Park energy double-layer optimization scheduling method considering electric vehicle and demand response

Abstract

The invention discloses a park energy double-layer optimization scheduling method considering electric vehicles and demand response, and the method comprises the following steps: constructing a park integrated energy system which comprises an energy supply side, an energy conversion side and a demand side; a double-layer optimization model is established, the upper layer is a multi-target optimization model with the target of minimizing the total cost of the system, minimizing the carbon emission and maximizing the renewable energy consumption rate, and the lower layer is a single-target optimization model with the target of minimizing the charging cost of the electric vehicle user and the load fluctuation of the power grid; based on the Pareto theory, designing a multi-objective litsea coreana optimization algorithm to solve an upper-layer multi-objective model; introducing chaos initialization, adaptive step length adjustment and a Gaussian disturbance strategy improved litsea coreana optimization algorithm to solve a lower-layer single-target model; and through a vehicle network interaction and demand response cooperation mechanism, real-time scheduling is executed, a day-ahead plan is fed back and corrected, and finally an optimal park energy optimization scheduling strategy is solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of comprehensive energy system optimal scheduling, and particularly relates to a park energy double-layer optimal scheduling method considering electric vehicles and demand response. BACKGROUND

[0002] With large-scale grid connection of renewable energy and popularization of electric vehicles (EVs), the operation mode of park energy system is facing profound changes. The traditional power grid is dominated by fossil energy, and the load characteristics are relatively stable. However, the access of high proportion of renewable energy and electric vehicles significantly aggravates the complexity of system operation. On the one hand, the disordered charging behavior of electric vehicles leads to the aggravation of spatial and temporal imbalance of power grid load, especially in the peak load period, and the centralized charging causes the peak load to rise, which makes it difficult for the traditional "peak load shifting" method to cope with it. On the other hand, the output of renewable energy such as wind power and photovoltaic power is affected by weather, and has strong intermittency and volatility. The existing scheduling method lacks flexibility, and it is difficult to realize dynamic matching of wind and light output and load demand, resulting in prominent wind and light abandonment, which restricts the efficient consumption of clean energy. In addition, the traditional scheduling strategy usually takes single economic cost as the target, and fails to fully utilize the flexibility of electric vehicle charging and discharging, demand side response and multi-energy coordination potential of electricity-heat-gas, resulting in high system power purchase cost, energy storage loss and equipment operation and maintenance cost, and poor overall economy.

[0003] In the prior art, single-layer single-target optimization model is difficult to coordinate the collaborative optimization of multi-energy coupling, electric vehicle scheduling and demand response, and the traditional optimization algorithm is low in efficiency when dealing with high-dimensional nonlinear problems, and cannot take into account economy, low carbon and flexibility. Therefore, it is urgent to develop a scheduling method based on double-layer multi-objective optimization, which coordinates electric vehicles and demand side resources through upper layer optimization of operation cost and carbon emission, and lower layer, and processes uncertainty by combining stochastic programming or robust optimization, so as to improve the economy, low carbon and reliability of park energy system. SUMMARY

[0004] The purpose of the application is to solve the defects in the prior art, and to provide a double-layer scheduling method of upper multi-objective optimization and lower single-objective optimization, which realizes the Pareto optimal solution set of the lowest total cost, the least carbon emission and the highest renewable energy consumption rate through the coordination of EV charging and discharging and demand response, and improves the solving efficiency by using the improved black eagle algorithm.

[0005] The park energy double-layer optimal scheduling method considering electric vehicles and demand response coordinates electric vehicle charging and discharging and demand response (DR), reduces the system operation cost, improves the renewable energy consumption rate, and realizes load peak clipping and valley filling under vehicle-to-grid (V2G).

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] The application discloses a park energy double-layer optimization scheduling method considering electric vehicles and demand response, and comprises the following steps:

[0008] S1: a park integrated energy system is constructed, including an energy supply side, an energy conversion side and a demand side;

[0009] S2: a double-layer optimization model is established, the upper layer is a multi-objective optimization model, the objectives are to minimize the total system cost, minimize the carbon emission and maximize the renewable energy consumption rate; the lower layer is a single-objective optimization model, the objective is to minimize the electric vehicle user charging cost and the power grid load fluctuation;

[0010] S3: a multi-objective secretary black eagle optimizer (MOSBEO) is designed based on the Pareto theory to solve the upper-layer multi-objective model; an improved black eagle optimizer (IBEO) is introduced to solve the lower-layer single-objective model, the IBEO is improved by introducing a chaotic initialization, an adaptive step adjustment and a Gaussian disturbance strategy;

[0011] S4: through a vehicle-grid interaction and demand response coordination mechanism, real-time scheduling is performed and the day-ahead plan is fed back and corrected.

[0012] Specifically, the energy supply side in S1 comprises wind power, photovoltaic power, grid power purchase, energy storage equipment, the energy conversion side comprises electric-gas conversion, combined heat and power and heat pump, and the demand side comprises electric / heat / gas load, electric vehicle cluster and adjustable load.

[0013] Specifically, the upper-layer multi-objective optimization model in S2 aims to minimize the total system cost F1, minimize the carbon emission F2 and maximize the renewable energy consumption rate F3, and the objective function is:

[0014]

[0015] In the formula, C purchase is the grid power purchase and gas purchase cost; C operation is the equipment operation and maintenance cost; C penalty is the penalty cost of abandoned wind and abandoned light; P CHP-e is the combined heat and power output; a is the carbon emission coefficient of the combined heat and power unit output; Q P2G is the electric-gas conversion equipment fuel consumption; Q gas-boiler is the fuel consumption of the gas boiler; b is the carbon emission coefficient of the gas boiler; P wind+PV-actualP wind+PV-predict P

[0016] The constraint conditions in the step S2 include the following: power balance constraint, equipment operation constraint, energy storage capacity constraint and carbon emission constraint.

[0017] The objective function of the lower-layer optimization model in the step S2 is to minimize the charging cost and load fluctuation, and the objective function is as follows:

[0018] min f = a P EV + b AP grid (2)

[0019] In the formula, P EV is the charging cost of the electric vehicle user; AP grid is the grid load fluctuation; a and b are weight coefficients, and a+b = 1, used to balance the charging cost and load fluctuation.

[0020] The lower-layer model constraint in the step S2 includes the following: battery capacity of the electric vehicle, charging and discharging power, and user comfort range.

[0021] In the step S3, the data related to the park are input, including the objective function of the double-layer multi-objective optimization model, demand response parameters and related variable constraint conditions, and further the parameters and decision variables of the park energy optimization scheduling strategy are obtained. The lower layer of the double-layer model is solved by using the IBEO algorithm, and the upper-layer model is solved by using the improved MOSBEO algorithm, to obtain the optimal Pareto non-dominated solution, so as to obtain the best park energy optimization scheduling scheme.

[0022] The solving step of the upper-layer model is as follows:

[0023] (1) Initialize the black eagle population: set the population size N, the maximum number of iterations T, the archive size |A| max , the initial step size a0, the chaos parameter r1 [0, 1]; use Latin hypercube sampling to generate the initial population:

[0024]

[0025] In the formula, π j (i) represents a random arrangement of the jth variable dimension; u i,j is a uniform random number, u i,j ~ U(0, 1), used to fine-tune the sampling position in the sub-interval.

[0026]

[0027] In the formula, X (i)λ represents all decision variables in the problem, including time-of-use electricity price λ t , combined heat and power output P CHP,t , energy storage charging and discharging power P ESS,t , and the operating state S of the electric-gas conversion P2G,t .

[0028] (2) Fitness calculation and constraint processing:

[0029] 1) Objective function calculation: Calculate the values of three objective functions, total cost F1, carbon emissions F2, and renewable energy consumption rate F3;

[0030] 2) Penalty function method for constraint processing:

[0031]

[0032] where: g j is the constraint violation value, such as power imbalance, SOC out-of-bounds value, etc.; λ = 10 5 , significantly increasing the fitness value of infeasible solutions, so that they are eliminated in evolution.

[0033] (3) Non-dominated sorting and crowding distance calculation:

[0034] 1) Non-dominated sorting: divide the population into multiple non-dominated layers, Front1, Front2,...; solutions in Front1 are the current Pareto optimal solution set;

[0035] 2) Calculate crowding distance:

[0036]

[0037] After sorting each objective function value, calculate the normalized distance between adjacent solutions to measure the sparsity of the solution.

[0038] (4) External archive update:

[0039] 1) Merge non-dominated solutions: add the non-dominated solutions Front1 of the current population to the external archive A;

[0040] 2) Remove duplicates and truncate: remove duplicate solutions, if |A| > |A| max , keep the first |A| max solutions in order of decreasing crowding distance to ensure solution set diversity;

[0041] (5) Dynamic weight adjustment: adjust the target weight according to the distribution of solutions in the archive:

[0042]

[0043] where: AvgCrowdingDistance kThe average congestion degree in the kth target direction.

[0044] (6) Black Hawk behavior model update:

[0045] 1) Tracking strategy: randomly select a reference solution X from the external archive ref ; update individual position:

[0046]

[0047] In the formula: X ref represents random selection from the archive A; α(t) = α0·exp(-t / T), the step size decays with iteration; t1 is generated by the Tent chaotic mapping, enhancing the global search ability.

[0048] 2) Hover strategy: perform rotational search in the target space, adjust the distribution direction of the solution:

[0049]

[0050] In the formula: m is a multi-objective rotation matrix, mapping to the principal component direction of the target space.

[0051] 3) Capture and plunder strategy: adaptively adjust the position of the solution, balance convergence and diversity:

[0052]

[0053] In the formula: D1 and D2 are global exploration and local development factors, balanced by dynamic weight.

[0054] (7) Population update and elite selection

[0055] Merge parent and child to generate 2N candidate solutions; elite selection, hierarchical according to non-dominated sorting, and sort within the same layer according to congestion degree, select the first N solutions as the new generation population.

[0056] (8) Convergence judgment and output

[0057] Termination condition, if the maximum iteration number T is reached or the Pareto front hyper volume change rate is less than 10 -4 , then terminate the algorithm, output the Pareto optimal solution set in the external storage A for decision maker selection; otherwise, return to step S322 and continue iteration.

[0058] Solution steps of the lower model:

[0059] (1) Chaos initialization of Black Hawk population: set population size N, maximum iteration number T, initial step size factor α 初始 , disturbance intensity η; chaos initialization:

[0060]

[0061] X = {X1, X2, X n} (12)

[0062] where each individual X i is a decision variable, including the charging and discharging rate P EV,t of the electric vehicle and the adjustment range AL DR,t of the DR.

[0063] (2) Fitness calculation and constraint handling:

[0064] 1) For each individual X i , the objective function value is calculated:

[0065] f(X i ) = a · C EV + b · AP load (13)

[0066] where P EV is the charging cost of the electric vehicle user; AP grid is the power load fluctuation; a and b are weight coefficients.

[0067] 2) The constraint violation degree is calculated, and the penalty function method is used to handle the constraint:

[0068]

[0069] where g j is the constraint violation value, such as the power imbalance, the battery capacity SOC out-of-bounds value, etc.; and l is the penalty factor.

[0070] (3) Black Hawk behavior model update:

[0071] 1) Tracking strategy, update individual position:

[0072]

[0073] where r1 is a random number in [0, 1]; t1 is a chaotic number generated by Tent mapping; X best is the current optimal solution; adaptive step size adjustment, step size factor a(t) is dynamically attenuated with iteration number, balancing global exploration and local development.

[0074] 2) Hover strategy, rotate search on individual position, enhance local development ability:

[0075]

[0076] where m is a rotation matrix, simulating the adjustment of search direction in high-dimensional space.

[0077] 3) Capture and Rob strategy, adjust the body position to balance exploration and development:

[0078]

[0079] In the formula: D1 is the global search direction correction factor, D2 is the local convergence factor.

[0080] (4) Local Gaussian disturbance:

[0081] 1) Gaussian disturbance to the current optimal solution X best , generate disturbance solution:

[0082]

[0083] In the formula: δ (t) = δ 初始 · (1-t / T), the standard deviation decreases with iteration, and the disturbance range is limited within the variable feasible region.

[0084] 2) Disturbance solution evaluation: calculate f penalty (X new ), if better than the current optimal solution, update X best .

[0085] (5) Population update and elite reservation:

[0086] Merge parent and child populations, combine current population, newly generated solution and disturbance solution to form 2N+1 candidate solutions; elite selection, sort by fitness value, reserve the first N optimal solutions.

[0087] (6) Termination condition:

[0088] If the maximum number of iterations T is reached, or if the optimal fitness change rate is less than 10 -4 for 10 generations in a row, the algorithm is terminated; output the optimal solution X best , and its corresponding P EV,t and ΔL DR,t .

[0089] Specifically, in the S4, the park energy double-layer optimal dispatching method, the dynamic battery loss compensation model calculation formula is:

[0090]

[0091] In the formula: E discharge,t is the discharge energy of period t; E capacity is the rated capacity of the electric vehicle battery; DOD t = 1-SOC t is the discharge depth; k is the battery compensation loss coefficient.

[0092] The step S4 is executed in real time, and the electric vehicle is guided to discharge to the power grid during the load peak period and is charged during the valley period.

[0093] The step S4 is executed in real time, and the electric vehicle is guided to discharge to the power grid during the load peak period and is charged during the valley period.

[0094] The price type demand response: the user is guided to adjust the power consumption behavior through the dynamic price signal, and the transferable load is transferred from the peak period to the valley period, so as to reduce the power consumption cost:

[0095]

[0096] ΔP = λ (P - Pavg) (1) DR1,t ΔP = λ (P - Pavg) (1) DR1,i λ is the price elasticity coefficient of the user i, which represents the sensitivity of the user to the price; λ t P is the time-of-use price of the period t; Pavg is the average price of the whole day.

[0097] The time-of-use price λ t guides the user to adjust the interruptible load, and the load is transferred from the peak period to the valley period, and the total amount of the load is unchanged.

[0098]

[0099] The alternative type demand response: in the heat load demand, the heat pump and the gas boiler are dynamically switched:

[0100]

[0101] ΔQ = λ (Q - Qavg) (2) DR2,t ΔQ = λ (Q - Qavg) (2) heat pump,t Q is the heat pump heating power, which is driven by electric energy; Q boiler,t Q is the gas boiler heating power, which is driven by gas; COP is the performance coefficient of the heat pump; η boiler η is the boiler efficiency; λ t λ is the price of the period t; λ gas Qavg is the gas price.

[0102] The step S4 is executed in real time, and the electric vehicle is guided to discharge to the power grid during the load peak period and is charged during the valley period.

[0103] The technical effect and advantage of the present application: compared with the prior art, the garden energy double-layer optimization scheduling method considering electric vehicles and demand response has the following advantages:

[0104] Firstly, the cooperative optimization of electric vehicle charging and discharging behavior and demand response mechanism is innovatively introduced in the traditional energy scheduling model, and through the cooperation of V2G technology and price type and substitution type DR strategy, the interaction efficiency of source-grid-load-storage is significantly improved; secondly, a double-layer optimization architecture is adopted to model the garden comprehensive energy system, the upper multi-objective optimization fully considers the wind and light output prediction and load characteristics, and the lower real-time scheduling accurately responds to uncertain factors, providing more accurate optimization decision for garden energy management; finally, aiming at the model solving problem, MOSBEO and IBEO are innovatively designed to solve the upper and lower models respectively, and through the improvement strategies such as chaotic initialization and dynamic weight adjustment, the algorithm solving efficiency is improved, and finally the optimal scheduling scheme considering economy, low carbon and flexibility is obtained, meeting the diversified operation demand of garden energy system. BRIEF DESCRIPTION OF DRAWINGS

[0105] Figure 1 It is a garden energy double-layer optimization scheduling method considering electric vehicles and demand response according to the present application.

[0106] Figure 2 It is a garden comprehensive energy system structure schematic diagram according to the present application.

[0107] Figure 3 It is a garden double-layer multi-objective optimization scheduling model solving flowchart according to the present application. CONCRETE IMPLEMENTATION METHOD

[0108] The related technical solutions will be described in detail below in combination with the drawings in the embodiments of the present application. Figure 1 The garden energy double-layer optimization scheduling method flowchart is introduced; Figure 2 It is a garden comprehensive energy system structure schematic diagram; Figure 3 The improved black eagle optimization algorithm and the double-layer scheduling solving flowchart of the multi-objective black eagle optimization algorithm are introduced, and the algorithm process is reasonably designed and improved, so that it can better solve the garden double-layer multi-objective optimization scheduling problem. It should be noted that the embodiments only represent some implementation cases of the present application, and do not cover all possible technical forms. The specific implementation cases listed in the present application are only used to explain and describe the present application, and should not be regarded as a limiting definition of the present application. Based on the embodiments disclosed in the present application, all technical solutions derived by the related technical personnel without creative labor belong to the protection scope of the present application.

[0109] With reference to the drawings shown, the present application proposes a kind of park energy double-layer optimization scheduling method considering electric vehicle and demand response, comprising the following steps:

[0110] S1: build park integrated energy system, including energy supply side, energy conversion side and demand side;

[0111] S2: establish double-layer optimization model, upper layer is multi-objective optimization model, goal is to minimize system total cost, minimize carbon emission and maximize renewable energy consumption rate;Lower layer is single-objective optimization model, goal is to minimize electric vehicle user charging cost and power grid load fluctuation;

[0112] In the embodiment of the present application, a double-layer optimization model is established, the upper layer is a multi-objective optimization model, the goal is to minimize the total cost of the system, minimize carbon emissions and maximize renewable energy consumption rate, the time-of-use electricity price, gas turbine output, energy storage charging and discharging power, and the operating state of the electric-gas conversion equipment are optimized, and the constraint conditions include power balance, device climbing rate, energy storage capacity limit, carbon emission constraint;The lower layer scheduling model minimizes the electric vehicle charging cost and power grid load fluctuation as the target, and optimizes the electric vehicle charging and discharging power and demand response load adjustment amount, and the constraint conditions include electric vehicle battery capacity, charging and discharging rate and user comfort.

[0113] S3: based on Pareto theory, design multi-objective secretary black eagle optimizer (MOSBEO) to solve the upper multi-objective model;Improved black eagle optimizer (IBEO) is introduced to improve the lower single-objective model by introducing chaos initialization, adaptive step adjustment and Gaussian disturbance strategy.

[0114] S4: through the vehicle-grid interaction and demand response coordination mechanism, execute scheduling and feedback correction day-ahead plan.

[0115] In the present application, in the peak period, EV is guided to discharge to power grid, and charging is arranged in the valley period, while combining price type DR to transfer adjustable load and replace type DR to switch electric heating energy form, dynamic fluctuation of power grid is suppressed;In the execution process, wind and light output deviation, load change and EV state data are collected in real time, are fed back to the upper model to correct time-of-use electricity price strategy, update equipment output plan and energy storage scheduling scheme, form "real-time response-data feedback-dynamic correction" closed-loop optimization mechanism, thereby improve the adaptability of system to uncertainty, Ensure the continuous optimization of economy, low carbon and power grid stability.

[0116] In the implementation of the present application, electric vehicles are not only consumers of electric energy, but also producers of electric energy, which can be fed back to the power grid. This interaction can improve the stability and reliability of the power grid, optimize energy distribution, reduce electricity costs, and create additional economic value for vehicle owners. Frequent charging and discharging can affect the life of electric vehicle batteries. The main purpose of introducing dynamic battery wear compensation is to reasonably assess and compensate for the wear and tear on battery life caused by charging and discharging behavior when electric vehicles participate in V2G, thereby protecting the interests of electric vehicle users and promoting the sustainable application of V2G mechanisms. DR strategies include price-based demand response, which guides users to shift interruptible loads to off-peak periods through time-of-use pricing; and alternative demand response, which uses heat pumps to replace gas boilers in thermal load demand to reduce gas consumption.

[0117] In the embodiment of the present application, a hierarchical progressive optimization solution strategy is adopted. First, MOSBEO is used to solve the upper day-ahead scheduling model, and the Pareto frontier solution set is obtained through non-dominated sorting and crowding distance calculation to determine the optimal time-of-use pricing strategy, device output plan and energy storage scheduling scheme. Then, based on the upper optimization result, IBEO is used to solve the lower real-time scheduling model to optimize the electric vehicle charging and discharging power and demand response load adjustment, thereby obtaining the best park energy optimization scheduling scheme.

[0118] Specifically, step S1 includes the following steps:

[0119] S11, the wind speed change of the wind power generation system satisfies the Weibull distribution, that is:

[0120]

[0121] In the formula: v represents the actual wind speed; k and c are the shape factor and scale factor of the Weibull distribution, and k is generally taken as 1.8 to 2.8; combined with the wind speed model, the mathematical model of the wind power generation system output can be obtained:

[0122]

[0123] In the formula, P wind (t) and P rated are the active output and rated output of the fan output; V cut-in , V cut-out , V rated are the cut-in, cut-out and rated wind speed of the fan, respectively.

[0124] S12, the mathematical model of the photovoltaic power generation system:

[0125]

[0126] In the formula: P pv(t) represents the output power of the photovoltaic device at time t; γ represents the temperature coefficient; P STC represents the rated power of the photovoltaic device; G(t) represents the standard irradiance of the photovoltaic plant; G STC represents the irradiance of the photovoltaic plant; T c (t) represents the ambient temperature at time t; T STC represents the standard ambient temperature.

[0127] S13, mathematical model of the gas boiler:

[0128] H boiler (t) = η boiler · Q gas-boiler (t) (4)

[0129] where: H boiler (t) is the thermal energy output at time t; Q gas-boiler (t) is the natural gas consumption at time t; η boiler is the efficiency of the gas boiler.

[0130] S14, mathematical model of the energy storage device:

[0131]

[0132] where: SOC(t) represents the state of charge of the energy storage device at time t η ch , η dis is the charge / discharge efficiency; P ch (t), P dis (t) is the charge / discharge power.

[0133] S15, mathematical model of the power-to-gas:

[0134] Q P2G (t) = η P2G · P P2G (t) (6)

[0135] where: Q P2G (t) is the gas production at time t; P P2G (t) is the electrical energy input at time t; η P2G is the conversion efficiency.

[0136] S16, mathematical model of the cogeneration:

[0137]

[0138] where: P CHP-e (t) is the electrical energy output at time t; H CHP (t) is the thermal energy output at time t; Q gas-CHP (t) is the natural gas consumption at time t; ηe η is the power generation efficiency h η is the heat recovery efficiency.

[0139] S17, mathematical model of the heat pump:

[0140] H HP (t) = COP P HP (t) (8)

[0141] H HP (t) is the thermal energy output at time period t; P HP (t) is the electrical energy input at time period t; COP is the coefficient of performance.

[0142] Specifically, step S2 includes the following steps:

[0143] S21, an upper-layer multi-objective optimization model is established, the objectives being to minimize the total system cost F1, to minimize the carbon emission F2, and to maximize the renewable energy consumption rate F3, and the objective function being:

[0144]

[0145] C purchase is the cost of purchasing electricity and gas from the power grid; C operation is the equipment operation and maintenance cost; C penalty is the penalty cost of abandoned wind and light; P CHP-e is the cogeneration output; a is the carbon emission coefficient of the cogeneration unit output; Q gas-boiler is the fuel consumption of the gas boiler; b is the carbon emission coefficient of the gas boiler; P wind+PV-actual is the actual output of wind power and photovoltaic; P wind+PV-predict is the predicted output of wind power and photovoltaic.

[0146] S211, the energy purchasing cost, including the cost of purchasing electricity from the power grid, purchasing gas and purchasing gas from external gas sources:

[0147]

[0148] T is the dispatching period, taken as T = 24 hours; λ grid,t is the electricity price of the power grid at time period t; P grid,t is the power grid electricity purchasing power at time period t; λ gas is the gas price; Q gas,t is the gas purchasing amount at time period t.

[0149] S212, the operation and maintenance cost, covering the operation and maintenance fees of the gas boiler, the cogeneration, the energy storage, the electric-to-gas equipment, etc.:

[0150]

[0151] kOM,i Pi is the unit power operation and maintenance cost of equipment i; P i,t Pi is the output power of equipment i at time period t.

[0152] S213, the penalty cost of abandoned wind and light, the penalty cost caused by insufficient renewable energy consumption.

[0153]

[0154] In the formula: γ wind ,γ PV are the unit penalty coefficients of abandoned wind and light; ΔP wind,t ,ΔP PV,t are the abandoned wind and light power at time period t.

[0155] S214, minimize carbon emissions, which mainly come from the operation of combined heat and power and gas boiler heating equipment:

[0156]

[0157] In the formula: T is the scheduling period, T=24 hours, P CHP-e is the power generation of combined heat and power at time period t; Q gas-boiler is the natural gas consumption of gas boiler at time period t.

[0158] S215, maximize renewable energy consumption rate:

[0159]

[0160] In the formula: T is the scheduling period, T=24 hours, are the actual output of wind power and photovoltaic at time period t; are the predicted output of wind power and photovoltaic at time period t.

[0161] S22, establish a single-objective optimization model at the lower level, the goal is to minimize electric vehicle user charging cost and grid load fluctuation:

[0162] min f=α·P EV +β·ΔP grid (15)

[0163] In the formula: P EV is the electric vehicle user charging cost; ΔP grid is the grid load fluctuation; α and β are weight coefficients, and α+β=1, used to balance the charging cost and load fluctuation.

[0164] S221, electric vehicle user charging cost:

[0165]

[0166] P(t) = P(t-1) + P(t) - P(t-1) (1) EV-charge,t P(t) = P(t-1) + P(t) - P(t-1) (1) EV-discharge,t P(t) = P(t-1) + P(t) - P(t-1) (1) EV-charge,t P(t) = P(t-1) + P(t) - P(t-1) (1) t P(t) = P(t-1) + P(t) - P(t-1) (1) loss P(t) = P(t-1) + P(t) - P(t-1) (1)

[0167] S222, Load fluctuation of the power grid:

[0168]

[0169] P(t) = P(t-1) + P(t) - P(t-1) (1) grid,t P(t) = P(t-1) + P(t) - P(t-1) (1) P(t) = P(t-1) + P(t) - P(t-1) (1)

[0170] S23, Specific constraints are as follows:

[0171] S231, Active power output constraint of distributed power supply:

[0172] P(t) = P(t-1) + P(t) - P(t-1) (1) min P(t) = P(t-1) + P(t) - P(t-1) (1) max (18)

[0173] P(t) = P(t-1) + P(t) - P(t-1) (1) min P(t) = P(t-1) + P(t) - P(t-1) (1) max P(t) = P(t-1) + P(t) - P(t-1) (1)

[0174] S232, Output constraint of gas boiler:

[0175]

[0176] P(t) = P(t-1) + P(t) - P(t-1) (1) P(t) = P(t-1) + P(t) - P(t-1) (1)

[0177] Ramp rate constraint of gas boiler:

[0178] P(t) = P(t-1) + P(t) - P(t-1) (1) P(t) = P(t-1) + P(t) - P(t-1) (1)

[0179] P(t) = P(t-1) + P(t) - P(t-1) (1) P(t) = P(t-1) + P(t) - P(t-1) (1)

[0180] S233, Constraint of energy storage device:

[0181] SOC(t) = SOC(t-1) + P(t) - P(t-1) (1) min SOC(t) = SOC(t-1) + P(t) - P(t-1) (1) max (21)

[0182]

[0183] SOC(t) = SOC(t-1) + P(t) - P(t-1) (1) minand SOC max respectively the upper and lower limits of the state of charge of the energy storage device; P ch (t) and P dis (t) are respectively the charge and discharge power of the energy storage device.

[0184] S234, operating constraints of the electric-gas conversion:

[0185]

[0186] where: is the maximum input power of the electric-gas conversion device; is the maximum ramp rate.

[0187] S235, power output constraints of the cogeneration:

[0188]

[0189] where: are respectively the minimum and maximum values of the electric energy output; are respectively the minimum and maximum values of the thermal energy output.

[0190] Ramp rate constraints of the cogeneration:

[0191]

[0192] where: are respectively the maximum ramp rates of the electric and thermal energy.

[0193] S236, operating constraints of the heat pump:

[0194]

[0195] where: is the maximum thermal energy output of the heat pump; is the maximum ramp rate of P

[0196] S237, total electric power balance constraint:

[0197]

[0198] where: P wind,t is the wind power active output at time t; P PV,t is the photovoltaic active output at time t; P grid,t is the total electricity purchase from the main grid at time t; P CHP-e,t is the cogeneration power at time t; is the energy storage discharge power at time t; is the basic electric load at time t; is the electric vehicle charging power at time t; P P2G,t is the electric-gas conversion device power consumption at time t; is the energy storage charging power at time t.

[0199] S238, thermal power balance:

[0200]

[0201] wherein: H boiler,t is the thermal energy output in period t; Q CHP,t is the cogeneration waste heat at time t; Q heat pump,t is the heat pump heating at time t. is the thermal load demand at time t.

[0202] S239, gas power balance:

[0203]

[0204] wherein: Q gas,t is the external gas purchase at time t; Q P2G,t is the gas production of the electric-gas conversion equipment at time t. is the industrial gas at time t; Q gas turbine,t is the gas consumption of the gas turbine at time t. Q gas-boiler,t is the natural gas consumption in period t.

[0205] S2310, carbon emission constraint, the total carbon emission of the park per day must not exceed the policy or planning limit:

[0206]

[0207] wherein: T is the dispatching period, taken as T = 24 hours, and a is the carbon emission coefficient of cogeneration; P CHP-e is the cogeneration output in period t; b is the carbon emission coefficient of the gas boiler; Q gas-boiler is the fuel consumption of the gas boiler.

[0208] S24, electric vehicle constraint:

[0209] S241, battery capacity limitation, to ensure that the state of charge (SOC) of the EV battery is within a reasonable range:

[0210] SOC min ≤ SOC t ≤ SOC max (31)

[0211] wherein: taken as typical values SOC min = 20%; and SOC max = 80%.

[0212] S242, the charging and discharging power limit, the EV charging and discharging power cannot exceed the rated value of the device:

[0213]

[0214] In the formula: is the maximum discharging power of the EV; is the maximum charging power of the EV.

[0215] S243, user travel demand, the SOC of the EV when leaving needs to meet the user travel demand:

[0216] SOC departure ≥ SOC required (33)

[0217] In the formula: SOC departure is the amount of electricity of the EV leaving the charging pile; SOC required is the amount of electricity required for the EV user to travel.

[0218] S25, demand response constraint:

[0219] S251, price type DR adjustment range, the adjustment amount of the transferable load:

[0220]

[0221] In the formula: ΔP DR1,t is the load adjustment amount at time t; is the maximum allowed load adjustment amount at time t.

[0222] Total load conservation, the load is transferred from the peak period to the valley period, and the total amount does not change:

[0223]

[0224] S252, alternative type DR thermal comfort, indoor temperature needs to meet the user comfort requirement:

[0225]

[0226] In the formula: take typical value

[0227] S26, time-of-use electricity price response constraint:

[0228] Electricity price signal transmission, the lower layer model receives the time-of-use electricity price signal transmitted by the upper layer, and dynamically adjusts the real-time electricity price:

[0229]

[0230] In the formula: is the current electricity price; Time-of-use price for upper layer transmission; Real-time price adjustment amount, dynamically adjusted according to actual wind and light output and load deviation.

[0231] Specifically, step S3 includes the following steps:

[0232] S31, input the relevant data of the park, wind and light historical output data, EV charging demand, basic electricity / heat / gas load, time-of-use electricity price initial value, equipment operation parameters: including ramp rate, energy storage capacity, etc.; output the optimal solution set through the upper multi-objective optimization: including real-time electricity price, equipment output plan, energy storage strategy. The lower layer receives the real-time electricity price of the upper layer, real-time data: including wind and light actual output, EV real-time SOC, user load demand, optimizes the EV charging and discharging plan and DR load adjustment amount with the minimum charging cost and load fluctuation as the target.

[0233] S32, solution of the upper layer model:

[0234] S321, initialize the black hawk population: set the population size N, the maximum number of iterations T, the archive size |A| max , the initial step size α0, the chaos parameter r1∈[0,1]; use Latin hypercube sampling to generate the initial population:

[0235]

[0236] In the formula: π j (i) represents the random arrangement of the jth variable dimension; u i,j is a uniform random number, u i,j ~U(0,1), used for fine-tuning the sampling position in the sub-interval.

[0237]

[0238] In the formula: X (i) represents all decision variables contained in the problem, including time-of-use electricity price λ t , combined heat and power output P CHP,t , energy storage charging and discharging power P ESS,t , and the operation state S P2G,t of electric-to-gas.

[0239] S322, fitness calculation and constraint processing:

[0240] S3221, objective function calculation: calculate the values of three objective functions total cost F1, carbon emission F2, and renewable energy consumption rate F3;

[0241] S3222, constraint processing by penalty function method:

[0242]

[0243] where g j is the constraint violation value, such as power imbalance, SOC out-of-bound value, etc.; λ = 10 5 , greatly increases the fitness value of infeasible solutions, making them eliminated in evolution.

[0244] S323, Non-dominated sorting and crowding distance calculation:

[0245] S3231, Non-dominated sorting: divide the population into multiple non-dominated layers, Front1, Front2,...; solutions in Front1 are the current Pareto optimal solution set;

[0246] S3232, Calculate crowding distance:

[0247]

[0248] After sorting each objective function value, calculate the normalized distance of adjacent solutions to measure the sparsity of the solution distribution.

[0249] S324, External archive update:

[0250] S3241, Merge non-dominated solutions: add the non-dominated solutions Front1 of the current population to the external archive A.

[0251] S3242, De-duplication and truncation: remove duplicate solutions, if |A| > |A| max , keep the first |A| max solutions in order of crowding distance from high to low to ensure solution set diversity.

[0252] S325, Dynamic weight adjustment: adjust the target weight adaptively according to the distribution of solutions in the archive:

[0253]

[0254] where AvgCrowdingDistance k is the average crowding distance in the kth objective direction.

[0255] S326, Black Hawk behavior model update:

[0256] S3261, Tracking strategy: randomly select a reference solution X ref from the external archive A; update the individual position:

[0257]

[0258] where X ref represents a random selection from the archive A; α(t) = α0·exp(-t / T), the step size decays with iteration; t1 is generated by the Tent chaotic mapping, enhancing the global search ability.

[0259] S3262, spiral strategy: search in the target space by rotating, adjust the distribution direction of the solution:

[0260]

[0261] In the formula: m is a multi-target rotation matrix, which is mapped to the principal component direction of the target space.

[0262] S3263, capture and seize strategy: adaptively adjust the position of the solution, balance convergence and diversity:

[0263]

[0264] In the formula: D1 and D2 are global exploration and local development factors, balanced by dynamic weight.

[0265] S327, population update and elite selection

[0266] Merge parent and child to generate 2N candidate solutions; elite selection, hierarchical according to non-dominated sorting, and sort according to crowding degree in the same layer, select the first N solutions as the new generation population.

[0267] S328, convergence judgment and output

[0268] Termination condition, if the maximum iteration number T is reached or the change rate of Pareto front hyper volume is less than 10 -4 , then terminate the algorithm, output the Pareto optimal solution set in the external storage A for decision maker selection; otherwise, return to step S322 and continue iteration.

[0269] S33, select the optimal scheduling scheme from the output Pareto optimal solution set in the upper layer and pass it to the lower layer to solve the lower layer model:

[0270] S331, chaotic initialization of black hawk population: set population size N, maximum iteration number T, initial step factor α 初始 , disturbance intensity η; chaotic initialization, Tent mapping:

[0271]

[0272] X = {X1, X2,... X n} (47)

[0273] In the formula: each individual X i is a decision variable, including the charging and discharging rate P EV,t of the electric vehicle and the adjustment range ΔL DR,t of the DR.

[0274] S332, fitness calculation and constraint processing:

[0275] S3321. For each individual X i Calculate the objective function value:

[0276] f(X i )=α·C EV +β·ΔP load (48)

[0277] In the formula: P EV Charging costs for electric vehicle users; ΔP grid The load fluctuation is represented by α and β, which are weighting coefficients.

[0278] S3322. Calculate the degree of constraint violation and use the penalty function method to handle constraints:

[0279]

[0280] Where: g j To constrain violations, such as power imbalance or battery capacity SOC exceeding limits; λ is the penalty factor.

[0281] S333, Black Hawk Behavior Model Update:

[0282] S3331. Tracking strategy, update individual location:

[0283]

[0284] In the formula: r1∈[0,1] random number; t1 is the chaotic number generated by the Tent mapping; X best This represents the current optimal solution; adaptive step size adjustment, where the step size factor α(t) dynamically decays with the number of iterations, balancing global exploration and local development.

[0285] S3332, hovering strategy: rotates to search for individual positions, enhancing local development capabilities.

[0286]

[0287] In the formula: m is the rotation matrix, which simulates the adjustment of the search direction in high-dimensional space.

[0288] S3333, Capture and plunder strategy: Adjust individual positions to balance exploration and development:

[0289]

[0290] In the formula: D1 is the global search direction correction factor, and D2 is the local convergence factor.

[0291] S334, Local Gaussian perturbation:

[0292] S3341, evaluate the current optimal solution X best S3342, perturb the solution: generate a perturbed solution X

[0293]

[0294] where δ(t) = δ 初始 (1-t / T), the standard deviation decreases with iteration, and the perturbation range is limited within the feasible region of the variable.

[0295] S3342, perturb the solution: generate a perturbed solution X penalty (X new ), if better than the current optimal solution, update X best .

[0296] S335, population update and elitist reservation:

[0297] Merge the parent and child populations, combine the current population, newly generated solutions and perturbed solutions to form 2N+1 candidate solutions; elitist selection, sort by fitness value, reserve the top N optimal solutions.

[0298] S336, termination condition:

[0299] If the maximum number of iterations T is reached, or if the optimal fitness change rate is less than 10 -4 for 10 consecutive generations, terminate the algorithm; output the optimal solution X best and its corresponding P EV,t and ΔL DR,t . Otherwise, return to step S332 and continue iteration.

[0300] Specifically, step S4 includes the following steps:

[0301] S41, V2G discharge and charging strategy:

[0302] Electric vehicles discharge to the grid during peak hours according to real-time price signals, reducing the cost of grid electricity purchases; electric vehicles charge during off-peak hours using low-cost electricity, reducing user costs.

[0303] S42, demand response execution:

[0304] S421, price-based DR, price-based DR guides users to adjust their electricity consumption behavior through dynamic price signals, shifting transferable loads from peak hours to off-peak hours to reduce electricity costs:

[0305]

[0306] where ΔP DR1,t is the price-based DR load adjustment amount for period t; λ DR1,i is the price elasticity coefficient of user i, indicating the sensitivity of the user to the price; λt The time-of-use electricity price for time period t; The average electricity price for the whole day.

[0307] S422, the alternative DR, in the heat load demand, compares the unit heating cost of electric-driven heat pump and gas boiler, selects the energy form with lower cost, realizes the flexible conversion and replacement of electricity-heat load, and dynamically switches the heat pump and gas boiler:

[0308]

[0309] ΔQ DR2,t represents the alternative DR heat load adjustment amount for time period t; Q heat pump,t represents the heat pump heating power, electric energy driven; Q boiler,t represents the gas boiler heating power, gas driven; COP represents the heat pump performance coefficient; η boiler represents the boiler efficiency; λ t represents the electricity price for time period t; λ gas represents the gas price.

[0310] S43, V2G dynamic compensation mechanism, when the electric vehicle discharges, the battery loss cost is calculated based on the discharge depth, and the user charging cost is calculated:

[0311]

[0312] k is the battery compensation consumption coefficient; E discharge,t is the discharge energy for time period t; E capacity the rated capacity of the electric vehicle battery; DOD t = 1-SOC t , is the discharge depth.

[0313] S44, real-time execution and feedback

[0314] During the execution process, the wind-solar output deviation, load change and EV state data are collected in real time, and are fed back to the upper model to correct the dynamic electricity price strategy, update the equipment output plan and the energy storage scheduling scheme. If both the double layers have converged, the final scheduling scheme is output, otherwise, step S232 is returned, and the iteration is continued.

[0315] Any processes or methods described in the flow charts or otherwise described herein can be understood as representing modules, segments, or portions of code that include one or more executable instructions for implementing the specified logical functions or steps, and the preferred embodiments of the application include additional or fewer steps, in other orders, with other functionality, in implementations of these preferred embodiments of the application. Thus, any of the steps, options, aspects, components, etc. discussed herein can be included or deleted in other embodiments of the application, and yet still be deemed to fall within the scope of the present application.

[0316] Although the embodiments of the present application have been shown and described above, it should be understood by those ordinary skilled in the art that the above embodiments are exemplary and cannot be construed as limiting the present application, and those ordinary skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1.A method for double-layer optimal scheduling of park energy considering electric vehicles and demand response, characterized in that, The method comprises the following steps: S1: constructing a park comprehensive energy system, including an energy supply side, an energy conversion side and a demand side; S2: establishing a double-layer optimization model, the upper layer being a multi-objective optimization model, the objectives being to minimize the total system cost, minimize carbon emissions and maximize renewable energy consumption rate; the lower layer being a single-objective optimization model, the objective being to minimize the electric vehicle user charging cost and grid load fluctuation; S3: based on the Pareto theory, designing a multi-objective secretary black eagle optimizer (MOSBEO) to solve the upper-layer multi-objective model; introducing a chaos initialization, an adaptive step adjustment and a Gaussian disturbance strategy to improve the improved black eagle optimizer (IBEO) to solve the lower-layer single-objective model; S4: through a vehicle-grid interaction and demand response coordination mechanism, performing real-time scheduling and feeding back and correcting the day-ahead plan. 2.The method of claim 1, wherein the method further comprises: The energy supply side of the park comprehensive energy system comprises wind power, photovoltaic units, grid electricity purchase and battery energy storage devices; the energy conversion side comprises electric-gas conversion, combined heat and power, heat pumps and gas boilers; and the energy demand side comprises basic load, adjustable load and electric vehicle clusters. 3.The method of claim 2, wherein the method further comprises: The upper layer is a multi-objective optimization model, the objectives being to minimize the total system cost F1, minimize carbon emissions F2 and maximize renewable energy consumption rate F3, and the objective function expression is: In the formula, C purchase is the cost of purchasing electricity and gas from the power grid; C operation is the equipment operation and maintenance cost; C penalty is the penalty cost of abandoned wind and light; P CHP-e is the cogeneration output; a is the carbon emission coefficient of the cogeneration unit output; Q gas-boiler is the fuel consumption of the gas boiler; b is the carbon emission coefficient of the gas boiler; P wind+PV-actual is the actual output of wind power and photovoltaic; P wind+PV-predict is the predicted output of wind power and photovoltaic. 4.The method of claim 2, wherein the method further comprises: The decision variables of the upper-layer objective function include time-of-use electricity price, gas boiler output, combined heat and power output, energy storage charging and discharging power and electric-gas conversion device operating state; and the constraint conditions include electric / thermal / gas power balance constraint, device ramp rate constraint, energy storage capacity constraint and carbon emission limit constraint. 5.The method of claim 2, wherein the method further comprises: The lower layer is a single-objective optimization model, the objective being to minimize the electric vehicle user charging cost and grid load fluctuation, and the objective function expression is: minf = a - P EV + β - ΔP grid (2) In the formula, P EV is the charging cost of the electric vehicle user; ΔP grid is the load fluctuation of the power grid; α and β are weight coefficients, and α+β=1, for balancing the charging cost and the load fluctuation. 6.The method of claim 2, wherein the method further comprises: The decision variables of the lower-layer objective function include electric vehicle charging and discharging power and adjustable load adjustment amount; and the constraint conditions include electric vehicle battery capacity constraint, charging and discharging rate constraint and user comfort constraint. 7.The method of claim 3, wherein: The improvements of the MOSBEO include: Dynamic weight adjustment: adaptively adjusting the target weight according to the population distribution to improve the uniformity of the Pareto frontier: where CrowdingDistance k is the average crowding distance of the kth objective. Non-dominated sorting and crowding degree screening: non-dominated sorting, stratifying the population according to the Pareto dominance relationship; and crowding degree calculation, measuring the distribution density of the solution in the objective space: External archive management: archive update, incorporate non-dominated solutions of current population, remove dominated solutions; truncation strategy, if archive size exceeds |A max , keep solutions in descending order of crowding distance. 8.The method of claim 3, wherein the method further comprises: The improvements of the IBEO include: The IBEO introduces chaos initialization adjustment, adaptive step and local search strategy on the basis of the original BEO to improve the solving efficiency and accuracy. Chaos initialization: to avoid uneven distribution of the initial population, a Tent mapping is used to generate a chaotic sequence: Adaptive step adjustment: the step factor alpha(t) is dynamically adjusted with the iteration number: The step is large in the initial stage to enhance the global search ability; and the step is reduced in the later stage to improve the local optimization accuracy. Local search strategy: applying Gaussian disturbance to the neighborhood of the current optimal solution to avoid falling into local optimum: Wherein, η is the disturbance intensity, usually set as 5% to 10% of the decision variable range; δ is the standard deviation of Gaussian distribution, which decreases with the iteration number. 9.The method of claim 4, wherein the method further comprises: The vehicle-to-grid interaction mechanism includes discharging of the electric vehicle to the power grid during a load peak period and charging during a load valley period. When the electric vehicle discharges, a compensation cost is dynamically calculated based on a battery discharge depth, and the formula is: where k is the battery compensation loss coefficient; E discharge,t is the discharge energy for time period t; E capacity is the electric vehicle battery rated capacity; DOD t = 1 - SOC t is the depth of discharge. 10.The method of claim 4, wherein the method further comprises: The demand response strategy includes a price-type demand response and a substitution-type demand response. The price-type demand response: through a dynamic price signal, a user adjusts an electricity consumption behavior, a transferable load is transferred from a peak period to a valley period, and an electricity consumption cost is reduced: where ΔP DR1,t is the price-type demand response load adjustment amount for time period t; λ DR1,i is the price elasticity coefficient of user i, which represents the sensitivity of the user to the electricity price; λ t Time-of-use price for time period t; Average price for the day. By time-of-use electricity price λ t Directing users to adjust interruptible loads, shifting loads from peak hours to off-peak hours, without changing the total amount of load. The substitution-type demand response: in a heat load demand, a heat pump and a gas boiler are dynamically switched. where ΔQ DR2,t denotes the alternative demand response heat load adjustment amount price for the time period t; Q heat pump,t denotes the heat pump heating power, electrically driven; Q boiler,t denotes the gas boiler heating power, gas driven; COP denotes the heat pump performance coefficient; η boiler denotes the boiler efficiency; λ t denotes the electricity price for the time period t; λ gas denotes the gas price.

Images