIPTION OF THE DRAWINGS
[0060] Figure 1 It is a flowchart of the multi-objective safety and economic optimization scheduling method for a multi-energy system of the present invention. DETAILED DESCRIPTION
[0061] The following will disclose the embodiments of the present invention with drawings. For the purpose of clear description, many practical details will be described together in the following description. However, it should be understood that these practical details should not be used to limit the present invention. That is, in some embodiments of the present invention, these practical details are not necessary. In addition, for the purpose of simplifying the drawings, some conventional structures and components will be depicted in a simple schematic manner in the drawings.
[0062] In the power grid environment, due to the access of large-scale intermittent energy such as wind turbines, the multi-energy safe and economic optimization scheduling presents the characteristics of multiple objective functions and complex constraints. Combined with demand response, the power generation cost and carbon emissions are considered, while meeting the constraints such as the safety of the power system, power balance constraints, rotating reserve constraints, generator operating power constraints and unit climbing rate constraints, a multi-objective safe and economic optimization scheduling model under hybrid multi-energy access is established.
[0063] like Figure 1 As shown, the present invention is a multi-objective safety and economic optimization scheduling method for a multi-energy system considering demand response, which specifically includes the following steps:
[0064] (I) The multi-objective safety and economic optimization dispatch model of a multi-energy system considering demand response is as follows:
[0065] (1) Objectives:
[0066] Power generation cost function:
[0067]
[0068] Carbon emission function:
[0069]
[0070] Line security functions:
[0071]
[0072] Among them, T is a scheduling cycle, N g and N DR are the number of thermal power units and the number of demand responses respectively; P i,t and P DRd,t are the power corresponding to thermal power unit i and demand response d at time t respectively; I i,t and λ d,t They are the switch operation status of the thermal power unit and the in-and-out scheduling of demand response, and their values switch between 0 and 1; a i 、b i and c i is the coefficient, and the power generation cost of the unit can be obtained by taking appropriate values; c d is the fixed coefficient of demand response cost; S i,t and D i,t is the switch operation cost; C DRO is the capacity cost of demand response, and its value changes with the change of demand response capacity; α i , β i and γ i is the carbon emission coefficient; N s is the number of feeder lines; is the safety distance of the lth feeder;
[0073] (2) Constraints
[0074] ① Power balance constraints:
[0075]
[0076] Among them, I i,t and λ d,t They are the switch operation status of the thermal power unit and the in-and-out scheduling of demand response, and their values switch between 0 and 1; P LNt is the net load of the system at time t.
[0077] ② Rotation reserve constraints:
[0078] Ascending rotation reserve constraint:
[0079]
[0080] R Uit =min(Pimax -P i,t RU i ) (6)
[0081] Descending Spin Reserve Constraint:
[0082]
[0083] R Dit =min(P i,t -P imin, RD i ) ⑻
[0084] Among them, P imax is the maximum output power of thermal power unit i; P DRdmax is the upper limit of demand response capacity; RU i and RD i are the maximum and minimum climbing limits of thermal power unit i respectively; P wjt is the predicted active power of wind turbine j at time t; u s % and d s % is the demand coefficient of wind power forecast error on positive and negative reserves.
[0085] ③Operation power constraints of generator sets:
[0086] P imax ≥P i,t I i,t ≥P imin i∈N g , t∈T ⑼
[0087]
[0088] Among them, P imin is the minimum output power of thermal power unit i; and are the minimum output power and maximum output power of the j-th wind turbine respectively.
[0089] ④ Generator set climbing rate constraint:
[0090] RU i ≥P i,t -P i,(t-1) ≥-RD i i∈N g , t∈T ⑾
[0091] RU wj ≥P j,t -P j,(t-1) ≥-RD wj j∈M,t∈T ⑿
[0092] Among them, RUi and RD i are the maximum and minimum ramp limits of thermal power unit i; RU wj ,RD wj are the maximum and minimum ramp limits of the j-th wind turbine.
[0093] ⑤ Transmission line power constraints:
[0094]
[0095] Among them, P i,t is the power of thermal power unit i at time t; P l,t is the line transmission loss at time t; G l2 and G l1 are the upper and lower limits of the power flow of transmission line l respectively; F l,i and F l,k is the matrix associated with the corresponding transmission line.
[0096] ⑥Start and stop time limit of thermal power units:
[0097]
[0098] in, and are the start-up and shutdown time of thermal power unit i at time t-1 respectively; and are the minimum startup and shutdown time of thermal power unit i respectively.
[0099] ⑦Demand response capacity constraints:
[0100] |P DRdmax |≥λ d,t |P DRd,t |≥0d∈N DR, t∈T ⒂
[0101] Among them, P DRdmax is the capacity cap for demand response.
[0102] (2) Solve the multi-objective safety and economic optimization scheduling model of multi-energy systems considering demand response.
[0103] (1) Initialization
[0104] ① Initialize the weight vector
[0105] Step 1.1.1: First, we need to generate a set of uniformly distributed weight vectors {ω 1 ,…,ω N}(N is the number of weight vectors, i.e., the population size). The weight vectors are generated based on the unit hyperplane, and the total number is Among them, m is the number of objective functions, H is the number of intervals of the coordinate axis of each dimension in the multi-objective problem, and is called are uniformly spaced values. Specifically, the unit hyperplane can be expressed as In this way, each weight vector ω can be expressed as and are the components of the weight vector ω;
[0106] Step 1.1.2: Next, calculate the Euclidean distance between any two weight vectors and calculate the T weight vectors closest to each weight vector. For i=1,…,N, set D(i)={i 1 ,…,i T},here There are T close to ω i The weight vector is the T particles close to the i-th particle. These T particles constitute the sub-population belonging to the i-th particle.
[0107] ② Initialize the particle group
[0108] Step 1.2.1: First, construct the target space according to the constraints in the above model, randomly initialize particles in the target space, and initialize the parameter values;
[0109] Step 1.2.2: Calculate the objective function value of the initialized particle and initialize the reference point z according to the objective function value. * and non-dominated solution set, where is the reference point and
[0110] Step 1.2.3: According to the Chebyshev aggregation method, calculate the aggregation function value of the particle, that is, the fitness value. For a multi-objective optimization problem, the Chebyshev method can be used to convert the problem into a scalar optimization problem, as follows:
[0111]
[0112] Step 1.2.4: Initialize the subpopulation p belonging to each particle according to the aggregation function value of the particle best and p gbest , where p best ={p i1 , p i2 ,…,p iD} is the best position of the i-th particle in the previous iteration; p gbest ={p g1 , p g2, …, p gD}, represents the optimal particle of all particles in the population; d = 1, 2, ..., D is the dimension of the particle;
[0113] (2) Update
[0114] ① Improved particle swarm optimization algorithm
[0115] Step 2.1.1: Use the particle swarm optimization algorithm to update particles and optimize them in their sub-populations. The speed and position update equations of the particle swarm optimization algorithm are:
[0116]
[0117] The position of the i-th particle is represented by x i ={x i1 , x i2 , …, x iD}, and x id ∈[x min,d , x max,d ], d = 1, 2, ..., D, x min,d and x max,d are the upper and lower bounds of the dimension d respectively; p best ={p i1 , p i2 ,…,p iD} is the best position of the i-th particle in the previous iteration; v i = {v i1 , v i2 , …, v iD} is the current speed, and v id ∈(v min , v max ), v min and v max are the upper and lower limits of speed respectively; p gbest ={p g1 , p g2 ,…,p gD}, represents the optimal particle of all particles in the sub-population; k is the number of iterations of the algorithm, rand 1d and rand 2d is a random number uniformly distributed between [0, 1], c 1 and c 2 is the learning factor.
[0118] Step 2.1.2: According to the constraints of the above model, modify the updated particle y′={y′ 1 , y′ 2 ,…,y′ D}, so that the particle is in the target space; for any d∈{1, 2, …, D}, if there is y′ d <x min,d, then let y′ d =x min,d ; If there is y′ d >x max,d , then let y′ d =x max,d . Where x min,d and x max,d are the upper and lower bounds of the dimension d, respectively;
[0119] Step 2.1.3: Update the p of the subpopulation gbest and the particle's p best , if for j∈D(i), if g tch (y'|ω j , z * )≤g tch (x j |ω j , z * ), then set p gbest =y′; if g tch (y'|ω j , z * )≤g tch (p best |ω j , z * ), then set p best =y′;
[0120] Step 2.1.4: Update reference point z * , for all j = 1, ..., m, if z j <f j (y′), then set z j =f j (y′), where m is the number of objective functions;
[0121] Step 2.1.5: Update the domain solution in the target space using a random trajectory correction based on the Cauchy-Lorentz distribution. For j∈D(i), we have:
[0122] x j (k+1)=x j (k)+β·L j (k)·||P g (k)-x j (k)|| 2 ⒅
[0123] Among them, β∈[-1, 1], indicating that the particle position changes from the positive direction or the negative direction; ||P g (k)-x j (k)|| 2 is the distance between particle j and the optimal particle; L j(k) is the density function of the introduced Cauchy distribution. According to the different positions of the particles, its update formula is as follows:
[0124]
[0125] in, γ 0 and τ are parameters, and k is the number of iterations.
[0126] Step 2.1.6: Update the inertia weight. The inertia weight of the k+1th iteration is as follows;
[0127]
[0128] Among them, k is the number of iterations; x best (k) is the optimal particle of the kth iteration; x worst (k) the worst particle in the kth iteration; x best (0) is the optimal particle at initialization; x worst (0) is the worst particle at initialization.
[0129] ② Update the non-dominated solution set
[0130] Step 2.2.1: Remove all solutions from the non-dominated solution set that are 1 (y′),…,f m (y′)};
[0131] Step 2.2.2: If there is no vector dominating the non-dominated solution set, F(y′) = {f 1 (y′),…,f m (y′)}, then add F(y′) to the non-dominated solution set;
[0132] (3) Stopping criteria
[0133] If the stopping criterion is met, the number of iterations is generally set to 200, then the process stops and the non-dominated solution set is output; if not, the process continues to go to step (2) for updating. In this way, the multi-objective safe and economical optimal dispatch model of a multi-energy system considering demand response is solved to obtain a non-dominated solution set that meets complex constraints. The solutions in this solution set can simultaneously meet the safe and economical optimal dispatch of the three indicators of economy, environmental protection and safety, thereby providing guidance for relevant dispatchers.
[0134] The multi-objective safe and economical optimization scheduling method for a multi-energy system considering demand response proposed in the present invention achieves the optimal configuration of reducing power generation costs, low carbon emissions and ensuring the safety of transmission lines, solves the problem of safe optimization scheduling of a multi-energy system that achieves low carbon emissions and power transmission reliability while considering demand response, and provides more solution guidance for relevant dispatching personnel.
[0135] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims
1. A multi-objective safety and economic optimization scheduling method for a multi-energy system considering demand response, characterized by: The multi-energy system multi-objective safety and economic optimization scheduling method comprises the following steps: Step 1: Establish a multi-objective, multi-constraint safety and economic optimization dispatch model for a multi-energy system with demand response; Step 2: Initialize the weight vector and initialize the particle population: According to the decomposition-based multi-objective evolutionary algorithm, a set of weight vectors uniformly distributed in the target space is preset to decompose the multi-objective model into a set of scalar subproblems, and then a number of neighbors are assigned to each subproblem. Each particle and several neighboring particles constitute a subpopulation of the particle. The optimization of each subproblem is optimized in its subpopulation through the replacement strategy of the particle swarm optimization algorithm based on the Chebyshev method; Step 3: Update the particle swarm optimization algorithm and update the non-dominated solution set: Adopt an adaptive weight adjustment strategy to balance the global and local exploration capabilities of the particle swarm optimization algorithm, and add a random trajectory correction part based on the Cauchy-Lorentz distribution to avoid premature convergence of the particle swarm optimization algorithm and ensure that the algorithm task is dynamically transformed from the early global search to the later local search. Add a non-dominated set, record and update the non-dominated solution obtained by each iteration of the particle swarm optimization algorithm; Step 4: Use the improved decomposition-based multi-objective evolutionary algorithm to solve the multi-objective overall optimization model and obtain the Pareto optimal solution set, where: The step 1 takes power generation cost, carbon emission and transmission line safety as objective functions, and establishes a multi-objective safety and economic optimization model under multi-energy access with demand response based on power balance constraints, rotation reserve constraints, generator set operating power constraints, unit ramp rate constraints, transmission line power constraints, generator start and stop time constraints and demand response capacity constraints: (1) Objectives Power generation cost function: Carbon emission function: Line security functions: (2) Constraints ① Power balance constraints: ② Rotation reserve constraints: Ascending rotation reserve constraint: R Uit =min(P imax -P i,t ,RU i ) Descending Spin Reserve Constraint: R Dit =min(P i,t -P imin ,RD i ) ③Operation power constraints of generator sets: P imax ≥P i,t I i,t ≥P imin i∈N g ,t∈T ④ Generator set climbing rate constraint: RU i ≥P i,t -P i,(t-1) ≥-RD i i∈N g ,t∈T RU wj ≥P j,t -P j,(t-1) ≥-RD wj j∈M,t∈T ⑤ Transmission line power constraints: ⑥Start and stop time limit of thermal power units: ⑦Demand response capacity constraints: |P DRdmax |≥λ d, t|P DRd,t |≥0d∈N DR, t∈T Among them, f1, f2 and f3 are the objective functions of power generation cost, carbon emission and transmission line safety respectively, T is a scheduling cycle, N g 、N DR and M are the number of thermal power units, the number of demand response units and the number of wind power units respectively, P i,t and P DRd,t are the power corresponding to thermal power unit i and demand response d at time t, P wjt is the predicted active power of wind turbine j at time t, and are the minimum output power and maximum output power of the j-th wind turbine, I i,t and λ d,t are the switch operation status of the thermal power unit and the in-and-out scheduling of demand response, and its value switches between 0 and 1. i 、b i and c i is the coefficient, c d is the fixed coefficient of demand response cost, S i,t and D i,t is the switching operation cost, C DR0 is the capacity cost of demand response, α i , β i and γ i is the carbon emission coefficient, P LNt is the net load of the system at time t, N s is the number of feeder lines, is the safety distance of the lth feeder, P imax and P imin is the maximum output power and minimum output power of thermal power unit i, RU i and RD i are the maximum and minimum ramp limits of thermal power unit i, RU wj ,RD wj is the maximum and minimum ramp limits of the j-th wind turbine, u s % and d s % is the demand coefficient of wind power forecast error for positive and negative reserves, and are the start-up and shutdown time of thermal power unit i at time t-1, P l,t is the line transmission loss at time t, G l2 and G l1 are the upper and lower limits of the power flow of transmission line l, respectively, and F l,i and F l,k is the matrix associated with the corresponding transmission line, and are the minimum start-up and shutdown time of thermal power unit i, P DRdmax is the capacity cap for demand response.
2. The multi-objective safety and economic optimization scheduling method for a multi-energy system considering demand response according to claim 1 is characterized by: The updating of particle swarm optimization in step 3 includes the following steps: Step 3-1-1: Use the particle swarm optimization algorithm to update particles and optimize them in their sub-populations. The speed and position update equations of the particle swarm optimization algorithm are: The position of the i-th particle is represented by x i ={x i1 , x i2 , …, x iD }, and x id ∈[x min,d , x max,d ], d = 1, 2, ..., D, x min,d and x max,d are the upper and lower bounds of the dimension d, respectively, and p best ={p i1 , p i2 ,…,p iD } is the best position of the last iteration of the i-th particle, v i = {v i1 , v i2 , …, v iD } is the current speed, and v id ∈(v min , v max ), v min and v max are the upper and lower limits of speed, p gbest ={p g1 , p g2 , ..., p gD }, represents the optimal particle of all particles in the sub-population, k is the number of iterations of the algorithm, rand 1d and rand 2d is a random number uniformly distributed between [0, 1], c1 and c2 are learning factors; Step 3-1-2: Modify the updated particle y′={y′1, y′2, ..., y′ according to the constraints of the multi-objective, multi-constraint safety and economic optimization dispatch model of the multi-energy system D }, so that the particle is in the target space; for any d∈{1, 2, ..., D}, if there is y′ d <x min,d , then let y′ d =x min,d ; If there is y′ d >x max,d , then let y′ d =x max,d , where x min,d and x max,d are the upper and lower bounds of the dimension d, respectively; Step 3-1-3: Update the p of the subpopulation gbest and the particle's p best , if for j∈D(i), if g tch (y'|ω j , z * )≤g tch (x j |ω j , z * ), then set p gbest =y′; if g tch (y'|ω j , z * )≤g tch (p best |ω j , z * ), then set p best =y′; Step 3-1-4: Update reference point z * , for all j = 1, ..., m, if z j <f j (y′), then set z j =f j (y′), where m is the number of objective functions; Step 3-1-5: Use the random trajectory correction based on the Cauchy-Lorentz distribution in the target space to update the domain solution. For j∈D(i), we have: x j (k+1)=x j (k)+β·L j (k)·||P g (k)-x j (k)||2 Among them, β∈[-1, 1] indicates that the particle position changes from the positive direction or the negative direction, ||P g (k)-x j (k)||2 is the distance between particle j and the optimal particle, L j (k) is the density function of the introduced Cauchy distribution. According to the different positions of the particles, its update formula is as follows: in, γ0 and τ are parameters, k is the number of iterations; Step 3-1-5: Update the inertia weight. The inertia weight of the k+1th iteration is as follows; Among them, k is the number of iterations, x best (k) is the optimal particle of the kth iteration, x worst (k) The worst particle in the kth iteration, x best (0) is the optimal particle at initialization, x worst (0) is the worst particle at initialization.
3. The multi-objective safety and economic optimization scheduling method for a multi-energy system considering demand response according to claim 1 or 2 is characterized in that: In step 3, updating the non-dominated solution set specifically includes the following steps: Step 3-2-1: Remove all solutions from the non-dominated solution set that are m (y′)}; Step 3-2-2: If there is no vector dominating the non-dominated solution set, F(y′)={f1(y′),…,f m (y′)}, then add F(y′) to the non-dominated solution set.
4. The multi-objective safety and economic optimization scheduling method for a multi-energy system considering demand response according to claim 1 is characterized by: In step 2, the initialization weight vector specifically includes the following steps: Step 2-1-1: First, you need to generate a set of uniformly distributed weight vectors {ω 1 ,…,ω N }, N is the number of weight vectors, that is, the population size, the weight vector is generated based on the unit hyperplane, and the total number is Among them, m is the number of objective functions, H is the number of intervals of the coordinate axis of each dimension in the multi-objective problem, and is called is a uniformly spaced value, and the unit hyperplane is represented by Each weight vector ω is expressed as and are the components of the weight vector ω; Step 2-1-2: After the weight vectors are generated, the nearest T weight vectors are assigned to each weight vector based on the Euclidean distance between the weight vectors.
5. The multi-objective safety and economic optimization scheduling method for a multi-energy system considering demand response according to claim 1 or 4 is characterized in that: In step 2, the initialization of the particle population includes the following steps: Step 2-2-1: Construct the target space according to the constraints of the multi-objective safety and economic optimization model in step 1, initialize the particles in the target space, and initialize the values of each parameter; Step 2-2-2: Calculate the objective function value based on the initialized particles, and initialize the reference point z based on the objective function value. * and non-dominated solution set, where is the reference point and Step 2-2-3: According to the Chebyshev aggregation method, calculate the aggregation function value of the particle, that is, the fitness value. For a multi-objective optimization problem, the problem can be converted into a scalar optimization problem based on the Chebyshev method: Step 2-2-4: Initialize the subpopulation p of each particle according to the aggregation function value of the particle best and p gbest , where p best ={p i1 , p i2 ,…,p iD } is the best position of the i-th particle in the previous iteration; p gbest ={p g1 , p g2 ,…,p gD } represents the optimal particle of all particles in the population; d = 1, 2,…, D is the dimension of the particle.
6. The multi-objective safety and economic optimization scheduling method for a multi-energy system considering demand response according to claim 1 is characterized by: The step 4 is specifically as follows: the particles in each population are updated and iterated 200 times, and the non-dominated solutions they searched are stored in the non-dominated solution set, and they are updated in each subsequent update until the iteration is completed. In this way, the multi-objective safety and economic optimization scheduling model under multi-energy access is solved, and a non-dominated solution set that satisfies complex constraints can be obtained.
