CN113258561A - Multi-attribute decision-based multi-distributed power supply micro-grid multi-objective optimization scheduling method - Google Patents

Abstract

The invention is a multi-objective optimal scheduling method for a micro-grid with multiple distributed power sources based on multi-attribute decision-making. For the micro-grid system including multiple distributed power sources, the invention establishes the lowest operating cost, the least pollution emission and the lowest power loss rate. The multi-objective microgrid optimization scheduling model is based on the particle swarm algorithm based on the penalty function to optimize separately with different objectives, and the candidate solution set is obtained after multiple optimizations. Analytic Hierarchy Process analyzes the weight of each objective, conducts confidence analysis on each objective function on its corresponding evaluation level, and uses evidence fusion and utility value analysis to obtain the optimal solution with the largest utility value. , improve the utilization rate of renewable energy and reduce system operating costs.

Description

Multi-attribute decision-based multi-distributed power supply micro-grid multi-objective optimization scheduling method

Technical Field

The invention belongs to the technical field of microgrid optimization, and particularly relates to a distributed multi-objective optimization scheduling control method for a microgrid.

Background

With the increasing environmental problems and energy crisis, there is an urgent need in the current society to change the energy structure and replace non-renewable energy with renewable clean energy. Microgrid technology arose from this background. The micro-grid technology can not only exert the advantages of clean power generation of new energy represented by wind and light, but also effectively reduce the adverse effects of the intermittency and volatility of the new energy on a large power grid. The reasonable scheduling is adopted, so that the utilization rate of the distributed power supply can be effectively improved, and the economic cost and the environmental cost of the operation of the micro-grid are reduced while the stability of the system is ensured. Due to the fact that the types of devices of the distributed power supply are various and the characteristics of the devices are different, the micro-grid optimization scheduling gradually changes from the former single target to the multi-target multi-constraint non-linear programming problem.

At present, multi-objective optimization scheduling for a microgrid is usually analyzed by an improved intelligent algorithm, however, sub-objectives of the microgrid scheduling are often mutually constrained, the improvement on the performance of one sub-objective possibly affects the performance of another sub-objective or even other sub-objectives, and finally, the multi-objective optimization scheduling for the microgrid can only select one scheme for decision making. In view of the above problems, the schemes obtained after the intelligent algorithm analysis are reasonably evaluated, and the selection of a unique and appropriate scheduling decision scheme from the candidate schemes becomes an important part of the scheduling method.

Disclosure of Invention

The invention mainly aims to solve the problems existing in the prior art background, and provides a multi-objective optimization scheduling method for a micro-grid based on multi-attribute decision.

Based on the purpose, the invention mainly adopts the following technical scheme: a multi-attribute decision-based micro-grid multi-objective optimization scheduling method mainly comprises the following steps:

step 1: carrying out optimization modeling on the microgrid according to the structure of the microgrid and the characteristics of all distributed power supplies, establishing each objective function with the lowest economic cost, the lowest emission of pollutant gas and the lowest loss rate of the microgrid, constructing constraint conditions of a system objective, and taking the microgrid as a research object in a grid-connected running state, wherein dynamic optimization is adopted and divided into 24 time periods, and each time period is 1 hour;

step 2: respectively optimizing the target function for multiple times by adopting a particle swarm algorithm based on a penalty function according to the instruction of a decision maker, taking the optimal solution of each optimization result as a candidate solution set, and obtaining a plurality of candidate schemes;

and step 3: and evaluating each objective function value in the obtained candidate objective schemes by adopting a multi-attribute decision theory, performing evidence fusion by adopting an evidence reasoning method, and selecting the candidate scheme with the maximum utility value as a final solution by adopting utility evaluation.

Further, the microgrid system in the step 1 mainly comprises a photovoltaic power generation system, a wind power generation system, a storage battery, a fuel cell and a control center, local daily typical photovoltaic output, wind power output and load data are obtained as system prediction data, and system operation parameters refer to local electricity price, pollution discharge treatment cost and system parameters of each distributed power supply;

the objective function of the system with the lowest economic cost of the microgrid is

mine₁＝F₁+F₂+F₃

Consideration of F₁Cost of purchasing electricity from the grid, F₂Fuel cell fuel cost and F₃The operating maintenance cost of the distributed power supply;

the objective function for minimum emission of pollutant gases is:

mine₂＝g_FcW_FC+g_GW_G

g_Fcand g_GEmission coefficients of fuel cell FC and grid pollution gas, W_FCAnd W_GRespectively discharging the pollution gas of the fuel cell FC and the pollution gas of the power grid;

the objective function for which the deficit rate is lowest is:

the LPSP is the ratio of the electric energy shortage and the load demand when the micro-grid is powered by only an internal power supply, reflects the capability of guaranteeing the power supply when the micro-grid fails, and E_LPSIs a shortage of electric energy, P_LThe total required electric quantity of the load is;

the distributed power supply comprises photovoltaic arrays, wind driven generators, storage batteries, fuel cells and other devices. The system constraints include:

1) and power balance constraint:

2) tie line constraint:

3) and limiting the capacity of the generator set:

4) and (4) restraining the storage capacity of the energy storage unit: SOC_min≤SOC(t)≤SOC_max

5) Charging and discharging balance constraint: t is_out＝T_in≤T_max

In the formula, N is the type of the generator set; p_DG,iIs the output power of genset type i; p_DNThe method is energy exchange with a power grid, and the specified electricity selling time is a negative value, and the electricity purchasing time is a positive value; p_LIs the load demand; p_GIs the output power of the tie line;

and

respectively, the minimum safe power output and the maximum safe power output of the tie line; p_DG.iIs the output power of the generator set i;

and

respectively outputting the minimum safe power and the maximum safe power of the generator set i; t is_outIs the time of day that the energy storage device is discharged; t is_inIs the time of day that the energy storage device is charging; t is_maxIs the maximum charge-discharge time of the energy storage device.

Further, the particle swarm optimization based on the penalty function is adopted in the step 2, optimization is carried out by using objective functions of economic cost, pollution emission and power shortage rate, and the optimal solution of the optimization result is selected as a candidate solution set after multiple times of optimization; setting a penalty function and control parameters of the particle swarm algorithm based on the penalty function, wherein the penalty function and the control parameters mainly comprise the particle number, the iteration number and the penalty function factors of the particle swarm algorithm; according to the optimization conditions of different objective functions, constraint conditions are combined to form a 'punishment' item, the 'punishment' item is loaded on the original objective function to force an iteration point to approach a feasible domain, and the optimization problem with constraint can be effectively processed; the penalty function is formulated as:

wherein F (x, M) is a penalty function; (x) is an objective function; m is a penalty factor; mp (x) is a penalty term; g_i(x) Is a constraint condition;

the method comprises the steps of coding the output of a storage battery, a fuel cell and a power distribution network in a micro-grid optimization scheduling system to generate a random initial population, and optimizing and obtaining an individual and global optimal scheme meeting constraint conditions and penalty function conditions through a particle swarm optimization.

Further, in the step 3, a multi-attribute decision theory is adopted for a candidate scheme solution set obtained by the multi-objective optimization intelligent algorithm to perform quality ranking and utility value evaluation on each objective function value in the obtained candidate target scheme, and the most reasonable and effective scheme is selected; determining the evaluation index set as E ═ E₁,e₂,e₃…,e_L}, evaluation gradeSet is H ═ H₁,H₂,H_3,…,H_DThe solution set of candidate schemes is a ═ a₁,a₂,a_3,…,a_M}; l is an evaluation index set, D is the number of evaluation grades and the number of M candidate schemes;

and performing confidence evaluation on different objective functions in each candidate solution set on corresponding different evaluation levels to obtain a confidence expression as follows:

S(e_i(a_j))＝{(H_n,β_n,i(a_j)),n＝1,2,…,N,i＝1,2,…,M}

in the formula, a_jIs the jth candidate solution set; beta is a_n,i(a_j) Is a_jConfidence on the ith evaluation index and the nth evaluation level;

confidence coefficient set of candidate schemes at each evaluation level can be finally obtained by adopting evidence fusion reasoning

S(a_j)＝{(H_n,β_n(a_j)),(n＝1,2,…,N)}

The evidence fusion reasoning result can only obtain the trust of each candidate scheme on each evaluation level, the quality of the candidate solution set cannot be judged visually, and then utility evaluation is applied to map the confidence coefficient distribution into utility values:

in the formula, u (a)_j) Representation scheme a_jA utility value of;

and selecting the scheme with the maximum utility value from all the schemes as the final scheme of the micro-grid multi-objective optimization scheduling, and using the scheme as a guiding decision for the micro-grid optimization scheduling operation in the next day.

The invention has the beneficial effects that: according to the method, a microgrid multi-target mathematical model is established by taking the minimum economic cost, the minimum emission of pollution gas and the minimum power loss rate of the operation of the microgrid as targets, a multi-attribute decision theory is combined with a penalty function particle swarm algorithm to solve the model to obtain a microgrid multi-target optimal scheduling optimal scheme, the problem that the system cannot achieve global optimization due to the fact that the performance of multiple targets changes caused by the change of one target is effectively solved, scientific decision theory basis and technical support are provided for the microgrid optimal scheduling of various distributed power supplies, and the method is beneficial to improving the renewable energy consumption capability of the microgrid and the economy of microgrid operation.

Drawings

FIG. 1 is a flow chart of the overall scheme of the present invention;

FIG. 2 is a flow chart of a penalty function particle swarm algorithm;

FIG. 3 is a schematic diagram of evidence fusion in accordance with the present invention.

Detailed Description

The invention adopts a multi-attribute decision method and a particle swarm algorithm of a penalty function to solve the model to obtain an optimal operation scheme, and mainly comprises the following steps: .

Step 1: establishing a micro-grid optimization scheduling objective function and constraint conditions of various distributed power supplies:

1. objective function

1) The objective function for minimizing the system operating cost is:

mine₁＝F₁+F₂+F₃

in the formula F₁Cost of purchasing electricity from the grid, F₂Fuel consumption cost of fuel cell, F₃The operating maintenance cost of the distributed power supply; p_G(t) average power on the microgrid and the main grid connecting line in the tth scheduling period; the electricity purchasing time of the micro-grid to the main grid is a positive value, and the electricity selling time is a negative value; Δ t denotes schedulingA time period duration; g_Pr(t) showing the price of power purchased and sold on the power grid in the tth time period; c_NGIs the gas price; q_LHVIs natural gas with low heat value; p_FC(t) is the average output power of the fuel cell FC during the tth period; eta_Fc(t) is the efficiency of the fuel cell FC during the tth period; is N_DGIs the DG number in the microgrid; k_OM,iOperating and maintaining a cost coefficient for the ith DG unit; p_i(t) outputting power for the ith DG during the tth scheduling period;

2) the objective function of the system for minimizing pollutant emission is as follows:

mine₂＝g_FcW_FC+g_GW_G

in the formula e₂The discharge amount of the pollution gas is reduced; g_FcAnd g_GThe emission coefficients of the fuel cell FC and the power grid pollution gas are respectively; w_FCAnd W_GRespectively discharging the pollution gas of the fuel cell FC and the pollution gas of the power grid; p_FC(t) is the average output power of the fuel cell FC during the tth period; eta_Fc(t) is the efficiency of the fuel cell FC during the tth period; p_GAnd (t) is the average power on the microgrid and the main network connecting line in the tth scheduling period.

3) The lowest objective function of the system power shortage rate is as follows:

in the formula, the power shortage rate of the LPSP adopted by the system can be used for representing the reliability of power supply of the microgrid system, and the smaller the LPSP is, the higher the reliability of power supply is; e_LPSThe power is insufficient.

2. The constraint conditions include:

1) and power balance constraint:

2) tie line constraint:

3) and limiting the capacity of the generator set:

4) and (4) restraining the storage capacity of the energy storage unit: SOC_min≤SOC(t)≤SOC_max

5) Charging and discharging balance constraint: t is_out＝T_in≤T_max

In the formula, N is the type of the generator set; p_DG,iIs the output power of genset type i; p_DNThe method is energy exchange with a power grid, and the specified electricity selling time is a negative value, and the electricity purchasing time is a positive value; p_LIs the load demand; p_GIs the output power of the tie line;

and

respectively, the minimum safe power output and the maximum safe power output of the tie line; p_DG.iIs the output power of the generator set i;

and

respectively outputting the minimum safe power and the maximum safe power of the generator set i; t is_outIs the time of day that the energy storage device is discharged; t is_inIs the time of day that the energy storage device is charging; t is_maxIs the maximum charge-discharge time of the energy storage device.

Step 2: performing simulation optimization on the micro-grid multi-objective optimization scheduling model by adopting a particle swarm algorithm based on a penalty function, performing multiple optimization simulation by mainly taking three different objectives, and selecting an optimal solution obtained by each optimization as a candidate scheme;

designing a penalty function particle swarm optimization algorithm, constructing a penalty function, combining constraint conditions of the constraint function into a penalty item according to a selected target function and the constraint conditions, loading the penalty item on the original target function to force an iteration point to approach a feasible domain, and effectively processing the optimization problem with constraint. The convergence precision and the convergence speed of the particle swarm algorithm can be effectively improved. The penalty function is formulated as:

wherein F (x, M) is a penalty function; (x) is an objective function; m is a penalty factor; mp (x) is a penalty term; g_i(x) Are constraints.

An iterative formula of the particle swarm optimization algorithm is as follows:

v_i(k+1)＝ων_i(k)+C₁r₁(p_ik-x_iD)+C₂r₂(p_gk-x_iD)

x_i(k+1)＝x_i(k)+V_i(k)

in the formula, v_i(k) Representing the current velocity of the particle; x is the number of_i(k) Representing the current position of the particle; p is a radical of_ikRepresenting the extreme value of the particle individual; p is a radical of_gkRepresenting a particle global extremum; omega is the weight of the particle swarm velocity position updating formula; c₁,C₂Is a learning factor; k is the number of iterations, r₁And r₂The values of mutually independent random number sequences before each other are uniformly distributed between (0, 1).

FIG. 2 is a flow chart of a penalty function particle swarm algorithm, which comprises the following specific flows:

1. and setting population parameters which mainly comprise the number of particles, penalty function factors and iteration numbers. An initial objective function, constraints and penalty functions are additionally determined.

2. And obtaining an initial population position meeting the constraint condition, taking the individual as the initial position if the constraint condition is met, keeping the initial speed unchanged, and updating the speed position of the individual according to an iterative formula if the constraint condition is not met until the constraint condition is met.

3. And (4) the individuals meeting the constraint conditions update the speed position of each individual through population iteration, judge whether the new population position meets the penalty function constraint, and if not, continuously update the speed position of the individual according to an iteration formula until the penalty function constraint is met.

4. And judging the convergence condition, stopping iteration when the final optimal value is met or the maximum iteration number is reached, outputting an optimization result, and continuing iteration if the final optimal value is not met, and turning to the step 2.

After the step of repeated optimization is carried out on each objective function, a candidate solution set of the micro-grid multi-objective optimization scheduling can be obtained.

And step 3: and after the candidate solution sets are obtained, analyzing and evaluating each candidate solution set by using a multi-attribute decision.

Firstly, analyzing each multi-attribute set { e ] according to an analytic hierarchy process and considering objective facts and willingness of a decision maker₁,e₂,e₃…,e_LCorresponding weight ω₁,ω₂,ω₃…,ω_LAnd are of

To indicate the relative importance between attributes. And then performing multi-attribute analysis on the candidate solution set to determine an evaluation index set, a candidate set and an evaluation grade. The evaluation index set is E ═ E₁,e₂,e₃…,e_LH, set of evaluation levels H ═ H₁,H₂,H_3,…,H_DThe solution set of candidate schemes is a ═ a₁,a₂,a_3,…,a_M}. L is an evaluation index set, D is the number of evaluation grades and the number of M candidate schemes. The evaluation index set refers to a multi-objective optimization function in the invention: economic operation cost, pollution gas emission and power shortage rate.

And performing confidence evaluation on different objective functions in each candidate solution set on corresponding different evaluation levels to obtain a confidence expression as follows:

S(e_i(a_j))＝{(H_n,β_n,i(a_j)),n＝1,2,…,N,i＝1,2,…,M}

in the formula, a_jIs the jth candidate solution set; beta is a_n,i(a_j) Is a_jThe confidence level at the ith evaluation index and the nth evaluation level.

FIG. 3 is a schematic diagram of evidence fusion according to the present invention, which is a set of evaluation indexes { e } of the multi-objective functions of the micro-grid₁,e₂,e₃…,e_LThe importance of each index is represented by the corresponding weight { omega } which is placed at the bottom layer of the evidence fusion₁,ω₂,ω₃…,ω_L}，{H₁,H₂,H_3,…,H_DThe decision maker evaluates the grade of each candidate scheme according to the objective function value of the candidate scheme by using a confidence evaluation method, and the confidence evaluation result (H)_n,β_n,i(a_j) The decision maker has difficulty in directly judging the optimal solution from the evidence matrix, and needs to fuse the evidence, and the process is as follows:

calculating to obtain basic confidence m according to the confidence_n,i＝ω_iβ_n,i(a_j) Because a decider has uncertainty on the multi-objective optimization scheduling cognition of the microgrid, an uncertain basic credibility expression is as follows:

the first i attributes are fused on the evaluation level, and the total basic trust degree m can be obtained_n,i

{H_n}:m_n,I(i+1)(a_j)＝K_I(i+1)(a_j)[m_n,I(i)(a_j)m_n,i+1(a_j)+m_H,I(i)(a_j)m_n,i+1(a_j)+m_n,I(i)(a_j)m_H,i+1(a_j)]

{H}:m_H,I(i+1)(a_j)＝K_I(i+1)(a_j)[m_H,I(i)(a_j)m_H,i+1(a_j)]

In the formula K_I(i+1)(a_j) The scale factor is used for expressing the conflict degree among the evidences; m is_n,I(i)Representing the total confidence coefficient obtained by fusing the confidence coefficients of the previous i evaluation indexes on the nth evaluation level; m is_H,I(i)Indicating a basic confidence level that is not assigned to the first i evaluation indexes.

Integrated confidence of available candidate solutions on the evaluation level set:

in the formula, beta_n(a_j) The confidence of the fused jth candidate scheme on the nth evaluation level; m is_n,I(L)(a_j) Representing the total confidence obtained by fusing the confidence of the L evaluation indexes on the nth evaluation level by the jth candidate scheme; m is_H,I(L)(a_j) Indicating that the jth candidate is not assigned to the basic confidence of the first L evaluation indexes.

Finally, the confidence coefficient set of the candidate schemes at each evaluation level is obtained

S(a_j)＝{(H_n,β_n(a_j)),(n＝1,2,…,N)}

The result of evidence fusion reasoning can only obtain the trust of each candidate scheme on each evaluation level, at the moment, the quality of the candidate solution set cannot be judged visually, and then utility evaluation is applied to map the confidence coefficient distribution into a utility value.

In the formula, u (a)_j) Representation scheme a_jThe utility value of (c).

And selecting the scheme with the maximum utility value from all the schemes as the final scheme of the micro-grid multi-objective optimization scheduling, and using the scheme as a guiding decision for the micro-grid optimization scheduling operation in the next day.

The scheme includes that a decision scheme of multi-objective optimization scheduling of a micro-grid of various distributed power supplies is obtained by a multi-attribute decision combined penalty function particle swarm optimization-based method, a limited number of candidate schemes are selected through multi-attribute decision, target weight is determined through an analytic hierarchy process, multi-evaluation-level confidence degree analysis is carried out on different targets, evidence fusion and utility evaluation are carried out, the problem of mutual influence performance of the targets in multi-objective optimization is effectively solved, and the system can stably run and achieve the optimal overall objective. The method can provide a scientific and effective method for selecting a future micro-grid multi-objective optimization scheduling scheme.

The above description is only of the preferred embodiments of the present invention, and it should be noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.

Claims (5)

1. A multi-attribute decision-based micro-grid multi-objective optimization scheduling method is characterized by mainly comprising the following steps:

step 1: carrying out optimization modeling on the microgrid according to the structure of the microgrid and the characteristics of all distributed power supplies, establishing each objective function with the lowest economic cost, the lowest emission of pollutant gas and the lowest loss rate of the microgrid, constructing constraint conditions of a system objective, and taking the microgrid as a research object in a grid-connected running state, wherein dynamic optimization is adopted and divided into 24 time periods, and each time period is 1 hour;

step 2: respectively optimizing the target function for multiple times by adopting a particle swarm algorithm based on a penalty function according to the instruction of a decision maker, taking the optimal solution of each optimization result as a candidate solution set, and obtaining a plurality of candidate schemes;

and step 3: and evaluating each objective function value in the obtained candidate objective schemes by adopting a multi-attribute decision theory, performing evidence fusion by adopting an evidence reasoning method, and selecting the candidate scheme with the maximum utility value as a final solution by adopting utility evaluation.

2. The multi-objective optimization scheduling method for the micro-grid based on the multi-attribute decision as claimed in claim 1, wherein the micro-grid system in the step 1 mainly comprises a photovoltaic power generation system, a wind power generation system, a storage battery, a fuel cell and a control center, local daily typical photovoltaic output, wind power output and load data are obtained as system prediction data, and system operation parameters refer to local electricity prices, pollution discharge treatment costs and system parameters of each distributed power supply;

the objective function of the system with the lowest economic cost of the microgrid is

mine₁＝F₁+F₂+F₃

Consideration of F₁Cost of purchasing electricity from the grid, F₂Fuel cell fuel cost and F₃The operating maintenance cost of the distributed power supply;

the objective function for minimum emission of pollutant gases is:

mine₂＝g_FcW_FC+g_GW_G

g_Fcand g_GEmission coefficients of fuel cell FC and grid pollution gas, W_FCAnd W_GRespectively discharging the pollution gas of the fuel cell FC and the pollution gas of the power grid;

the objective function for which the deficit rate is lowest is:

the LPSP reflects the ratio of the electric energy shortage and the load demand when the micro-grid is powered by only an internal power supplyEnsuring the power supply capability of the grid in the event of a grid fault, E_LPSIs a shortage of electric energy, P_LThe total required electric quantity of the load is;

the system constraint conditions comprise power balance constraint, tie line constraint, generator set capacity limit constraint, energy storage unit storage capacity constraint and charge-discharge balance constraint; distributed power sources include photovoltaic arrays, wind generators, batteries, fuel cells, gas turbines, and diesel generator devices.

3. The multi-objective optimization scheduling method for the micro-grid based on the multi-attribute decision as claimed in claim 1, wherein the constraint conditions of the system include:

1) and power balance constraint:

2) tie line constraint:

3) and limiting the capacity of the generator set:

4) and (4) restraining the storage capacity of the energy storage unit: SOC_min≤SOC(t)≤SOC_max

5) Charging and discharging balance constraint: t is_out＝T_in≤T_max

In the formula, N is the type of the generator set; p_DG,iIs the output power of genset type i; p_DNThe method is energy exchange with a power grid, and the specified electricity selling time is a negative value, and the electricity purchasing time is a positive value; p_LIs the load demand; p_GIs the output power of the tie line;

and

respectively, the minimum safe power output and the maximum safe power output of the tie line; p_DG.iIs the output power of the generator set i;

and

respectively outputting the minimum safe power and the maximum safe power of the generator set i; t is_outIs the time of day that the energy storage device is discharged; t is_inIs the time of day that the energy storage device is charging; t is_maxIs the maximum charge-discharge time of the energy storage device.

4. The multi-objective optimization scheduling method for the micro-grid based on the multi-attribute decision as claimed in claim 1, wherein the particle swarm algorithm based on the penalty function is adopted in the step 2 to perform optimization respectively by objective functions of economic cost, pollution emission and power shortage rate, and an optimal solution of an optimization result is selected as a candidate solution set after multiple times of optimization; setting a penalty function and control parameters of the particle swarm algorithm based on the penalty function, wherein the penalty function and the control parameters mainly comprise the particle number, the iteration number and the penalty function factors of the particle swarm algorithm; according to the optimization conditions of different objective functions, constraint conditions are combined to form a 'punishment' item, the 'punishment' item is loaded on the original objective function to force an iteration point to approach a feasible domain, and the optimization problem with constraint can be effectively processed; the penalty function is formulated as:

wherein F (x, M) is a penalty function; (x) is an objective function; m is a penalty factor; mp (x) is a penalty term; g_i(x) Is a constraint condition;

the method comprises the steps of coding the output of a storage battery, a fuel cell and a power distribution network in a micro-grid optimization scheduling system to generate a random initial population, and optimizing and obtaining an individual and global optimal scheme meeting constraint conditions and penalty function conditions through a particle swarm optimization.

5. The multi-attribute decision-based micro-grid multi-objective optimization scheduling method according to claim 1, wherein in the step 3, a multi-attribute decision theory is adopted for a candidate scheme solution set obtained by a multi-objective optimization intelligent algorithm to perform ranking of the advantages and disadvantages and evaluation of utility values on each objective function value in the obtained candidate target schemes, so as to select the most reasonable and effective scheme; determining the evaluation index set as E ═ E₁,e₂,e₃…,e_LH, set of evaluation levels H ═ H₁,H₂,H₃,…,H_DThe solution set of candidate schemes is a ═ a₁,a₂,a_3,…,a_M}; l is an evaluation index set, D is the number of evaluation grades and the number of M candidate schemes;

and performing confidence evaluation on different objective functions in each candidate solution set on corresponding different evaluation levels to obtain a confidence expression as follows:

S(e_i(a_j))＝{(H_n,β_n,i(a_j)),n＝1,2,…,N,i＝1,2,…,M}

in the formula, a_jIs the jth candidate solution set; beta is a_n,i(a_j) Is a_jConfidence on the ith evaluation index and the nth evaluation level;

confidence coefficient set of candidate schemes at each evaluation level can be finally obtained by adopting evidence fusion reasoning

S(a_j)＝{(H_n,β_n(a_j)),(n＝1,2,…,N)}

β_n(a_j) Is a_jConfidence at the nth rating level; the evidence fusion reasoning result can only obtain the trust of each candidate scheme on each evaluation level, the quality of the candidate solution set cannot be judged visually, and then utility evaluation is applied to map the confidence coefficient distribution into utility values:

in the formula, u (a)_j) Representation scheme a_jA utility value of; beta is a_nThe confidence level of the nth evaluation level;

and selecting the scheme with the maximum utility value from all the schemes as the final scheme of the micro-grid multi-objective optimization scheduling, and using the scheme as a guiding decision for the micro-grid optimization scheduling operation in the next day.

Images