CN117291315B

CN117291315B - Carbon recycling electric-gas-thermal multi-energy combined supply network cooperative operation method

Info

Publication number: CN117291315B
Application number: CN202311576668.6A
Authority: CN
Inventors: 周斌; 陈煜�; 郑玲; 陈亚鹏; 曾琬胭; 吴雨恒; 颜鑫
Original assignee: Hunan University
Current assignee: Hunan University
Priority date: 2023-11-24
Filing date: 2023-11-24
Publication date: 2024-02-20
Anticipated expiration: 2043-11-24
Also published as: CN117291315A

Abstract

The carbon recycling electricity-gas-heat multi-energy combined supply network cooperative operation method comprises the following steps of S1, constructing an electricity-gas-heat multi-energy combined supply network heterogeneous topology model under high-proportion distributed photovoltaic access; s2, identifying a heterogeneous topology model of the multi-energy co-generation network by adopting a heterogeneous structure multi-task learning method; s3, constructing a carbon recycling model of linkage of the electric conversion gas, the oxygen-enriched fuel gas and the carbon dioxide capturing device; s4, constructing a cost efficiency model of the multi-energy co-supply system considering carbon recycling; and S5, constructing a multi-energy cooperative regulation optimization model aiming at minimizing the operation cost so as to realize the zero-carbon self-circulation multi-energy cooperative operation of the distributed energy supply network. The method provided by the invention can improve the accuracy of the multi-energy combined supply network topology identification, improve the photovoltaic digestion capability and reduce the carbon emission.

Description

Carbon recycling electric-gas-thermal multi-energy combined supply network cooperative operation method

Technical Field

The invention relates to the technical field of distributed energy supply systems, in particular to an electric-gas-heat multi-energy combined supply network collaborative operation method for carbon recycling.

Background

The wide access of distributed photovoltaics brings great challenges to the operation regulation and control of the multi-energy co-generation network. On the one hand, the uncertain characteristics of the distributed photovoltaic and the complex inherent characteristics such as strong coupling and spatial correlation of the multi-energy co-supply network lead to the change of the multi-energy flow, the multi-energy co-supply network topology identification is difficult, and the wrong topology file brings a plurality of inconveniences to the optimal configuration and the collaborative operation of the multi-energy co-supply network, and influences the interconnection and mutual utilization of the multi-energy co-supply network, so that the development of the multi-energy co-supply network topology identification research suitable for the high-proportion distributed photovoltaic access is imperative. On the other hand, the distributed photovoltaic has the inverse peak regulation characteristic, and the light rejection phenomenon occurs sometimes, so that the research on the cooperative operation method of the multi-energy co-supply network has important significance in realizing multi-energy co-ordination planning, and the mutual conversion and mutual complementation of different energy sources fully excavate the mutual potential of interconnection among the multi-energy sources, improve the photovoltaic digestion capacity of the network, maximally utilize solar energy resources, further improve the carbon emission structure, reduce the carbon emission of the multi-energy co-supply network and achieve the effects of energy conservation and emission reduction.

At present, researches on electric-gas-heat multi-energy combined supply network topology identification only focus on an electric power network, and an intelligent ammeter deployed in a distribution area is utilized to measure internal relations of a data mining topological structure, wherein the internal relations comprise a model fitting method, a feature learning method and a correlation analysis method 3-type method, and in addition, an artificial intelligence technology is widely applied to the field of topology identification, such as a Bayesian network, a Markov random field, a knowledge graph technology and the like. However, the above research does not relate to a multi-energy co-generation network with various energy coupling and conversion characteristics, and does not consider the influence of randomness and fluctuation of high-proportion distributed photovoltaic access on multi-energy flow. The multi-energy co-supplied network collaborative optimization operation is also paid attention to by a plurality of students, the low-carbon economic operation of the multi-energy co-supplied network is realized through the optimal configuration of multi-type energy sources, wherein the emission reduction of network carbon mainly depends on a carbon capture technology, and the multi-energy co-supplied network can be divided into a pre-combustion capture technology, a post-combustion capture technology and an oxygen-enriched combustion decarburization technology, the latter has the best emission reduction effect, but an air separation device for manufacturing oxygen has high energy consumption, and the oxygen-enriched condition forms a great challenge for practical application. It is needed to study an electric-gas-heat multi-energy co-supply network collaborative operation method for carbon recycling, and improve new energy consumption and carbon emission reduction capability of a system on the basis of accurately identifying a multi-energy co-supply network topological structure.

Disclosure of Invention

The technical problem to be solved by the invention is to provide the carbon recycling electric-gas-heat multi-energy co-supply network collaborative operation method, which can improve the accuracy of multi-energy co-supply network topology identification, improve the photovoltaic digestion capability and reduce the carbon emission.

In order to solve the technical problems, the invention adopts the following technical methods: a carbon recycling electricity-gas-heat multi-energy combined supply network cooperative operation method comprises the following steps:

step S1, constructing an electric-gas-heat multi-energy combined supply network heterogeneous topology model under high-proportion distributed photovoltaic access;

s2, identifying a heterogeneous topology model of the multi-energy co-generation network by adopting a heterogeneous structure multi-task learning method;

s3, constructing a carbon recycling model of linkage of the electric conversion gas, the oxygen-enriched fuel gas and the carbon dioxide capturing device;

s4, constructing a cost efficiency model of the multi-energy co-supply system considering carbon recycling;

and S5, constructing a multi-energy cooperative regulation optimization model aiming at minimizing the operation cost so as to realize the zero-carbon self-circulation multi-energy cooperative operation of the distributed energy supply network.

Further, the step S1 includes:

s101, regarding various devices in a power distribution network, a gas supply network and a heat supply network as nodes in a topological graph, wherein the nodes are divided into four types of source nodes, storage nodes, load nodes and coupling nodes, and the nodes under the types of the nodes are respectively:

1) Source node: a power source, a PV, a gas source;

2) Storage node: the device comprises an electricity storage device, a gas storage device and a heat storage device;

3) Load node: electrical, gas, and thermal loads;

4) Coupling node: CHP unit, electric boiler, gas boiler, P2G;

s102, regarding the circuit of the connection node as an edge in the topological graph, wherein the edge adopts the following adjacency matrixTo describe:

(1)

s103, forming 3 two-dimensional adjacency matrixes for describing topological structures of a power distribution network, a gas supply network and a heat supply network respectively by using the determined nodes and side information, constructing a multi-dimensional adjacency matrix by using the coupling characteristics of the coupling nodes as hinges among different energy networks, and establishing an electric-gas-heat energy combined supply network heterogeneous topological model under high-proportion distributed photovoltaic access.

Further, in the step S1, in the process of constructing the heterogeneous topology model of the electric-gas-heat multi-energy combined supply network under the high-proportion distributed photovoltaic access, the influence of the high-proportion distributed photovoltaic access on the electric-gas-heat multi-energy combined supply network multi-energy flow is considered, and the topology is simplified;

the model of the electric-gas-heat multi-energy combined supply network multi-energy flow is as follows in the formula (2) -formula (4):

(2)

an alternating current power flow model of a 1 st-2 nd action power distribution network of (2), wherein And->Nodes +.>Active power and reactive power, +.>、/>Nodes +.>、/>Voltage amplitude of>、/>、/>Nodes +.>、/>Conductance, susceptance, phase angle difference between each other;

a 3 rd behavioral air supply network energy flow model of (2), whereinFor node->The volume flow is converted into the power value of the power flow, +.>Is a node to node of the air sourceiFlow of injection->For node->Neighbor node set,/->For the characteristic parameters of the pipeline>Is +.>、/>Constant coefficient related to inter-pipeline parameters, +.>、/>Nodes +.>、/>Is air pressure of->For node->Load flow of>Is the heat value of natural gas;

the 4 th-7 th behavioural heating network energy flow model of formula (2), whereinIs the specific heat capacity of the thermal mass>For node branch association matrix, < >>For the thermal flow rate in the pipeline, < > is->、/>The node water supply temperature and the thermal mass outlet temperature, respectively, < >>Load thermal power for node +.>For the pipeline loop correlation matrix, < > is provided>Is the impedance coefficient of the pipeline>、/>The water supply network and the backwater network coefficient matrixes are respectively +.>、/>Respectively representing the temperature obtained by subtracting the ambient temperature from the water supply temperature and the backwater temperature;

(3)

the formula (3) is a coupling link model, wherein the 1 st-2 nd behavior CHP model is CHP, and CHP is cogeneration; lines 3-5 are respectively an electric boiler model, a gas boiler model and a P2G model, wherein P2G is electric conversion gas; 、/>、/>、/>、/>The output thermal power, the thermoelectric ratio, the electric power, the natural gas input flow and the efficiency coefficient of the CHP unit are respectively>、/>The lower limit and the upper limit of the heat generation power of the CHP unit are respectively set; />、/>The lower limit and the upper limit of the power generated by the CHP unit are respectively set; />、/>、/>The electric heating conversion efficiency, the output heat power and the consumption power of the electric boiler are respectively +.>An upper limit of output thermal power for the electric boiler;、/>、/>、/>the natural gas flow consumption, the heat output, the heat value utilization rate and the natural gas low heat value of the gas boiler are respectively +.>An upper limit of output heat power of the gas boiler; />、/>、/>Natural gas flow, consumed electric power, conversion efficiency, respectively P2G produced,/->An upper limit of the output of the P2G equipment;

(4)

the PV node is (4)Wherein>For node->Net injection power, < >>For node->Photovoltaic output of>For node->Total power to P2G node, +.>For node->Total power to electrical load node, +.>For node->Store power store total power, +.>For node->The electricity storage releases the total power;

the simplification process includes:

1) Deleting branch nodes and redundant measuring nodes, and only reserving a source node, a storage node, a load node and a coupling node;

2) Combining various redundant connection lines, and reserving only one connection line between node pairs;

3) The numbers and types of the node devices are defined, so that the judgment of the node types is not needed during topology identification, and only the node connection relation is needed to be identified.

Still further, the step S2 includes:

s201, constructing a heterostructure multi-task learning model, wherein the heterostructure multi-task learning model comprises three subtask input layers, a sharing layer, three subtask output layers and a coupling layer, and the sharing layer is a heterogeneous network convolution layer formed by combining a plurality of heterogeneous network convolution neurons;

s202, respectively sending power distribution network voltage data, air supply network air pressure data and heat supply network water pressure data in a heterogeneous topology model of a multi-energy combined supply network into three subtask input layers of a heterogeneous structure multi-task learning model, setting time windows for the three subtask input layers to intercept equal-length time sequences, and then carrying out normalization processing on the obtained data to respectively generate node feature matrixes of the power distribution network, the air supply network and the heat supply network;

s203, the node characteristic matrixes of the power distribution network, the air supply network and the heat supply network are sent into a sharing layer together for iterative processing, and each processing process is as follows: firstly, obtaining hidden feature matrixes of a power distribution network, a gas supply network and a heat supply network through operation of heterogeneous network convolution neurons, then carrying out inner product operation on node feature vectors in the hidden feature matrixes to obtain prediction scores representing association degrees among nodes, and then optimizing prediction scoring results by utilizing cross entropy loss to obtain two-dimensional adjacent matrixes of the power distribution network, the gas supply network and the heat supply network;

S204, coupling the two-dimensional adjacent matrixes of the power distribution network, the air supply network and the heat supply network, which are finally obtained after the iterative processing of S203, by a coupling layer to form a multi-dimensional adjacent matrix of the multi-energy combined supply network;

the heterogeneous network convolution neuron calculates a hidden feature matrix by adopting the following formula (5):

(5)

in the method, in the process of the invention,is->Hidden feature matrix of convolutional neuron output of heterogeneous network, < ->Wherein->Is thatDimension node feature matrix, < >>And->The number of network nodes and the dimension of the network node feature vector are respectively; />Is an activation function; />Normalized Laplacian matrix for network symmetry, +.>Wherein->For adding the network adjacency matrix after the self-edge connection, < >>，/>For the network adjacency matrix>Is->Identity matrix of>Is a network degree matrix; />Is->Trainable weight matrix of convolutional neurons of heterogeneous network;

the heterogeneous network convolution neuron optimizes the prediction scoring result by adopting the following cross entropy loss function:

(6)

in the method, in the process of the invention,is one of networksThe total loss function value of the nodes is provided; />Is->The true value of the individual node; />Is->Predicted values of the individual nodes; />The method is the number of nodes in the multi-energy combined supply network, namely the sum of the numbers of all nodes of each energy network.

Still further, step S2 is performed to perform optimization training on the heterostructure multi-task learning model by using the historical power distribution network voltage data, the air supply network air pressure data and the heat supply network water pressure data before identifying the multi-energy co-supply network heterogeneous topology model, so as to obtain an optimal heterostructure multi-task learning model.

Further, in the step S3, the constructed carbon recycling model of linkage of the electric conversion gas, the oxygen-enriched fuel gas and the carbon dioxide capturing device is represented by the following formulas (7) - (13):

(7)

formula (7) represents a methane production process in whichPower consumed for electrical conversion; />The efficiency of converting electrical energy into hydrogen; />Is the low heating value of hydrogen; />、/>、/>The volume of carbon dioxide, hydrogen and methane are required and generated respectively; />Conversion of carbon dioxide to->Is not limited by the efficiency of (2);

(8)

formula (8) represents a methane storage model in which、/>Methane tank->Net output and storage capacity at time; />、/>Methane tank->Net output and storage capacity at time; />Is a time interval; />、Respectively->Lower bound, upper bound of (2); />、/>Respectively->Lower bound, upper bound of (2);

(9)

formula (9) represents an oxygen storage model in which、/>Oxygen cylinders are respectively->Net output and storage capacity at the moment; / >、/>Oxygen cylinders are respectively->Net output and storage capacity at the moment; />、/>Respectively->Lower bound, upper bound of (2); />、/>Respectively->Lower bound, upper bound of (2);

(10)

equation (10) represents a methane-fueled cogeneration unit operating constraint whereinMethane volume consumed for cogeneration; />Is the calorific value of methane; />、/>、/>The power output is the electric active power output, the reactive power output and the thermal power output of the cogeneration respectively; />、/>Methane-electric efficiency and methane-thermal efficiency of cogeneration respectively; />Representing a correlation coefficient between reactive power and active power; />、/>Respectively the carbon dioxide emission amount of the cogeneration and the oxygen required by the total oxygen combustion of methane; />、/>Respectively representing carbon dioxide emission coefficient and oxygen consumption coefficient;

(11)

the formula (11) represents a carbon dioxide capturing device model, whereinIs the volume of carbon dioxide captured; />Is carbon dioxide trapping efficiency; />Is carbon dioxide density; />Power consumed for the capture system; />Is the power consumption coefficient;

(12)

(13)

formulas (12) - (13) represent equilibrium constraints of oxygen and methane, respectively, whereinIs the volume of oxygen generated.

Further, in step S4, the constructed multi-energy co-supply system cost efficiency model considering carbon recycling is used for calculating degradation cost of the electric conversion equipment, life degradation cost of the carbon dioxide capturing device, carbon dioxide purchasing cost, degradation cost of the energy storage system, power purchasing cost and discarding cost, and the calculation formulas of the costs are as follows:

1) Degradation cost of electric conversion gas equipment；

(14)

Wherein,、/>respectively representing the fund cost and the service life of the electrolytic tank; />Is->Bivariate of the moment on or off state; />Is->Bivariate of the moment on or off state; />、/>Respectively representing the unit degradation cost when the electrolytic cell is started and closed;

2) Carbon dioxideCost of life degradation of trapping device：

(15)

Wherein,、/>respectively representing the inherent capital cost and the service life of the carbon dioxide capturing device; />Equivalent operation and management costs per hour for the system; />The unit trapping cost; />Power consumed for the capture system;

3) Carbon dioxide purchasing cost：

(16)

Wherein,unit cost for purchasing carbon dioxide;

4) Cost of energy storage system degradation：

(17)

Wherein,is a cost factor associated with life degradation; />、/>Respectively ESSbCharging power and discharging power of (a); />、/>Respectively ESSbCharging efficiency and discharging efficiency of (a); />，/>Is an energy storage node set;

5) Cost of electricity purchasing：

(18)

Wherein,representing active power traded in the market; />、/>Respectively representing purchase prices and selling prices;is an operator, express->；

6) Cost of discarding light：

(19)

In the method, in the process of the invention,is a light discarding cost coefficient; />For uncertainty, the prediction error of the total output of the distributed photovoltaic is expressed, Wherein->、/>Respectively representing the predicted value and the actual value of the total output of the distributed photovoltaic.

Furthermore, the objective function of the multi-energy cooperative regulation optimization model is as follows:

(20)

the multi-energy collaborative regulation optimization model comprises an energy storage system operation constraint, a coupling equipment operation constraint, a distributed photovoltaic operation constraint, a power distribution network constraint and a heat supply network constraint, and the multi-energy collaborative regulation optimization model comprises the following steps of:

1) Energy storage system operation constraints

And the power of the energy storage system is adjusted by adopting a single-factor affine adjustable strategy without considering the prediction error of the distributed photovoltaic, and the following formula is adopted:

(21)

(22)

(23)

in the formulae (21) to (23),ESS for energy storage systembA power adjustment amount; />As a corresponding affine factor, the affine factor is constrained by equation (22); formula (23) ensures ESSbThe reserve provided does not exceed its capacity, wherein ∈>、/>Respectively represent ESSbLower spare capacity, upper spare capacity;

ESSbthe operating constraints of (2) are as follows:

formulas (24) - (25) are SOC constraints:

(24)

(25)

formulae (26) - (29) are ESSbCharging and discharging power constraint and upper and lower standby capacity constraint:

(26)

(27)

(28)

(29)

in the formulae (24) to (29),、/>respectively indicate->、/>Time ESSbState of charge value of (2); />、/>Respectively represent ESSbUpper and lower states of charge limits; />Representing the energy storage capacity; />、/>Respectively ESSbA charging power upper limit value and a discharging power upper limit value; / >、/>Respectively represent the charge state and the discharge state of the energy storage battery0-1 variable;

2) Coupling device operation constraints

The coupling device operation constraint comprises an output constraint and a conversion constraint of the coupling device, namely formula (3);

3) Distributed photovoltaic operation constraints

The output reactive power constraint of the distributed photovoltaic is as shown in formula (30):

(30)

wherein,indicating PV->Output reactive power, < >>、/>Respectively represent PV->Outputting an upper limit and a lower limit of reactive power; />A collection of distributed light Fu Jiedian;

4) Distribution network constraints

The constraint of the power network is represented by a linear power flow model, as shown in formulas (31) - (32):

(31)

(32)

wherein,、/>respectively represent bus bars->To bus bar->Active power flow and reactive power flow of (a); />、/>Respectively represent bus bars->To bus bar->Resistance and reactance of (a); />、/>、/>、/>Respectively represent bus bars->To bus bar->Voltage amplitude and phase angle of (a);

the branch power flow and voltage magnitude constraints are shown in equations (33) - (34):

(33)

(34)

wherein,indicating line->Maximum apparent power of (a); />Indicating busbar->Voltage amplitude of>、/>Respectively representing the upper limit and the lower limit thereof;

the active and reactive power balance of the bus is shown in formulas (35) - (36):

(35)

(36)

wherein,、/>、/>、/>、/>is 0-1 variable, which respectively indicates PV, ESS, CHP, electric boiler, electric converter is positioned on bus +. >A place; />Representing a bus set, wherein the subscript of a bus connected with a main network is 0; />、/>Respectively express and bus bar->A connected upstream and downstream line set; />、/>Respectively represent bus bars->Active and reactive loads of (a);

5) Heating network constraints

The heat supply and demand balance constraint of the heat supply network is shown as a formula (37):

(37)

wherein,is the heat load of the heating network.

Preferably, in the step S5, after the multi-energy cooperative regulation optimization model is built:

processing uncertainty of photovoltaic output by using distributed robust optimization and using Wasserstein distanceThe fuzzy set constructs a true probability distribution with high probability level, and transforms the objective function of the multi-energy collaborative regulation optimization model into a formula (38): (38)

wherein,indicating uncertainty +.>Distribution of->Indicating uncertainty +.>Is (are) fuzzy set, < >>Representing uncertainty fuzzy sets->Is the worst scene distribution in->For uncertain amount +.>At->Cost of discarding light under distributionIs a desired value of (2);

and meanwhile, linearizing the bilinear term in the formula (14) and the secondary constraint nonlinear term in the formula (33) by adopting a linearizing method, converting the multi-energy collaborative regulation and control optimization problem containing uncertain quantity into a mixed integer linear programming problem by linearizing, and solving the mixed integer linear programming problem to realize the optimal collaborative regulation and control operation of zero-carbon self-circulation of the distributed multi-energy combined supply network.

According to the carbon recycling electric-gas-heat multi-energy co-supplied network collaborative operation method provided by the invention, the influence of high-proportion distributed photovoltaic access on the dynamic distribution of electric-gas-heat multi-energy co-supplied network energy is analyzed, the multi-energy co-supplied network topology identification problem is decomposed into power distribution, gas supply and heat supply network topology identification subtasks based on heterostructure multi-task learning, heterogeneous energy flow network node association characteristics considering the operation characteristics of multi-energy coupling equipment are extracted through a heterogeneous network convolution layer by utilizing a shared learning mechanism, and a multi-dimensional topology adjacency matrix of the multi-energy co-supplied network is constructed by a multi-task regression layer. The invention also provides a carbon recycling model of linkage of the electric gas conversion device, the oxygen-enriched fuel gas and the carbon dioxide capturing device, and the zero-carbon energy supply system is different from the prior art in that the electrolysis water generates hydrogen and carbon dioxide in the electric gas conversion device to synthesize methane as cogeneration fuel, the oxygen generated by the electrolysis water is used for supporting the total oxygen combustion of methane to generate concentrated carbon dioxide, and the concentrated carbon dioxide is recycled by the carbon dioxide capturing device for methane production, so that the carbon dioxide is discharged and utilized to realize a closed zero-carbon self-circulation, and the process does not need a high-energy air separation device, thereby greatly improving the energy utilization efficiency. According to the invention, the degradation cost of the electric gas conversion equipment and the life degradation cost of the carbon dioxide capture device are further considered, a distributed multi-energy co-generation system cost efficiency model considering carbon recycling and a multi-energy co-regulation optimization method are constructed, and based on the cost model, the total operation cost of a multi-energy co-generation network can be effectively reduced, the photovoltaic digestion capacity is improved, and the carbon emission is reduced.

Drawings

FIG. 1 is a flow chart of a method of co-operating an electrical-gas-thermal multi-energy co-generation network for carbon recycling in accordance with the present invention;

FIG. 2 is a heterogeneous topology of an electrical-thermal multi-energy co-generation network in accordance with an embodiment of the present invention;

FIG. 3 is a schematic diagram of a adjacency matrix of an electrical-thermal multi-energy co-generation network heterogeneous topology in an embodiment of the present invention;

FIG. 4 is a schematic diagram of a heterostructure multitask learning model in accordance with the present invention;

FIG. 5 is a schematic diagram of a carbon recycling model of the invention in which the electrotransport, oxygen-enriched gas and carbon dioxide capture device are linked;

FIG. 6 is a graph comparing topology identification results of various algorithms in an embodiment of the present invention;

FIG. 7 is a graph showing P2G carbon emission reduction in accordance with various embodiments of the present invention;

fig. 8 is a graph showing P2G photovoltaic power consumption versus power consumption in various methods according to embodiments of the present invention.

Detailed Description

The invention will be further described with reference to examples and drawings, to which reference is made, but which are not intended to limit the scope of the invention.

As shown in FIG. 1, the method for the cooperative operation of the electric-gas-heat multi-energy combined supply network for recycling carbon comprises the following steps.

And S1, constructing an electric-gas-heat multi-energy combined supply network heterogeneous topology model under high-proportion distributed photovoltaic access.

The main energy supply types of the multi-energy combined supply network comprise electricity, gas and heat, and the power distribution, gas supply and heat supply networks are tightly coupled through equipment such as a combined heat and power unit (CHP), an electric boiler, a gas boiler, an electric conversion gas (P2G) and the like, so that interaction and conversion between different energy sources are realized. Meanwhile, the system is also provided with electric energy storage, gas energy storage and thermal energy storage equipment so as to further improve the operation flexibility and economy of the multi-energy combined supply network. The wide access of high-proportion distributed Photovoltaics (PV) causes a plurality of uncertainty factors to be generated by a multi-energy co-generation network, wherein one part of the output is directly supplied to an electric load, and the other part of the output is used for energy conversion gas supply load or thermal load through a coupling device, and the direct supply to electric conversion gas equipment is mainly considered.

The model of the distributed photovoltaic access-considered electric-gas-heat multi-energy combined supply network multi-energy flow is as follows formula (2) -formula (4):

(2)

an alternating current power flow model of a 1 st-2 nd action power distribution network of (2), whereinAnd->Nodes +.>Active power and reactive power, +.>、/>Nodes +.>、/>Voltage amplitude of>、/>、/>Nodes +.>、/>Conductance, susceptance, phase angle difference between each other;

a 3 rd behavioral air supply network energy flow model of (2), whereinFor node- >The volume flow is converted into the power value of the power flow, +.>Is a node to node of the air sourceiFlow of injection->For node->Neighbor node set,/->For the characteristic parameters of the pipeline>Is +.>、/>Constant coefficient related to inter-pipeline parameters, +.>、/>Nodes +.>、/>Is air pressure of->For node->Load flow of>Is the heat value of natural gas;

(3)

the formula (3) is a coupling link model, wherein the 1 st-2 nd behavior CHP model is CHP, and CHP is cogeneration; lines 3-5 are respectively an electric boiler model, a gas boiler model and a P2G model, wherein P2G is electric conversion gas;、/>、/>、/>、/>the output thermal power, the thermoelectric ratio, the electric power, the natural gas input flow and the efficiency coefficient of the CHP unit are respectively>、/>The lower limit and the upper limit of the heat generation power of the CHP unit are respectively set; / >、/>The lower limit and the upper limit of the power generated by the CHP unit are respectively set; />、/>、/>The electric heating conversion efficiency, the output heat power and the consumption power of the electric boiler are respectively +.>An upper limit of output thermal power for the electric boiler;、/>、/>、/>the natural gas flow consumption, the heat output, the heat value utilization rate and the natural gas low heat value of the gas boiler are respectively +.>An upper limit of output heat power of the gas boiler; />、/>、/>Natural gas flow, consumed electric power, conversion efficiency, respectively P2G produced,/->An upper limit of the output of the P2G equipment;

(4)

the PV node is (4)Wherein>For node->Net injection power, < >>For node->Photovoltaic output of>For node->Total power to P2G node, +.>For node->Total power to electrical load node, +.>For node->Store power store total power, +.>For node->The electricity storage releases the total power.

According to the analysis of the formulas (2) - (4), the high-proportion distributed photovoltaic large-scale access to the multi-energy combined supply network can increase multi-energy supply, the randomness and fluctuation of the output of the multi-energy combined supply network can cause the power flow change in the power distribution network, meanwhile, the air supply network and the heat supply network can be influenced indefinitely through the energy coupling link, the energy supply and demand balance is optimized, and the energy storage equipment is put into the network to operate for power adjustment. Therefore, the complex characteristics of the multi-energy flows after the distributed photovoltaic access are aggravated, quantitative analysis is difficult to carry out, coupling association analysis among the multi-type nodes is difficult, and new challenges are brought to the identification of the multi-energy co-supply network topology.

The accurate topology structure information is an important basis for the intelligent planning, operation and management of the multi-energy co-generation network, so that a heterogeneous topology model of the multi-energy co-generation network is built.

Different from the topology of an independent power distribution network, the electric-gas-heat multi-energy combined supply network comprises various types of nodes of different energy networks, node characteristic information is more complicated, a topological structure is changed from a homogeneous diagram to a heterogeneous diagram, and various devices in the power distribution network, a gas supply network and a heat supply network are regarded as 'nodes' in the topological diagram;

the lines of the distribution line, the air supply line, the heat supply line and the like which are connected with network nodes are regarded as 'edges' in the topological graph, and the edges adopt the following adjacent matrixesTo describe:

(1)

and 3 two-dimensional adjacency matrixes respectively used for describing the topological structures of the power distribution network, the air supply network and the heat supply network are formed by utilizing the determined node and side information, the coupling characteristic of the coupling node is utilized to take the coupling node as a junction among different energy networks, a multi-dimensional adjacency matrix (the multi-energy co-generation network topology can be restored through the multi-dimensional adjacency matrix) is constructed, and an electric-heat energy co-generation network heterogeneous topology model under high-proportion distributed photovoltaic access is established. The diagonal elements of the multi-dimensional adjacency matrix contain coupling node information, and the specific formula (39) is as follows:

(39)

In order to avoid the redundancy of information in the graph and facilitate the subsequent topology identification, the built multi-energy co-supply network heterogeneous topology makes the following simplified treatment:

The heterogeneous topology model of the multi-energy co-generation network established in the embodiment of the invention is shown in fig. 2, it is not difficult to see that the coupling node exists in a plurality of energy networks at the same time, and the coupling node may be a source node in one energy network, and become a charge node in another energy network, so that different energy networks are connected together, wherein the cogeneration unit is a source node in a power distribution and heating network, a charge node in a gas supply network, a charge boiler is a source node in a heating network, a charge node in a power distribution network, a gas boiler is a source node in a heating network, a charge node in a gas supply network, and an electric conversion device is a source node in a gas supply network, and is a charge node in a power distribution network. Therefore, the single energy network adjacent matrix topology model is expanded into a multidimensional adjacent matrix topology model suitable for the multi-energy combined supply network, as shown in fig. 3, main information of the heterogeneous topology of the electric-gas-thermal multi-energy combined supply network comprises the adjacent matrix of each node type, the number, the distribution, the gas supply and the heat supply network, and the heterogeneous topology of the multi-energy combined supply network can be obtained through the position relations of 3 sub-network adjacent matrixes and the coupling nodes. Specifically, 3 two-dimensional adjacency matrixes for describing the topological structures of the power distribution network, the air supply network and the heat supply network are formed by using the determined nodes and the side information, and in figure 3, Two-dimensional adjacent matrixes of topological structures of the power distribution network, the air supply network and the heat supply network are respectively used, the two-dimensional adjacent matrixes are used as hubs to break barriers among different energy networks by utilizing the coupling characteristics of coupling nodes, and the multi-energy combined supply network topology is restored to be a picture, and in fig. 3, two-dimensional adjacent matrixes of the topological structures of the power distribution network, the air supply network and the heat supply network are visibleThe multidimensional adjacency matrix formed by the arrays is used for describing heterogeneous topology information of the established multi-energy co-supply network.

And S2, identifying a heterogeneous topology model of the multi-energy co-generation network by adopting a heterogeneous structure multi-task learning method.

Aiming at a heterogeneous topology model of a multi-energy co-generation network, because the energy coupling conversion characteristic and the distributed photovoltaic access, a large amount of energy coupling shared information are hidden in data and are difficult to summarize through a traditional artificial feature extraction method, the invention provides a multi-task learning method based on a heterostructure, which is used for identifying the heterogeneous topology model of an electric-thermal multi-energy co-generation network.

Based on the fact that the multi-task learning (MTL) mainly learns through a sharing layer and obtains auxiliary coupling information provided by other subtasks, a hard parameter sharing mechanism is adopted, a heterostructure multi-task learning model for identifying a multi-energy-combined network heterogeneous topology model is established, and as shown in fig. 4, the model comprises three subtask input layers, a sharing layer, three subtask output layers and a coupling layer, the sharing layer is a heterogeneous network convolution layer formed by combining a plurality of heterogeneous network convolution neurons, the sharing layer is a key of the model and is used for extracting heterogeneous energy flow network node association characteristics considering the operation characteristics of the multi-energy-combined network heterogeneous topology model, and whether links exist between two nodes is judged in a link prediction mode.

The multi-task learning loss function of the heterostructure multi-task learning model is shown in equation (40):

(40)

the overall optimization loss function may be defined as equation (41):

(41)

wherein,as a loss function; />The number of the data samples; />Is a prediction function; />Is->Data point>True values for the individual tasks; />Identifying a set of tasks for a multi-energy co-supplied network topology, here +.>= [ power distribution network topology identification subtask, air supply network topology identification subtask, and heat supply network topology identification subtask ]；/>The shared parameters of each task; />Is a task weight coefficient.

The process of identifying the heterogeneous topology model of the multi-energy co-supply network by the heterogeneous structure multi-task learning model is as follows:

s201, constructing the heterostructure multi-task learning model.

S202, respectively sending power distribution network voltage data, air supply network air pressure data and heat supply network water pressure data in the heterogeneous topology model of the multi-energy combined supply network into three subtask input layers of a multi-task learning model of a heterostructure, setting time windows by the three subtask input layers to perform equal-length time sequence interception, and then performing normalization processing on the obtained data to respectively generate node feature matrixes of the power distribution network, the air supply network and the heat supply network.

S203, the node characteristic matrixes of the power distribution network, the air supply network and the heat supply network are sent into a sharing layer together for iterative processing, and each processing process is as follows: firstly, obtaining hidden feature matrixes of a power distribution network, a gas supply network and a heat supply network through operation of heterogeneous network convolution neurons, then carrying out inner product operation on node feature vectors in the hidden feature matrixes to obtain prediction scores representing association degrees among nodes, and then optimizing prediction scoring results by utilizing cross entropy loss to obtain two-dimensional adjacent matrixes of the power distribution network, the gas supply network and the heat supply network.

And S204, coupling the two-dimensional adjacent matrixes of the power distribution network, the air supply network and the heat supply network, which are finally obtained after the iterative processing of the step 203, by the coupling layer to form a multi-dimensional adjacent matrix of the multi-energy combined supply network.

(5)

in the method, in the process of the invention,is->Hidden feature matrix of convolutional neuron output of heterogeneous network, < ->Wherein->Is thatDimension node feature matrix, < >>And->The number of network nodes and the dimension of the network node feature vector are respectively; />Is an activation function; />Normalized Laplacian matrix for network symmetry, +.>Wherein->For adding the network adjacency matrix after the self-edge connection, < >>，/>For the network adjacency matrix>Is->Identity matrix of>Is a network degree matrix; />Is->A trainable weight matrix of individual heterogeneous networks convolving neurons.

The heterogeneous network convolution neurons optimize the prediction scoring result by adopting the following cross entropy loss function:

(6)

in the method, in the process of the invention,the total loss function value of all nodes in a certain network; />Is->The true value of the individual node; />Is->Predicted values of the individual nodes; />The method is the number of nodes in the multi-energy combined supply network, namely the sum of the numbers of all nodes of each energy network.

It is noted that before the heterostructure multi-task learning model is used for identifying the multi-energy co-generation network heterogeneous topology model, the heterostructure multi-task learning model is firstly optimally trained by using historical distribution network voltage data, air supply network air pressure data and heat supply network water pressure data (the training process refers to the multi-energy co-generation network heterogeneous topology model identification process), until termination conditions are met, the heterostructure multi-task learning model is shown to learn the mapping rule of data to topology, then the trained heterostructure multi-task learning model can be saved and used for real-time multi-energy co-generation network topology identification, and finally the coupling nodes are used as hinges to couple each energy sub-network topology into the multi-energy co-generation network topology and perform visual output, so that a precise topological structure is provided for the follow-up collaborative regulation optimization.

Introducing confusion matrix to evaluate accuracy of multi-energy joint supply network topology identification, simplifying identification result between two nodes into binary classification,i.e. "associated" and "unassociated". Topology identification accuracyACCThe definition is as follows:

(42)

in the method, in the process of the invention,TPa number of users that are actually associated and identified as being associated; FPA number of users that are actually unassociated and identified as associated;FNa number of users that are actually associated and identified as unassociated;TNis the number of users that are actually unassociated and identified as unassociated.

And S3, constructing a carbon recycling model of linkage of the electric conversion gas, the oxygen-enriched fuel gas and the carbon dioxide capturing device.

According to the invention, an electrolytic tank, a methanation reactor and a carbon dioxide trapping device are additionally arranged on the basis of multi-energy combined supply network electricity conversion, and a carbon recycling model of linkage of electricity conversion, oxygen-enriched fuel gas and the carbon dioxide trapping device is established, as shown in fig. 5.

In a distributed photovoltaic driven electrolytic tank, electrolyzed water generates hydrogen and oxygen, and the reaction principle is 2H ₂ O2H ₂ +O ₂ And the photovoltaic digestion capacity of the system is improved. The hydrogen produced by electrolysis of water and carbon dioxide are used for generating Sabatier reaction to synthesize methane, and the reaction principle is CO ₂ +4H ₂ CH ₄ +2H ₂ And O, taking the generated methane as hydrocarbon fuel of the cogeneration unit, and fully excavating the value of hydrogen energy in the multi-energy cogeneration system. While oxygen generated by electrolysis of water is used to support oxy-fuel combustion of methane to obtain concentrated carbon dioxide. Among them, the full oxygen combustion technology has the advantages that the high-purity carbon dioxide is easy to separate and recycle, but must be carried out in an environment rich in oxygen, and the high-purity oxygen is not easy to prepare, and is usually provided by an air separation device with high energy consumption, so that a great challenge is brought to the practical application of the oxygen-enriched combustion decarburization technology, and the problem can be solved by utilizing electrolyzed water to generate oxygen. The concentrated carbon dioxide stream can then be readily recycled through the carbon dioxide capture device for methane synthesis, thereby achieving a closed zero carbon dioxide emissions and utilization The carbon self-circulation greatly reduces carbon emission while greatly improving the energy utilization efficiency.

The carbon recycling model of the linkage of the electric gas conversion, the oxygen-enriched fuel gas and the carbon dioxide capturing device is as follows formulas (7) - (13):

(7)

(8)

formula (8) represents a methane storage model in which、/>Methane tank->Net output and storage capacity at time; />、/>Methane tank->Net output and storage capacity at time; />Is a time interval; />、Respectively->Lower bound, upper bound of (2); />、/>Respectively->Lower bound, upper bound of (2); />

(9)

Formula (9) represents an oxygen storage model in which、/>Oxygen cylinders are respectively->Net output and storage capacity at the moment; />、/>Oxygen cylinders are respectively->Net output and storage capacity at the moment; />、/>Respectively->Lower bound, upper bound of (2); />、/>Respectively->Lower bound, upper bound of (2);

(10)

equation (10) represents a methane-fueled cogeneration unit operating constraint whereinMethane volume consumed for cogeneration; / >Is the calorific value of methane; />、/>、/>The power output is the electric active power output, the reactive power output and the thermal power output of the cogeneration respectively; />、/>Methane-electric efficiency and methane-thermal efficiency of cogeneration respectively; />Representing a correlation coefficient between reactive power and active power; />、/>Respectively the carbon dioxide emission amount of the cogeneration and the oxygen required by the total oxygen combustion of methane; />、/>Respectively representing carbon dioxide emission coefficient and oxygen consumption coefficient;

(11)

(12)

(13)

And S4, constructing a cost efficiency model of the multi-energy co-supply system considering carbon recycling.

The cost efficiency model of the multi-energy combined supply system considering carbon recycling consists of degradation and degradation cost of electric gas conversion equipment, service life degradation cost of a carbon dioxide capturing device, carbon dioxide purchasing cost, energy storage system degradation cost, electric power purchasing cost and abandon light cost.

1) Since the on/off cycle causes degradation of the electrolytic cell, degradation cost of the electric power conversion apparatus for operating degradation and on/off cycle degradation per hour can be calculated by the formula (14) ；

(14)

2) Considering life degradation cost of carbon dioxide capture device：

(15)/>

3) If the captured carbon dioxide is insufficient to produce methane, the insufficient portion will be purchased from a carbon dioxide supplier and the cost of carbon dioxide procurementThe method comprises the following steps:

(16)

wherein,unit cost for purchasing carbon dioxide;

4) Considering degradation cost of energy storage system generated by charge-discharge cycle：

(17)

5) The power transaction amount in the multi-energy co-generation system is determined in the day-ahead scheduling stage and is determined at time intervalsThe cost/profit of the electricity purchased/sold from the electricity market is the electricity purchasing cost +. >：

(18)

6) To make the multi-energy combined supply system fully consume the distributed photovoltaic as much as possible, the cost of discarding light is introduced：

(19)

In the method, in the process of the invention,is a light discarding cost coefficient; />For uncertainty, the prediction error of the total output of the distributed photovoltaic is expressed,wherein->、/>Respectively representing the predicted value and the actual value of the total output of the distributed photovoltaic.

The objective function of the multi-energy cooperative regulation optimization model is as follows:

(20)

constraint conditions of the multi-energy cooperative regulation optimization model comprise energy storage system operation constraint, coupling equipment operation constraint, distributed photovoltaic operation constraint, distribution network constraint and heating network constraint. It is worth noting that the methane produced in the distributed multi-energy co-supply system taking carbon recycling into consideration is used as hydrocarbon fuel for cogeneration, in addition, no other gas load exists, and the gas transmission distance in the researched multi-energy co-supply system is short, so that the gas supply network is not considered in the constraint condition.

1) Energy storage system operation constraints

Because the invention does not consider the prediction error of the distributed photovoltaic, the power of the energy storage system is adjusted by adopting a single-factor affine adjustable strategy, and the following formula is adopted:

(21)

(22)/>

(23)

in consideration of uncertainty of distributed photovoltaic and user load, the system reduces the running cost of the system through optimizing the ESS in the day-ahead scheduling and provides sufficient reserve for real-time adjustment. ESS (ESS)bThe operating constraints of (2) are as follows:

formulas (24) - (25) are SOC constraints:

(24)

(25)

(26)

(27)

(28)

(29)

in the formulae (24) to (29),、/>respectively indicate->、/>Time ESSbState of charge value of (2); />、/>Respectively represent ESSbUpper and lower states of charge limits; />Representing the energy storage capacity; />、/>Respectively ESSbA charging power upper limit value and a discharging power upper limit value; />、/>0-1 variables respectively representing the charge state and the discharge state of the energy storage battery;

2) Coupling device operation constraints

3) Distributed photovoltaic operation constraints

In order to adapt to the access of high-proportion distributed photovoltaic, the phenomenon of light rejection is avoided, and the distributed photovoltaic is usually enabled to maintain the maximum output of active power. The output reactive power constraint of the distributed photovoltaic is as shown in formula (30):

(30)

4) Distribution network constraints

(31)

(32)

(33)

(34)

(35)

(36)

wherein,、/>、/>、/>、/>is 0-1 variable, which respectively indicates PV, ESS, CHP, electric boiler, electric converter is positioned on bus +.>A place; />Representing a bus set, wherein the subscript of a bus connected with a main network is 0; / >、/>Respectively express and bus bar->A connected upstream and downstream line set; />、/>Respectively represent bus bars->Active and reactive loads of (a);

5) Heating network constraints

(37)

wherein,is the heat load of the heating network.

Because distributed photovoltaic has uncertainty, both the objective function and the decision quantity contain uncertain parametersAnd nonlinear terms exist in the proposed model, which brings difficulty to solving the model. Therefore, uncertainty of photovoltaic output is processed by adopting Distribution Robust Optimization (DRO), real probability distribution with high probability level is constructed by using fuzzy set based on Wasserstein distance (also called Earth-river distance for measuring distance between two distributions), and an objective function is converted into a form which is easy to process, as shown in a formula (38), so that solving difficulty of a distribution robust optimization model is reduced:

(38)

wherein,indicating uncertainty +.>Distribution of->Indicating uncertainty +.>Is (are) fuzzy set, < >>Representing uncertainty fuzzy sets->Is the worst scene distribution in->For uncertain amount +.>At->Cost of discarding light under distributionThe original optimization objective translates into minimizing the total cost of operation in the worst scenario that takes into account the distributed photovoltaic uncertainty. And meanwhile, linearizing nonlinear items such as bilinear items in a formula (14) and secondary constraints in a formula (33) by adopting a linearizing method, converting a multi-energy collaborative regulation optimization problem containing an uncertain amount into a mixed integer linear programming problem by linearizing, and solving the mixed integer linear programming problem to obtain an optimal collaborative regulation operation strategy of zero-carbon self-circulation of the distributed multi-energy combined network.

The model and the method provided by the steps S1-S5 are applied to a multi-energy co-supply network with a certain high photovoltaic permeability, the topology identification result pairs of various algorithms are shown in fig. 6, DNN (Deep Neural Networks) in fig. 6 represents a deep neural network, SVM (Support Vector Machines) represents a support vector machine, GA (Greedy Algorithm) represents a greedy algorithm, and as can be seen from fig. 6, the coupling characteristics among the multi-energy networks are fully considered, the topology identification accuracy in each energy network is higher than that of other common algorithms, and the identification accuracy can be improved by at least 12.9%.

On the basis, the collaborative regulation and control optimization of the multi-energy co-generation network is carried out, and 4 operation methods are considered for comparison: 1) The method 1 adopts the multi-energy co-supply network cooperative operation method provided by the invention; 2) Method 2 is similar to method 1, using an oxycombustion technique, but without improving the carbon recovery path; 3) Method 3 is similar to method 1, but does not utilize carbon recovery by post-combustion carbon capture technology; 4) Method 4 does not consider carbon capture and is a conventional method of operation. The P2G carbon emission reduction capacity and the distributed photovoltaic power consumption under different operation methods are respectively shown in fig. 7 and 8, and the operation result statistical data are shown in table 2, so that the multi-energy co-supply network collaborative operation method provided by the invention can be found to reduce the total operation cost of the multi-energy co-supply network, improve the photovoltaic power consumption capability, remarkably reduce the carbon emission, have better effects in the aspects of economic benefit, environmental benefit and the like, and compared with the operation method adopting carbon capture after combustion, the operation cost is reduced by 26.61%, and the carbon emission reduction capacity is improved by 20.34%.

The foregoing embodiments are preferred embodiments of the present invention, and in addition, the present invention may be implemented in other ways, and any obvious substitution is within the scope of the present invention without departing from the concept of the present invention.

In order to facilitate understanding of the improvements of the present invention over the prior art, some of the figures and descriptions of the present invention have been simplified, and some other elements have been omitted from this document for clarity, as will be appreciated by those of ordinary skill in the art.

Claims

1. The carbon recycling electricity-gas-heat multi-energy combined supply network cooperative operation method is characterized by comprising the following steps of:

step S3, constructing a carbon recycling model of linkage of the electrotransformation gas, the oxygen-enriched fuel gas and the carbon dioxide capturing device, wherein the carbon recycling model is represented by the following formulas (7) - (13):

formula (7) represents a methane production process in which p _P2G,t Power consumed for electrical conversion;the efficiency of converting electrical energy into hydrogen; />Is the low heating value of hydrogen; />The volume of carbon dioxide, hydrogen and methane are required and generated respectively; />Conversion of carbon dioxide to CH ₄ Is not limited by the efficiency of (2);

formula (8) represents a methane storage model in whichNet output and storage capacity of the methane tank at the moment t respectively;net output and storage capacity at the moment t-delta t of the methane tank respectively; Δt (delta t) _t Is a time interval; />Respectively->Lower bound, upper bound of (2); />Respectively->Lower bound, upper bound of (2);

formula (9) represents an oxygen storage model in whichNet output and storage capacity of the oxygen cylinder at the moment t are respectively;net output and storage capacity of the oxygen cylinder at the time t-delta t are respectively; />Respectively->Lower bound, upper bound of (2); />Respectively->Lower bound, upper bound of (2);

equation (10) represents a methane-fueled cogeneration unit operating constraint whereinMethane volume consumed for cogeneration; />Is the calorific value of methane; />The power output is the electric active power output, the reactive power output and the thermal power output of the cogeneration respectively; />Methane-electric efficiency and methane-thermal efficiency of cogeneration respectively; x-shaped articles ^C Representing a correlation coefficient between reactive power and active power; / >Respectively the carbon dioxide emission amount of the cogeneration and the oxygen required by the total oxygen combustion of methane; delta and omicron respectively represent carbon dioxide emission coefficients and oxygen consumption coefficients;

the formula (11) represents a carbon dioxide capturing device model, whereinIs the volume of carbon dioxide captured; gamma is carbon dioxide trapping efficiency; />Is carbon dioxide density; />Power consumed for the capture system; beta is the power consumption coefficient;

formulas (12) - (13) represent equilibrium constraints of oxygen and methane, respectively, whereinFor the volume of oxygen generated;

2. The carbon recycling electric-gas-thermal multi-energy co-supplied network co-operation method according to claim 1, wherein: the step S1 includes:

1) Source node: a power source, a PV, a gas source;

3) Load node: electrical, gas, and thermal loads;

4) Coupling node: CHP unit, electric boiler, gas boiler, P2G;

s102, regarding the circuit of the connection node as an edge in the topological graph, wherein the edge adopts the following adjacency matrix y _ij To describe:

s103, forming 3 two-dimensional adjacency matrixes for describing topological structures of a power distribution network, a gas supply network and a heat supply network respectively by using the determined nodes and side information, constructing a multi-dimensional adjacency matrix by using the coupling characteristics of the coupling nodes as hinges among different energy networks, and establishing an electric-gas-heat multi-energy combined supply network heterogeneous topological model under high-proportion distributed photovoltaic access.

3. The carbon recycling electric-gas-thermal multi-energy co-supplied network co-operation method according to claim 2, wherein: step S1, in the process of constructing an electric-gas-heat multi-energy combined supply network heterogeneous topology model under high-proportion distributed photovoltaic access, the influence of the high-proportion distributed photovoltaic access on the electric-gas-heat multi-energy combined supply network multi-energy flow is considered, and topology is simplified;

an alternating current power flow model of a 1 st-2 nd behavioral power distribution network of (2), wherein P _e,i And Q _e,i Active power and reactive power of node i, U _i 、U _j The voltage amplitudes of the nodes i and j are respectively G _ij 、B _ij 、θ _ij The conductance, susceptance and phase angle difference between the nodes i and j are respectively;

a 3 rd behavior air supply network energy flow model of (2), wherein P _g,i Power value converted to power flow for node i volume flow, S _i For the flow injected from the air source node to the node i, N _i For node i neighbor node set, k _ij Is a pipeline characteristic parameter, Y _g,ij Is a constant coefficient related to pipeline parameters between the nodes i and j, y _i 、y _j The air pressures of the nodes i and j are respectively L _g,i The load flow of the node i is represented by c, and the heating value of the natural gas is represented by c;

the 4 th to 7 th behavioral heating network energy flow model of formula (2), wherein R' is the thermal mass specific heat capacity, A is the node branch correlation matrix, m is the thermal mass flow rate in the pipeline, F _s 、F _o Respectively supplying water to the node and outlet temperature of the thermal mass, P _h For node load thermal power, K is a pipeline loop correlation matrix, ρ is a pipeline impedance coefficient, C _s 、C _r The water supply network and the backwater network coefficient matrixes are respectively F _s ′、F _r ' represents the temperature obtained by subtracting the ambient temperature from the water supply temperature and the return water temperature, respectively;

The formula (3) is a coupling link model, wherein the 1 st-2 nd behavior CHP model is CHP, and CHP is cogeneration; lines 3-5 are respectively an electric boiler model, a gas boiler model and a P2G model, wherein P2G is electric conversion gas; phi _CHP 、λ _m 、P _CHP 、F _in 、η _e Respectively the output thermal power, the thermoelectric ratio, the electric power, the natural gas input flow and the efficiency coefficient of the CHP unit,Φ _CHP 、the lower limit and the upper limit of the heat generation power of the CHP unit are respectively set;P _CHP 、/>the lower limit and the upper limit of the power generated by the CHP unit are respectively set; η (eta) _EB 、Φ _EB 、P _EB The electric heating conversion efficiency, the output heat power and the consumption power of the electric boiler are respectively +.>An upper limit of output thermal power for the electric boiler; f (F) _GB 、Φ _GB 、η _GB 、L _HV The natural gas flow consumption, the heat output, the heat value utilization rate and the natural gas low heat value of the gas boiler are respectively +.>An upper limit of output heat power of the gas boiler; f (F) _P2G 、P _P2G 、μ _P2G Natural gas flow, consumed electric power, conversion efficiency, respectively P2G produced,/->An upper limit of the output of the P2G equipment;

equation (4) is the electric power balance equation of the PV node P, where P _inject,p For the net injection power of node p,for the photovoltaic output of node p, +.>Total power flowing to P2G node for node P, +.>The total power flowing to the electrical load node for node p,storing the total power for node p power store, +.>Releasing the total power for node p power storage;

The simplification process includes:

4. The carbon recycling electric-gas-thermal multi-energy co-generation network co-operation method according to claim 3, wherein: the step S2 includes:

s204, outputting two-dimensional adjacency matrixes of the power distribution network, the air supply network and the heat supply network, which are finally obtained after the iterative processing of S203, from three subtask output layers respectively, and then entering a coupling layer together for coupling to form a multi-dimensional adjacency matrix of the multi-energy co-generation network;

H _l ＝σ(L _sym H _l-1 W _l ) (5)

wherein H is _l For the first heterogeneous network volumeHidden feature matrix of the output of the product neuron, H ₀ X, where X is an n×m-dimensional node feature matrix, N and M being the number of network nodes and the dimension of the network node feature vector, respectively; sigma (·) is the activation function; l (L) _sym The laplace matrix is normalized for network symmetry,wherein->To add the network adjacency matrix after the self-edge,a is a network adjacency matrix, I _N N is N multiplied by N identity matrix, D is network degree matrix; w (W) _l Convolving a trainable weight matrix of neurons for a first heterogeneous network;

wherein L is _C The total loss function value of all nodes in a certain network; u (u) _i Is the true value of the ith node;a predicted value for the i-th node; n is the number of nodes in the multi-energy combined supply network, namely the sum of the number of all nodes of each energy network.

5. The carbon recycling electric-gas-heat multi-energy co-supplied network co-operation method according to claim 4, wherein the method comprises the following steps: and step S2, before the heterogeneous topology model of the multi-energy combined supply network is identified, the multi-task learning model of the heterostructure is optimized and trained by utilizing historical power distribution network voltage data, air supply network air pressure data and heat supply network water pressure data, and the optimal multi-task learning model of the heterostructure is obtained.

6. The carbon recycling electric-gas-heat multi-energy co-supplied network co-operation method according to claim 5, wherein the method comprises the following steps: in the step S4, the constructed multi-energy co-generation system cost efficiency model considering carbon recycling is used for calculating degradation and degradation cost of the electric gas conversion equipment, life degradation cost of the carbon dioxide capturing device, carbon dioxide purchasing cost, degradation cost of the energy storage system, electric power purchasing cost and light discarding cost, and the calculation formulas of the costs are as follows:

1) Degradation cost f of electric conversion gas equipment _ED,t ；

Wherein alpha is _EC 、τ _EC Respectively representing the fund cost and the service life of the electrolytic tank; w (w) _E,t A bivariate of on or off state at time t; w (w) _E,t-1 A bivariate of on or off state at time t-1; mu (mu) _ESU 、μ _ESD Respectively representing the unit degradation cost when the electrolytic cell is started and closed;

2) Cost of life degradation of carbon dioxide capture device f _CA,t ：

Wherein alpha is _CA 、τ _CA Respectively representing the inherent capital cost and the service life of the carbon dioxide capturing device; v (v) _CA,t Equivalent operation and management costs per hour for the system; mu (mu) _CA The unit trapping cost;power consumed for the capture system;

3) Carbon dioxide purchasing cost

Wherein,unit cost for purchasing carbon dioxide;

4) Cost of energy storage system degradation

Wherein omega _d Is a cost factor associated with life degradation;charging power and discharging power of the ESS b respectively; />Charging efficiency and discharging efficiency of the ESS b respectively; />B is an energy storage node set;

5) Electric power purchasing cost f _e,t (P _0,t )：

Wherein P is _0,t Representing active power traded in the market;respectively representing purchase prices and selling prices; (. Cndot. ⁺ Is an operator, represents (x) ⁺ ＝max(x,0)；

6) Cost of discarding light

Wherein omega is _p,t Is a light discarding cost coefficient;for uncertainty, the prediction error of the total output of the distributed photovoltaic is expressed, Wherein->Respectively representing the predicted value and the actual value of the total output of the distributed photovoltaic.

7. The carbon recycling electric-gas-thermal multi-energy co-supplied network co-operation method according to claim 6, wherein: the objective function of the multi-energy cooperative regulation optimization model is as follows:

1) Energy storage system operation constraints

in the formulae (21) to (23),the power adjustment quantity of the ESS b is used for the energy storage system; />As a corresponding affine factor, the affine factor is constrained by equation (22); formula (23) ensures that the reserve provided by ESS b does not exceed its capacity, wherein +.> The lower standby capacity and the upper standby capacity of the ESS b are respectively represented;

the operational constraints of ESS b are as follows:

formulas (24) - (25) are SOC constraints:

equations (26) - (29) are ESS b charge and discharge power constraints and upper and lower reserve capacity constraints:

in the formulae (24) - (29), SOC _b,t 、SOC _b,t-1 The charge state values of the ESS b at the time t and the time t-1 are respectively shown; SOC _b 、Respectively representing an upper limit value and a lower limit value of the charge state of the ESS b; e (E) _R,b Representing the energy storage capacity; /> The upper limit value of charging power and upper limit value of discharging power of ESS b respectively; />Respectively represent the charge state and discharge state of the energy storage battery0-1 variable of state;

2) Coupling device operation constraints

3) Distributed photovoltaic operation constraints

wherein,output reactive power indicative of PV r, +.>Respectively representing an upper limit and a lower limit of the output reactive power of the PV r; r is a collection of distributed light Fu Jiedian;

4) Distribution network constraints

wherein P is _ij,t 、Q _ij,t Active power flow and reactive power flow of bus i to bus j are respectively represented; r is (r) _ij 、x _ij The resistances and reactances of bus i through bus j are respectively represented; v (V) _i,t 、δ _i,t 、V _j,t 、δ _j,t The voltage amplitude and phase angle of the bus i and the bus j are respectively shown;

wherein,representing the maximum apparent power of line ij; v (V) _j,t Representing the voltage amplitude of busbar j, < >> Respectively representing the upper limit and the lower limit thereof;

Wherein,a variable of 0-1, which respectively indicates whether PV, ESS, CHP, an electric boiler and electric converter are positioned at a bus j; omega shape _n Representing a bus set, wherein the subscript of a bus connected with a main network is 0; omega shape _j+ 、Ω _j- Respectively representing an upstream line set and a downstream line set connected with a bus j; />Active load and reactive load of the bus j are respectively represented;

5) Heating network constraints

Φ _CHP,t +Φ _EB,t +Φ _GB,t ＝Φ _L,t (37)

wherein phi is _L,t Is the heat load of the heating network.

8. The carbon recycling electric-gas-thermal multi-energy co-supplied network co-operation method according to claim 7, wherein: step S5, after constructing the multi-energy cooperative regulation optimization model:

adopting distributed robust optimization to process uncertainty of photovoltaic output, constructing real probability distribution containing high probability level by using fuzzy set based on Wasserstein distance, and converting an objective function of the multi-energy collaborative regulation optimization model into a formula (38):

wherein,indicating uncertainty +.>Distribution of->Indicating uncertainty +.>Is (are) fuzzy set, < >>Representing nothingDetermining an amount of fuzzy set +.>Is the worst scene distribution in->For uncertain amount +.>At->Distributed reject cost->Is a desired value of (2);