Multi-energy microgrid scheduling method based on target robustness optimization

Abstract

The invention discloses a multi-energy microgrid scheduling method based on target robustness optimization by establishing empirical distribution of energy demands according to historical data of various types of energy consumption. The method comprises the following specific steps: step 1, establishing a multi-energy micro-grid system; 2, establishing a multi-energy micro-grid energy management certainty optimization model; step 3, establishing a target robustness optimization model of micro-grid energy management; and step 4, solving the target robustness optimization model of micro-grid energy management. All potential distributions are searched from energy demand empirical distribution, energy demand uncertainty can be effectively dealt with compared with deterministic optimization, and decision economy and feasibility are improved to a certain extent compared with distributed robust optimization. Empirical distribution of energy demands is established based on historical data of various types of energy consumption, all potential distribution is searched from the energy demand empirical distribution, and a target robustness optimization model based on an adjustable decision rule and a solvable dual model thereof are established.

Description

Technical Field

[0001] The invention belongs to the technical field of microgrid operation and dispatching, establishes empirical distribution of energy demand based on historical data of various energy consumptions, and proposes a multi-energy microgrid dispatching method based on target robustness optimization. Background Art

[0002] Developing and connecting to distributed photovoltaic and other new energy power generation systems is an effective way for the power and energy systems to cope with challenges. Concepts such as integrated energy systems and multi-energy systems emphasize cooperation between the power industry and other energy sectors, and through energy conversion, achieve coordinated planning, integrated operation, collaborative management, interactive response and mutual complementation of power systems and other energy systems, which can promote the consumption of new energy output and improve the comprehensive utilization efficiency of energy.

[0003] With the penetration of distributed new energy, the multi-energy microgrid system connects medium and low voltage power distribution networks, regional cooling systems, regional heating networks, natural gas networks and other energy networks and various energy markets to form a user-level integrated energy system, and multiple microgrids are interconnected to form a regional integrated energy system. Therefore, the energy management of multi-energy microgrids is an important prerequisite for the safe and economic operation of various energy networks and the flexible and reliable co-production of multiple energy sources.

[0004] With the in-depth development of the concept of integrated energy systems, the research on multi-energy microgrids has become increasingly rich and systematic, focusing on general modeling and comprehensive simulation calculations, planning and design, energy management, and engineering application and implementation. Microgrid energy management refers to the formulation of distributed energy equipment operation plans such as micro diesel engines, photovoltaic arrays, and heat storage tanks supporting combined heat and power units, as well as their interaction plans with energy networks and other microgrid energies, so as to achieve safe and economic operation, new energy consumption, and energy demand guarantee.

[0005] Microgrid energy management technology is mainly based on model-driven and learning-driven optimization methods. The present invention further divides model-driven methods into deterministic and uncertain optimization methods. Deterministic optimization methods usually assume that model parameters are given / stable or can be accurately predicted. The modeling ideas are relatively intuitive, and the relevant theories are relatively complete. There are already a large number of commercial software or program modules that provide solution tools. Uncertainty optimization models model uncertainties that are difficult to predict and give the most appropriate decision-making solutions. In terms of uncertainty processing methods, uncertainty optimization mainly includes two types of models: stochastic optimization and robust optimization. For stochastic optimization, it is difficult to estimate the probability distribution of uncertainties through multidimensional integration. There may still be a large difference between the estimated empirical probability distribution of uncertainties and the true probability distribution, resulting in poor model performance. In order to make the decision more robust and make the optimization model that takes uncertainty into account easier to solve, robust optimization limits the uncertainty value to a bounded uncertainty set, facing the worst case, so that the optimal solution is sufficiently robust to all changes in the uncertainty. Existing research mainly uses Box sets to construct a two-stage robust optimization model to characterize the uncertainties in new energy consumption, comprehensive demand-side response, multi-energy coupling, and market energy trading prices. How to construct uncertainty sets more reasonably and improve decision-making flexibility is one of the difficult problems that urgently needs to be solved in microgrid robust energy management.

[0006] The existing multi-energy microgrid energy management technology initially combines stochastic optimization and robust optimization to characterize uncertainty, hoping to effectively deal with the problem of over-conservative robust optimization solutions. The day-ahead and real-time settlement prices are described as random scenarios, and the objective function is constructed based on stochastic optimization and conditional risk value; at the same time, an uncertain set of new energy output is established, and constraints are constructed based on robust optimization, and a stochastic robust coordination optimization model for multi-energy microgrids participating in spot market transactions is established. In fact, there is a relatively blank in the research on the distributed robust optimization model of microgrid energy management. At the same time, data-driven distributed robust optimization has been widely used in the optimal dispatch of power and energy systems, and the fuzzy set of probability distribution of uncertain quantities is constructed through historical data of uncertain quantities (Ambiguity Set of Distributions). The fuzzy set construction methods mainly include two methods: based on the statistical information of uncertain quantity moments (such as mean and covariance) and based on the metric distance between the true distribution and the empirical distribution of uncertain quantities (such as Kullback-Leibler divergence and Wasserstein distance). A fuzzy set of wind power output distribution based on Kullback-Leibler divergence is constructed, and a distributed robust optimization model for economic dispatch of integrated energy systems is established. It is then transformed into a deterministic mixed integer linear optimization model that can be directly solved using a solver.

[0007] The distributed robust optimization model specifies the maximum distance between the actual uncertainty distribution and the empirical distribution of uncertainty, limits the range of the uncertainty fuzzy set to the maximum distance, and ignores the actual distribution of uncertainty outside the fuzzy set. However, this approach does not take into account the situation where the empirical distribution of uncertainty is far away from the true distribution of uncertainty. The fuzzy set that limits the probability distribution range of uncertainty actually ignores the actual distribution of uncertainty outside the fuzzy set, that is, the true distribution of uncertainty may exist outside the constructed fuzzy set. Summary of the invention

[0008] The present invention establishes the empirical distribution of energy demand based on the historical data of various energy consumptions, proposes a target robustness optimization model for energy management of multi-energy microgrids and its solution method, searches all potential distributions based on the empirical distribution of energy demand, and can effectively deal with the uncertainty of energy demand compared to deterministic optimization, and improves the economic and feasibility of decision-making to a certain extent compared to distributed robust optimization. In order to fully consider all potential probability distributions of uncertain quantities, the present invention proposes a target robustness optimization model for energy management of multi-energy microgrids and its solution method based on the uncertainty of energy demand. Considering a microgrid system containing distributed energy such as new energy, micro diesel engines, heat storage tanks supporting cogeneration units, and cogeneration units, a deterministic optimization model for energy management of microgrids under grid-connected operation is established. The empirical distribution of energy demand is established based on the historical data of various energy consumptions, and all potential distributions are searched based on the empirical distribution of energy demand, and a target robustness optimization model based on adjustable decision rules and its solvable dual model are established.

[0009] In order to achieve the purpose of the present invention, the present invention provides a multi-energy microgrid scheduling method based on target robustness optimization, comprising the following steps: Step 1: Establish a multi-energy microgrid system; Step 2: Establish a deterministic optimization model for energy management of multi-energy microgrids; Step 3: Establish a target robustness optimization model for microgrid energy management; Step 4: Solve the target robustness optimization model of microgrid energy management.

[0010] The step 2 specifically includes the following steps: Step 2.1: Establish the objective function of the microgrid energy management deterministic optimization model; The deterministic optimization model of multi-energy microgrid energy management takes minimizing the system operating cost as the objective function, involving the start-stop cost of the micro diesel engine, the fuel cost of the micro diesel engine, the operation and maintenance cost of the micro diesel engine, the operation and maintenance cost of the combined heating and cooling unit, the operation and maintenance cost of the photovoltaic array, the cost / income of electricity purchase and sale, and the cost of purchasing natural gas: (1) in, (2) (3) (4) (5) (6) (7) (8) In the formula: min represents the minimization of the objective function; Indicates the running time; Indicates the operation cycle; Indicates The rolling moment of the moment, i.e. ; Indicates the number of the micro diesel engine; Indicates the number of micro diesel engines; express The rolling number of a micro diesel engine, i.e. ; Indicates the CCHP unit number; Indicates the number of combined cooling, heating and power units; express The rolling number of each cogeneration unit, i.e. ; Indicates the photovoltaic array number; Indicates the number of photovoltaic arrays; express The rolling number of the photovoltaic array, i.e. ; Indicates A micro diesel engine The start and stop costs at the dispatch time; Indicates A micro diesel engine Fuel cost at the dispatch time; Indicates A micro diesel engine Operation and maintenance costs at the time of dispatch; Indicates Combined cooling, heating and power units in Operation and maintenance costs at the time of dispatch; Indicates Photovoltaic arrays in Operation and maintenance costs at the time of dispatch; Indicated in Natural gas fuel cost at the dispatch time; Indicated in The cost of purchasing and selling electricity at the dispatch time; Indicates Unit start-stop cost of a micro diesel engine; Indicates A micro diesel engine The start and stop plan of the dispatch time, Indicates A micro diesel engine Start and stop plan at dispatch time; Indicates Unit fuel cost of a micro diesel engine; Indicates A micro diesel engine Active power output plan at the dispatching time; Indicates unit duration, used to convert power units into energy units; Indicates Unit operation and maintenance cost of a micro diesel engine; Indicates Unit operation and maintenance cost of each CHP unit; Indicates Combined cooling, heating and power units in Output plan at dispatch time; Indicates Unit operation and maintenance cost of a photovoltaic array; Indicates Photovoltaic arrays in Output plan at dispatch time; Indicates Unit natural gas fuel cost per CHP unit; Indicates Combined cooling, heating and power units in Gas usage plan at dispatch time; Indicated in The electricity purchase price at the dispatch time; Indicated in Power purchase plan at dispatch time; Indicated in The electricity sales price at the dispatch time; Indicated in Electricity sales plan at dispatch time; Step 2.2: Determine the constraints of the microgrid energy management deterministic optimization model; The DistFlow (distributed power flow) model is used to describe the active power, reactive power and voltage balance of each node: (9) In the formula, Indicates the node number, Indicates the number of nodes; Indicates until The rolling number of the node, i.e. ; Indicates that the The node Micro diesel engine Timely meritorious contribution plan; Indicates that the The node Photovoltaic arrays in Timely output plan; Indicates that the The node Combined cooling, heating and power units in Timely output plan; Indicated in Scheduling time from node Outflow and inflow nodes The amount of active power; Representation Node exist Active power load demand at the dispatching time; (10) In the formula, Indicated in Scheduling time from node Outflow and inflow nodes The amount of reactive power; Indicates that the The node Micro diesel engine Reactive power output plan at all times; Indicates that the The node Photovoltaic arrays in The reactive power compensation at the moment is provided by the inverter installed in conjunction with the photovoltaic array; Representation Node exist Reactive power / load at the dispatch time; It is worth noting that Indicates access The node The active power output of a micro diesel engine, Indicates access The node The reactive power output of a micro diesel engine is If there are no devices on the nodes, ; , and Similarly; (11) In the formula, Representation Node exist Voltage at the dispatch time; Representation Node and nodes The resistance of the line between them; Representation Node and nodes The reactance of the line between them; Indicates the reference voltage; Among them, the voltage limit constraint is: (12) In the formula, Indicates the maximum voltage offset; The line capacity constraint is: (13) In the formula, Representation Node and nodes The maximum apparent power of electric energy transmitted between the lines; The output constraints of micro diesel engines are: (14) In the formula, Indicates The maximum apparent power output of a micro diesel engine; (15) In the formula, Indicates The minimum active output of a micro diesel engine; The mini diesel engine climbing constraint is: (16) In the formula, Indicates The gradeability of the output of a micro diesel engine; The reactive power compensation constraint is: (17) In the formula, Indicates The maximum apparent power output of each PV array and its inverter; Indicates Rated output of each photovoltaic array; Indicates Reactive power compensation amount of each photovoltaic array inverter; Indicates The maximum reactive power compensation of a photovoltaic array inverter.

[0011] In the multi-energy microgrid scheduling method based on target robustness optimization, the node that interacts with the grid energy is defined as node 1, and its electric energy purchase / sale volume satisfies: (18) Among them, the power purchase and sales constraints meet: (19) (20).

[0012] The multi-energy microgrid scheduling method based on target robustness optimization, wherein the combined cooling, heating and power unit consumes natural gas output and recovers the remaining energy for heating: (twenty one) (twenty two) In the formula, Indicates Combined cooling, heating and power units in The total energy generated at any moment, Indicates The output of each combined cooling, heating and power unit accounts for the proportion of total energy; Indicates Combined cooling, heating and power units in Heat recovered at all times; Indicates The proportion of waste heat recovery of each cogeneration unit to the remaining heat; Among them, the output constraint of the combined cooling, heating and power unit is: (twenty three) In the formula, Indicates Minimum output of each combined cooling, heating and power unit; Indicates Maximum output of each combined cooling, heating and power unit; The ramping constraint of the combined cooling, heating and power unit is: (twenty four) In the formula, Indicates Output ramp rate of each CHP unit; The waste heat recovered by the combined cooling, heating and power unit meets the constraints: (25) In the formula, Indicates Combined cooling, heating and power units in Maximum heat recovered at all times; In order to promote the consumption of photovoltaic power and meet the peak demand of various types of energy, the heat balance equation of the node connected to the combined heating and cooling unit is as follows: (26) In the formula, Indicates The hot boilers of the combined cooling, heating and power units are Heat production at each moment; Indicates The heat production efficiency of the hot boilers supporting each combined cooling, heating and power unit; Indicates The heat storage tanks of the combined cooling, heating and power units are Heat release at any time; Indicates The heat storage tanks of the combined cooling, heating and power units are The amount of heat at any time; Indicates The combined cooling, heating and power units are equipped with absorption chillers. Cooling capacity at the time; Indicates Combined cooling, heating and power units in The corresponding heating demand at the time.

[0013] In the multi-energy microgrid scheduling method based on target robustness optimization, a binary variable is used to represent the charging and discharging state of the heat storage tank supporting the combined heat, cooling and power unit: (27) Where: Indicates The heat storage tanks of the combined cooling, heating and power units are The charging status at all times; Indicates The heat storage tanks of the combined cooling, heating and power units are The exothermic state at all times; The heat storage tank charging and discharging constraints of the combined cooling, heating and power unit are: (28) (29) Where: Indicates Maximum heat capacity of the heat storage tank supporting each combined cooling, heating and power unit; Indicates The maximum heat release of the heat storage tank supporting each combined heating, cooling and power unit.

[0014] The heat stored in the heat storage tank of the combined heat and power unit satisfies the state equation: (30) Where: Indicates The heat storage tanks of the combined cooling, heating and power units are The amount of heat stored at all times; Indicates The charging efficiency of the heat storage tanks supporting each cogeneration unit; Indicates Heat release efficiency of the heat storage tank supporting each combined heating, cooling and power unit.

[0015] The initial state of the heat stored in the heat storage tank supporting the combined cooling, heating and power unit is consistent with the end time of the previous scheduling cycle: (31) Then the heat stored in the heat storage tank of the combined cooling, heating and power unit meets the constraints: (32) Where: Indicates The upper limit of the amount of heat that can be stored in the heat storage tank of each combined cooling, heating and power unit; Indicates The lower limit of heat storage in the heat storage tank of each combined heating, cooling and power unit.

[0016] The multi-energy microgrid scheduling method based on target robustness optimization, wherein the cooling capacity of the absorption refrigeration mechanism supporting the combined heat and power unit satisfies the constraint: (33) Where: Indicates Refrigeration efficiency of absorption chillers for each combined cooling, heating and power unit; Indicates The combined cooling, heating and power units are equipped with absorption chillers. Refrigeration needs at all times.

[0017] In the multi-energy microgrid scheduling method based on target robustness optimization, the heat production of the hot boiler supporting the combined heat and power unit satisfies the constraint: (34) In the formula, Indicates The maximum heat output of the hot boiler supporting each combined cooling, heating and power unit.

[0018] In the multi-energy microgrid scheduling method based on target robustness optimization, the amount of natural gas consumed by the combined heat and power unit and the supporting hot boiler of the combined heat and power unit is: (35) In the formula, Indicates the total energy that can be generated by a unit amount of natural gas.

[0019] The multi-energy microgrid scheduling method based on target robustness optimization, wherein step 3 specifically includes the following sub-steps: Step 3.1: Uncertain set of microgrid energy demand; Focusing on the uncertainty of user electricity, heating, and cooling demand and the coupling effect between multiple types of energy demand, the energy demand random variable is expressed as the sum of the energy demand expectation and the energy demand prediction error random variable: (36) (37) (38) In the formula, Representation Node exist The random variable of active load energy demand at the dispatching time, Representation Node exist The expected active load demand at the dispatch time, Representation Node exist Random variable of active power load demand forecast error at the dispatching time; Indicates Combined cooling, heating and power units in The corresponding heat demand random variable at time, Indicates Combined cooling, heating and power units in The corresponding heat demand expectation at the time, Indicates Combined cooling, heating and power units in The corresponding heat demand forecast error random variable at the time; Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand random variable at time, Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand expectations at all times, Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand forecast error at time is a random variable; Among them, the uncertainty of energy demand forecast error satisfies the uncertainty set: (39) (40) (41) In the formula, The uncertainty set represents the uncertainty of the electric energy demand of the electric load; The uncertainty set represents the uncertainty of heat energy demand; The uncertainty set represents the uncertainty of cooling demand; Representation Node exist The uncertainty of the electric load energy demand at the dispatching time, Representation Node exist The lower limit of the forecast error of the electric load demand interval at the dispatching time, Representation Node exist The upper limit of the forecast error of the electric load demand interval at the dispatching time; Indicates Combined cooling, heating and power units in The uncertainty of heat demand at each moment, Indicates Combined cooling, heating and power units in The lower limit of the forecast error of the heat demand interval at the moment, Indicates Combined cooling, heating and power units in The upper limit of the forecast error of the heat demand interval at the moment; Indicates Combined cooling, heating and power units in The uncertainty of cooling demand at each moment, Indicates Combined cooling, heating and power units in The lower limit of the forecast error of the cooling demand interval at time, Indicates Combined cooling, heating and power units in The upper limit of the forecast error of the cooling demand interval at the time; represents the field of real numbers, indicating , and belongs to the field of real numbers; Step 3.2: Target robustness optimization model; The deterministic optimization model of multi-energy microgrid energy management is represented as a two-stage problem for modeling; the day-ahead decision is the first-stage optimization problem: (42) ; In the formula, represents the micro diesel engine start-stop schedule variable; Indicates the variable of the charging state of the heat storage tank supporting the combined heat and power unit; Indicates the heat release state variable of the heat storage tank supporting the combined heat and power unit; Indicates the variable of the amount of heat stored in the heat storage tank supporting the combined cooling, heating and power unit; It represents the variable of heat release of the heat storage tank supporting the combined cooling, heating and power unit; represents the optimal value of the second stage optimization problem, whose input includes , , , and ; Intraday decision making is a second-stage optimization problem: (43) ; The day-ahead decision-making and intraday decision-making variables mainly include the start-stop status of the micro-diesel engine, the active output of the micro-diesel engine, the reactive output of the micro-diesel engine, the output of the cogeneration unit, the reactive compensation amount, the electricity purchase / sale amount, the natural gas purchase amount, the waste heat recovery amount of the cogeneration unit, the heat output of the cogeneration unit's supporting hot boiler, the charging amount of the cogeneration unit's supporting heat storage tank, the heat release of the cogeneration unit's supporting heat storage tank, the cooling capacity of the cogeneration unit's supporting absorption refrigeration machine and the charging and discharging status of the cogeneration unit's supporting heat storage tank; the state variables mainly include the line active power distribution, the line reactive power distribution, the node voltage distribution and the heat stored in the cogeneration unit's supporting heat storage tank.

[0020] Considering the uncertainty of energy demand, the deterministic optimization model based on microgrid energy management is expressed as a distributed robust optimization model based on Wasserstein distance; the optimization problem in the first stage is: (44) ; In the formula, represents the energy demand forecast error distribution; represents a fuzzy set that defines the distribution of energy demand forecast errors, whose range depends on the parameter ; Represents the distribution of forecast error for a given energy demand The expectation of the optimal value of the second-stage optimization problem; represents the comprehensive energy demand forecast error random variable, including electricity, heat, and cooling demand: , In the formula, Represents the transposition of the electric energy demand parameter quantity; Represents the transposition of the heat energy demand parameter quantity; Represents the transposition of the cooling demand parameter quantity; The objective function of the first stage is to minimize the sum of the expected value of the start-stop cost of the micro diesel engine and the daily operation cost of the system. Based on the uncertainty set of energy demand forecast error, the fuzzy set is defined as: , In the formula, represents all potential distributions of energy demand forecast errors; represents the empirical distribution of energy demand forecast errors; Representation distribution and distribution The distance between The uncertainty set representing the energy demand forecast error is: ; In order to consider all potential energy demand forecast error distributions, the target robustness optimization model of multi-energy microgrid energy management starts from the energy demand empirical distribution, allows all potential distributions of energy demand, imposes penalties on the energy demand forecast error distribution that deviates from the energy demand forecast error empirical distribution, minimizes the penalty term, that is, minimizes the impact of energy demand forecast error uncertainty, and maximizes the robustness of the specified microgrid system operation cost; the first stage optimization model is expressed as: (45) (46) , (47) In the formula, represents the penalty for energy demand forecast error distribution shift; It represents the sufficiently satisfactory operation cost target of the microgrid system specified by experts based on domain knowledge; The measurement function of the distance between the empirical distribution and the actual distribution of energy demand forecast error has not been clarified, and the target robustness optimization model of multi-energy microgrid energy management is still unsolvable; therefore, the joint distribution of the actual distribution and the empirical distribution of energy demand forecast error is established, and the Wasserstein distance is used to measure the distance between the actual distribution and the empirical distribution of energy demand forecast error: , In the formula, represents Wasserstein distance; represents the joint distribution of the actual distribution and the empirical distribution of energy demand forecast errors; Represents the distribution of forecast error for a given energy demand Down and The expectation of the difference-one norm of ; A random variable representing the historical error in the combined energy demand forecast: , In the formula, The transpose of the equivalent random variable representing the historical error in the electricity demand forecast; The transpose of the equivalent random variable representing the historical error in the thermal energy demand forecast; The transpose of the equivalent random variable representing the historical error in cooling demand forecasts; Then formula (46) is expressed as: (48) In the formula, Represents the joint distribution range of the actual distribution and empirical distribution of energy demand forecast error: , In order to consider the coupling effect between energy demand such as electricity, cooling and heating energy, the distance between the actual distribution and the empirical distribution of comprehensive energy demand forecast error is expressed as a random variable in combination with Wasserstein distance. : (49) In the formula, represents a norm; Introducing historical data on energy demand forecast errors: (50) In the formula, Indicates historical data / sample number, such as timestamp; Indicates the number of historical scenes; express The rolling number of historical data, that is, ; Indicates The joint distribution corresponding to the historical errors of the group comprehensive energy demand forecast; Indicates The distribution distance corresponding to the historical error of the comprehensive energy demand forecast is: (51) In the formula, Indicates Historical data of group power demand forecast errors; Indicates Historical data of forecast errors of heat demand; Indicates Historical data of group cooling demand forecast errors; The joint distribution corresponding to the historical errors of the comprehensive energy demand forecast is expressed as a tractable uncertainty: (52) In order to simplify the objective robustness optimization problem into a more tractable form, an auxiliary variable of distribution distance is introduced to represent the distance between the actual distribution and the empirical distribution of the comprehensive energy demand forecast error: (53) in, Auxiliary variables representing the vicinity of the empirical distribution; Indicates that as of The moment-to-moment energy demand forecast error is a random variable; Indicated in Time about auxiliary variables and as of Energy demand forecast error random variable and its Natural gas fuel cost function for historical scenarios; Indicates Micro diesel engine Time about auxiliary variables and as of Energy demand forecast error random variable and its Fuel cost function for historical scenarios; Indicates Micro diesel engine Time about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; Indicates The combined heat and power unit Time about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; Indicated in Time about auxiliary variables and as of Energy demand forecast error random variable and its The cost function of electricity purchase and sale for each historical scenario; Indicates PV arrays about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; About The uncertainty set of energy demand forecast error based on historical data is expressed as: .

[0021] The multi-energy microgrid scheduling method based on target robustness optimization, wherein step 4 specifically includes the following sub-steps: Step 4.1: Adjustable decision rules; The target robust optimization model for energy management of multi-energy microgrid systems does not deal with the relationship between the second-stage decision variables and the uncertainty of the comprehensive energy demand forecast error and auxiliary variables. In order to improve the flexibility of the target robust optimization model solution, the present invention introduces an adjustable decision rule to adjust the second-stage decision variables and state variables according to the actual energy demand and Wasserstein distance; taking the purchased electricity as an example, the electricity purchase / sale amount can be expressed in a similar form: (54) (55) In the formula, Indicates the running time; Indicates that Uncertainty of energy demand forecast error at each moment and its The electricity purchase plan function for each historical scenario; Indicates that Uncertainty of energy demand forecast error at each moment and its The electricity purchase plan function for each historical scenario; They are The intercept coefficient of They are The slope coefficient of for The intercept coefficient of for The slope coefficient of for The auxiliary variable coefficient of ; By solving the adjustable decision rule coefficients through the target robust optimization model, the energy management strategy of the multi-energy microgrid system can be obtained according to the actual energy demand and Wasserstein distance; Step 4.2: Model linearization; The quadratic constraints, second-order cone constraints, and constraints containing absolute values ​​of the objective robust optimization model based on adjustable decision rules will be handled; First, introduce binary auxiliary variables The constraint containing absolute values, formula (2) is expressed as: (56) (57) (58) Based on the Cartesian coordinate system, the dodecagon interior approximation method is adopted. Taking the second-order cone constraint (14) as an example, the quadratic constraint (13) and the second-order cone constraint (14) are expressed in a similar form: (59) in, Indicates Micro diesel engine Energy demand forecast error random variable and its Reactive power output plan for each historical scenario; Indicates Micro diesel engine Energy demand forecast error random variable and its A meritorious contribution plan for each historical scene; Indicates The horizontal coordinate of the nth point of the approximate dodecagon inside the original feasible region of the micro diesel engine; It represents the ordinate of the nth point of the approximate dodecagon inside the original feasible region of the vth micro diesel engine; Indicates The horizontal coordinate of the mth point of the approximate dodecagon in the original feasible region of a micro diesel engine Indicates The ordinate of the mth point of the approximate dodecagon inside the original feasible region of the micro diesel engine; (60) (61) Step 4.3: Solvable form of the target robustness optimization model; The energy demand forecast error has infinite values ​​in the uncertain set and fuzzy set. The constraints of the energy demand forecast error uncertainty and random variables are actually infinite constraints, which makes the target robust optimization model a semi-infinite programming model. Therefore, it is necessary to express it as a solvable form with a finite number of constraints. Specifically, each constraint of the target robust optimization model needs to be treated as a separate optimization problem and the duality theorem of linear programming theory is applied to it, so as to obtain a linear programming model with a finite number of constraints for solution. Taking the key formula (53) as an example, based on the adjustable decision rule, the sub-problem containing uncertainty in formula (53) is expressed in a compact form: (62) (63) (64) (65) In the formula, Indicates the device number; represents the first stage coefficient vector; represents the second stage coefficient vector; represents the decision variables of the first-stage optimization problem; represents the intercept coefficient of the decision variable of the second stage optimization problem; represents the slope coefficient of the decision variable of the second-stage optimization problem; The coefficients of auxiliary variables representing the decision variables of the second-stage optimization problem; Represents the lower limit of interval prediction error; Represents the upper limit of the interval prediction error; About Devices in The uncertainty of energy demand forecast error at each moment; The energy demand forecast error is Historical data of each scene; Formula (65) can be expressed as: (66) In the formula, express No. A historical scene; Further introduction of auxiliary variables , formula (66) is equivalent to: (67) (68) Mode In linear programming form, when the uncertain set When not empty, the formula There is a feasible solution. According to the duality theorem of linear programming theory, strong duality holds, and the dual subproblem can be obtained: (69) (70) (71) (72) (73) In the formula, is the dual variable matrix corresponding to constraint (63); is the dual variable matrix corresponding to constraint (64); is the dual variable matrix corresponding to the left side of constraint (68); is the dual variable matrix corresponding to the right side of constraint (68); is a single dual variable corresponding to constraint (67); For the Devices in The dual variable of the moment constraint (63); For the Devices in Dual variable of the moment constraint (64); For the Devices in The dual variable on the left side of the moment constraint (68); For the Devices in The dual variable on the right side of moment constraint (68); To satisfy An indicator function that outputs 1 when .

[0022] Compared with the prior art, the present invention can achieve at least the following beneficial effects: Different from distributed robust optimization, target robustness optimization does not describe the worst scenario of uncertainty in the microgrid system, specifies sufficiently satisfactory system operation cost / benefit targets, and maximizes the robustness of achieving system operation cost / benefit targets under the action of uncertainty. The present invention establishes the empirical distribution of energy demand based on historical data of various energy consumption, proposes a target robustness optimization model for energy management of multi-energy microgrids and its solution method, searches for all potential distributions based on the empirical distribution of energy demand, and can effectively deal with energy demand uncertainty compared to deterministic optimization, and improves decision-making economy and feasibility to a certain extent compared to distributed robust optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 The present invention provides a multi-energy microgrid system architecture and the electrical and thermal energy flows it dispatches. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] The present invention provides a multi-energy microgrid scheduling method based on target robustness optimization, comprising: Step 1: Establish a multi-energy microgrid system.

[0026] like Figure 1 As shown, the present invention includes a multi-energy microgrid system of distributed energy such as electricity, refrigeration, and heating. The resources on the power supply side involve grid interaction, distributed photovoltaic output, micro diesel engine output, and combined heat and power unit output, and meet the user's electricity demand through transmission lines. Among them, the micro diesel engine can be used as an active power source and a reactive power source; the photovoltaic array and the combined heat and power unit can only be used as an active power source. The thermal energy supply side involves the combined heat boiler supporting the combined heat and power unit to produce heat, the combined heat storage tank supporting the combined heat and power unit to charge and release heat, and the combined heat and power unit to produce heat to meet the heating demand. Due to the small scale of the system and thermal inertia, the dynamic characteristics of the circulating pipeline are ignored. The combined heat and power unit refers to a combined heat and power unit that inputs natural gas and can recover waste heat. Among them, due to the small scale of the system and the inertia of the fluid, the gas pipeline is ignored. In addition to meeting the heating demand, the recovered heat is also used for the combined heat and power unit on the cold energy supply side to match the absorption chiller to meet the cooling demand. The multi-energy microgrid system can simultaneously meet users' electricity, heating and cooling needs.

[0027] Step 2: Establish a deterministic optimization model for energy management of multi-energy microgrids.

[0028] In the present invention, step 2 specifically includes the following sub-steps: Step 2.1: Objective function of the deterministic optimization model for microgrid energy management; The deterministic optimization model of multi-energy microgrid energy management takes minimizing the system operating cost as the objective function, involving the start-stop cost of the micro diesel engine, the fuel cost of the micro diesel engine, the operation and maintenance cost of the micro diesel engine, the operation and maintenance cost of the combined heating and cooling unit, the operation and maintenance cost of the photovoltaic array, the cost / income of electricity purchase and sale, and the cost of purchasing natural gas: (1) in, (2) (3) (4) (5) (6) (7) (8) Where: Indicates the running time; Indicates the operation cycle; Indicates The rolling moment of the moment, i.e. ; Indicates the number of the micro diesel engine; Indicates the number of micro diesel engines; express The rolling number of a micro diesel engine, i.e. ; Indicates the CCHP unit number; Indicates the number of combined cooling, heating and power units; express The rolling number of each cogeneration unit, i.e. ; Indicates the photovoltaic array number; Indicates the number of photovoltaic arrays; express The rolling number of the photovoltaic array, i.e. ; Indicates A micro diesel engine The start and stop costs at the dispatch time; Indicates A micro diesel engine Fuel cost at the dispatch time; Indicates A micro diesel engine Operation and maintenance costs at the time of dispatch; Indicates Combined cooling, heating and power units in Operation and maintenance costs at the time of dispatch; Indicates Photovoltaic arrays in Operation and maintenance costs at the time of dispatch; Indicated in Natural gas fuel cost at the dispatch time; Indicated in The cost of purchasing and selling electricity at the dispatch time; Indicates Unit start-stop cost of a micro diesel engine; Indicates A micro diesel engine The start and stop plan of the dispatch time, Indicates A micro diesel engine Start and stop plan at dispatch time; Indicates Unit fuel cost of a micro diesel engine; Indicates A micro diesel engine Active power output plan at the dispatching time; Indicates unit duration, used to convert power units into energy units; Indicates Unit operation and maintenance cost of a micro diesel engine; Indicates Unit operation and maintenance cost of each CHP unit; Indicates Combined cooling, heating and power units in Output plan at dispatch time; Indicates Unit operation and maintenance cost of a photovoltaic array; Indicates Photovoltaic arrays in Output plan at dispatch time; Indicates Unit natural gas fuel cost per CHP unit; Indicates Combined cooling, heating and power units in Gas usage plan at dispatch time; Indicated in The electricity purchase price at the dispatch time; Indicated in Power purchase plan at dispatch time; Indicated in The electricity sales price at the dispatch time; Indicated in Electricity sales plan at dispatch time.

[0029] Step 2.2: Constraints of the deterministic optimization model for microgrid energy management.

[0030] The present invention adopts the DistFlow (distributed power flow) model to describe the active power, reactive power and voltage balance of each node: (9) In the formula, Indicates the node number, Indicates the number of nodes; Indicates until The rolling number of the node, i.e. ; Indicates that the The node Micro diesel engine Timely meritorious contribution plan; Indicates that the The node Photovoltaic arrays in Timely output plan; Indicates that the The node Combined cooling, heating and power units in Timely output plan; Indicated in Scheduling time from node Outflow and inflow nodes The amount of active power; Representation Node exist Active power load demand at the dispatching time.

[0031] (10) In the formula, Indicated in Scheduling time from node Outflow and inflow nodes The amount of reactive power; Indicates that the The node Micro diesel engine Reactive power output plan at all times; Indicates that the The node Photovoltaic arrays in The reactive power compensation at the moment is provided by the inverter installed in conjunction with the photovoltaic array; Representation Node exist Reactive power / load at the dispatch time.

[0032] It is worth noting that Indicates access The node The active power output of a micro diesel engine, Indicates access The node The reactive power output of a micro diesel engine is If there are no devices on the nodes, ; , and Same reason.

[0033] (11) In the formula, Representation Node exist Voltage at the dispatch time; Representation Node and nodes The resistance of the line between them; Representation Node and nodes The reactance of the line between them; Indicates the reference voltage.

[0034] Among them, the voltage limit constraint is: (12) In the formula, Indicates the maximum voltage offset.

[0035] The line capacity constraint is: (13) In the formula, Representation Node and nodes The maximum apparent power of electrical energy transmitted between the lines.

[0036] The output constraints of micro diesel engines are: (14) In the formula, Indicates The maximum apparent power output of a micro diesel engine.

[0037] (15) In the formula, Indicates The minimum active output of a micro diesel engine.

[0038] The mini diesel engine climbing constraint is: (16) In the formula, Indicates The gradeability of the output of a micro diesel engine.

[0039] The reactive power compensation constraint is: (17) In the formula, Indicates The maximum apparent power output of each PV array and its inverter; Indicates Rated output of each photovoltaic array; Indicates Reactive power compensation amount of each photovoltaic array inverter; Indicates The maximum reactive power compensation of a photovoltaic array inverter.

[0040] The present invention defines the node that interacts with the grid energy as node 1, and its electric energy purchase / sale volume satisfies: (18) Among them, the power purchase and sales constraints meet: (19) (20) Combined cooling, heating and power (CHP) converts photovoltaic output / waste heat recovery that is difficult to consume in time into heat and cold energy that is easier to store / electricity with high immediate demand. For example, at night in winter, the combined cooling, heating and power (CHP) unit can generate 30%-50% of the total energy while meeting the heating needs of the people. Together with local wind power generation, these excess outputs are difficult to consume in time at night and are usually directly wasted, with up to 20% of the wind being abandoned. Excess electricity should be converted into other energy that is needed immediately or heat and cold energy that is easier to store to improve the comprehensive efficiency of energy utilization. Specifically, the present invention realizes combined cooling, heating and power and electric-heat coupling through a combined cooling, heating and power (CHP) unit. Unlike a micro diesel engine, a combined cooling, heating and power (CHP) unit consumes natural gas output and recovers excess energy for heating: (twenty one) (twenty two) In the formula, Indicates Combined cooling, heating and power units in The total energy generated at any moment, Indicates The output of each combined cooling, heating and power unit accounts for the proportion of total energy; Indicates Combined cooling, heating and power units in Heat recovered at all times; Indicates The waste heat recovery amount of each cogeneration unit accounts for the proportion of residual heat.

[0041] Among them, the output constraint of the combined cooling, heating and power unit is: (twenty three) In the formula, Indicates Minimum output of each combined cooling, heating and power unit; Indicates Maximum output of a combined cooling, heating and power unit.

[0042] The ramping constraint of the combined cooling, heating and power unit is: (twenty four) In the formula, Indicates The output ramp rate of each combined heating, cooling and power unit.

[0043] The waste heat recovered by the combined cooling, heating and power unit meets the constraints: (25) In the formula, Indicates Combined cooling, heating and power units in Maximum heat recovered at all times; In order to promote the consumption of photovoltaic power and meet the peak demand of multiple types of energy, the present invention focuses on the heat balance equation of the node connected to the combined heating and cooling unit: (26) In the formula, Indicates The hot boilers of the combined cooling, heating and power units are Heat production at each moment; Indicates The heat production efficiency of the hot boilers supporting each combined cooling, heating and power unit; Indicates The heat storage tanks of the combined cooling, heating and power units are Heat release at any time; Indicates The heat storage tanks of the combined cooling, heating and power units are The amount of heat at any time; Indicates The combined cooling, heating and power units are equipped with absorption chillers. Cooling capacity at the time; Indicates Combined cooling, heating and power units in The corresponding heating demand at the time.

[0044] The present invention uses binary variables to represent the heat storage tank charging and discharging state of the combined heat and power unit: (27) Where: Indicates The heat storage tanks of the combined cooling, heating and power units are The charging status at all times; Indicates The heat storage tanks of the combined cooling, heating and power units are Exothermic state at all times.

[0045] The heat storage tank charging and discharging constraints of the combined cooling, heating and power unit are: (28) (29) Where: Indicates Maximum heat capacity of the heat storage tank supporting each combined cooling, heating and power unit; Indicates The maximum heat release of the heat storage tank supporting each combined heating, cooling and power unit.

[0046] The heat stored in the heat storage tank of the combined heat and power unit satisfies the state equation: (30) Where: Indicates The heat storage tanks of the combined cooling, heating and power units are The amount of heat stored at all times; Indicates The charging efficiency of the heat storage tanks supporting each cogeneration unit; Indicates Heat release efficiency of the heat storage tank supporting each combined heating, cooling and power unit.

[0047] The present invention assumes that the initial state of the heat stored in the heat storage tank supporting the combined heat and power unit is consistent with the end time of the previous scheduling cycle: (31) Then the heat stored in the heat storage tank of the combined cooling, heating and power unit meets the constraints: (32) Where: Indicates The upper limit of the amount of heat that can be stored in the heat storage tank of each combined cooling, heating and power unit; Indicates The lower limit of heat storage in the heat storage tank of each combined heating, cooling and power unit.

[0048] The cooling capacity of the absorption refrigeration mechanism supporting the combined cooling, heating and power unit meets the constraints: (33) Where: Indicates Refrigeration efficiency of absorption chillers for each combined cooling, heating and power unit; Indicates The combined cooling, heating and power units are equipped with absorption chillers. Refrigeration needs at all times.

[0049] The heat output of the hot boiler supporting the combined cooling, heating and power unit meets the constraints: (34) In the formula, Indicates The maximum heat output of the hot boiler supporting each combined cooling, heating and power unit.

[0050] In addition, the amount of natural gas consumed by the combined heat and power unit and the supporting heat boiler of the combined heat and power unit is: (35) In the formula, Indicates the total energy that can be generated by a unit amount of natural gas.

[0051] Step 3: Establish a target robustness optimization model for microgrid energy management.

[0052] In the present invention, step 3 specifically includes the following sub-steps: Step 3.1: Uncertain set of microgrid energy demand; The user's energy demand is affected by complex factors such as meteorological conditions, energy prices, social and economic development levels, day type information (people's daily activities), etc., and the prediction is difficult. Therefore, the present invention focuses on the uncertainty of user's electricity, heating, and cooling demand and the coupling effect between multiple types of energy demand, and expresses the energy demand random variable as the sum of the energy demand expectation and the energy demand prediction error random variable: (36) (37) (38) In the formula, Representation Node exist The random variable of active load energy demand at the dispatching time, Representation Node exist The expected active load demand at the dispatch time, Representation Node exist Random variable of active power load demand forecast error at the dispatching time; Indicates Combined cooling, heating and power units in The corresponding heat demand random variable at time, Indicates Combined cooling, heating and power units in The corresponding heat demand expectation at the time, Indicates Combined cooling, heating and power units in The corresponding heat demand forecast error random variable at the time; Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand random variable at time, Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand expectations at all times, Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand forecast error at time is a random variable.

[0053] Among them, the uncertainty of energy demand forecast error satisfies the uncertainty set: (39) (40) (41) In the formula, The uncertainty set represents the uncertainty of the electric energy demand of the electric load; The uncertainty set represents the uncertainty of heat energy demand; The uncertainty set represents the uncertainty of cooling demand; Representation Node exist The uncertainty of the electric load energy demand at the dispatching time, Representation Node exist The lower limit of the forecast error of the electric load demand interval at the dispatching time, Representation Node exist The upper limit of the forecast error of the electric load demand interval at the dispatching time; Indicates Combined cooling, heating and power units in The uncertainty of heat demand at each moment, Indicates Combined cooling, heating and power units in The lower limit of the forecast error of the heat demand interval at the moment, Indicates Combined cooling, heating and power units in The upper limit of the forecast error of the heat demand interval at the moment; Indicates Combined cooling, heating and power units in The uncertainty of cooling demand at each moment, Indicates Combined cooling, heating and power units in The lower limit of the forecast error of the cooling demand interval at time, Indicates Combined cooling, heating and power units in The upper limit of the forecast error of the cooling demand interval at the time; represents the field of real numbers, indicating , and belongs to the field of real numbers.

[0054] Step 3.2: Target robustness optimization model; The present invention represents the deterministic optimization model of multi-energy microgrid energy management as a two-stage problem for modeling. The day-ahead decision is the first-stage optimization problem: (42) ; In the formula, represents the micro diesel engine start-stop schedule variable; Indicates the variable of the charging state of the heat storage tank supporting the combined heat and power unit; Indicates the heat release state variable of the heat storage tank supporting the combined heat and power unit; Indicates the variable of the amount of heat stored in the heat storage tank supporting the combined cooling, heating and power unit; It represents the variable of heat release of the heat storage tank supporting the combined cooling, heating and power unit; represents the optimal value of the second stage optimization problem, whose input includes , , , and ; Intraday decision making is a second-stage optimization problem: (43) ; The decision variables of the present invention mainly include the start and stop state of the micro diesel engine, the active output of the micro diesel engine, the reactive output of the micro diesel engine, the output of the combined heat and power unit, the reactive compensation amount, the electricity purchase / sale amount, the natural gas purchase amount, the waste heat recovery amount of the combined heat and power unit, the heat output of the combined heat and power unit supporting the hot boiler, the charging amount of the heat storage tank supporting the combined heat and power unit, the heat release of the heat storage tank supporting the combined heat and power unit, the cooling capacity of the absorption refrigeration machine supporting the combined heat and power unit, and the charging and releasing state of the heat storage tank supporting the combined heat and power unit; the state variables mainly include the line active power distribution, the line reactive power distribution, the node voltage distribution and the heat stored in the heat storage tank supporting the combined heat and power unit.

[0055] Considering the uncertainty of energy demand, the deterministic optimization model based on microgrid energy management can be expressed as a distributed robust optimization model based on Wasserstein distance. The first stage optimization problem is: (44) ; In the formula, represents the energy demand forecast error distribution; represents a fuzzy set that defines the distribution of energy demand forecast errors, whose range depends on the parameter ; Represents the distribution of forecast error for a given energy demand The expectation of the optimal value of the second-stage optimization problem; represents the expectation of the distribution of errors in a given energy demand forecast; represents the comprehensive energy demand forecast error random variable, including electricity, heat, and cooling demand: , In the formula, Represents the transposition of the electric energy demand parameter quantity; Represents the transposition of the heat energy demand parameter quantity; Represents the transposition of the cooling demand parameter quantity; The objective function of the first stage is to minimize the sum of the expected value of the start-stop cost of the micro diesel engine and the daily operation cost of the system. Based on the uncertainty set of energy demand forecast error, the fuzzy set is defined as: , In the formula, represents all potential distributions of energy demand forecast errors; represents the empirical distribution of energy demand forecast errors; Representation distribution and distribution The distance between The uncertainty set representing the energy demand forecast error is: ; In order to consider all potential energy demand forecast error distributions, the target robustness optimization model of multi-energy microgrid energy management starts from the energy demand empirical distribution, allows all potential distributions of energy demand, imposes penalties on the energy demand forecast error distribution that deviates from the energy demand forecast error empirical distribution, minimizes the penalty term, that is, minimizes the impact of energy demand forecast error uncertainty, and maximizes the robustness of the specified microgrid system operation cost; the first stage optimization model is expressed as: ; In order to consider all potential energy demand forecast error distributions, the target robustness optimization model of multi-energy microgrid energy management starts from the energy demand empirical distribution, allows all potential distributions of energy demand, imposes penalties on energy demand forecast error distributions that deviate from the energy demand forecast error empirical distribution, minimizes the penalty term, that is, minimizes the impact of energy demand forecast error uncertainty, and maximizes the robustness of the specified microgrid system operating cost. The first stage optimization model can be expressed as: (45) (46) , (47) In the formula, represents the penalty for energy demand forecast error distribution shift; It represents the sufficiently satisfactory operation cost target of the microgrid system specified by experts based on domain knowledge.

[0056] The measurement function of the distance between the empirical distribution and the actual distribution of energy demand forecast error has not been clarified, and the target robustness optimization model of multi-energy microgrid energy management is still unsolvable. To this end, the joint distribution of the actual distribution and empirical distribution of energy demand forecast error is established, and the Wasserstein distance is used to measure the distance between the actual distribution and empirical distribution of energy demand forecast error: , In the formula, represents Wasserstein distance; represents the joint distribution of the actual distribution and the empirical distribution of energy demand forecast errors; Represents the distribution of forecast error for a given energy demand Down and The expectation of the difference-one norm of ; A random variable representing the historical error in the combined energy demand forecast: , In the formula, The transpose of the equivalent random variable representing the historical error in the electricity demand forecast; The transpose of the equivalent random variable representing the historical error in the thermal energy demand forecast; The transpose of the equivalent random variable representing the historical error in cooling demand forecasts; Then formula (46) is expressed as: (48) In the formula, Represents the joint distribution range of the actual distribution and empirical distribution of energy demand forecast error: , In order to consider the coupling effect between energy demand such as electricity, cooling and heating energy, the distance between the actual distribution and the empirical distribution of comprehensive energy demand forecast error is expressed as a random variable in combination with Wasserstein distance. : (49) In the formula, Represents a norm.

[0057] Introducing historical data on energy demand forecast errors: (50) In the formula, Indicates historical data / sample number, such as timestamp; Indicates the number of historical scenes; express The rolling number of historical data, that is, ; Indicates The joint distribution corresponding to the historical errors of the group comprehensive energy demand forecast; Indicates The distribution distance corresponding to the historical error of the comprehensive energy demand forecast is: (51) In the formula, Indicates Historical data of group power demand forecast errors; Indicates Historical data of forecast errors of heat demand; Indicates Historical data of cooling demand forecast errors.

[0058] The joint distribution corresponding to the historical errors of the comprehensive energy demand forecast is expressed as a tractable uncertainty: (52) In order to simplify the objective robustness optimization problem into a more tractable form, an auxiliary variable of distribution distance is introduced to represent the distance between the actual distribution and the empirical distribution of the comprehensive energy demand forecast error: (53) in, Auxiliary variables representing the vicinity of the empirical distribution; Indicates that as of The moment-to-moment energy demand forecast error is a random variable; Indicated in Time about auxiliary variables and as of Energy demand forecast error random variable and its Natural gas fuel cost function for historical scenarios; Indicates Micro diesel engine Time about auxiliary variables and as of Energy demand forecast error random variable and its Fuel cost function for historical scenarios; Indicates Micro diesel engine Time about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; Indicates The combined heat and power unit Time about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; Indicated in Time about auxiliary variables and as of Energy demand forecast error random variable and its The cost function of electricity purchase and sale for each historical scenario; Indicates PV arrays about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; About The uncertainty set of energy demand forecast error based on historical data is expressed as: .

[0059] Step 4: Solve the target robustness optimization model of microgrid energy management.

[0060] In the present invention, step 4 specifically includes the following sub-steps: Step 4.1: Adjustable decision rules; The target robust optimization model for energy management of multi-energy microgrid system does not deal with the relationship between the second-stage decision variables and the uncertainty of comprehensive energy demand forecast error and auxiliary variables. In order to improve the flexibility of the target robust optimization model solution, the present invention introduces an adjustable decision rule

[24] , the second-stage decision variables and state variables are adjusted according to the actual energy demand and Wasserstein distance. Taking the amount of electricity purchased as an example, the amount of electricity purchased / sold can be expressed in a similar form: (54) (55) In the formula, Indicates the running time; Indicates that Uncertainty of energy demand forecast error at each moment and its The electricity purchase plan function for each historical scenario; Indicates that Uncertainty of energy demand forecast error at each moment and its The electricity purchase plan function for each historical scenario; They are The intercept coefficient of They are The slope coefficient of for The intercept coefficient of for The slope coefficient of for The coefficient of the auxiliary variable .

[0061] By solving the adjustable decision rule coefficients through the target robust optimization model, the energy management strategy of the multi-energy microgrid system can be obtained according to the actual energy demand and Wasserstein distance. For example, the relevant coefficients of electricity purchase / sale include , , , , In other words, this microgrid energy management strategy changes with the changes in the actual demand for electricity, heat, and cooling, and belongs to dynamic microgrid energy management, and takes into account the historical electricity, heat, and cooling energy consumption. Compared with not dealing with the relationship between the second-stage decision variables and uncertainties, the adjustable decision rules can fully tap the historical information of comprehensive energy consumption, enhance the adaptability of the target robustness optimization model to changes in energy demand, and thus improve the flexibility of the model solution.

[0062] Step 4.2: Model linearization; The present invention will process quadratic constraints, second-order cone constraints, and constraints containing absolute values ​​in the target robust optimization model based on adjustable decision rules. First, the binary auxiliary variable The constraint (2) with absolute values ​​can be expressed as: (56) (57) (58) Based on the Cartesian coordinate system, the dodecagon interior approximation method is adopted. Taking the second-order cone constraint (14) as an example, the quadratic constraint (13) and the second-order cone constraint (14) are expressed in a similar form: (59) in, Indicates Micro diesel engine Energy demand forecast error random variable and its Reactive power output plan for each historical scenario; Indicates Micro diesel engine Energy demand forecast error random variable and its A meritorious contribution plan for each historical scene; Indicates The horizontal coordinate of the nth point of the approximate dodecagon inside the original feasible region of the micro diesel engine; It represents the ordinate of the nth point of the approximate dodecagon inside the original feasible region of the vth micro diesel engine; Indicates The horizontal coordinate of the mth point of the approximate dodecagon in the original feasible region of a micro diesel engine Indicates The ordinate of the mth point of the approximate dodecagon inside the original feasible region of the micro diesel engine; (60) (61).

[0063] Step 4.3: Solvable form of the target robustness optimization model; The energy demand forecast error has infinite values ​​in the uncertain set and fuzzy set. The constraints of the energy demand forecast error uncertainty and random variables are actually infinite constraints, which makes the target robust optimization model a semi-infinite programming model. Therefore, it is necessary to express it as a solvable form with a finite number of constraints. Specifically, each constraint of the target robust optimization model needs to be treated as a separate optimization problem and the duality theorem of linear programming theory is applied to obtain a linear programming model with a finite number of constraints for solution.

[0064] Taking the key formula (53) as an example, based on the adjustable decision rule, the sub-problem containing uncertainty in formula (53) is expressed in a compact form: (62) (63) (64) (65) In the formula, Indicates the device number; represents the first stage coefficient vector; represents the second stage coefficient vector; represents the decision variables of the first-stage optimization problem; represents the intercept coefficient of the decision variable of the second stage optimization problem; represents the slope coefficient of the decision variable of the second-stage optimization problem; The coefficients of auxiliary variables representing the decision variables of the second-stage optimization problem; Represents the lower limit of interval prediction error; Represents the upper limit of the interval prediction error; About Devices in The uncertainty of energy demand forecast error at each moment; The energy demand forecast error is Historical data of each scene; Formula (65) can be expressed as: (66) In the formula, express No. A historical scene; Further introduction of auxiliary variables , formula (66) is equivalent to: (67) (68) Mode In linear programming form, when the uncertain set When not empty, the formula There is a feasible solution. According to the duality theorem of linear programming theory, strong duality holds, and the dual subproblem can be obtained: (69) (70) (71) (72) (73) In the formula, is the dual variable matrix corresponding to constraint (63); is the dual variable matrix corresponding to constraint (64); is the dual variable matrix corresponding to the left side of constraint (68); is the dual variable matrix corresponding to the right side of constraint (68); is a single dual variable corresponding to constraint (67); For the Devices in dual variable of the moment constraint (63); For the Devices in Dual variable of the moment constraint (64); For the Devices in The dual variable on the left side of the moment constraint (68); For the Devices in The dual variable on the right side of moment constraint (68); To satisfy An indicator function that outputs 1 when .

[0065] Other constraints can also be processed by applying the duality theorem to obtain the corresponding solvable form. According to the actual power, heat and cooling demand, the target robustness decision-making scheme of the energy management of the multi-energy microgrid system in the first and second stages can be obtained based on the solvable form.

Claims

1. A multi-energy microgrid scheduling method based on target robustness optimization, characterized in that: The following steps are involved: Step 1: Establish a multi-energy microgrid system; Step 2: Establish a deterministic optimization model for energy management of multi-energy microgrids; Step 3: Establish a target robustness optimization model for microgrid energy management; Step 4: Solve the target robustness optimization model of microgrid energy management; Wherein, the step 2 specifically includes the following steps: Step 2.1: Establish the objective function of the microgrid energy management deterministic optimization model; The deterministic optimization model of multi-energy microgrid energy management takes minimizing the system operating cost as the objective function, involving the start-stop cost of the micro diesel engine, the fuel cost of the micro diesel engine, the operation and maintenance cost of the micro diesel engine, the operation and maintenance cost of the combined heating and cooling unit, the operation and maintenance cost of the photovoltaic array, the cost / income of electricity purchase and sale, and the cost of purchasing natural gas: （1） in, （2） （3） （4） （5） （6） （7） （8） In the formula: min represents the minimization of the objective function; Indicates the running time; Indicates the operation cycle; Indicates The rolling moment of the moment, i.e. ; Indicates the number of the micro diesel engine; Indicates the number of micro diesel engines; express The rolling number of a micro diesel engine, i.e. ; Indicates the CCHP unit number; Indicates the number of combined cooling, heating and power units; express The rolling number of each cogeneration unit, i.e. ; Indicates the photovoltaic array number; Indicates the number of photovoltaic arrays; express The rolling number of the photovoltaic array, i.e. ; Indicates A micro diesel engine The start and stop costs at the dispatch time; Indicates A micro diesel engine Fuel cost at the dispatch time; Indicates A micro diesel engine Operation and maintenance costs at the time of dispatch; Indicates Combined cooling, heating and power units in Operation and maintenance costs at the time of dispatch; Indicates Photovoltaic arrays in Operation and maintenance costs at the time of dispatch; Indicated in Natural gas fuel cost at the dispatch time; Indicated in The cost of purchasing and selling electricity at the dispatch time; Indicates Unit start-stop cost of a micro diesel engine; Indicates A micro diesel engine The start and stop plan at the dispatch time, Indicates A micro diesel engine Start and stop plan at dispatch time; Indicates Unit fuel cost of a micro diesel engine; Indicates A micro diesel engine Active power output plan at the dispatching time; Indicates unit duration, used to convert power units into energy units; Indicates Unit operation and maintenance cost of a micro diesel engine; Indicates Unit operation and maintenance cost of each CHP unit; Indicates Combined cooling, heating and power units in Output plan at dispatch time; Indicates Unit operation and maintenance cost of a photovoltaic array; Indicates Photovoltaic arrays in Output plan at dispatch time; Indicates Unit natural gas fuel cost per CHP unit; Indicates Combined cooling, heating and power units in Gas usage plan at dispatch time; Indicated in The electricity purchase price at the dispatch time; Indicated in Power purchase plan at dispatch time; Indicated in The electricity sales price at the dispatch time; Indicated in Electricity sales plan at dispatch time; Step 2.2: Determine the constraints of the microgrid energy management deterministic optimization model; The DistFlow model is used to describe the active power, reactive power and voltage balance of each node: （9） In the formula, Indicates the node number, Indicates the number of nodes; Indicates until The rolling number of the node, i.e. ; Indicates that the The node Micro diesel engine Timely meritorious contribution plan; Indicates that the The node Photovoltaic arrays in Timely output plan; Indicates that the The node Combined cooling, heating and power units in Timely output plan; Indicated in Scheduling time from node Outflow and inflow nodes The amount of active power; Representation Node exist Active power load demand at the dispatching time; （10） In the formula, Indicated in Scheduling time from node Outflow and inflow nodes The amount of reactive power; Indicates that the The node Micro diesel engine Reactive power output plan at all times; Indicates that the The node Photovoltaic arrays in The reactive power compensation at the moment is provided by the inverter installed in conjunction with the photovoltaic array; Representation Node exist Reactive power / load at the dispatch time; It is worth noting that Indicates access The node The active power output of a micro diesel engine, Indicates access The node The reactive power output of a micro diesel engine is If there are no devices on the nodes, ; , and Similarly; （11） In the formula, Representation Node exist Voltage at the dispatch time; Representation Node and nodes The resistance of the line between them; Representation Node and nodes The reactance of the line between them; Indicates the reference voltage; Among them, the voltage limit constraint is: （12） In the formula, Indicates the maximum voltage offset; The line capacity constraint is: （13） In the formula, Representation Node and nodes The maximum apparent power of electric energy transmitted between the lines; The output constraints of micro diesel engines are: （14） In the formula, Indicates The maximum apparent power output of a micro diesel engine; （15） In the formula, Indicates The minimum active output of a micro diesel engine; The mini diesel engine climbing constraint is: （16） In the formula, Indicates The gradeability of the output of a micro diesel engine; The reactive power compensation constraint is: （17） In the formula, Indicates The maximum apparent power output of each PV array and its inverter; Indicates Rated output of each photovoltaic array; Indicates Reactive power compensation amount of each photovoltaic array inverter; Indicates The maximum reactive power compensation of a photovoltaic array inverter.

2. According to a multi-energy microgrid scheduling method based on target robustness optimization according to claim 1, it is characterized in that: The node that interacts with the grid energy is defined as node 1, and its power purchase / sale quantity satisfies: （18） Among them, the power purchase and sales constraints meet: （19） （20）。 3. The multi-energy microgrid scheduling method based on target robustness optimization according to claim 2 is characterized in that: The combined cooling, heating and power unit consumes natural gas output and recovers the remaining energy for heating: （21） （22） In the formula, Indicates Combined cooling, heating and power units in The total energy generated at any moment, Indicates The output of each combined cooling, heating and power unit accounts for the proportion of total energy; Indicates Combined cooling, heating and power units in Heat recovered at all times; Indicates The proportion of waste heat recovery of each cogeneration unit to the remaining heat; Among them, the output constraint of the combined cooling, heating and power unit is: （23） In the formula, Indicates Minimum output of each combined cooling, heating and power unit; Indicates Maximum output of each combined cooling, heating and power unit; The ramping constraint of the combined cooling, heating and power unit is: （24） In the formula, Indicates Output ramp rate of each CHP unit; The waste heat recovered by the combined cooling, heating and power unit meets the constraints: （25） In the formula, Indicates Combined cooling, heating and power units in Maximum heat recovered at all times; In order to promote the consumption of photovoltaic power and meet the peak demand of various types of energy, the heat balance equation of the node connected to the combined heating and cooling unit is as follows: （26） In the formula, Indicates The hot boilers of the combined cooling, heating and power units are Heat production at each moment; Indicates The heat production efficiency of the hot boilers supporting each combined cooling, heating and power unit; Indicates The heat storage tanks of the combined cooling, heating and power units are Heat release at any time; Indicates The heat storage tanks of the combined cooling, heating and power units are The amount of heat at any time; Indicates The combined cooling, heating and power units are equipped with absorption chillers. Cooling capacity at the time; Indicates Combined cooling, heating and power units in The corresponding heating demand at the time.

4. The multi-energy microgrid scheduling method based on target robustness optimization according to claim 3 is characterized in that: A binary variable is used to represent the charging and discharging status of the heat storage tank supporting the combined heat and power unit: （27） Where: Indicates The heat storage tanks of the combined cooling, heating and power units are The charging status at all times; Indicates The heat storage tanks of the combined cooling, heating and power units are The exothermic state at all times; The heat storage tank charging and discharging constraints of the combined cooling, heating and power unit are: （28） （29） Where: Indicates Maximum heat capacity of the heat storage tank supporting each combined cooling, heating and power unit; Indicates The maximum heat release of the heat storage tank supporting each combined cooling, heating and power unit; The heat stored in the heat storage tank of the combined heat and power unit satisfies the state equation: （30） Where: Indicates The heat storage tanks of the combined cooling, heating and power units are The amount of heat stored at all times; Indicates The charging efficiency of the heat storage tanks supporting each combined cooling, heating and power unit; Indicates Heat release efficiency of the heat storage tanks supporting each combined cooling, heating and power unit; The initial state of the heat stored in the heat storage tank supporting the combined cooling, heating and power unit is consistent with the end time of the previous scheduling cycle: （31） Then the heat stored in the heat storage tank of the combined cooling, heating and power unit meets the constraints: （32） Where: Indicates The upper limit of the amount of heat that can be stored in the heat storage tank of each combined cooling, heating and power unit; Indicates The lower limit of heat storage in the heat storage tank of each combined heating, cooling and power unit.

5. The multi-energy microgrid scheduling method based on target robustness optimization according to claim 4 is characterized in that: The cooling capacity of the absorption refrigeration mechanism supporting the combined cooling, heating and power unit meets the constraints: （33） Where: Indicates Refrigeration efficiency of absorption chillers for each combined cooling, heating and power unit; Indicates The combined cooling, heating and power units are equipped with absorption chillers. Refrigeration needs at all times.

6. The multi-energy microgrid scheduling method based on target robustness optimization according to claim 5 is characterized in that: The heat output of the hot boiler supporting the combined cooling, heating and power unit meets the constraints: （34） In the formula, Indicates The maximum heat output of the hot boiler supporting each combined cooling, heating and power unit.

7. The multi-energy microgrid scheduling method based on target robustness optimization according to claim 6 is characterized in that: The amount of natural gas consumed by the combined cooling, heating and power unit and its supporting heat boiler is: （35） In the formula, Indicates the total energy that can be generated by a unit amount of natural gas.

8. The multi-energy microgrid scheduling method based on target robustness optimization according to claim 7 is characterized in that: The step 3 specifically includes the following sub-steps: Step 3.1: Uncertain set of microgrid energy demand; Focusing on the uncertainty of user electricity, heating, and cooling demand and the coupling effect between multiple types of energy demand, the energy demand random variable is expressed as the sum of the energy demand expectation and the energy demand prediction error random variable: （36） （37） （38） In the formula, Representation Node exist The random variable of active load energy demand at the dispatching time, Representation Node exist The expected active load demand at the dispatch time, Representation Node exist Random variable of active power load demand forecast error at the dispatching time; Indicates Combined cooling, heating and power units in The corresponding heat demand random variable at time, Indicates Combined cooling, heating and power units in The corresponding heat demand expectation at the time, Indicates Combined cooling, heating and power units in The corresponding heat demand forecast error random variable at the time; Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand random variable at time, Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand expectations at all times, Indicates The combined cooling, heating and power units are equipped with absorption chillers. The cooling demand forecast error at time is a random variable; Among them, the uncertainty of energy demand forecast error satisfies the uncertainty set: （39） （40) (41) In the formula, The uncertainty set represents the uncertainty of the electric energy demand of the electric load; The uncertainty set represents the uncertainty of heat energy demand; The uncertainty set represents the uncertainty of cooling demand; Representation Node exist The uncertainty of the electric load energy demand at the dispatching time, Representation Node exist The lower limit of the forecast error of the electric load demand interval at the dispatching time, Representation Node exist The upper limit of the forecast error of the electric load demand interval at the dispatching time; Indicates Combined cooling, heating and power units in The uncertainty of heat demand at each moment, Indicates Combined cooling, heating and power units in The lower limit of the forecast error of the heat demand interval at the moment, Indicates Combined cooling, heating and power units in The upper limit of the forecast error of the heat demand interval at the moment; Indicates Combined cooling, heating and power units in The uncertainty of cooling demand at each moment, Indicates Combined cooling, heating and power units in The lower limit of the forecast error of the cooling demand interval at time, Indicates Combined cooling, heating and power units in The upper limit of the forecast error of the cooling demand interval at the time; represents the field of real numbers, indicating , and belongs to the field of real numbers; Step 3.2: Target robustness optimization model; The deterministic optimization model of multi-energy microgrid energy management is represented as a two-stage problem for modeling; the day-ahead decision is the first-stage optimization problem: （42） ； In the formula, represents the micro diesel engine start-stop schedule variable; Indicates the variable of the charging state of the heat storage tank supporting the combined heat and power unit; Indicates the heat release state variable of the heat storage tank supporting the combined heat and power unit; Indicates the variable of the amount of heat stored in the heat storage tank supporting the combined cooling, heating and power unit; It represents the variable of heat release of the heat storage tank supporting the combined cooling, heating and power unit; represents the optimal value of the second stage optimization problem, whose input includes , , , and ; Intraday decision making is a second-stage optimization problem: （43） ； The day-ahead decision-making and intraday decision-making variables mainly include the start-stop status of the micro-diesel engine, the active output of the micro-diesel engine, the reactive output of the micro-diesel engine, the output of the combined heat and power unit, the reactive compensation amount, the purchase / sale of electricity, the purchase of natural gas, the waste heat recovery of the combined heat and power unit, the heat output of the supporting hot boiler of the combined heat and power unit, the charging heat of the supporting heat storage tank of the combined heat and power unit, the heat release of the supporting heat storage tank of the combined heat and power unit, the cooling capacity of the supporting absorption refrigeration machine of the combined heat and power unit, and the charging and releasing heat status of the supporting heat storage tank of the combined heat and power unit; the state variables mainly include the line active power distribution, the line reactive power distribution, the node voltage distribution and the heat stored in the supporting heat storage tank of the combined heat and power unit; Considering the uncertainty of energy demand, the deterministic optimization model based on microgrid energy management is expressed as a distributed robust optimization model based on Wasserstein distance; the optimization problem in the first stage is: （44） ； In the formula, represents the energy demand forecast error distribution; represents a fuzzy set that defines the distribution of energy demand forecast errors, whose range depends on the parameter ; Represents the distribution of forecast error for a given energy demand The expectation of the optimal value of the second-stage optimization problem; represents the comprehensive energy demand forecast error random variable, including electricity, heat, and cooling demand: ， In the formula, Represents the transposition of the electric energy demand parameter quantity; Represents the transposition of the heat energy demand parameter quantity; Represents the transposition of the cooling demand parameter quantity; The objective function of the first stage is to minimize the sum of the expected value of the start-stop cost of the micro diesel engine and the daily operation cost of the system. Based on the uncertainty set of energy demand forecast error, the fuzzy set is defined as: ， In the formula, represents all potential distributions of energy demand forecast errors; represents the empirical distribution of energy demand forecast errors; Representation distribution and distribution The distance between The uncertainty set representing the energy demand forecast error is: ； In order to consider all potential energy demand forecast error distributions, the target robustness optimization model of multi-energy microgrid energy management starts from the energy demand empirical distribution, allows all potential distributions of energy demand, imposes penalties on the energy demand forecast error distribution that deviates from the energy demand forecast error empirical distribution, minimizes the penalty term, that is, minimizes the impact of energy demand forecast error uncertainty, and maximizes the robustness of the specified microgrid system operation cost; the first stage optimization model is expressed as: （45） （46） ， （47） In the formula, represents the penalty for energy demand forecast error distribution shift; It represents the sufficiently satisfactory operation cost target of the microgrid system specified by experts based on domain knowledge; The measurement function of the distance between the empirical distribution and the actual distribution of energy demand forecast error has not been clarified, and the target robustness optimization model of multi-energy microgrid energy management is still unsolvable; therefore, the joint distribution of the actual distribution and the empirical distribution of energy demand forecast error is established, and the Wasserstein distance is used to measure the distance between the actual distribution and the empirical distribution of energy demand forecast error: ， In the formula, represents Wasserstein distance; represents the joint distribution of the actual distribution and the empirical distribution of energy demand forecast errors; Represents the distribution of forecast error for a given energy demand Down and The expectation of the difference-one norm of ; A random variable representing the historical error in the combined energy demand forecast: ， In the formula, The transpose of the equivalent random variable representing the historical error in the electricity demand forecast; The transpose of the equivalent random variable representing the historical error in the thermal energy demand forecast; The transpose of the equivalent random variable representing the historical error in cooling demand forecasts; Then formula (46) is expressed as: (48) In the formula, Represents the joint distribution range of the actual distribution and empirical distribution of energy demand forecast error: , In order to consider the coupling effect between energy demand such as electricity, cooling and heating energy, the distance between the actual distribution and the empirical distribution of comprehensive energy demand forecast error is expressed as a random variable in combination with Wasserstein distance. : (49) In the formula, represents a norm; Introducing historical data on energy demand forecast errors: (50) In the formula, Indicates historical data / sample number, such as timestamp; Indicates the number of historical scenes; express The rolling number of historical data, that is, ; Indicates The joint distribution corresponding to the historical errors of the group comprehensive energy demand forecast; Indicates The distribution distance corresponding to the historical error of the comprehensive energy demand forecast is: (51) In the formula, Indicates Historical data of group power demand forecast errors; Indicates Historical data of forecast errors of heat demand; Indicates Historical data of group cooling demand forecast errors; The joint distribution corresponding to the historical errors of the comprehensive energy demand forecast is expressed as a tractable uncertainty: (52) In order to simplify the objective robustness optimization problem into a more tractable form, an auxiliary variable of distribution distance is introduced to represent the distance between the actual distribution and the empirical distribution of the comprehensive energy demand forecast error: (53) in, Auxiliary variables representing the vicinity of the empirical distribution; Indicates that as of The moment-to-moment energy demand forecast error is a random variable; Indicated in Time about auxiliary variables and as of Energy demand forecast error random variable and its Natural gas fuel cost function for historical scenarios; Indicates Micro diesel engine Time about auxiliary variables and as of Energy demand forecast error random variable and its Fuel cost function for historical scenarios; Indicates Micro diesel engine Time about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; Indicates The combined heat and power unit Time about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; Indicated in Time about auxiliary variables and as of Energy demand forecast error random variable and its The cost function of electricity purchase and sale for each historical scenario; Indicates PV arrays about auxiliary variables and as of Energy demand forecast error random variable and its The operation and maintenance cost function of the historical scenario; About The uncertainty set of energy demand forecast error based on historical data is expressed as: 。 9. The multi-energy microgrid scheduling method based on target robustness optimization according to claim 8 is characterized in that: The step 4 specifically includes the following sub-steps: Step 4.1: Adjustable decision rules; The target robust optimization model for energy management of multi-energy microgrid system does not deal with the relationship between the second-stage decision variables and the uncertainty of comprehensive energy demand forecast error and auxiliary variables. In order to improve the flexibility of the solution of the target robust optimization model, an adjustable decision rule is introduced to adjust the second-stage decision variables and state variables according to the actual energy demand and Wasserstein distance. Taking the purchased electricity as an example, the electricity purchase / sale amount can be expressed in a similar form: （54） （55） In the formula, Indicates the running time; Indicates that Uncertainty of energy demand forecast error at each moment and its The electricity purchase plan function for each historical scenario; Indicates that Uncertainty of energy demand forecast error at each moment and its The electricity purchase plan function for each historical scenario; They are The intercept coefficient of They are The slope coefficient of for The intercept coefficient of for The slope coefficient of for The auxiliary variable coefficient of ; By solving the adjustable decision rule coefficients through the target robust optimization model, the energy management strategy of the multi-energy microgrid system can be obtained according to the actual energy demand and Wasserstein distance; Step 4.2: Model linearization; The quadratic constraints, second-order cone constraints, and constraints containing absolute values ​​of the objective robust optimization model based on adjustable decision rules will be handled; First, introduce binary auxiliary variables The constraint (2) with absolute values ​​can be expressed as: （56） （57) (58) Based on the Cartesian coordinate system, the dodecagon interior approximation method is adopted. Taking the second-order cone constraint (14) as an example, the quadratic constraint (13) and the second-order cone constraint (14) are expressed in a similar form: (59) in, Indicates Micro diesel engine Energy demand forecast error random variable and its Reactive power output plan for each historical scenario; Indicates Micro diesel engine Energy demand forecast error random variable and its A meritorious contribution plan for each historical scene; Indicates The horizontal coordinate of the nth point of the approximate dodecagon inside the original feasible region of the micro diesel engine; It represents the ordinate of the nth point of the approximate dodecagon inside the original feasible region of the vth micro diesel engine; Indicates The horizontal coordinate of the mth point of the approximate dodecagon in the original feasible region of a micro diesel engine Indicates The ordinate of the mth point of the approximate dodecagon inside the original feasible region of the micro diesel engine; （60） （61） Step 4.3: Solvable form of the target robustness optimization model; The energy demand forecast error has infinite values ​​in the uncertain set and fuzzy set. The constraints of the energy demand forecast error uncertainty and random variables are actually infinite constraints, which makes the target robust optimization model a semi-infinite programming model. Therefore, it is necessary to express it as a solvable form with a finite number of constraints. Specifically, each constraint of the target robust optimization model needs to be treated as a separate optimization problem and the duality theorem of linear programming theory is applied to it, so as to obtain a linear programming model with a finite number of constraints for solution. Taking the key formula (53) as an example, based on the adjustable decision rule, the sub-problem containing uncertainty in formula (53) is expressed in a compact form: （62） （63） (64) (65) In the formula, Indicates the device number; represents the first stage coefficient vector; represents the second stage coefficient vector; represents the decision variables of the first-stage optimization problem; represents the intercept coefficient of the decision variable of the second stage optimization problem; represents the slope coefficient of the decision variable of the second-stage optimization problem; The coefficients of auxiliary variables representing the decision variables of the second-stage optimization problem; Represents the lower limit of interval prediction error; Represents the upper limit of the interval prediction error; About Devices in The uncertainty of energy demand forecast error at each moment; The energy demand forecast error is Historical data of each scene; Formula (65) can be expressed as: （66） In the formula, express No. A historical scene; Further introduction of auxiliary variables , formula (66) is equivalent to: （67) (68) Mode In linear programming form, when the uncertain set When not empty, the formula There is a feasible solution. According to the duality theorem of linear programming theory, strong duality holds, and the dual subproblem can be obtained: (69) (70) (71） （72） （73） In the formula, is the dual variable matrix corresponding to constraint (63); is the dual variable matrix corresponding to constraint (64); is the dual variable matrix corresponding to the left side of constraint (68); is the dual variable matrix corresponding to the right side of constraint (68); is a single dual variable corresponding to constraint (67); For the Devices in The dual variable of the moment constraint (63); For the Devices in Dual variable of the moment constraint (64); For the Devices in The dual variable on the left side of the moment constraint (68); For the Devices in The dual variable on the right side of moment constraint (68); To satisfy An indicator function that outputs 1 when .

Images