Low-carbon scheduling method for integrated energy system considering carbon coupling

Abstract

The application discloses a kind of low-carbon scheduling methods of integrated energy system considering energy-carbon coupling, which is based on carbon emission flow theory, combined with the strong fitting ability of neural network, proposes the method of carbon flow constraint learning, converts the complex mapping relationship between power flow and carbon flow into mixed integer linear constraint, realizes the effective embedding of carbon flow constraint in optimization model. At the same time, in order to reduce the structural complexity of neural network, the sparse training strategy is introduced, the model parameter size is effectively compressed, and the ReLU activation function is linearized by the improved big-M method, the feasible region is gradually tightened by introducing the cut plane constraint, so as to significantly improve the solving efficiency of optimization model. Finally, the carbon flow constraint model is embedded in the integrated energy system optimization scheduling problem, the carbon emission reduction consciousness of load side is stimulated, the demand response behavior of load side based on its own carbon signal is guided, low-carbon energy adjustment is promoted, and low-carbon scheduling under energy-carbon cooperation is realized, to reduce the overall carbon emission level of the system.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system optimization scheduling, and in particular to a low-carbon scheduling method for integrated energy systems that considers energy-carbon coupling. Technical Background

[0002] With increasing global attention on climate change, reducing greenhouse gas emissions has become a crucial trend in global energy development. Integrated energy systems, with their advantages of cascaded energy utilization and multi-energy complementarity, can efficiently utilize various energy types, thereby reducing carbon emissions and serving as a vital pathway to accelerating the achievement of carbon neutrality. In promoting low-carbon sustainable development and transitioning to a low-carbon economy, the analysis and statistics of carbon emissions are particularly important. Carbon emission flow theory is a carbon emission tracking method based on energy flow distribution in a network, viewing carbon emissions as a virtual network flow accompanying energy flow from the source to the load. Compared to traditional macro-statistical methods and life-cycle methods, carbon emission flow theory provides a clearer definition of carbon emission responsibility on the energy-consuming side, enabling accurate tracking and source tracing of specific carbon emission flows.

[0003] Currently, carbon emission flow models are widely used to analyze load-side carbon emissions to stimulate the carbon reduction potential of demand-side resources and promote low-carbon system operation. Most existing studies calculate carbon emission flows based on optimal power flow solutions, without co-optimizing carbon emission flows with power flow. While carbon emission flows and power flow are closely related, the complex nonlinear coupling between them makes it difficult to explicitly model carbon flow as a tractable constraint in optimization models. Therefore, existing carbon emission flow analyses mainly focus on calculating nodal carbon potential. Scheduling optimization algorithms considering energy-carbon flow constraints iteratively separate low-carbon demand response and energy flow scheduling. That is, they obtain power flow results based on system cost scheduling, then calculate the corresponding carbon emission flow, generate carbon signals to guide demand-side responses, and then re-schedule. This approach fails to achieve co-optimization of carbon emission flows and power flow, and the resulting scheduling scheme is difficult to guarantee global optimality under energy-carbon flow constraints.

[0004] To address the aforementioned issues, introducing data-driven energy and carbon flow constraints into the optimization scheduling model is one of the research directions in energy scheduling algorithms aimed at a low-carbon economy. In recent years, the "Optimization with Constraint Learning" (OCL) framework and technology have emerged, aiming to solve constraints in real-world optimization problems that are difficult to explicitly model. This method utilizes deep neural networks with ReLU activation functions to learn these constraints from data and accurately converts the trained neural network model into a mixed-integer linear programming problem, thereby effectively embedding the fitting relationship of the constraints into the optimization model. However, as the scale of fully connected neural networks increases, the number of parameters grows exponentially, significantly increasing the computational complexity of subsequent energy and carbon co-optimization. Therefore, sparse neural networks have become an effective alternative, effectively reducing redundant connections in the network by introducing structural pruning or sparse constraints during training. Compared to fully connected structures, sparse neural networks reduce the parameter scale and computational cost while achieving performance comparable to or even better than the original neural networks. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a low-carbon scheduling method for integrated energy systems that considers energy-carbon coupling.

[0006] The objective of this invention is achieved through the following technical solution: a low-carbon dispatching method for an integrated energy system considering energy-carbon coupling, the method comprising:

[0007] By coupling the AC power flow model of the power grid, the equivalent power flow model of the heating network supply and return water, and the natural gas network model through the energy conversion equipment model, an energy flow model coupled with the power flow model of the power grid, the operation power flow constraints of the energy flow model are used as constraints, and the goal is to minimize the operating cost, thus constructing a comprehensive energy system optimization model that does not consider the carbon cost on the load side.

[0008] Based on carbon emission flow theory, and according to the energy flow distribution of the integrated energy system, a system carbon emission flow model is constructed by applying the principles of proportional sharing and energy merging. The system carbon emission flow model is used to calculate the carbon flow rate of each load node according to the load and unit output to obtain a training dataset.

[0009] Based on the training dataset, a sparse neural network model was used to fit the nonlinear coupling relationship between tidal current and carbon flow.

[0010] A comprehensive energy system optimization model considering demand response is constructed, including system operation power flow constraints, carbon flow constraints, and demand response load constraints as constraints. The power flow constraints are based on an energy flow model coupled with an electricity-heat-gas network, while the carbon flow constraints are mixed-integer linear constraints obtained from the equivalent transformation of the sparse network model using the Big-M method. A tiered carbon pricing mechanism is used to describe the carbon cost of the load at each node.

[0011] The optimization model of the integrated energy system taking into account demand response is solved: for each node relaxation solution obtained, a cutting plane constraint is added according to the linearization improvement method of ReLU neuron big-M method; it is judged whether the error of the current optimization solution meets the standard. If it does not meet the standard, the solution is continued. If it meets the standard, the final optimization result is obtained.

[0012] The beneficial effects of this invention are:

[0013] This invention proposes a low-carbon scheduling method for integrated energy systems considering energy-carbon coupling. Since carbon emission flow calculation is highly nonlinear and nonconvex, the complex mapping relationship between power flow and carbon flow in integrated energy system optimization scheduling cannot be directly converted into easily manageable constraints. This invention utilizes the strong fitting capability of neural networks to propose a carbon flow constraint learning method, transforming the complex mapping relationship into mixed-integer linear constraints, thereby effectively embedding carbon flow constraints into the optimization model. Simultaneously, the fully connected neural network is sparsely trained to reduce the parameter size and model complexity, alleviating the computational burden on subsequent optimization models. Furthermore, the Big-M method's linearized ReLU neurons are improved, effectively and reasonably adding cutting plane constraints during the scheduling optimization solution process, gradually tightening the feasible region of the problem, thereby reducing the optimization search space. Compared to the traditional Big-M method, the solution speed of the optimization model is significantly improved. Finally, the carbon flow constraint model is embedded into the optimization scheduling problem of the integrated energy system. The tiered carbon price mechanism is used to describe the carbon cost of the load at each node, stimulate the carbon emission reduction awareness of the load side, guide the load side to respond to low-carbon demand under its own carbon signal perception, realize low-carbon scheduling with energy and carbon coordination, and thus reduce the overall carbon emissions of the integrated energy system. Attached Figure Description

[0014] Figure 1 The flowchart illustrates the calculation process of a low-carbon scheduling method for an integrated energy system that considers energy-carbon coupling, as provided in an embodiment of the present invention.

[0015] Figure 2 A schematic diagram of a low-carbon dispatching device for an integrated energy system considering energy-carbon coupling, provided as an embodiment of the present invention. Detailed Implementation

[0016] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0017] like Figure 1 As shown, the present invention provides a low-carbon scheduling method for a comprehensive energy system considering energy-carbon coupling, the specific process of which includes:

[0018] S1. Under different load parameter settings, solve the comprehensive energy system optimization model without considering the carbon cost on the load side, and calculate the carbon flow rate of each load node based on the carbon emission flow theory to generate a training dataset.

[0019] S2. Train a sparse neural network to accurately fit the mapping relationship between tidal current and carbon flow. After training, determine whether the test error meets the standard. If it does not meet the standard, adjust the training parameters and retrain. If it meets the standard, use the Big-M method to convert the model into a mixed integer linear constraint and embed it into the integrated energy system optimization model.

[0020] S3. Solve the integrated energy system optimization model that takes demand response into account. During the solution process, for each node relaxation solution obtained, add cutting plane constraints according to the ReLU neuron big-M method linearization improvement method; determine whether the error of the current optimization solution meets the standard. If it does not meet the standard, continue to solve. If it meets the standard, obtain the final optimization result.

[0021] S1 includes:

[0022] S1.1 Solving the comprehensive energy system optimization model without considering load-side carbon costs under different load parameter settings:

[0023] A comprehensive energy system optimization scheduling model is established without considering load-side carbon costs. The objective is to minimize the total system cost, constrained by system operation flow constraints. The model's objective function is shown below:

[0024] minC=C Buy,e +C Buy,g +C Gen

[0025] In the formula, C Buy,e The cost of electricity purchase is calculated by multiplying the purchased power capacity in the power subsystem by the time-of-use price; C Buy,g The cost of gas purchase is calculated as the amount of gas purchased in the natural gas subsystem multiplied by the gas price; C Gen This refers to the unit operating costs, including the CHP unit, EB, etc. in the system.

[0026] The system's power flow constraints are as follows:

[0027] The AC power flow model of the power grid includes node power balance constraints as follows:

[0028]

[0029] In the formula, P PG,i With Q PG,i P represents the active and reactive power output of the unit at node i. L,i With Q L,i P represents the active and reactive power of the load at node i. ij With Q ij These represent the active and reactive power flowing from node i to node j, respectively. ki r is the square of the current in line ki.ki With x ki Let N represent the resistance and reactance of line ki, respectively; j: i→j means that node j is an outflow node from node i; k: k→i means that node k is an inflow node from node i; N e This represents the number of nodes in the power grid.

[0030] The voltage relationship between the two ends of the line is as follows:

[0031] In the formula, v i The square of the voltage at node i;

[0032] The formula for the power at the beginning of a branch is: In radial networks, the lossless relaxation using second-order cone relaxation is as follows:

[0033] The equivalent power flow model for unified supply and return water in a heating network includes:

[0034] Based on the heat network flow model including thermal and hydraulic models, an equivalent flow model unifying the supply and return water of the heat network is established. First, flow calculations are performed on the heat network based on the thermal and hydraulic models. The delay of the heat network pipeline is not considered, and a quality regulation method is adopted, that is, the water flow rate is constant, and the heating demand is met by changing the supply water temperature.

[0035] In the hydraulic model, the nodal flow balance is represented as:

[0036] A·m P =m N

[0037] In the formula, m P For pipeline mass flow rate, m N For node quality flow, in the node-pipe association matrix A, "0" indicates that the node is not connected to the pipe, and +1 (-1) indicates that the node is an outflow (inflow) node of the pipe.

[0038] A thermal model describes the relationship between supply and return water temperatures and the amount of heat transferred:

[0039] Φ=C p ·m N ⊙(T s -T r )

[0040] In the formula, Φ is the heat power vector of the heat source or load node, and C p T is the specific heat capacity of water. s and T r These are the supply water temperature and return water temperature of the node, respectively.

[0041] The relationship between the inlet and outlet temperatures of the pipe is expressed by the temperature drop equation:

[0042]

[0043] In the formula, T f and T t These are the temperatures at the pipe inlet and outlet, respectively, T. a Let L be the ambient temperature, L be the pipe length, and λ be the total heat transfer coefficient per unit length of each pipe.

[0044] In the equivalent power flow model that unifies the supply and return water of a heating network, the equivalent heat power of the heating network pipes is expressed as the difference in heat power between the supply and return water pipes:

[0045]

[0046] In the formula, P h,P,f and P h,P,t These represent the equivalent heat power at the inlet and outlet of the heating network pipeline, respectively, T. s,f and T s,t These are the inlet and outlet temperatures of the water supply pipe, T. r,f and T r,t These are the inlet and outlet temperatures of the return water pipe, respectively.

[0047] The natural gas network model includes:

[0048] Similar to power and heating networks, the flow rates at each natural gas node satisfy an input-output balance. The gas flow balance equation for each natural gas network node is:

[0049] G·f P =f N

[0050] In the formula, G is the node-pipeline correlation matrix, f P f is the gas flow rate in the pipeline. N This represents the gas flow rate at the node.

[0051] Ignoring the slow inertia of natural gas transmission, a steady-state model of the natural gas network based on the Weymouth equation is adopted, i.e., the gas flow rate f in the natural gas pipeline. P It is a nonlinear function of the pressure at both ends of the pipeline:

[0052]

[0053] In the formula, K ij Let s be a constant for the natural gas pipeline ij. P = +1 or -1 indicates the direction of flow in the pipe, π i Let be the pressure value of node i.

[0054] The equivalent power vector P of c pipelines in a natural gas network g for:

[0055] P g=Bf P

[0056] In the formula, B is the calorific value of natural gas, taken as 10.45 kWh / m³. 3 .

[0057] Energy conversion equipment models include:

[0058] CHP units are important components that couple electrical and thermal energy. They function as the load end in the natural gas subsystem and as the source end in the power and heat subsystems. Their operating characteristics are as follows:

[0059]

[0060] In the formula, η CHP,e and η CHP,h For the electrical and thermal efficiency of the CHP unit, P CHP and H CHP For the electrical and thermal power output of the CHP unit, G CHP The energy contained in the natural gas supplied to the CHP unit. and Constraints on CHP unit output and ramp-up rate.

[0061] Electric boilers operate as loads in the power subsystem and as power sources in the thermal subsystem. Their operating characteristics are as follows:

[0062]

[0063] In the formula, H EB and P EB For the output heat power and input electrical power of the electric boiler, and For the electric boiler input power constraint, η EB The thermal efficiency of the electric boiler.

[0064] After constructing an integrated energy system optimization model that does not consider load-side carbon costs, different load parameters are set for solving the model, resulting in different scheduling results, which are then used as input for the training dataset.

[0065] S1.2. Based on carbon emission flow theory, calculate the carbon flow rate at each load node and generate a training dataset:

[0066] Based on the energy flow distribution of an integrated energy system, carbon emission flows are calculated using the principles of proportional sharing and energy merging. The formula for calculating the carbon emission intensity of grid nodes is as follows:

[0067]

[0068] In the formula, Carbon emission intensity of grid node i For the carbon emission intensity of branch road l, ρ e,S P represents the carbon emission intensity of the generator set. e,l The active power on branch l, Let P be the set of branches flowing into node i. e,S,i Let i be the output of the unit at node i.

[0069] Based on the principle of proportional sharing, the carbon emission intensity of branch lines in the power grid Equal to the carbon emission intensity of node i flowing into this branch:

[0070]

[0071] Similarly, the formula for calculating the nodal carbon intensity of the equivalent heating network is:

[0072]

[0073] In the formula, Carbon emission intensity of heating network node i For pipeline carbon emission intensity, ρ h,S For the carbon emission intensity of heating units, P h,S,i P provides heating power to the heat source at node i. h,l The equivalent heat power on pipe l, Let be the set of pipelines flowing into node i.

[0074] The formula for calculating the nodal carbon intensity of a natural gas network is:

[0075]

[0076] In the formula, Carbon emission intensity of node i in the gas grid. For pipeline carbon emission intensity, ρ g,S The carbon emission intensity of the gas source is typically taken as 0.2 kg CO2 / kWh, P g,S,i P is the equivalent power of the gas source at node i. g,l The equivalent power on pipe l, Let be the set of pipelines flowing into node i.

[0077] For energy conversion devices that couple multiple energy sources in an integrated energy system, they are divided into single-input single-output devices and single-input multiple-output devices, and their CEF characteristics are analyzed respectively.

[0078] 1) Single-input single-output device: Taking EB as an example, all input carbon emissions of the device are allocated to the output energy flow, that is:

[0079]

[0080] In the formula, and The carbon emission intensity of the EB input and output ports are respectively. and These represent the input electrical power and output thermal power of EB, respectively, η EB This refers to the EB energy conversion efficiency.

[0081] 2) Single-input multiple-output equipment: Taking the CHP unit as an example, the carbon emission balance between equipment input and output is as follows:

[0082]

[0083] In the formula, and These are the carbon intensity and equivalent gas power at the input port of the CHP unit, respectively. and These refer to the carbon intensity at the electrical and thermal output ports of the CHP unit. and These represent the electrical and thermal power outputs of the CHP unit, respectively.

[0084] For the allocation of carbon emissions among multiple outputs, the efficiency method is used, which assumes that the carbon emission intensity of electrical output ports and thermal output ports is inversely proportional to the corresponding energy efficiency:

[0085]

[0086] In the formula, η e,CHP and η h,CHP These are the electrical and thermal efficiencies of the CHP unit, respectively.

[0087] According to the above formula, the carbon intensity at the electrical and thermal output ports of the CHP unit are respectively:

[0088]

[0089] The specific formula for calculating the unified carbon flow matrix of the electricity-heat-gas network is as follows:

[0090] According to the principle of proportional sharing, the carbon intensity of all outflow lines from each node is equal to the carbon intensity of that node. If the carbon intensity of each node in the integrated energy system can be calculated, then the carbon flow rate of all lines and load nodes can be obtained from the carbon intensity of each node and the power flow results. Therefore, solving for the carbon intensity of each node is the primary objective. For the carbon emission flow of the integrated electricity-heat-gas energy system, a unified matrix calculation method is used.

[0091] As can be seen from the system carbon flow model, the formulas for calculating carbon intensity at nodes in the electricity, heat, and gas networks are identical. Therefore, the formulas for calculating carbon intensity at each node are:

[0092]

[0093] In the formula, Π∈{E,H,G} represents the integrated energy system including electricity, heat, and gas networks.

[0094] Considering the relationship between line carbon intensity and node carbon intensity, the numerator in the above equation can be rewritten in matrix form as follows:

[0095]

[0096] In the formula, Let i be an N-dimensional row vector whose i-th element is 1 and all other elements are 0. Represents the branch power flow distribution matrix, for If active power p flows from node i to node j, then Represents an N-dimensional column vector of carbon intensity at nodes. This represents the K×N dimensional source-end power distribution matrix, where K is the number of sources. This represents a K-dimensional source-end carbon intensity column vector.

[0097] Define the active flux matrix of the nodes. Its diagonal elements represent the absolute value of the active power flowing into the nodes under the power flow direction:

[0098]

[0099] In the formula, ξ N+K It is an N+K dimensional row vector with all elements being 1.

[0100] Using matrices The formula for calculating nodal carbon strength can be rewritten as:

[0101]

[0102] Extending the above equation to an N-node system yields:

[0103]

[0104] In the formula, and These are the power vector required by the load and its corresponding carbon flow rate vector, respectively. ⊙ indicates the element-wise multiplication of the matrices.

[0105] Based on the above calculation results, a training dataset is constructed by taking the load and unit output as inputs and the load carbon flow rate as output.

[0106] S2 includes:

[0107] S2.1 Training a sparse neural network to accurately fit the mapping relationship between tidal current and carbon flow, and judging whether the test error meets the standard after training:

[0108] The carbon flow rate of each load can be determined from the unified matrix calculation formula of the system. From the matrix The matrix M is determined by the power output and load demand of each node's source end. The distribution of network power flow is determined by the established network power flow distribution. Therefore, and The mapping relationship between them is as follows:

[0109]

[0110] In the formula, f c Represents the carbon flow rate from matrix M to the load. The function f of the mapping relationship p Representation matrix A function that maps a matrix M to its matrix M.

[0111] Because carbon flow calculation is highly nonlinear and nonconvex, and The mapping relationship between tidal current and carbon flow cannot be converted into easily manageable constraints in the optimization model. By utilizing the strong fitting ability of neural networks, a carbon flow constraint learning method is proposed to convert the complex mapping relationship between tidal current and carbon flow into easily manageable mixed integer linear constraints.

[0112] Using sparse neural network models to accurately fit the complex nonlinear coupling relationship between tidal currents and carbon flows includes:

[0113] Neural network models mainly include an input layer. Hidden layer (ReLU-based neurons), output layer The specific formula for the model is shown below:

[0114]

[0115] Where: x = v 0 , ReLU(x): = max{0, x}

[0116] In the formula, Let w be the output of the i-th neuron in the l-th layer of the network, and w and b be the weight and bias parameters of the neural network, respectively. [L] represents the set of L hidden layers of the neural network, and N... l [N] is the number of neurons in the l-th layer. l ] represents the set of neurons in the l-th layer, x is the input of the neural network, and y is the output of the neural network.

[0117] Based on the neural network model structure, the loss function L is minimized by the mean square error (MSE). MSE To achieve the goal, the stochastic gradient descent algorithm is used to optimize the parameters in the neural network model, making the predicted values... It gradually approaches the actual value y, as shown in the following formula:

[0118]

[0119] In the formula, N MSE f is the number of training data samples. NN A function that represents the mapping relationship from input x to output y.

[0120] Learning through training a neural network model and The complex mapping relationship between them enables accurate inference of the carbon flow rate of each node load, as shown in the following equation:

[0121]

[0122] In the formula, The results are from the neural network model inference of the carbon flow rate.

[0123] In the training process of a fully connected neural network model, internal connections of the neural network are iteratively discarded and activated to reduce the number of parameters in the equivalent model and alleviate the computational burden of optimizing and scheduling the model. The sparse training algorithm described above discards some connections based on the weights and activates new connections using instantaneous gradient information; after updating the connections, training continues with the new neural network until the next update. The main parts of sparse training include: 1) sparse distribution, 2) update strategy, 3) connection discarding, and 4) connection activation.

[0124] 1) Sparse distribution: sparsity s l ∈(0,1) is defined as the connection dropout rate of the l-th layer, and the sparsity s of each layer is... l To maintain consistency with the overall sparsity S of the model, the first layer is initialized as a dense layer, i.e., s. 1 =0, the remaining layers are initialized as sparse layers, and the sparse training of each layer adopts a unified method.

[0125] 2) Update Plan: After model initialization, with the number of iterations ΔT as the update interval, a portion of connections are discarded based on the magnitude of the neural network weights, and new connections are activated based on the magnitude of the gradient of the neural network parameters. A fractional function f for updating connections is defined using the cosine annealing algorithm. decay :

[0126]

[0127] In the formula, α is the initial score for updating the connection, and T end This represents the number of iterations for sparse training.

[0128] 3) Connection Dropping: During training, connections with smaller weight magnitudes have a relatively smaller impact on the training loss. These unnecessary connections in the neural network can be dropped according to the update plan. Specifically, every ΔT step, ArgTopK(-|w l |,f decay (t)(1-s l )N l Select the connections to discard, where ArgTopK(ν,k) returns the set of indices of the k largest elements in vector ν.

[0129] 4) Connection Activation: After the connection drop-off step, inactive connections with high-amplitude gradients are selected for reactivation. These weights reflect the strong responsive connections between neurons during backpropagation. Specifically, this is achieved through ArgTopK(|grad(w...)... l )|,f decay (t)(1-s l )N l Select the connection to activate. Newly activated connections are initialized to 0, so they will not affect the network output.

[0130] After training, evaluate whether the test error of the neural network model meets the standard. If it does not meet the standard, adjust the training parameters and retrain. If it meets the standard, continue with the subsequent process.

[0131] S2.2 If the model test error meets the standard, the model is equivalently converted into mixed integer linear constraints using the Big-M method, and then embedded into the integrated energy system optimization model.

[0132] Due to the strong nonlinearity of neural networks, these models are difficult to directly participate in the operation and scheduling of integrated energy systems. The nonlinear part of the neural network model is the ReLU activation function of the hidden layer neurons. The Big-M method, an optimization relaxation strategy, can be used to convert the ReLU activation function into a mixed-integer linear constraint, thus enabling its application in optimization scheduling models. Specifically, for neurons based on the ReLU function, the independent variable of the ReLU function is an affine function f(x) = b + w. T x, where w is a column vector of weight parameters, is linearized using the Big-M method:

[0133]

[0134] In the formula, M L and M U Let f(x) be the upper and lower limits of all possible values ​​of x, and z be an auxiliary 0-1 variable. When f(x) ≥ 0, z takes the value 1, and the output v is f(x); when f(x) < 0, z takes the value 0, and the output v is 0.

[0135] For a trained sparse neural network model, the Big-M equivalent transformation based on the ReLU activation function converts ReLU(x) into a set of mixed integer linear constraints, thereby... and The complex mapping relationship between them is transformed into easily tractable mixed-integer linear constraints, enabling carbon flow constraint learning. The specific formula is shown below, where the input x represents Output represent

[0136]

[0137] S3. Solve the integrated energy system optimization model that takes demand response into account. During the solution process, for each node relaxation solution obtained, add cutting plane constraints according to the ReLU neuron big-M method linearization improvement method; determine whether the error of the current optimization solution meets the standard. If it does not meet the standard, continue to solve. If it meets the standard, obtain the final optimization result.

[0138] S3 includes:

[0139] S3.1 Construct and solve the integrated energy system optimization model that considers demand response:

[0140] A comprehensive energy system optimization model considering demand response is established, aiming to minimize the total system cost. Constraints include system operation flow constraints, carbon flow constraints, and demand response load constraints. A tiered carbon pricing mechanism is incorporated to explore the carbon reduction potential on the load side and guide demand response under carbon potential awareness. The model's objective function is shown in the following equation:

[0141] minC=C Buy,e +C Buy,g +C Gen +C DR +C CB

[0142] In the formula, C Buy,e For electricity purchase costs, C Buy,g For gas purchase costs, C Gen For unit operating costs, C DR To compensate for demand response costs, C CB This refers to the carbon emission costs on the load side.

[0143] 1) Demand response compensation cost C DR :

[0144]

[0145] In the formula, and These are the compensation cost coefficients for reducing and transferring unit power load, respectively. and These represent the load reduction and transfer power of node n at time t, respectively.

[0146] Based on actual needs, the allowable load adjustment amount for demand response is constrained, i.e., a range of adjustable load proportions is given, as shown in the following formula:

[0147]

[0148] In the formula, λ represents the load at node n in time period t before demand response is implemented; cut and λ mov These are the proportional coefficients for load reduction and transfer, respectively, which are determined by the specific load characteristics.

[0149] 2) Load-side carbon emission cost C CB :

[0150] By adopting a tiered carbon pricing mechanism, the carbon emission responsibility on the load side is divided into multiple consecutive intervals. The greater the carbon emissions, the higher the carbon price in the corresponding interval. This "high emissions, high penalties" pricing mechanism for load-side carbon responsibility is more likely to stimulate its carbon reduction potential and reduce carbon emissions through demand response, as detailed below:

[0151]

[0152] δ CB,m =δ CB,base (1+(m-1)·σ)

[0153] In the formula, C CB For load-side carbon emission costs, The number of load nodes in the system. For the carbon emission cost of load node n, N CB The load-side carbon price tier number is given by T, where T is a scheduling cycle (taken as 24 hours), Δt is 1 hour, and δ is... CB,m Let be the carbon price at step m. Let δ be the carbon flow rate of load node n in one scheduling cycle. CB,base Let σ be the carbon benchmark price, and σ be the step change rate.

[0154] The system power flow constraints in this model are the same as those in S1.1.

[0155] S3.2 In the process of solving the integrated energy system optimization model that takes demand response into account, for each node relaxation solution obtained, a cutting plane constraint is added according to the ReLU neuron big-M method linearization improvement method; it is judged whether the error of the current optimization solution meets the standard. If it does not meet the standard, the solution continues. If it meets the standard, the final optimization result is obtained.

[0156] Improved linearization method for ReLU neurons using the Big-M method:

[0157] For neurons based on the ReLU function, when linearizing them using the Big-M method, the activation states of ReLU are constructed only based on the upper and lower bounds of the affine function f(x), without considering the range of the input vector x. This results in a convex relaxation space that is not tight enough and not precise enough for the Big-M linearized ReLU neuron, significantly increasing the time required to search for the optimal solution. Therefore, an improved Big-M linearization constraint for ReLU neurons is constructed using the upper and lower bounds of the input vector x:

[0158] v≥b+w T x

[0159]

[0160] In the formula, supp(w) is the weight w. i A set that is not equal to 0. These are the sets consisting of the minimum and maximum values ​​of the input x, respectively.

[0161] To ensure definition and gather:

[0162]

[0163] In the formula, They are respectively and Elements in the set Input x i The minimum and maximum values.

[0164] However, due to the excessive number of constraints in the improved Big-M method, direct application would lead to difficulties in solving the optimization model; it was observed that when I = supp(w) and The first two equations are equivalent to the second and third equations in the original large-M method, respectively. Therefore, we can start with the original large-M method and, during the solution process, add cutting plane constraints for each relaxed solution obtained at a node to progressively tighten the feasible region of the problem and improve solution efficiency. Define the set of violated constraints corresponding to the current relaxed solution.

[0165]

[0166] In the formula, and The input x is the current relaxation solution. i And the value of the 0-1 variable z.

[0167] Determine if the current relaxed solution violates the rules. If the constraint of the improved Big-M method corresponding to the set is violated, then this constraint is the most violated constraint among the constraints corresponding to the family of sets I. The specific mathematical expression of this constraint is as follows:

[0168] if:

[0169]

[0170] Then add constraints:

[0171]

[0172] Finally, determine whether the error of the current optimized solution meets the standard. If it does not meet the standard, continue to solve the problem. If it meets the standard, obtain the final low-carbon scheduling result of the integrated energy system considering energy-carbon coupling.

[0173] Corresponding to the aforementioned embodiment of a low-carbon scheduling method for an integrated energy system that considers energy-carbon coupling, the present invention also provides an embodiment of a low-carbon scheduling device for an integrated energy system that considers energy-carbon coupling.

[0174] See Figure 2 The present invention provides a low-carbon scheduling device for an integrated energy system that considers energy-carbon coupling, comprising a memory and one or more processors. The memory stores executable code, and when the processor executes the executable code, it is used to implement a low-carbon scheduling method for an integrated energy system that considers energy-carbon coupling as described in the above embodiment.

[0175] The embodiment of the low-carbon dispatching device for an integrated energy system considering energy-carbon coupling provided by this invention can be applied to any device with data processing capabilities, such as a computer. The device embodiment can be implemented through software, hardware, or a combination of both. Taking software implementation as an example, as a logical device, it is formed by the processor of any data processing device loading the corresponding computer program instructions from non-volatile memory into memory for execution. From a hardware perspective, such as... Figure 2 The diagram shown is a hardware structure diagram of any data processing-capable device in a comprehensive energy system low-carbon dispatching device considering energy-carbon coupling, provided by the present invention. (Except for...) Figure 2 In addition to the processor, memory, network interface, and non-volatile memory shown, any data processing device in the embodiment may also include other hardware depending on the actual function of the data processing device, which will not be described in detail here.

[0176] The specific implementation process of the functions and roles of each unit in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0177] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0178] This invention also provides a computer-readable storage medium storing a program thereon, which, when executed by a processor, implements a low-carbon scheduling method for an integrated energy system considering energy-carbon coupling as described in the above embodiments.

[0179] The computer-readable storage medium can be an internal storage unit of any data processing device as described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device of any data processing device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units and external storage devices of any data processing device. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0180] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the aforementioned low-carbon scheduling method for an integrated energy system considering energy-carbon coupling.

[0181] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this application are indicated by the claims.

[0182] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. This application is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling, characterized in that, The method includes: By coupling the AC power flow model of the power grid, the equivalent power flow model of the heating network supply and return water, and the natural gas network model through the energy conversion equipment model, an energy flow model coupled with the power flow of the electricity-heat-gas network is constructed. The operation power flow constraints of the energy flow model are used as constraints, and the objective function is to minimize the operating cost. This leads to the construction of a comprehensive energy system optimization model that does not consider the carbon cost on the load side. Based on carbon emission flow theory, and according to the energy flow distribution of the integrated energy system, a system carbon emission flow model is constructed by applying the principles of proportional sharing and energy merging. The system carbon emission flow model is used to calculate the carbon flow rate of each load node according to the load and unit output to obtain a training dataset. Based on the training dataset, a sparse neural network model was used to fit the nonlinear coupling relationship between tidal current and carbon flow. A comprehensive energy system optimization model considering demand response is constructed, including system operation power flow constraints, carbon flow constraints, and demand response load constraints as constraints. The power flow constraints are based on an energy flow model coupled with an electricity-heat-gas network, while the carbon flow constraints are mixed-integer linear constraints obtained from the equivalent transformation of the sparse network model using the Big-M method. A tiered carbon pricing mechanism is used to describe the carbon cost of the load at each node. The optimization model of the integrated energy system taking into account demand response is solved: for each node relaxation solution obtained, a cutting plane constraint is added according to the linearization improvement method of ReLU neuron big-M method; it is judged whether the error of the current optimization solution meets the standard. If it does not meet the standard, the solution is continued. If it meets the standard, the final optimization result is obtained.

2. The low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 1, characterized in that, The energy conversion equipment model specifically includes the operating constraints of the CHP unit and the electric boiler. The CHP unit acts as a load in the natural gas subsystem, as a source in the power and heat subsystem, and as a load in the power subsystem and as a source in the heat subsystem. The operating characteristics of the CHP unit include: In the formula, and For the electrical and thermal efficiency of the CHP unit, and For the electrical and thermal power output of the CHP unit, The energy contained in the natural gas supplied to the CHP unit. , , and Constraints on CHP unit output and ramp-up rate; The operating characteristics of the electric boiler are as follows: In the formula, and For the output heat power and input electrical power of the electric boiler, and For the electric boiler input power constraint, The thermal efficiency of the electric boiler.

3. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 2, characterized in that, The specific AC power flow model of the power grid is as follows: In the formula, and This represents the active and reactive power output of the unit at node i. and This represents the active and reactive power of the load at node i. and These represent the active and reactive power flowing from node i to node j, respectively. Let be the square of the current in line ki. and Let ki represent the resistance and reactance of the line, respectively. This indicates that node j is the outflow node of node i. This indicates that node k is the inflow node of node i. This represents the number of nodes in the power grid. The voltage relationship between the two ends of the line is as follows: In the formula, The square of the voltage at node i; The formula for the power at the beginning of a branch is: In radial networks, this formula utilizes a lossless relaxation technique using second-order cone relaxation: .

4. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 2, characterized in that, The equivalent power flow model for unified supply and return water in a heating network includes: The power flow calculations for the heating network are performed based on thermal and hydraulic models, without considering the delay in the heating network pipelines (i.e., the water flow rate is constant). The heating demand is met by changing the supply water temperature. In the hydraulic model, the nodal flow balance is represented as: In the formula, For pipeline mass flow rate, For node quality flow, in the node-pipe association matrix A, "0" indicates that the node is not connected to the pipe, and +1 (-1) indicates that the node is an outflow (inflow) node of the pipe; A thermal model describes the relationship between supply and return water temperatures and the amount of heat transferred: In the formula, C is the heat power vector of the heat source or load node. p The specific heat capacity of water, and These are the supply water temperature and return water temperature of the node, respectively; The relationship between the inlet and outlet temperatures of the pipe is expressed by the temperature drop equation: In the formula, and These are the temperatures at the pipe inlet and outlet, respectively. Where L is the ambient temperature and L is the pipe length. This represents the overall heat transfer coefficient per unit length of each pipe; In the equivalent power flow model that unifies the supply and return water of a heating network, the equivalent heat power of the heating network pipes is expressed as the difference in heat power between the supply and return water pipes: In the formula, and These represent the equivalent heat power at the inlet and outlet of the heating network pipeline, respectively. and These are the inlet and outlet temperatures of the water supply pipe, respectively. and These are the inlet and outlet temperatures of the return water pipe, respectively.

5. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 2, characterized in that, The natural gas network model includes: The gas flow balance equations for the natural gas network nodes are as follows: In the formula, G is the node-pipeline correlation matrix. For pipeline gas flow rate, For node gas flow rate; Ignoring the slow inertia of natural gas transmission, a steady-state model of the natural gas network based on the Weymouth equation is adopted, i.e., the gas flow rate in the natural gas pipeline. It is a nonlinear function of the pressure at both ends of the pipeline: In the formula, Let be a constant of the natural gas pipeline ij. =+1 or -1 indicates the direction of flow in the pipe. The pressure value for node i; The equivalent power vector of c pipelines in a natural gas network for: In the formula, This refers to the calorific value of natural gas.

6. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 1, characterized in that, The system carbon emission flow model includes: Formula for calculating nodal carbon strength: in, For node carbon emission intensity, Let i be an N-dimensional row vector whose i-th element is 1 and all other elements are 0. Represents the branch power flow distribution matrix. Represents an N-dimensional column vector of nodal carbon intensity. This represents a K-dimensional column vector of source-end carbon intensity. express The source-end power distribution matrix of dimension, Let be the active power flux matrix at the nodes, where the diagonal elements represent the absolute values ​​of the active power flowing into the nodes in the power flow direction: In the formula, It is an N+K dimensional row vector with all elements being 1; Extending the above equation to an N-node system yields: In the formula, and These are the power vector required by the load and its corresponding carbon flow rate vector, respectively. This indicates element-wise multiplication of matrices.

7. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 1, characterized in that, The specific steps of using a sparse neural network model to fit the nonlinear coupling relationship between tidal current and carbon flow include: Neural network models include an input layer. Hidden layer, output layer The specific formula for the model is shown below: In the formula, This represents the output of the i-th neuron in the l-th layer of the network. and These are the neural network weight parameters and bias parameters, respectively. This represents the set of L hidden layers in a neural network. The number of neurons in the l-th layer. Let x represent the set of neurons in the l-th layer, x be the input to the neural network, and y be the output of the neural network. Based on the neural network model structure, to minimize the mean squared error loss function To achieve the goal, the stochastic gradient descent algorithm is used to optimize the parameters in the neural network model, making the predicted values... Gradually approaching the actual value As shown in the following formula: In the formula, The number of training data samples, Indicates input To output Functions relating mapping relationships; Learning through training a neural network model and The complex mapping relationship between them enables accurate inference of the carbon flow rate of each node load, as shown in the following equation: In the formula, The results are from the neural network model inference of the carbon flow rate.

8. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 7, characterized in that, During the training process of the sparse neural network model, the internal connections of the neural network are iteratively discarded and activated, specifically including: Sparse distribution: sparsity Defined as the connection drop rate of layer l, and the sparsity of each layer. Maintain consistency with the overall sparsity S of the model; initialize the first layer as a dense layer, i.e. The remaining layers are initialized as sparse layers, and the sparse training of each layer adopts a unified method; Update plan: After model initialization, based on the number of iterations... To update the connection interval, a portion of connections are discarded based on the magnitude of the neural network weights, and new connections are activated based on the magnitude of the gradients of the neural network parameters. A fractional function for updating connections is defined using the cosine annealing algorithm. : In the formula, To update the initial score of the connection, This represents the number of iterations for sparse training. Connection discarding: During training, connections with smaller weight magnitudes have a relatively smaller impact on the training loss. Therefore, these unnecessary connections in the neural network are discarded according to the update plan. Specifically, each... Step, through Select the connections to discard, where Return vector The set of indices of the k largest elements in the middle; Connection activation: After the connection discarding step, inactive connections with high-amplitude gradients are selected for reactivation. These weights reflect the strong responsive connections between neurons during backpropagation; specifically, through... Select the connection to activate; newly activated connections are initialized to 0, so they do not affect the network output.

9. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 1, characterized in that, The integrated energy system optimization model that takes demand response into account includes: Objective function: In the formula, For electricity purchase costs, For gas purchase costs, For unit operating costs, To compensate for the cost of demand response, Cost of carbon emissions on the load side; Demand response compensation cost : In the formula, and These are the compensation cost coefficients for reducing and transferring unit power load, respectively. and These represent the load reduction and transfer power of node n at time t, respectively. Based on actual needs, the allowable load adjustment amount for demand response is constrained, i.e., a range of adjustable load proportions is given, as shown in the following formula: In the formula, This represents the load size of node n in time period t before demand response is implemented. and These are the proportional coefficients for load reduction and transfer, respectively, which are determined by the specific load characteristics. Load-side carbon emission costs : A tiered carbon pricing mechanism is adopted to divide the carbon emission responsibility on the load side into multiple consecutive intervals, and carbon emissions are reduced through demand response, as detailed below: In the formula, For load-side carbon emission costs, The number of load nodes in the system. The carbon emission cost of load node n, The threshold for carbon price on the load side is represented by T, which is a scheduling cycle of 24 hours. For 1 hour, Let be the carbon price at step m. Let be the carbon flow rate of load node n in one scheduling cycle. As a benchmark price for carbon, This represents the step change rate.

10. A low-carbon dispatching method for a comprehensive energy system considering energy-carbon coupling according to claim 1, characterized in that, For each node relaxation solution obtained, according to the improved ReLU neuron big-M method linearization method, cutting plane constraints are added, including: The set of violated constraints corresponding to the current relaxed solution : In the formula, and The inputs for the current relaxation solution are respectively and 0-1 variables The value; , Inputs The minimum and maximum values, Network weights; Determine if the current relaxed solution violates the rules. The constraints of the improved Big-M method corresponding to the set, if violated, then the constraint is... The constraint that is most violated in a family of sets is added to the set family. The specific mathematical expression is as follows: if: Then add constraints: 。

Images