Random distributed robust optimization scheduling method for power distribution network of integrated energy microgrid group

Abstract

The invention is suitable for the technical field of integrated energy microgrid scheduling, and provides a power distribution network random distributed robust optimization scheduling method of an integrated energy microgrid group, comprising the following steps: constructing a DN structure comprising a plurality of microgrids; the DN day-ahead scheduling optimization model containing the IEMGs pays attention to operation cost and constraint conditions; for the DN, the problem of non-convex nonlinear optimization including line loss is solved; for the IEMG, the model is used for processing a convex optimization problem considering WT and PV output power uncertainty; power exchange is carried out between the DN and the IEMG through a connecting line, so that a double-layer model framework is constructed to analyze the system; the solving process is a double-layer optimization iteration process, an outer layer obtains an interaction power value between the DN and the IEMG by using an ATC method, and an inner layer uses Camp; the CG method solves the problem of random distribution robust optimization of IEMG. According to the method, the carbon emission and the operation cost of the system can be effectively reduced, meanwhile, stable operation of the DN is ensured, and a theoretical basis is provided for efficient utilization of ammonia energy in the DN.

Description

Technical Field

[0001] The present invention belongs to the technical field of integrated energy microgrid scheduling, and particularly relates to a stochastic distributed robust optimal scheduling method for a distribution network of an integrated energy microgrid cluster. Background Art

[0002] The increasing diversification of energy demand and the serious environmental pollution problem have promoted the development of integrated energy microgrid (IEMG) cluster technology. IEMG can realize the large-scale grid connection of renewable energy (RE), and introducing it into the distribution network (DN) can improve the local consumption capacity of renewable energy in the system and provide new opportunities for the sustainable economic operation of DN. However, the joint operation system of IEMG cluster and DN involves multiple stakeholders and various distributed units, and these stakeholders will not unconditionally support the interests of other parties, resulting in an increasingly complex energy management strategy. At the same time, with the increase of RE penetration rate in IEMG, higher requirements are put forward for the stable operation ability of the coordinated operation system of DN and IEMG cluster. The power to ammonia (P2A) technology stores surplus renewable electric energy by converting electric energy into zero-carbon fuel, providing a new way for the efficient utilization of RE in the system. Therefore, considering the fluctuations of system uncertainty variables, studying the multi-stakeholder energy management strategy of DN and IEMG cluster including ammonia energy can break the traditional mode of independent operation of each energy supply system and achieve comprehensive operation optimization while taking into account the interests of all parties.

[0003] Currently, most of the research on DN modeling and energy management of IEMG cluster mainly focuses on centralized and distributed modeling. In contrast, distributed modeling regards DN and microgrids as independent entities with different interests, and models by restricting the interconnection lines of relevant parties, which can better reflect the actual situation and conform to the current development trend. In the prior art, most of the research uses time-of-use electricity price to coordinate and optimize DN and IEMG. Due to the fixed electricity price and time period of the time-of-use electricity price mechanism, the flexibility of energy interaction between different stakeholders is insufficient, and the flexibility of energy interaction between multiple stakeholders still needs to be improved. In addition, the uncertainty of renewable energy output power makes the energy management of IEMG and DN more challenging.

[0004] So far, the research on resisting the uncertainty of renewable energy mainly adopts the Stochastic Optimization (SO) method and the Robust Optimization (RO) method. Although the existing research has effectively offset the impact of the volatility of wind power and photovoltaic power in the system on the system's stable operation ability by using distributed robust optimization, the system often increases the output of coal-fired and gas-fired units after adopting the distributed robust optimization method, reducing the system's flexible operation ability and resulting in an increase in the system's carbon emissions. Therefore, the low-carbon operation ability of the system needs to be further explored.

[0005] On the other hand, if the IEMG does not have enough flexible alternative options, it may lead to a significant reduction in RE, thus exacerbating the impact of RE uncertainty on the economic operation of the microgrid. The Power-to-Ammonia (P2A) technology has the characteristics of high efficiency and flexible operation, which can convert excess electric energy into ammonia. At the same time, it can be combined with ammonia co-firing technology to provide an effective way for the large-scale storage and utilization of RE. However, the current research on ammonia doping mainly focuses on the operating characteristics of ammonia-modified gas turbines, lacking an analysis of the system's operation economics and low-carbon performance at the macro level after introducing P2A and ammonia co-firing.

[0006] Although many constructive ideas have been put forward in the current research on the coordinated operation of DN and IEMG, the following problems still need to be solved:

[0007] (1) In the field of the coordinated operation of DN and IEMG, most research aims to minimize the operating costs of DN and IEMG, construct a coordinated optimal scheduling model for multiple stakeholders, and achieve multiple benefits for different stakeholders. However, most of the current research ignores the impact of electricity price fluctuations on the energy usage behavior of IEMG users.

[0008] (2) In terms of resisting the uncertainty factors of the system, most research adopts SO and RO. Although the above methods can effectively improve the system's ability to resist the fluctuations of uncertainty factors, SO and RO respectively have problems of poor practicability and economy. Therefore, it is necessary to improve the existing scheduling optimization model.

[0009] (3) Currently, most research resists the fluctuations of system uncertainty variables by increasing the power generation of gas-fired units. However, increasing the power generation of gas-fired units will lead to an increase in the system's carbon emissions and also compress the space for renewable energy consumption in the system. P2A and ammonia co-firing technology can effectively alleviate these problems, but the feasibility of incorporating P2A into IEMG still needs to be comprehensively and multi-dimensionally evaluated. Summary of the Invention

[0010] The purpose of the embodiment of the present invention is to provide a stochastic distributed robust optimal scheduling method for the distribution network of an integrated energy microgrid group, aiming to solve the problems proposed in the above background technology.

[0011] The embodiment of the present invention is implemented as follows. A stochastic distributed robust optimal scheduling method for the distribution network of an integrated energy microgrid group includes the following steps:

[0012] Step 1: Construct a DN structure including multiple microgrids;

[0013] The DN includes multiple IEMGs. Power interaction is achieved between each IEMG and the DN through connection lines. By strategically arranging the interactive power with the DN, planning the output of distributed energy, and managing the output of different devices, the IEMG reduces the overall operating cost while meeting various load demands. The DN is used to manage the collaborative capabilities of each IEMG, strategically plan the allocation of distributed generation capacity, and optimize the power purchase from the main grid, minimizing the operating cost while meeting the power load demand.

[0014] Step 2: Hierarchical energy optimization management;

[0015] The day-ahead scheduling optimization model of the DN containing IEMGs focuses on operating costs and constraints. For the DN, this model aims to solve a non-convex and non-linear optimization problem including line losses. For the IEMG, this model is used to handle a convex optimization problem considering the uncertainty of the output power of wind turbines (WT) and photovoltaic (PV). Power exchange occurs between the DN and the IEMG through connection lines, so a two-layer model framework is constructed to analyze this system. This optimization problem is a two-layer optimization problem involving the DN layer and the IEMG layer, and the two layers are interconnected through coupling variables. First, the scheduling models of the DN and the IEMG are established, and then the Analysis Target Cascade (ATC) method is used to solve this combined model.

[0016] Step 3: Model solution;

[0017] The solution process of the constructed DN-IEMG collaborative operation model is a two-layer optimization iteration process. The outer layer uses the ATC method to obtain the interactive power value between the DN and the IEMG, while the inner layer uses the C&CG method to solve the stochastic distributed robust optimization problem of the IEMG.

[0018] A further technical solution is that in the above step 1, the IEMG including the P2A technology has a RE, a P2A module, a thermoelectric energy supply and conversion device, and an Integrated Demand Response (IDR) model;

[0019] The constructed device model is as follows:

[0020] 1) Operating model of the P2A module:

[0021] The technical route of green ammonia includes electrolytic water (EL), cryogenic air separation, and the Haber-Bosch thermochemical synthesis pathway; the electrochemical production of ammonia can be divided into the following three stages:

[0022] The first stage: Renewable energy provides electrical energy for EL. The electrolyzer electrolyzes water as raw material to produce hydrogen and oxygen. The chemical reaction formula in EL is shown in formula (1); formula (2) describes the operating model of EL:

[0023]

[0024] In the formula, is the hydrogen production rate of EL at time period ; η EL represents the hydrogen production efficiency of EL; P EL,e,t represents the power consumption of the electrolyzer at time period ; R EL,down and R EL,up represent the power increase and power decrease constraints of EL respectively;

[0025] The second stage: The air separation unit (ASU) uses pressure swing adsorption technology to extract nitrogen from the air; the ammonia production in this model can be adjusted according to the hydrogen production rate of EL. The power consumption of the pressure swing adsorption nitrogen production device is proportional to the hydrogen production rate of EL; the power consumption and operating constraints of ASU are as follows:

[0026]

[0027] Among them, P ASU,e,t represents the power consumption of ASU at time period ; P ASU,e,r represents the power consumption of ASU for producing unit volume or mass of nitrogen; represents the amount of nitrogen produced by ASU at time period ; P ASU,e,max represents the maximum power consumption of ASU;

[0028] The third stage: The ammonia synthesis device synthesizes ammonia from hydrogen and nitrogen by the Haber-Bosch process, as shown in the chemical reaction formula (4); in addition, in order to maintain the pressure in the ammonia synthesis device at 300 bar, part of the electrical energy is required to drive the internal air compressor to increase the pressure; the energy consumption of the ammonia production device is related to the ammonia synthesis rate, as shown in formula (5):

[0029]

[0030] Among them, ΔH represents the waste heat generated by synthesizing one mole of ammonia;

[0031]

[0032] Among them, P Haber，e,t represents the power consumption during the ammonia synthesis process in a time period; P Haber,e,r represents the power consumption per unit of ammonia produced; represents the ammonia production rate in a time period; represents the lower calorific value of ammonia;

[0033] 2) Ammonia co - combustion model of gas turbine (GT):

[0034] For a GT retrofitted with ammonia co - combustion technology, ammonia will replace a certain proportion of the natural gas entering the GT. The actual gas consumption of the GT is given by the following formula:

[0035]

[0036] Among them, F GT,gas,t represents the natural gas consumption of the GT in a time period; a gas , b gas and c gas are the combustion coefficients of natural gas; L gas represents the calorific value of natural gas; is the ammonia mixing ratio; P WHB,GT,t represents the waste heat power released by the GT in time period t; P GT,e,t represents the electric power output of the GT in time period t; η GT represents the operating efficiency of the GT; η loss,GT is the heat loss coefficient of the GT;

[0037] In the model, the upper limit of the ammonia mixing ratio is set to 20%.

[0038] 3) Integrated demand response model

[0039] IEMG adopts an incentive mechanism, including time - of - use market pricing signals or economic incentive measures; the demand response constraints of the IEMG power load are expressed as follows:

[0040]

[0041] Among them, P IL,e,max and P IL,e,min represent the maximum and minimum values of the interruptible power load respectively;

[0042] The demand response constraints of the IEMG heat load are expressed as follows:

[0043]

[0044] Among them, P IL,h,maxand P IL,h,min respectively represent the maximum and minimum values of the interruptible heat load.

[0045] For a further technical solution, in the said step 2:

[0046] 1) The overall structure of the model is as follows:

[0047]

[0048] In the above model, C dn represents the total operating cost before DN days; x dn and x line,dn correspond to the scheduling plan of DN and the tie-line power transmitted to the lower-layer IEMG respectively; the former involves the injection power, branch current and node voltage of the DN root node, while the latter describes the optimal power exchange scheme between DN and each IEMG based on the day-ahead scheduling strategy; G and H respectively represent the equality constraint and inequality constraint of DN operation; C iemg represents the total operating cost of IEMG; x iemg and x line,iemg correspond to the scheduling plan of IEMG and the tie-line power transmitted to the upper-layer DN respectively; the former mainly calculates the output power of the controllable distributed units in IEMG, while the latter describes the optimal power exchange scheme between each IEMG and DN in the day-ahead scheduling plan;

[0049] 2) The optimal scheduling model of DN:

[0050] The premise of DN scheduling is to meet its load demand, and its goal is to formulate an optimal strategy for purchasing electricity from the main power grid in advance and achieve effective power exchange with the microgrid; the core goal of scheduling is to minimize the operating cost;

[0051] (1) The objective function of DN operation

[0052] The specific objectives of the upper-layer DN are as follows:

[0053]

[0054] In this formula, the first term represents the cost of DN purchasing electricity from the main power grid; c main,t is the trading electricity price; P main,t represents the active power injected into the root node; the second term is the network loss cost, where C loss is the network loss price, I ij,t is the branch current, is the square of the branch current, r ij is the resistance of the corresponding branch; the third term represents the income obtained by DN through transmitting electricity to the microgrid, where C buy,k,t and C sell,k,tThey are the electricity purchase and sale prices from IEMG to DN; P DN-IEMG,k,t and P IEMG-DN,k,t are the corresponding electricity purchase and sale quantities; is the total number of nodes in DN; T refers to the operation period; i and j are node indices; k is the identifier of IEMG; M is the number of IEMGs in DN;

[0055] (2) Constraints

[0056] By introducing substitution variables and slack variables, and adjusting the variables, the original non-convex and non-linear optimization problem is transformed into a second-order cone optimization problem. The operation constraints of DN after introducing substitution variables and slack variables are as follows:

[0057]

[0058] Among them, P load,i,t and respectively represent the active power consumption and reactive power consumption of the node at the moment; g ij and b ij respectively represent the conductance and susceptance between nodes; is the reactive power injected into the root node; U i,max and U i,min are the minimum and maximum voltage values allowed for the node respectively; I ij,max is the maximum current allowed for the branch connecting nodes and nodes; U i,t represents the voltage value of the node at the moment; θ ij,t represents the phase angle difference between nodes and nodes at the moment; Z ij,t and Y ij,t respectively represent the impedance resistance and conductance of the branch at the moment; X i,t and X j,t respectively represent the reactance of nodes and nodes at the moment;

[0059] By introducing slack variables into Equation (11) and transforming it into the second-order cone form to meet the requirements of second-order cone optimization, it is specifically shown in Equation (12):

[0060]

[0061] 3) Optimization scheduling model of IEMG:

[0062] (1) Objective function

[0063] The optimization objective of IEMGs is to minimize the total operation cost, including the energy purchase cost, curtailment cost, unit operation cost, unit start-stop cost, distribution network interaction cost, and demand response compensation cost of IEMG;

[0064]

[0065] Among them, C gas,k,t represents the amount of natural gas purchased by the k-th IEMG; F gas,GT,k,t and F gas,GB,k,t respectively represent the natural gas consumption of the GT and the gas boiler (GB); η ab,WT and η ab,PV are respectively the cost coefficients of the WT and PV curtailment; β WT , β PV , β EB , β GT and β GB respectively represent the operating cost coefficients of the WT, PV, electric boiler (EB), GB, and GT; α WT , α PV , α EB , α GT and α GB are respectively the operating states of the WT, PV, EB, GT, and GB, as binary coefficients, where 1 represents operating and 0 represents shutdown. C IL,h,k,t and C IL,e,k,t are respectively the compensation cost coefficients for heat load interruption and power load interruption. k is used to identify the IEMG, M is the total number of IEMGs in the DN; t represents the time interval;

[0066] (2) Constraint conditions

[0067] Power balance constraint:

[0068]

[0069] Among them, P DN-IEMG,k,t represents the electricity sold by the DN to the IEMG; P IEMG-DN,k,t represents the electricity sold by the IEMG to the DN; P load,e,k,t and P load,h,k,t respectively represent the electrical load and heat load of the k-th IEMG at the moment;

[0070] Generator set constraint:

[0071] The generator sets in the IEMG include the GT and the GB; they follow the output upper and lower limits and the ramp rate constraint conditions:

[0072]

[0073] Among them, gas-unit is the general expression of the generator set, which is the general representation of the output power type of the generator set; R gas-unit,down and R gas-unit,up respectively represent the lower and upper limits of the ramp rate of the generator set; α gas-unit,off-on,k,t represents the start-stop state of the generator set;

[0074] Multi - energy coupling unit constraint:

[0075] The multi - energy coupling unit in IEMG includes a waste heat boiler (WHB) and an EB. The unified expression of the multi - energy coupling unit constraint is as follows:

[0076]

[0077] Among them, tran - unit is the general expression of the multi - energy coupling unit; and respectively represent the input power and output power of the energy conversion unit;

[0078] Energy storage unit constraint:

[0079] The energy storage unit of the micro - grid includes a battery storage (BS), heat storage tanks (HST), and an ammonia storage tank (AST). Its operation constraints include charge - discharge energy constraints, capacity continuity constraints, upper and lower limit constraints, and initial and final energy balance constraints, which are specifically as follows:

[0080]

[0081] Among them, storge is the general expression of the energy storage unit; α storge,ch,k,t and α storge,dis,k,t respectively represent the charging and discharging efficiencies of the energy storage unit; η storge,ch and η storge,dis respectively represent the charging and discharging states of the energy storage device, as binary coefficients, where 1 represents operation and 0 represents shutdown; E storge,t represents the energy storage capacity of the energy storage unit;

[0082] Inter - connection line power constraint:

[0083]

[0084] For a further technical solution, step 3 includes the following specific steps:

[0085] Step 3.1: IEMG stochastic distribution robust optimization model;

[0086] Objective function:

[0087] Divide the key variables in the deterministic optimal scheduling model in Step 1 into first-stage variables and second-stage variables; in the first stage, which is the day-ahead stage, formulate the optimal daily operation strategy to minimize the total cost of IEMG operation; the first stage consists of the decision variables of GT, GB, and energy storage devices, and these variables are not affected by the uncertainty of RE in day-ahead decisions; the second-stage variable is denoted by q, which can be adjusted according to the actual output of RE; the variables in these two stages are both represented in matrix form:

[0088]

[0089] In the formula, s represents the severe operation scenario of wind and light; P S is the probability of the occurrence of the severe wind and light scenario, and the occurrence probability can be used as a reference to estimate the extreme wind and light scenario; the historical occurrence probability of a certain type of extreme disaster scenario can be approximated as its future occurrence probability; Ψ Π is the expected value; Π is the probability distribution of the wind and light operation state; Υ is the fuzzy set of the wind and light operation state; is the additional operation cost of the system under the extreme scenario s;

[0090] Fuzzy set of wind and light fluctuations:

[0091] Construct a fuzzy set of wind and light fluctuations using the moment information of the wind and light output under the extreme scenario s, and describe the probability distribution of the wind and light output under any severe scenario through the fuzzy set, which is specifically expressed as follows:

[0092]

[0093] In the formula, represents the wind and light operation state under scenario s; R represents all operation scenarios of wind and light; P(R) is the probability distribution of all operation scenarios of wind and light; U is the uncertainty set of wind and light;

[0094] Transformation and solution of the stochastic distribution robust optimization model:

[0095] According to the definition of the wind and light fuzzy set, formulate the upper bound problem of the objective function (19) as a semi-infinite optimization problem:

[0096]

[0097] In the formula, is the maximum probability of the severe wind and light scenario; dΠ is the probability density function; ψ s and Υ s,wt-pv,t are dual variables; using the strong duality theory, transform formula (21) from a semi-infinite optimization problem into a finite optimization problem:

[0098]

[0099] Formula (23) is further rewritten as an expression for the most severe wind and light scenarios:

[0100]

[0101] Substitute Formulas (22) and (24) into the objective function (13) of IEMG to obtain an equivalent form of Formula (13):

[0102]

[0103] A three-layer optimization problem in the min-max-min form:

[0104] The converted integrated energy stochastic distribution robust optimization model is a min-max-min three-layer optimization problem; it is expressed in the matrix / vector form of Cai Yutong:

[0105]

[0106] s.t. Ap ≤ b (27)

[0107]

[0108] Bp + Cq s + Du s ≤ d (29)

[0109] In the formula, p is the necklace form of the first-stage decision variable, including the normal operation variables of GT, GB, and energy storage devices; q s is the vector form of the second-stage decision variable; ψ s and Υ s are the vector expression forms of the dual variables; u s is the vector form of the renewable energy operation state; a, b, c, d, and ρ s are the vector expression forms of constant coefficients; A, B, C, and D are matrices of constant coefficients;

[0110] Formula (27) is the first-stage unit normal operation constraint; Formula (29) is the second-stage operation constraint;

[0111] The three-layer model (26)-(29) is solved using the C&CG algorithm. First, the original problem is decomposed into a master problem and a sub-problem, and solved iteratively; since the first-stage decision of the stochastic distribution robust model is based on the line fault situations under N s scenarios, in each iteration process of the C&CG algorithm, N s wind and light operation scenarios need to be fed back to the master problem; after decomposition, the master problem is as follows:

[0112]

[0113] such that \(Ap\leq b\) (31)

[0114]

[0115] where \(j\) represents the number of times the sub - problem returns to the master problem; \(J\) represents the number of iterations, \(q\) s,j represents the second - stage decision variable added by the sub - problem to the master problem, represents the renewable energy operation scenario added when the \(j\) - th sub - problem returns to the master problem;

[0116] During each iteration, the C&CG algorithm needs to solve \(N\) S sub - problems, and each sub - problem corresponds to the wind - solar operation scenario; in addition to the first - stage decision variable \(p\), the dual variable \(\Upsilon\) obtained from the solution of the master problem s also needs to be passed to the sub - problems. The decomposed sub - problems are as follows:

[0117]

[0118] where \(p^*\) and are the results obtained by solving the master problem in the \(J\) - th iteration, \(V\) s is the dual variable of formula (35);

[0119] The sub - problems mentioned above belong to a bilevel optimization problem. With the help of the strong duality theory, the inner - layer min problem is transformed into a max problem, and then a single - layer optimization problem is obtained, which is specifically expressed as follows:

[0120]

[0121] C T V s \(\leq c\) (37)

[0122] Step 3.2: Solve the overall model based on ATC;

[0123] For the optimization coordination problem between the DN and the IEMG in the bilevel system, the ATC method is used to handle this optimization problem; between the DN and the IEMG, the tie - line power is set as \(P\) DN,load,k,t ; in the DN, the tie - line power is regarded as a virtual load \(P\) DN-IEMG,k,t ; in the IEMG, \(P\) IEMG-DN,k,t is regarded as the output of the virtual generator.

[0124] Using the ATC method to solve the overall scheduling model of DN - IEMG involves the following steps:

[0125] Step a: Input the relevant parameters of the distribution network and the integrated energy micro - grid, and the penalty function multiplier Set the initial value

[0126] Step b: Add the tie-line penalty factor to the objective function of the integrated energy microgrid according to formula (38); solve the data-driven stochastic distributed robust scheduling problem of each microgrid, and then transfer the obtained P IEMG-DN,k,t to the upper-level distribution network;

[0127]

[0128] Step c: During the DN scheduling process, introduce the tie-line penalty function K according to formula (39) into the DN objective function, solve the DN scheduling problem with relevant constraints, and transfer the obtained virtual load transfer P DN to the IEMG layer; DN-IEMG,k to the IEMG layer;

[0129]

[0130] Step d: Perform multiple optimization iterations on the DN and IEMG until the convergence criterion is met.

[0131] The distribution network stochastic distributed robust optimization scheduling method for the integrated energy microgrid group provided by the embodiment of the present invention has the following beneficial effects:

[0132] (1) To coordinate the interests of the distribution network and the integrated energy microgrid, a dynamic pricing mechanism is proposed. Based on the distribution network loss index and the load distribution characteristics of the integrated energy microgrid, this mechanism can effectively balance the interest conflicts between the distribution network and the microgrid, and is in line with the current development trend of the energy field. The dynamic pricing mechanism optimizes the load distribution by adjusting the electricity price in real time, reduces the power supply pressure during peak hours, and simultaneously reduces the system operation cost and carbon emissions.

[0133] (2) To improve the consumption rate of renewable energy, an integrated energy microgrid model containing P2A is constructed. This model converts excess renewable energy into ammonia through the power-to-ammonia technology and uses the ammonia co-firing technology to provide clean fuel for gas turbines, significantly improving the utilization rate of renewable energy. At the same time, the model optimizes the electro-thermal coupling relationship by recovering the waste heat in the power-to-ammonia process, improving the flexibility and efficiency of energy use.

[0134] (3) To alleviate the power supply pressure of the distribution network during the peak load period of the integrated energy microgrid, an integrated demand response mechanism is introduced. By adjusting the electricity consumption behavior of users, this mechanism transfers part of the peak load to the low valley period, significantly reducing the power purchase demand of the integrated energy microgrid. This not only alleviates the power supply pressure of the distribution network, but also reduces the system network loss, minimizing the impact of the microgrid peak load on the distribution network.

[0135] (4) To address the uncertainty of renewable energy, the ATC and C&CG algorithms are used to model and optimize renewable energy. This method ensures the accuracy of the second-order cone relaxation of the distribution network while effectively achieving convergence and meeting the requirements of system operation and scheduling. By introducing stochastic distributed robust optimization, the adaptability to the volatility of renewable energy is improved while ensuring the economy of the system. Brief Description of the Drawings

[0136] Figure 1 is the distributed two-layer optimization framework for the distribution network with an integrated energy microgrid;

[0137] Figure 2 is the schematic diagram of IEMG;

[0138] Figure 3 is the flow chart of green power to ammonia;

[0139] Figure 4 is the flow chart of the solution process. Specific Embodiments

[0140] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0141] The following describes the specific implementation of the present invention in detail with reference to specific embodiments.

[0142] A stochastic distributed robust optimization scheduling method for the distribution network of an integrated energy microgrid group provided by an embodiment of the present invention includes the following steps:

[0143] Step 1: Construct the distribution network structure including multiple microgrids:

[0144] The distribution network (DN) includes multiple integrated energy microgrids (IEMG). Power interaction is achieved between each IEMG and the DN through connecting lines. By strategically arranging the interactive power with the DN, planning the output of distributed energy, and managing the output of different devices, the IEMG reduces the overall operating cost while meeting various load demands. The distribution network effectively manages the collaborative capabilities of each IEMG, strategically plans the allocation of distributed generation capacity, and optimizes the power purchase from the main grid to minimize the operating cost while meeting the power load demand.

[0145] Among them, the IEMG structure including P2A technology is as Figure 2As shown, it includes RE, P2A module, thermoelectric energy supply and conversion equipment, as well as integrated demand response.

[0146] The constructed equipment model is as follows:

[0147] 1) Operating model of P2A module:

[0148] The technical route of green ammonia mainly involves water electrolysis, cryogenic air separation, and Haber-Bosch thermochemical synthesis pathway. The electrochemical production of ammonia can be divided into the following three stages:

[0149] The first stage: RE provides electrical energy for the electrolytic cell (EL). The electrolytic cell electrolyzes water as raw material to produce hydrogen and oxygen. Equation (1) represents the chemical reaction equation in EL, and Equation (2) describes the operating model of EL.

[0150]

[0151]

[0152] In the formula, is the hydrogen production rate of EL in the time period; η EL represents the hydrogen production efficiency of EL; P EL,e,t represents the power consumption of the electrolytic cell in the time period; R EL,down and R EL,up represent the power increase and power decrease constraints of EL respectively.

[0153] The second stage: The air separation unit (ASU) uses pressure swing adsorption technology to extract nitrogen from the air. The ammonia production in this model can be adjusted according to the hydrogen production rate of EL to prevent ASU from consuming too much electrical energy. The power consumption of the pressure swing adsorption nitrogen production device is proportional to the hydrogen production rate of EL. The power consumption and operating constraints of ASU are as follows:

[0154]

[0155] Among them, P ASU,e,t represents the power consumption of ASU in the time period; P ASU,e,r represents the power consumption of ASU for producing per unit volume or mass of nitrogen; represents the amount of nitrogen produced by ASU in the time period; P ASU,e,max represents the maximum power consumption of ASU.

[0156] Third stage: The ammonia synthesis unit synthesizes ammonia from hydrogen and nitrogen using the Haber-Bosch process, as shown in the chemical reaction formula (4). In addition, in order to maintain the pressure in the ammonia synthesis unit at 300 bar, a part of the electrical energy is required to drive the internal air compressor to increase the pressure. The energy consumption of the ammonia production unit is related to the ammonia synthesis rate, as shown in formula (5).

[0157]

[0158] Among them, ΔH represents the waste heat generated by synthesizing one mole (1 mol) of ammonia.

[0159]

[0160] Among them, P Haber，e,t represents the power consumption during the ammonia synthesis process in the time period; P Haber,e,r represents the power consumption per unit of ammonia produced; represents the ammonia production rate in the time period; represents the lower calorific value of ammonia.

[0161] The flow directions of different forms of energy during the electrochemical ammonia synthesis process are as Figure 3 shown.

[0162] 2) Ammonia co-firing model of gas turbine (GT):

[0163] For GT retrofitted with ammonia co-firing technology, ammonia will replace a certain proportion of the natural gas entering GT. The actual gas consumption of GT is given by the following formula:

[0164]

[0165] Among them, F GT,gas,t represents the natural gas consumption of GT in the time period; a gas , b gas and c gas are the combustion coefficients of natural gas; L gas represents the calorific value of natural gas; is the ammonia mixing ratio; P WHB,GT,t represents the waste heat power released by GT in the time period t; P GT,e,t represents the electrical power output of GT in the time period t; η GT represents the operating efficiency of GT; η loss,GT is the heat loss coefficient of GT.

[0166] Currently, IHI Corporation in Japan has successfully achieved a thermal ratio of natural gas-ammonia co-firing in gas turbines reaching 20%. Therefore, in the model, the upper limit of the ammonia mixing ratio is set to 20%.

[0167] 3) Integrated Demand Response Model

[0168] The IEMG promotes and encourages energy consumers to change their energy consumption patterns by adopting incentive mechanisms, including time-of-use market pricing signals or economic incentives. This move aims to enhance the balance between power supply and demand, thus ensuring the stability of the power grid. The demand response constraints of the IEMG power load are expressed as follows:

[0169]

[0170] Among them, P IL,e,max and P IL,e,min represent the maximum and minimum values of the interruptible power load respectively.

[0171] Due to the ambiguity in users' perception of thermal comfort, within the range of human comfort, users may not be greatly affected by changes in heating temperature, which improves the adjustability of the thermal load. The demand response constraints of the IEMG thermal load are expressed as follows:

[0172]

[0173] Among them, P IL,h,max and P IL,h,min represent the maximum and minimum values of the interruptible thermal load respectively.

[0174] Step 2: Hierarchical Energy Optimization Management

[0175] 1) Overall Model Framework:

[0176] The day-ahead scheduling optimization model of the DN with IEMGs mainly focuses on operating costs and constraints. For the DN, this model aims to solve the non-convex and non-linear optimization problem including line losses. At the same time, for the IEMG, this model mainly deals with the convex optimization problem considering the uncertainty of the output power of wind turbines (WT) and photovoltaic (PV). Power exchange occurs between the DN and the IEMG through connecting lines. Therefore, a two-layer model framework is constructed to analyze this system.

[0177] This optimization problem is a two-layer optimization problem involving the DN layer and the IEMG layer, and the two layers are connected through coupling variables. First, the scheduling models of the DN and the IEMG are established, and then the Analysis Target Cascade (ATC) method is used to solve this joint model. The overall structure is as follows:

[0178]

[0179] In the above proposed model, C dn represents the total operating cost of the DN day-ahead; xdn and x line,dn respectively correspond to the scheduling plan of the DN and the tie-line power transmitted to the lower-layer IEMG. The former involves the injection power, branch current, and node voltage of the DN root node, while the latter describes the optimal power exchange scheme between the DN and each IEMG based on the day-ahead scheduling strategy. G and H respectively represent the equality constraint and inequality constraint of the DN operation. Similarly, C iemg represents the total operating cost of the IEMG; x iemg and x line,iemg respectively correspond to the scheduling plan of the IEMG and the tie-line power transmitted to the upper-layer DN. The former mainly calculates the output power of the controllable distributed units in the IEMG (such as gas turbines, gas boilers, electric boilers, and energy storage systems, etc.), while the latter describes the optimal power exchange scheme between each IEMG and the DN in the day-ahead scheduling plan.

[0180] 2) Optimization scheduling model of the DN:

[0181] The basic premise of DN scheduling is to meet its load demand. Its goal is to formulate an optimal strategy for purchasing electricity from the main grid in advance and achieve an effective power exchange with the microgrid. The core goal of scheduling is to minimize the operating cost.

[0182] (1) Objective function of DN operation

[0183] The specific objectives of the upper-layer DN are as follows:

[0184]

[0185] In this formula, the first term represents the cost of the DN purchasing electricity from the main grid; c main,t is the trading electricity price; P main,t represents the active power injected into the root node; the second term is the network loss cost, where C loss is the network loss price, I ij,t is the branch current, is the square of the branch current, r ij is the resistance of the corresponding branch. The third term represents the revenue obtained by the DN from delivering electricity to the microgrid, where C buy,k,t and C sell,k,t are respectively the electricity purchase and sale prices from the IEMG to the DN, P DN-IEMG,k,t and P IEMG-DN,k,t are the corresponding electricity purchase and sale quantities. is the total number of nodes in the DN; T refers to the operating period; i and j are node indices; k is the identifier of the IEMG; M is the number of IEMGs in the DN.

[0186] (2) Constraint conditions

[0187] This method introduces substitution variables and slack variables to efficiently solve the optimization problem of DN operation and obtain the global optimal solution. By adjusting the variables, the original non-convex and non-linear optimization problem is transformed into a second-order cone optimization problem. The operation constraints of DN after introducing substitution variables and slack variables are as follows:

[0188]

[0189] Among them, P load,i,t and respectively represent the active power consumption and reactive power consumption of the node at the moment. g ij and b ij respectively represent the conductance and susceptance between the node and the node. is the reactive power injected into the root node. U i,max and U i,min are respectively the minimum and maximum voltage values allowed for the node. I ij,max is the maximum current allowed for the branch connecting the node and the node. U i,t represents the voltage value of the node at the moment; θ ij,t represents the phase angle difference between the node and the node at the moment. Z ij,t and Y ij,t respectively represent the impedance resistance and conductance of the branch at the moment; X i,t and X j,t respectively represent the reactance of the node and the node at the moment.

[0190] Slack variables are introduced into Equation (11), and it is successfully transformed into the second-order cone form to meet the requirements of second-order cone optimization, as shown in Equation (12):

[0191]

[0192] 3) Optimization scheduling model of IEMG:

[0193] (1) Objective function

[0194] The optimization objective of the integrated energy microgrid group is to minimize the total operation cost, including the energy purchase cost, energy abandonment cost, unit operation cost, unit start-stop cost, distribution network interaction cost and demand response compensation cost of the integrated energy microgrid.

[0195]

[0196] Among them, C gas,k,t represents the amount of natural gas purchased by the kth IEMG; F gas,GT,k,t and F gas,GB,k,t respectively represent the natural gas consumption of the GT and the gas boiler (GB); η ab,WT and η ab,PVThey are the cost coefficients of WT and PV curtailment; β WT , β PV , β EB , β GT and β GB represent the operating cost coefficients of WT, PV, electric boiler (EB), GB, and GT respectively; α WT , α PV , α EB , α GT and α GB are the operating states of WT, PV, EB, GT, and GB respectively, as binary coefficients, where 1 represents operation and 0 represents shutdown. C IL,h,k,t and C IL,e,k,t are the compensation cost coefficients for heat load interruption and power load interruption respectively. k is used to identify IEMG, and M is the total number of IEMGs in DN; t represents the time interval.

[0197] (2) Constraints

[0198] Power balance constraint:

[0199]

[0200] Among them, P DN-IEMG, k ,t represents the electricity sold by DN to IEMG. P IEMG-DN, k ,t represents the electricity sold by IEMG to DN. P load,e,k,t and P load,h,k,t represent the electrical load and heat load of the k-th IEMG at the moment respectively.

[0201] Generator set constraint:

[0202] The controllable units that can actively supply energy in IEMG are collectively referred to as generator sets, including GT and GB. GT generates electricity by burning gas fuel, while GB mainly generates heat by burning natural gas. They mainly follow conditions such as output upper and lower limits and ramp constraints:

[0203]

[0204] Among them, gas-unit is the generalized expression of the generator set and the generalized representation of the output power type of the generator set. R gas-unit,down and R gas-unit,up represent the lower and upper limits of the ramp power of the generator set respectively. α gas-unit,off-on,k,t represents the start-stop state of the generator set.

[0205] Multi-energy coupling unit constraint:

[0206] The controllable units in IEMG that can convert thermal energy and electrical energy are collectively referred to as multi - energy coupling units, including waste heat boilers (WHB) and EB. EB consumes electrical energy to generate thermal energy, realizing the energy conversion between electrical energy and thermal energy; WHB recovers the waste heat generated during the operation of GT and generates high - quality thermal energy. The unified expression of the constraints of the multi - energy coupling unit is as follows:

[0207]

[0208] Among them, tran - unit is the general expression of the multi - energy coupling unit. And respectively represent the input power and output power of the energy conversion unit.

[0209] Constraints of the energy storage unit:

[0210] The energy storage units of the micro - grid mainly include battery storage (BS), heat storage tanks (HST) and ammonia storage tanks (AST). Its operation constraints mainly include charge - discharge energy constraints, capacity continuity constraints, upper and lower limit constraints, initial and final energy balance constraints, etc., as follows:

[0211]

[0212] Among them, storge is the general expression of the energy storage unit. α storge,ch,k,t And α storge,dis,k,t respectively represent the charging and discharging efficiencies of the energy storage unit. η storge,ch And η storge,dis respectively represent the charging and discharging states of the energy storage device, as binary coefficients, where 1 represents operation and 0 represents shutdown; E storge,t represents the energy storage capacity of the energy storage unit.

[0213] Constraints on tie - line power:

[0214]

[0215] Step 3: Model solution

[0216] The solution process of the constructed DN - IEMG coordinated operation model is essentially a two - layer optimization iteration process. The outer layer uses the ATC method to obtain the interaction power value between DN and IEMG, while the inner layer uses the C&CG method to solve the stochastic distribution robust optimization problem of IEMG. The specific applications of these two methods will be elaborated in detail below.

[0217] Step 3.1: Stochastic distribution robust optimization model of IEMG;

[0218] Objective function:

[0219] For the convenience of subsequent analysis of the model, the key variables in the deterministic optimal scheduling model in Step 1 are divided into first-stage variables and second-stage variables. In the first stage, the day-ahead stage, the best daily operation strategy is formulated to minimize the total cost of IEMG operation. The first stage consists of decision variables such as GT, GB, and energy storage devices, and these variables are not affected by the uncertainty of RE in day-ahead decisions. On the contrary, the second-stage variable is denoted by q, which can be adjusted according to the actual output of RE. The variables in these two stages are both represented in matrix form:

[0220]

[0221] In the formula, s represents the harsh operation scenario of wind and light; P S is the probability of the occurrence of the harsh scenario of wind and light. The occurrence probability can be used as a reference to estimate the extreme scenario of wind and light. The historical occurrence probability of a certain type of extreme disaster scenario can be approximated as its future occurrence probability. Ψ Π is the expected value; Π is the probability distribution of the wind and light operation state; Υ is the fuzzy set of the wind and light operation state; is the additional operation cost of the system under the extreme scenario s.

[0222] Fuzzy set of wind and light fluctuations:

[0223] The moment information of the wind and light output under the extreme scenario s is used to construct the fuzzy set of wind and light fluctuations. The probability distribution of the wind and light output under any harsh scenario is described by the fuzzy set, which is specifically expressed as follows:

[0224]

[0225] In the formula, represents the wind and light operation state under scenario s; R represents all operation scenarios of wind and light; P(R) is the probability distribution of all operation scenarios of wind and light; U is the uncertainty set of wind and light.

[0226] Transformation and solution of the stochastic distributionally robust optimization model:

[0227] The transformation of the upper bound sup problem of the objective function (19) is the key to model solution. According to the definition of the wind and light fuzzy set, the upper bound problem is formulated as a semi-infinite optimization problem:

[0228]

[0229] In the formula, is the maximum probability of the harsh scenario of wind and light; dΠ is the probability density function; ψ s and Υ s,wt-pv,tis the dual variable; by using the strong duality theory, formula (21) is transformed from a semi-infinite optimization problem into a finite optimization problem:

[0230]

[0231] Formula (23) can be further rewritten as the expression of the worst wind and light scenarios:

[0232]

[0233] Substituting formulas (22) and (24) into the objective function (13) of IEMG, the equivalent form of formula (13) can be obtained:

[0234]

[0235] A three-layer optimization problem in the form of min-max-min:

[0236] The converted integrated energy stochastic distribution robust optimization model is a three-layer optimization problem in the form of min-max-min. To facilitate the description of the solution process of this model, Cai Yutong expresses it in matrix / vector form:

[0237]

[0238] s.t. Ap ≤ b (27)

[0239]

[0240] Bp + Cq s + Du s ≤ d (29)

[0241] where p is the necklace form of the first-stage decision variable, including the normal operation variables of GT, GB, and energy storage devices; q s is the vector form of the second-stage decision variable; ψ s and Υ s are the vector expression forms of the dual variables; u s is the vector form of the operation state of renewable energy; a, b, c, d, and ρ s are the vector expression forms of constant coefficients; A, B, C, and D are matrices of constant coefficients.

[0242] Formula (27) is the normal operation constraint of the unit in the first stage; formula (29) is the operation constraint in the second stage.

[0243] The three - layer models (26)-(29) can be solved by using the Benders decomposition or the C&CG algorithm. Given that the C&CG algorithm has better convergence, the C&CG algorithm is selected for the solution operation. The C&CG algorithm first decomposes the original problem into a master problem and a sub - problem, and solves it iteratively. Since the first - stage decision of the stochastic distribution - robust model is formulated based on the line - fault situations under N s scenarios, the C&CG algorithm feeds back N s wind - solar operation scenarios to the master problem in each iteration process. After decomposition, the master problem is as follows:

[0244]

[0245] s.t. Ap ≤ b (31)

[0246]

[0247] where j represents the number of times of returning between the sub - problem and the master problem; J represents the number of iterations, q s,j represents the second - stage decision variables added to the master problem by the sub - problem, represents the renewable - energy operation scenarios added when the j - th sub - problem returns to the master problem.

[0248] In each iteration process, the C&CG algorithm needs to solve N S sub - problems, and each sub - problem corresponds to a wind - solar operation scenario. In addition to the first - stage decision variable p, the dual variable Υ s obtained from solving the master problem also needs to be passed to the sub - problem. After decomposition, the sub - problem is as follows:

[0249]

[0250] where p* and are the results obtained by solving the master problem in the J - th iteration, and V s is the dual variable of formula (35).

[0251] The above - mentioned sub - problem belongs to a two - layer optimization problem. With the help of the strong duality theory, the inner - layer min problem is transformed into a max problem, and then a single - layer optimization problem is obtained, which is specifically expressed as follows:

[0252]

[0253] C T V s ≤ c (37)

[0254] Step 3.2: Solve the overall model based on ATC;

[0255] Research on the optimization coordination problem between DN and IEMG in a two - layer system. The ATC method is used to handle this optimization problem to achieve the parallel solution of DN and IEMG. Between DN and IEMG, the tie - line power is set to P DN,load,k,t . In DN, the tie - line power is regarded as a virtual load P DN-IEMG,k,t ; meanwhile, in IEMG, P IEMG-DN,k,t is regarded as the output of a virtual generator.

[0256] Solving the overall scheduling model of DN - IEMG using the ATC method involves the following steps:

[0257] Step a: Input the relevant parameters of the distribution network and the integrated energy micro - grid. Set the initial value for the penalty function multiplier

[0258] Step b: As shown in formula (38), add the tie - line penalty factor to the objective function of the integrated energy micro - grid. Solve the data - driven stochastic distributed robust scheduling problem of each micro - grid, and then transfer the obtained P IEMG-DN,k,t to the upper - layer distribution network.

[0259]

[0260] Step c: During the DN scheduling process, as shown in formula (39), introduce the tie - line penalty function K DN into the DN objective function. Solve the DN scheduling problem with relevant constraints and transfer the obtained virtual load P DN-IEMG,k to the IEMG layer.

[0261]

[0262] Step d: Conduct multiple optimization iterations on DN and IEMG until the convergence criterion is met.

[0263] The solution steps of the overall model in step 3 are as Figure 4 shown.

[0264] The above - mentioned is only the preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.​

Claims

1. A random distributed robust optimization scheduling method for distribution networks of integrated energy microgrids, characterized in that: The following steps are involved: Step 1: Construct a DN structure with multiple microgrids; DN contains multiple IEMGs, and each IEMG and DN realize power interaction through connecting lines. By strategically arranging the interactive power with DN, planning the output of distributed energy, and managing the output of different devices, IEMG can reduce the overall operating cost while meeting various load demands. DN is used to manage the coordination capabilities of each IEMG, strategically plan the allocation of decentralized power generation capacity, and optimize the purchase of electricity from the main power grid, so as to minimize the operating cost while meeting the power load demand. Step 2: Hierarchical energy optimization management; The day-ahead dispatch optimization model of DN with IEMGs focuses on operating costs and constraints. For DN, the model aims to solve a non-convex nonlinear optimization problem with line losses. For IEMG, the model is used to deal with a convex optimization problem with uncertainty in WT and PV output power. Power is exchanged between DN and IEMG through connecting lines, so a two-layer model framework is constructed for analysis. Step 3: Model solution; The solution process of the constructed DN-IEMG collaborative operation model is a two-layer optimization iterative process. The outer layer uses the ATC method to obtain the interaction power value between DN and IEMG, while the inner layer uses the C&CG method to solve the random distribution robust optimization problem of IEMG.

2. The random distributed robust optimization scheduling method for distribution network of integrated energy microgrid group according to claim 1 is characterized in that: In the step 1, wherein the IEMG including the P2A technology has RE, P2A modules, thermal power energy supply and conversion equipment, and integrated demand response; The device model constructed is as follows: 1) Operation model of P2A module: The technical route of green ammonia includes water electrolysis, cryogenic air separation and Haber-Bosch thermochemical synthesis. The electrochemical production of ammonia is divided into the following three stages: Phase 1: RE provides electrical energy to EL. The electrolyzer uses water as raw material for electrolysis to produce hydrogen and oxygen. The chemical reaction formula in EL is shown in formula (1); formula (2) describes the operation model of EL: 0≤P EL,e,t ≤P EL,e,max R EL,down P EL,e,max ≤P EL,e,t -P EL,e,t-1 ≤R EL,up P EL,e,max (2) In the formula, is the hydrogen production rate of EL in the time period; η EL represents the hydrogen production efficiency of EL; P EL,e,t Indicates the power consumption of the electrolytic cell during the period; R EL,down and R EL,up denote the power up and power down constraints of EL respectively; Phase II: ASU uses pressure swing adsorption technology to extract nitrogen from the air; the ammonia production in this model can be adjusted according to the hydrogen production rate of EL, and the power consumption of the pressure swing adsorption nitrogen production unit is proportional to the hydrogen production rate of EL; the power consumption and operation constraints of ASU are as follows: Among them, P ASU,e,t Indicates the power consumption of ASU in the time period; P ASU,e,r It indicates the power consumption of ASU per unit volume or mass of nitrogen produced; represents the amount of nitrogen produced by ASU during the period; P ASU,e,max Indicates the maximum power consumption of ASU; The third stage: The ammonia synthesis unit uses the Haber-Bosch process to synthesize ammonia from hydrogen and nitrogen, as shown in the chemical reaction formula (4). In addition, in order to maintain the pressure in the ammonia synthesis unit at 300 bar, some electrical energy is required to drive the internal air compressor to increase the pressure. The energy consumption of the ammonia production unit is related to the ammonia synthesis rate, as shown in formula (5): Where ΔH represents the waste heat generated by synthesizing one mole of ammonia; Among them, P Haber，e,t Indicates the power consumption during the ammonia synthesis process during the period; P Haber,e,r It indicates the power consumption per unit of ammonia produced; represents the ammonia production rate during the time period; Indicates the low calorific value of ammonia; 2) GT's ammonia co-combustion model: For GTs modified with ammonia co-firing technology, ammonia will replace a certain proportion of the natural gas entering the GT. The actual gas consumption of the GT is given by the following formula: Among them, F GT,gas,t Indicates the natural gas consumption of GT in the period; a gas , b gas and c gas is the combustion coefficient of natural gas; L gas Indicates the calorific value of natural gas; is the ammonia mixing ratio; P WHB,GT,t represents the waste heat power released by GT in period t; P GT,e,t represents the electric power output of GT in time period t; η GT Represents the operating efficiency of GT; η loss,GT is the heat loss coefficient of GT; In the model, the upper limit of the ammonia mixing ratio is set at 20%; 3) IDR Model IEMG adopts incentive mechanisms, including time-of-use market pricing signals or economic incentives; the demand response constraints of IEMG power loads are expressed as follows: P IL,e,min ≤P IL,e,t ≤P IL,e,max Among them, P IL,e,max and P IL,e,min They represent the maximum and minimum values ​​of the interruptible power load respectively; The demand response constraint for the IEMG heating load is expressed as follows: P IL,h,min ≤P IL,h,t ≤P IL,h,max Among them, P IL,h,max and P IL,h,min Respectively represent the maximum and minimum values ​​of the interruptible heat load.

3. The random distributed robust optimization scheduling method for distribution network of integrated energy microgrid group according to claim 2 is characterized in that: In step 2: 1) The overall structure of the model is as follows: In the above model, C dn represents the total operating cost of DN before the current day; x dn and x line,dn They correspond to the dispatching plan of DN and the tie line power transmitted to the lower IEMG respectively; the former involves the injected power, branch current and node voltage of the DN root node, while the latter describes the optimal power exchange scheme between DN and each IEMG based on the day-ahead dispatching strategy; G and H represent the equality constraints and inequality constraints of DN operation respectively; C iemg represents the total operating cost of IEMG; x iemg and x line,iemg They correspond to the dispatch plan of IEMG and the power of the tie line transmitted to the upper DN respectively; the former mainly calculates the output power of the controllable distributed units in the IEMG, while the latter describes the optimal power exchange scheme between each IEMG and DN in the day-ahead dispatch plan; 2) DN optimization scheduling model: The premise of DN dispatch is to meet its load demand, and its goal is to formulate the optimal strategy for purchasing electricity from the main grid in advance and realize effective power exchange with the microgrid; the core goal of dispatch is to minimize operating costs; (1) Objective function of DN operation The specific goals of the upper DN are as follows: In this formula, the first term represents the cost of DN purchasing electricity from the main grid; c main,t is the transaction electricity price; P main,t represents the active power injected into the root node; the second term is the network loss cost, where C loss is the network loss price, I ij,t is the branch current, is the square of the branch current, r ij is the resistance of the corresponding branch; the third term represents the benefit obtained by DN by delivering power to the microgrid, where C buy,k,t and C sell,k,t are the electricity purchase and sales prices from IEMG to DN; P DN-IEMG,k,t and P IEMG-DN,k,t is the corresponding amount of electricity purchased and sold; is the total number of nodes in the DN; T refers to the runtime period; i and j are node indices; k is the identifier of the IEMG; M is the number of IEMGs in the DN; (2) Constraints Substitution variables and slack variables are introduced. By adjusting the variables, the original non-convex and nonlinear optimization problem is transformed into a second-order cone optimization problem. After the substitution variables and slack variables are introduced, the operating constraints of DN are as follows: Among them, P load,i,t and They represent the active power consumption and reactive power consumption of the node at the time; g ij and b ij Respectively represent the conductance and susceptance between nodes; is the reactive power injected into the root node; U i,max and U i,min are the minimum and maximum voltage values ​​allowed by the node; I ij,max is the maximum current allowed by the branch connecting the node and the node; U i,t Represents the voltage value of the node at the time; θ ij,t Represents the phase angle difference between nodes at time; Z ij,t and Y ij,t They represent the impedance resistance and conductance of the branch at the time; X i,t and X j,t denote the reactance of the node and the node at time respectively; Slack variables are introduced into equation (11) and converted into a second-order cone form to meet the requirements of second-order cone optimization, as shown in equation (12): 3) IEMG’s optimized scheduling model: (1) Objective function The optimization goal of IEMGs is to minimize the total operating cost, including the IEMG's energy purchase cost, energy abandonment cost, unit operation cost, unit start-up and shutdown cost, distribution network interaction cost, and demand response compensation cost; Among them, C gas,k,t represents the amount of natural gas purchased by the kth IEMG; F gas,GT,k,t and F gas,GB,k,t Respectively represent the natural gas consumption of GT and GB; η ab,WT and η ab,PV are the cost coefficients of WT and PV energy abandonment, respectively; β WT , β PV , β EB , β GT and β GB Respectively represent the operating cost coefficients of WT, PV, EB, GB and GT; α WT , α PV , α EB , α GT and α GB are the operating states of WT, PV, EB, GT and GB, respectively, as binary coefficients, where 1 indicates operation and 0 indicates shutdown; C IL,h,k,t and C IL,e,k,t are the compensation cost coefficients for thermal load interruption and electrical load interruption, respectively; k is used to identify IEMG, M is the total number of IEMGs in DN; t represents the time interval; (2) Constraints Power balance constraints: Among them, P DN-IEMG,k,t represents the amount of electricity sold by DN to IEMG; P IEMG-DN,k,t represents the amount of electricity sold by IEMG to DN; P load,e,k,t and P load,h,k,t They represent the electrical load and thermal load of the kth IEMG at time instant; Generator set constraints: The generator sets in IEMG include GT and GB; they follow the output upper and lower limits and ramp constraints: Among them, gas-unit is a generalized expression of the generator set, which is a generalized expression of the output power type of the generator set; R gas-unit,down and R gas-unit,up Respectively represent the lower and upper limits of the generator set climbing power; α gas-unit,off-on,k,t Indicates the start and stop status of the generator set; Multi-energy coupling unit constraints: The multi-energy coupling unit in IEMG includes WHB and EB, and the unified expression of the multi-energy coupling unit constraint is as follows: Among them, tran-unit is the generalized expression of the multi-energy coupling unit; and represent the input power and output power of the energy conversion unit respectively; Energy storage unit constraints: The energy storage units of the microgrid include BS, HST and AST. Their operation constraints include charging and discharging energy constraints, capacity continuity constraints, upper and lower limit constraints, and initial and final energy balance constraints, as follows: Where storge is a generalized expression of the energy storage unit; α storge,ch,k,t and α storge,dis,k,t Respectively represent the charging and discharging efficiency of the energy storage unit; η storge,ch and η storge,dis They represent the charging and discharging states of the energy storage device, respectively, as binary coefficients, where 1 indicates operation and 0 indicates shutdown; E storge,t Indicates the energy storage capacity of the energy storage unit; Tie line power constraints:

4. The random distributed robust optimization scheduling method for distribution network of integrated energy microgrid group according to claim 3 is characterized in that: The step 3 comprises the following specific steps: Step 3.1: IEMG stochastic distribution robust optimization model; Objective function: The key variables in the deterministic optimization scheduling model in step 1 are divided into first-stage variables and second-stage variables; in the first stage, the optimal daily operation strategy is formulated to minimize the total cost of IEMG operation; the first stage consists of decision variables of GT, GB and energy storage equipment, which are not affected by RE uncertainty in day-ahead decision-making; The second stage variable is denoted by q, which can be adjusted according to the actual output of RE; the variables of these two stages are expressed in matrix form: Where, s represents the operation scenario with bad wind and solar conditions; P S is the probability of occurrence of severe wind and solar disaster scenarios, which can be used as a reference to estimate extreme wind and solar disaster scenarios; the historical probability of a certain type of extreme disaster scenario is approximately its future probability of occurrence; Π is the expected value; Π is the probability distribution of wind and solar operation status; Υ is the fuzzy set of wind and solar operation status; is the additional operating cost of the system under extreme scenario s; Scenery fluctuation fuzzy set: The moment information of wind-solar output under extreme scenario s is used to construct wind-solar fluctuation fuzzy sets, and the probability distribution of wind-solar output under any severe scenario is described by fuzzy sets, which is specifically expressed as follows: In the formula, represents the wind and solar operation status under scenario s; R represents all wind and solar operation scenarios; P(R) is the probability distribution of all wind and solar operation scenarios; U is the uncertainty set of wind and solar; Conversion and solution of stochastic distributed robust optimization model: According to the definition of scenery fuzzy set, the upper bound problem of objective function (19) is expressed as a semi-infinite optimization problem: In the formula, is the maximum probability of a bad scene; dΠ is the probability density function; ψ s With Υ s,wt-pv,t is the dual variable; using the strong duality theory, formula (21) is transformed from a semi-infinite optimization problem to a finite optimization problem: Formula (23) can be further rewritten as the expression for the worst scene: Substituting formulas (22) and (24) into the objective function (13) of IEMG, we obtain the equivalent form of formula (13): Three-level optimization problem of min-max-min form: The converted comprehensive energy stochastic distribution blue bar optimization model is a min-max-min three-level optimization problem; it is expressed in the matrix / vector form of Cai Yutong: stAp≤b (27) Bp+Cq s +You s ≤d (29) Where p is the necklace form of the first-stage decision variables, including the normal operation variables of GT, GB, and energy storage equipment; q s is the vector form of the second-stage decision variable; ψ s With Υ s is the vector expression of the dual variable; u s is the vector form of the operating state of renewable energy; a, b, c, d and ρ s is a constant coefficient vector expression; A, B, C and D are matrices of constant coefficients; Formula (27) is the normal operation constraint of the unit in the first stage; Formula (29) is the operation constraint in the second stage; The three-layer model (26)-(29) uses the C&CG algorithm to solve the problem. First, the original problem is decomposed into a main problem and sub-problems, and then solved by iteration. Since the first stage decision of the random distribution robust model is based on N s Therefore, the C&CG algorithm must feedback N to the main problem in each iteration process. s There are several wind and solar operation scenarios. After decomposition, the main problems are as follows: stAp≤b (31) In the formula, j represents the number of returns between the sub-problem and the main problem; J represents the number of iterations, q s,j represents the second-stage decision variable added by the subproblem to the main problem, It means that the j-th sub-problem returns to the main problem with the added renewable energy operation scenario; In each iteration, the C&CG algorithm needs to S sub-problems are solved, and each sub-problem corresponds to the wind and solar operation scenario; in addition to the decision variable p in the first stage, the dual variable Υ obtained by solving the main problem s The same needs to be passed to the sub-problem, and the decomposed sub-problem is as follows: Where p * and is the result of solving the main problem in the Jth iteration, V s is the dual variable of formula (35); The subproblem mentioned belongs to a two-layer optimization problem. With the help of strong duality theory, the inner min problem is transformed into a max problem, and then a single-layer optimization problem is obtained, which is specifically expressed as follows: C T In s ≤c (37) Step 3.2: Solve the overall model based on ATC; Aiming at the optimization coordination problem between DN and IEMG in the two-layer system, the ATC method is used to deal with the optimization problem. Between DN and IEMG, the tie line power is set to P DN,load,k,t ; In DN, the tie line power is regarded as a virtual load P DN-IEMG,k,t ; In IEMG, P IEMG-DN,k,t is considered as the output of a virtual generator; Solving the DN-IEMG overall scheduling model using the ATC method involves the following steps: Step a: Input the relevant parameters of the distribution network and the integrated energy microgrid, which is the penalty function multiplier Setting initial values Step b: Add the tie line penalty factor to the objective function of the integrated energy microgrid as shown in formula (38); solve the data-driven random distributed robust scheduling problem for each microgrid, and then use the obtained P IEMG-DN,k,t Transmit to upper distribution network; Step c: In the DN scheduling process, according to formula (39), the tie line penalty function K DN Introduce the DN objective function, solve the DN scheduling problem with relevant constraints, and transfer the obtained virtual load to P DN-IEMG,k To IEMG layer; minC DN +K DN Step d: Perform multiple optimization iterations on DN and IEMG until the convergence criteria are met.

Images