Optimization method of green hydrogen metallurgy comprehensive energy system

Abstract

The invention discloses an optimization method for a green hydrogen metallurgy comprehensive energy system, and the method comprises the steps: constructing a network matrix model of a standardized matrix for the multi-energy mass flow characteristics in the green hydrogen metallurgy comprehensive energy system, and carrying out the multi-energy mass flow distribution calculation; an extended resource-task network model is constructed through multi-time scale characteristic analysis of flexible and adjustable resources, and then energy quality flow multi-link system regulation and control are achieved; an operation optimization model is constructed based on uncertainty variables in the system, association between key indexes in an operation scene is mined, and then a transaction strategy of energy blocks and carbon blocks is output by constructing a multi-market adaptive transaction aid decision-making system. The method solves the problems that in the prior art, multi-energy-mass-flow modeling is not accurate, the system regulation and control efficiency is low, and operation is difficult to deal with uncertainty; through combination of accurate modeling, cooperative regulation and efficient operation, the overall operation efficiency of the system is improved, multi-market transaction is optimized, and sustainable development of the green hydrogen metallurgy system is promoted.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy system optimization, and specifically relates to an optimization method for a green hydrogen metallurgical integrated energy system. Background Art

[0002] Driven by global carbon emission reduction, green hydrogen, as a clean energy carrier, has gradually shown important application prospects in the metallurgical industry. The traditional metallurgical industry relies on fossil fuels and has huge carbon emissions. However, by introducing green hydrogen to replace carbon-based reducing agents, carbon emissions can be significantly reduced, and even zero-carbon production can be achieved. With the development of renewable energy technology, the widespread application of green hydrogen provides a key path for the green transformation of the metallurgical industry.

[0003] At present, the research on green hydrogen metallurgical integrated energy system mainly focuses on the preparation, storage and system integration of green hydrogen. In the transmission and regulation of multiple energy and material flows, some progress has been made, such as single-link analysis and optimization methods for different energy carriers and material flows. However, since the green hydrogen metallurgical system involves the interaction and coordinated regulation of multiple heterogeneous energy and material flows, existing research still has great limitations in the overall modeling, cross-link coordinated regulation and dynamic optimization of multiple energy and material flows.

[0004] The existing technology mainly faces three shortcomings:

[0005] First, it is impossible to accurately describe the transmission characteristics of heterogeneous energy and material flows, which affects the overall optimization of the system. Specifically, the green hydrogen metallurgical integrated energy system involves multiple heterogeneous energy and material flows such as electricity, hydrogen, and heat, and these energy and material flows show strong dynamics and uncertainty in spatial and temporal dimensions. Existing modeling methods usually target a single energy flow or use simplified models, which cannot accurately describe the complex coupling and transmission process of multiple energy and material flows, resulting in low energy scheduling accuracy and difficulty in achieving efficient operation of the system.

[0006] Second, there is a lack of effective coordinated control mechanisms, which makes it impossible to fully realize efficient coordination of all links within the system. Specifically, most metallurgical systems still use isolated resource scheduling methods, lacking coordinated control methods at multiple time scales and multiple links, especially between green hydrogen production, storage, transportation and metallurgical links. Flexible adjustment and dynamic response cannot be achieved, resulting in the overall operation efficiency and stability of the system being affected when load fluctuates or renewable energy output is unstable.

[0007] Third, in the face of uncertain energy supply, the existing optimization methods are not robust enough to meet the comprehensive needs of system safety, economy and flexible operation; specifically, the current optimization methods focus on single goals such as economy or energy efficiency, ignoring the uncertainty risks in the system, especially the impact of the volatility of renewable energy supply on the system. In addition, traditional optimization methods are difficult to deal with complex multi-objective and multi-constraint problems, and lack robust optimization mechanisms for uncertain scenarios, resulting in poor adaptability and low safety of the system when dealing with different operating scenarios.

[0008] These problems lead to low system energy efficiency and poor carbon emission reduction effects, and breakthrough optimization methods are urgently needed to solve them. Summary of the invention

[0009] In view of the above-mentioned deficiencies in the prior art, the optimization method of the green hydrogen metallurgical integrated energy system provided by the present invention solves the problems in the prior art of uncertainty in multi-energy and mass flow modeling and difficulty in coping with uncertainty due to low-level operation of the system regulation rate.

[0010] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a method for optimizing a green hydrogen metallurgical integrated energy system, comprising:

[0011] Aiming at the multi-energy and mass flow characteristics in the green hydrogen metallurgical integrated energy system, a multi-energy and mass flow network matrix model based on differential dynamics equations that unifies heterogeneous energy flows and material flows is constructed, and the multi-energy and mass flow distribution calculation is performed based on it;

[0012] Analyze the multi-time scale characteristics of flexible and adjustable resources in the green hydrogen metallurgical integrated energy system, construct an extended resource-task network model that takes into account the distribution of multiple energy and mass flows, and combine the multi-model predictive control mechanism to coordinate the multi-link regulation of energy and mass flows;

[0013] Based on the uncertainty variables in the green hydrogen metallurgical integrated energy system, a robust optimization method is used to construct a multi-objective optimization operation model and randomly generate massive operation scenarios. The energy efficiency-flexibility-risk dynamic quantitative analysis of the operation scenario data is carried out through the oblique decision tree model. According to the results of the energy efficiency-flexibility-risk dynamic quantitative analysis, a multi-market adaptive transaction auxiliary decision-making system based on a two-layer optimization model is constructed. The scheduling and trading strategies of the energy blocks and carbon blocks are output to achieve efficient operation of the system. The energy blocks and carbon blocks represent the physical energy transmission and carbon emission quota in the green hydrogen metallurgical integrated energy system, respectively.

[0014] During the system optimization process, the multi-energy mass flow distribution calculated by the multi-energy mass flow network matrix model provides the required multi-energy mass flow distribution state for constructing the extended resource-task network model; the extended resource-task network model serves as a constraint condition for constructing a multi-objective optimization operation model; the energy mass flow regulation strategy obtained by the multi-model predictive control mechanism serves as the production regulation strategy required for constructing the multi-energy mass flow network matrix model and generating massive operation scenarios; the scheduling and trading strategy of energy blocks and carbon blocks output by the multi-market adaptive transaction auxiliary decision-making system provides optimized decision information for the energy mass flow regulation strategy output by the multi-model predictive control mechanism;

[0015] Through the system optimization process, multi-energy and mass flow distribution calculation in seconds, multi-link coordinated regulation of energy and mass flow in minutes, and efficient operation of the system in hours are achieved.

[0016] Furthermore, the method of constructing a multi-energy-mass flow network matrix model based on differential dynamics equations to unify heterogeneous energy flows and material flows is as follows:

[0017] S1. Analyze the characteristics of multi-energy-mass flow in the green hydrogen metallurgical integrated energy system, classify the physical quantities involved into intensive quantities and extensive quantities, and then construct a unified dynamic equation of multi-energy-mass flow based on differential dynamic equations;

[0018] The multi-energy-mass-flow unified dynamic equation includes a multi-energy-mass-flow transmission unified dynamic equation and a multi-energy-mass-flow coupling transformation unified dynamic equation;

[0019] S2. The constructed unified dynamic equations of multi-energy mass flow are subjected to static algebraic analysis based on Laplace transform and external port equivalence based on distributed parameter approximation in turn, and a unified model of multi-energy mass flow is constructed;

[0020] The multi-energy-mass-flow unified model includes a multi-energy-mass-flow transmission unified model obtained by sequentially performing transmission static algebraic analysis and branch external port equivalence, and a multi-energy-mass-flow coupling transformation unified model obtained by sequentially performing energy-mass transformation static algebraic analysis and coupling node external port equivalence;

[0021] S3. Based on the unified model of multi-energy mass flow, a dynamic characteristic model of the whole network of multi-energy mass flow network is constructed;

[0022] S4. Reduce the dimension of the dynamic characteristic model of the multi-energy mass flow network to a boundary equivalent model that only contains key boundary nodes, and then construct its corresponding unified Jacobian matrix as the matrix model of the multi-energy mass flow network.

[0023] Furthermore, in the step S1, in the constructed unified dynamics equation of multi-energy-mass flow, the spatial differential of the intensity quantity and the extension quantity is used to represent the left side of the equal sign, and the time differential of the intensity quantity and the extension quantity and related terms are used to represent the right side of the equal sign, which is expressed as:

[0024]

[0025] In the formula, ψ represents the generalized intensity quantity, ζ represents the extension quantity, and α 1 , α 2 , β 1 and β 2 All represent branch parameters;

[0026] The step S4 is specifically as follows:

[0027] According to the importance and physical properties of network nodes, they are divided into key boundary nodes and non-key nodes. The intensity and extension injection values ​​of the nodes involved in the dynamic characteristics model of the multi-energy mass flow network are subjected to matrix operations according to the divided node types, and a block matrix is ​​constructed. The constructed block matrix is ​​subjected to Gaussian elimination to obtain a boundary equivalent model containing only key boundary nodes, and then the corresponding unified Jacobian matrix is ​​constructed as the matrix model of the multi-energy mass flow network.

[0028] Among them, the constructed block matrix is ​​expressed as:

[0029]

[0030] Where Y BB , Y BI , Y IB , Y II represents the block matrix of Y, Y represents the energy node admittance matrix, ψ B (s) and ψ I (s) represents two sets of node strength quantities divided according to node types, ζ B (s) and ζ I (s) represents two sets of values ​​injected into the extensive quantity divided according to the node type;

[0031] The constructed boundary equivalent model is expressed as:

[0032]

[0033] The constructed unified Jacobian matrix is ​​expressed as:

[0034]

[0035] In the formula, J ψ1ζ1 ,...,J ψmζn It represents the extensive and intensive quantity characterization of process J in the green hydrogen metallurgical integrated energy system. The subscripts m and n represent the parameter index in the green hydrogen metallurgical integrated energy system.

[0036] Furthermore, the method of constructing an extended resource-task network model taking into account the distribution of multiple energy and mass flows is as follows:

[0037] X1. Analyze the regulation response characteristics of energy supply, demand, energy conversion and storage links in the green hydrogen metallurgical integrated energy system at different time scales, construct a regulation response time scale matrix, and analyze the cross-link regulation characteristics in multi-energy interaction based on it, and construct an equivalent model of regulation external characteristics based on heterogeneous links;

[0038] X2. Based on the equivalent model of external regulation characteristics corresponding to different regulation characteristics of various types of equipment, dynamic correlation analysis is carried out to analyze the impact of different production tasks on the distribution of energy and mass flow, and an extended resource-task network model is established to characterize the relationship between production tasks and the distribution of supply and demand of multiple energy and mass flows.

[0039] Furthermore, the step X2 comprises the following sub-steps:

[0040] X21. Classify the production tasks in the green hydrogen metallurgical integrated energy system according to production demand, determine the demand characteristics of various production tasks for energy and mass flow in different periods, fit their historical data, establish dynamic demand curves for various production tasks, and then produce multiple association models for different types of production tasks;

[0041] X22. Model the energy quality demand characteristics of each production process of each production task in the green hydrogen metallurgical integrated energy system through dynamic programming or optimization algorithm, adjust the execution order and energy quality flow distribution of the production process of the production task in real time, so that the coordinated scheduling between production processes is optimal, and construct a coordinated regulation model of energy quality flow between processes;

[0042] The inter-process energy and mass flow collaborative control model takes the time scheduling cost between production processes, the dependency between production processes and resource limitations as constraints;

[0043] X23. Based on the constructed production task dynamic association model and the energy-mass flow coordinated control model between processes, and by introducing the dynamic characteristics of time and space energy-mass flow, an extended resource-task network model is constructed;

[0044] The extended resource-task network model is a model for dynamically adjusting resource supply and task demand in the green hydrogen metallurgical integrated energy system according to time series data.

[0045] Furthermore, based on the multi-model predictive control mechanism, the multi-link coordinated regulation of energy and mass flow is carried out, and the method of outputting the energy and mass flow regulation strategy is specifically as follows:

[0046] Based on the task-resource allocation results obtained by the extended resource-task network model, the random variables involved in the production process at future moments are predicted and input into the multi-model predictive control mechanism to output the energy and mass flow control strategy for the current working conditions;

[0047] The multi-model predictive control mechanism includes a multi-model predictive mechanism based on a combination of local models and a multi-model predictive mechanism based on a combination of multiple controllers;

[0048] The multi-model prediction mechanism based on local model combination refers to:

[0049] Construct a local model for predicting energy and mass flow distribution under different working conditions, and switch the optimal local model to perform corresponding energy and mass flow distribution prediction according to the operating status of the green hydrogen metallurgical integrated energy system based on the conditional model switching mechanism;

[0050] The multi-model prediction mechanism based on multi-controller combination refers to:

[0051] On the basis of the energy flow allocation strategy predicted by the multi-model prediction mechanism of the local model combination output, the corresponding independent controllers are designed according to different energy flow characteristics, and according to different working conditions and energy flow control requirements, a collaborative mechanism based on constraint optimization is adopted, and the global objective function is introduced to optimize the control target of each independent controller to obtain the optimal energy flow control strategy for the system as a whole; wherein the outputs of each independent controller are interrelated;

[0052] The basic model of the multi-model predictive control mechanism is expressed as:

[0053]

[0054] In the formula, x(t+k|t) represents the predicted value of state x at time t+k, u(t+k|t) represents the predicted value of control input u at time t+k, Q and R are the weight matrices of state and control respectively. u(t) represents the control input function and N represents the number of predictions or time steps of the optimization process.

[0055] Furthermore, based on the uncertainty variables in the green hydrogen metallurgical integrated energy system, a multi-objective optimization operation model is constructed using a robust optimization method to randomly generate massive operation scenarios. The method for dynamic quantitative analysis of energy efficiency, flexibility and risk of the operation scenario data is carried out through an oblique decision tree model as follows:

[0056] T1. Characterize the uncertainty variables in the green hydrogen metallurgical integrated energy system as uncertainty sets through probability distribution fitting. Use robust optimization methods based on the uncertainty sets to build a multi-objective optimization operation model for random production simulation. Use continuous convex envelope approximation strategy to transform the multi-objective optimization operation model with nonlinear constraints into a linear model.

[0057] The uncertain variables include power supply fluctuations and market price fluctuations; the constraints of the multi-objective optimization operation model are an extended resource-task network model of the green hydrogen metallurgical integrated energy system and multi-energy and mass flow distribution;

[0058] T2. Randomly generate uncertainty variables contained in different operation scenarios, and take maximizing the system's operating efficiency and minimizing risks as the optimization goal of the operation scenario, thereby generating a large number of operation scenarios;

[0059] T3. Analyze the generated operation scenarios, extract the main features that affect the key indicators of system operation, construct an oblique decision tree based on multi-indicator sparse weights, and regularize its sparse weights to obtain an oblique decision tree model;

[0060] The key indicators mentioned include energy efficiency, flexibility and risk;

[0061] T4. Use the oblique decision tree model to extract the association rules between energy efficiency, flexibility and risk in the operating scenario data to achieve dynamic quantitative analysis of energy efficiency-flexibility-risk.

[0062] Furthermore, in step T1, the multi-objective optimization operation model for implementing random production simulation is expressed as:

[0063]

[0064] In the formula, x is the decision variable of the system, ξ is the uncertainty variable, f(x,ξ) is the profit and loss function of the system corresponding to the optimization target, and θ represents the robustness parameter, which is used to control the trade-off between the desirability and conservatism of the system. Indicates based on The expected function of represents the uncertainty set, P ξ represents the probability distribution of the uncertainty variable ξ, represents the reference distribution obtained by fitting the operational history data;

[0065] In step T2, the optimization target of the running scenario is expressed as:

[0066]

[0067] In the formula, x s represents the decision variable of running scenario s, f(x s ,ξ s ,η s ) represents the system’s operating performance under operating scenario s, R(x s ,ξ s ,η s ) represents the system risk, η s and s They represent the power supply fluctuation and market price fluctuation in the operating scenario s respectively, λ represents the weight coefficient for balancing system performance and risk, s∈S, and S represents the operating scenario set.

[0068] Furthermore, the two-layer optimization model constructed based on the results of the dynamic quantitative analysis of energy efficiency-flexibility-risk includes an upper model and a lower model;

[0069] The optimization goal of the upper model is to minimize the energy cost and carbon emission cost of the system while meeting the production demand; the optimization goal of the lower model is to purchase the required energy and carbon emission quotas at the lowest price in the energy market and carbon market according to the resource scheduling requirements of the upper model, or sell the excess energy and carbon quotas to obtain profits.

[0070] The objective function of the upper model is:

[0071]

[0072] Among them, C energy (P) represents the cost of energy blocks, represents the cost of carbon blocks, P is the energy consumption of the system, is carbon emissions;

[0073] The constraints of the upper model include energy balance constraints and carbon emission constraints, which are respectively expressed as:

[0074] P generation +P import =P demand +P storage

[0075]

[0076] Where P generation represents the local power generation, P import represents the energy imported from the market, P demand and P storage represent the demand and storage capacity respectively, Carbon credits purchased for the system;

[0077] The objective function of the lower model is expressed as:

[0078]

[0079] In the formula, λ energy (t) represents the price of the energy market in the tth period, λ carbon (t) is the price of carbon market, P import (t) is the energy purchased from the market in period t, E buy (t) and E sell (t) are the carbon emission quotas purchased and sold in period t;

[0080] The constraints of the lower model include market supply and demand balance constraints and carbon market rating constraints, which are respectively expressed as:

[0081] P import (t)+P local_generation (t) = P demand (t)

[0082]

[0083] Where P import (t) represents the energy imported from the market in period t, P local_generation (t) represents the energy generated locally by the system in the tth period, P demand (t) represents the demand in the tth period, represents the carbon dioxide emissions at time t, represents the capital cost or constraint on CO2 emissions at time t.

[0084] Furthermore, a multi-market adaptive trading auxiliary decision-making system based on a two-layer optimization model is constructed, and the method of outputting the scheduling and trading strategies of energy blocks and carbon blocks is specifically as follows:

[0085] G1. Based on the results of the dynamic quantitative analysis of energy efficiency, flexibility and risk, the time series coupling analysis and confidence interval analysis of the energy block and the carbon block are carried out to quantitatively infer the coupling characteristics of the energy block and the carbon block;

[0086] G2. Based on the inferred coupling characteristics of energy blocks and carbon blocks, the optimal bidding strategy of energy blocks and carbon blocks in market transactions is explored, and combined with the constructed two-layer optimization model to form power trading agents, carbon emission rights trading agents and green hydrogen trading agents;

[0087] G3. Based on the power trading agent, carbon emission trading agent and green hydrogen trading agent, a multi-agent reinforcement learning adaptive trading auxiliary decision-making system is constructed to continuously learn and optimize the trading strategies of energy blocks and carbon blocks to adapt to different market changes and output the scheduling and trading strategies of energy blocks and carbon blocks;

[0088] The optimization goal of the adaptive trading auxiliary decision-making system is to maximize the cumulative reward of each intelligent agent through the scheduling and trading strategies of the output energy blocks and carbon blocks.

[0089] The beneficial effects of the present invention are:

[0090] (1) This paper proposes a unified mathematical model of heterogeneous energy and material flows based on differential dynamics to solve the problem of large differences in the transmission characteristics of different mass flows. By combining the distributed parameter generalized transmission model, the time domain dynamic problem is transformed into an algebraic problem, which simplifies the complex dynamic calculation of multi-energy and mass flows. The modeling challenges of heterogeneous branches and nodes are solved through an equivalent modeling method, forming a standardized matrix modeling paradigm.

[0091] (2) Based on the multi-time scale regulation characteristics, this paper proposes a cross-link multi-energy mass flow coordinated control method to optimize the allocation of system resources. By expanding the resource-task network model to analyze the dynamic relationship between production tasks and multi-energy mass flow distribution, the system's responsiveness in the interaction of multi-energy mass flows is enhanced. Combined with the multi-model predictive control strategy, the control accuracy of the system under multi-variable disturbances is improved.

[0092] (3) In the present invention, a method for characterizing power supply and market uncertainties is combined with a distributed robust optimization strategy to propose an optimization method for balancing system safety, economy, and flexibility. A continuous convex envelope approximation strategy is used to optimize nonlinear constraint problems, and adaptive trading decisions are achieved through multi-agent reinforcement learning, thereby optimizing the market trading efficiency of the system.

[0093] (4) The present invention innovatively introduces a multi-agent reinforcement learning algorithm to optimize the temporal coupling characteristics of energy blocks and carbon blocks in multi-market transactions, and proposes an adaptive trading decision-making assistance method that can adjust trading strategies in real time in a dynamic market environment, improve the adaptability of market transactions and the accuracy of decision-making, and support efficient trading and resource allocation of green hydrogen metallurgical systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] Figure 1 This is a principle block diagram of the optimization of the green hydrogen metallurgical integrated energy system provided by the present invention. DETAILED DESCRIPTION

[0095] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0096] The embodiment of the present invention provides a method for optimizing a green hydrogen metallurgical integrated energy system. Figure 1 As shown, including:

[0097] Aiming at the multi-energy and mass flow characteristics in the green hydrogen metallurgical integrated energy system, a multi-energy and mass flow network matrix model based on differential dynamics equations that unifies heterogeneous energy flows and material flows is constructed, and the multi-energy and mass flow distribution calculation is performed based on it;

[0098] Analyze the multi-time scale characteristics of flexible and adjustable resources in the green hydrogen metallurgical integrated energy system, construct an extended resource-task network model that takes into account the distribution of multiple energy and mass flows, and combine the multi-model predictive control mechanism to coordinate the multi-link regulation of energy and mass flows;

[0099] Based on the uncertainty variables in the green hydrogen metallurgical integrated energy system, a robust optimization method is used to construct a multi-objective optimization operation model and randomly generate massive operation scenarios. The energy efficiency-flexibility-risk dynamic quantitative analysis of the operation scenario data is carried out through the oblique decision tree model. According to the results of the energy efficiency-flexibility-risk dynamic quantitative analysis, a multi-market adaptive transaction auxiliary decision-making system based on a two-layer optimization model is constructed. The scheduling and trading strategies of the energy blocks and carbon blocks are output to achieve efficient operation of the system. The energy blocks and carbon blocks represent the physical energy transmission and carbon emission quota in the green hydrogen metallurgical integrated energy system, respectively.

[0100] During the system optimization process, the multi-energy mass flow distribution calculated by the multi-energy mass flow network matrix model provides the required multi-energy mass flow distribution state for constructing the extended resource-task network model; the extended resource-task network model serves as a constraint condition for constructing a multi-objective optimization operation model; the energy mass flow regulation strategy obtained by the multi-model predictive control mechanism serves as the production regulation strategy required for constructing the multi-energy mass flow network matrix model and generating massive operation scenarios; the scheduling and trading strategy of energy blocks and carbon blocks output by the multi-market adaptive transaction auxiliary decision-making system provides optimized decision information for the energy mass flow regulation strategy output by the multi-model predictive control mechanism;

[0101] Through the system optimization process, multi-energy and mass flow distribution calculation in seconds, multi-link coordinated regulation of energy and mass flow in minutes, and efficient operation of the system in hours are achieved.

[0102] The method of constructing a multi-energy-mass flow network matrix model based on differential dynamics equations to unify heterogeneous energy flows and material flows in an embodiment of the present invention is specifically as follows:

[0103] S1. Analyze the characteristics of multi-energy-mass flow in the green hydrogen metallurgical integrated energy system, classify the physical quantities involved into intensive quantities and extensive quantities, and then construct a unified dynamic equation of multi-energy-mass flow based on differential dynamic equations;

[0104] The multi-energy-mass-flow unified dynamic equation includes a multi-energy-mass-flow transmission unified dynamic equation and a multi-energy-mass-flow coupling transformation unified dynamic equation;

[0105] S2. The constructed unified dynamic equations of multi-energy mass flow are subjected to static algebraic analysis based on Laplace transform and external port equivalence based on distributed parameter approximation in turn, and a unified model of multi-energy mass flow is constructed;

[0106] The multi-energy-mass-flow unified model includes a multi-energy-mass-flow transmission unified model obtained by sequentially performing transmission static algebraic analysis and branch external port equivalence, and a multi-energy-mass-flow coupling transformation unified model obtained by sequentially performing energy-mass transformation static algebraic analysis and coupling node external port equivalence;

[0107] S3. Based on the unified model of multi-energy mass flow, a dynamic characteristic model of the whole network of multi-energy mass flow network is constructed;

[0108] S4. Reduce the dimension of the dynamic characteristic model of the multi-energy mass flow network to a boundary equivalent model that only contains key boundary nodes, and then construct its corresponding unified Jacobian matrix as the matrix model of the multi-energy mass flow network.

[0109] In step S1 of this embodiment, the green hydrogen metallurgical integrated energy system involves multiple heterogeneous energy and material flows such as electricity, hydrogen, and heat. Each energy and material flow has different physical and dynamic characteristics during the transmission process; for example, the power flow uses voltage and current as basic parameters, while the hydrogen flow depends on pressure and mass flow rate. In order to accurately describe these differentiated flow characteristics, a multi-energy and material flow dynamic modeling method based on differential dynamics is proposed in this embodiment.

[0110] Specifically, in this embodiment, the transmission dynamic characteristics and coupled transformation dynamic characteristics of multiple energy and mass flows in the green hydrogen metallurgical integrated energy system are analyzed, wherein the transmission characteristic model of multiple energy and mass flows includes the current (carbon) flow equation, the thermodynamic (carbon) flow equation, the hydrogen (carbon) flow equation and the material flow equation; the coupled transformation process model of multiple energy and mass flows includes the electric-thermal conversion process model, the electric-hydrogen conversion process model and the material conversion process model.

[0111] In step S1 of this embodiment, the physical quantities in the above-mentioned transmission characteristic model and the conversion process model are classified into intensity quantities and extension quantities. In the constructed unified dynamics equation of multi-energy-mass flow, the spatial differentials of the intensity quantities and extension quantities are used to represent the left side of the equal sign, and the time differentials of the intensity quantities and extension quantities and related terms are used to represent the right side of the equal sign, which is expressed as follows:

[0112]

[0113] In the formula, ψ represents the generalized intensity quantity, ζ represents the extension quantity, and α 1 , α 2 , β 1 and β 2 Both represent branch parameters.

[0114] Specifically, in the unified dynamics equation of multi-energy and mass flow constructed above, intensive quantity refers to physical quantities that do not change with the scale of the system or the amount of matter, such as the voltage U in the power grid, the temperature difference T in the thermal grid, and the gas pressure change Δπ in the gas grid; extensive quantity changes proportionally with the scale of the system or the amount of matter, including the current I in the power grid, the heat flow power φ in the thermal grid, and the flow change Δf in the gas grid.

[0115] In step S2 of this embodiment, during the static algebraic analysis based on Laplace transform, the research on the green hydrogen metallurgical multi-energy system mainly focuses on the mutual influence caused by "change" between systems, and therefore focuses on the zero-state response of the system. Since the generalized energy flow equation is a linear equation, the initial state influence of the system in this embodiment can be handled by the "superposition theorem", then when the transmission static algebraic analysis based on Laplace transform is performed on the unified dynamic equation of multi-energy flow transmission, and when the energy-mass conversion static algebraic analysis based on Laplace transform is performed on the unified dynamic equation of multi-energy flow coupling conversion, it is expressed in the Laplace domain as:

[0116]

[0117] Wherein, ψ(x,s)=L(ψ(x,t)), ζ(x,s)=L(ζ(x,t)), L(·) represents the Laplace transform operator, and the complex number s represents the Laplace transform parameter.

[0118] In this embodiment, through the above-mentioned Laplace transform, the distributed parameter circuit in the time domain can be simplified to a Laplace domain distributed parameter circuit containing only spatial differentials, and the resistance, inductance and capacitance are uniformly represented. However, the distributed parameter circuit is still difficult to analyze directly, so it is necessary to further simplify or approximate it to a concentrated parameter circuit, so as to analyze the pipeline branch in the multi-energy network as a whole. This is similar to the approximate equivalent method of using π-shaped and T-shaped equivalent circuits in the power system; based on this, in this embodiment, the unified dynamic equation of multi-energy mass flow transmission after Laplace transformation is equivalent to the branch external port based on distributed parameter approximation, and the unified dynamic equation of multi-energy mass flow coupling transformation after Laplace transformation is equivalent to the coupling node external port based on distributed parameter approximation. The obtained unified model of multi-energy mass flow and the unified model of multi-energy mass flow coupling transformation are the same in form, and are expressed as:

[0119] ψ(x,s)=aexp(lx)+bexp(-lx) (4)

[0120]

[0121] In the formula, l and ε represent approximate operators, a and b are integration constants determined by boundary conditions.

[0122] Step S4 of this embodiment is specifically as follows:

[0123] According to the importance and physical properties of network nodes, they are divided into key boundary nodes and non-key nodes. The intensity and extension injection values ​​of the nodes involved in the dynamic characteristics model of the multi-energy mass flow network are subjected to matrix operations according to the divided node types, and a block matrix is ​​constructed. The constructed block matrix is ​​subjected to Gaussian elimination to obtain a boundary equivalent model containing only key boundary nodes, and then the corresponding unified Jacobian matrix is ​​constructed as the matrix model of the multi-energy mass flow network.

[0124] Specifically, in this embodiment, for key boundary nodes, the changes in the extensive quantity and intensive quantity of such nodes are more critical to the system or other systems coupled therewith; for example, the busbars connected to the tie lines in the power system, the nodes where the gas generator sets are located in the gas grid, etc.; for non-key nodes, such nodes are system nodes other than key boundary nodes, and can be further divided in combination with specific system characteristics and application requirements; for example, they can be divided according to whether they are connected to key nodes, or whether the intensive quantity / extensive quantity is constant, etc.;

[0125] In this embodiment, the constructed block matrix is ​​expressed as:

[0126]

[0127] Where Y BB , Y BI , Y IB , Y II represents the block matrix of Y, Y represents the energy node admittance matrix, ψ B (s) and ψ I (s) represents two sets of node strength quantities divided according to node types, ζ B (s) and ζ I (s) represents two sets of values ​​injected into the extensive quantity divided according to the node type;

[0128] In this embodiment, the constructed boundary equivalent model is expressed as:

[0129]

[0130] In this embodiment, the constructed unified Jacobian matrix is ​​expressed as:

[0131]

[0132] In the formula, It represents the extensive and intensive quantity characterization of process J in the green hydrogen metallurgical integrated energy system. The subscripts m and n represent the parameter index in the green hydrogen metallurgical integrated energy system.

[0133] The unified Jacobian matrix in this embodiment, as a standardized matrix model, simplifies the solution of complex dynamic problems, realizes the efficient computing capability of the system through fast algebraic operations, and can adapt to energy and mass flow networks of different scales and complexities, greatly improving the computing efficiency of the system operation.

[0134] In an embodiment of the present invention, the method for constructing an extended resource-task network model taking into account the distribution of multiple energy and mass flows is specifically as follows:

[0135] X1. Analyze the regulation response characteristics of energy supply, demand, energy conversion and storage links in the green hydrogen metallurgical integrated energy system at different time scales, construct a regulation response time scale matrix, and analyze the cross-link regulation characteristics in multi-energy interaction based on it, and construct an equivalent model of regulation external characteristics based on heterogeneous links;

[0136] X2. Based on the equivalent model of external regulation characteristics corresponding to different regulation characteristics of various types of equipment, dynamic correlation analysis is carried out to analyze the impact of different production tasks on the distribution of energy and mass flow, and an extended resource-task network model is established to characterize the relationship between production tasks and the distribution of supply and demand of multiple energy and mass flows.

[0137] In step X1 of this embodiment, the operation of the green hydrogen metallurgical integrated energy system involves supply, demand, energy conversion and storage, and these links show different dynamic regulation characteristics at different time scales. In order to achieve multi-scale coordinated regulation, the regulation response characteristics of each link are analyzed in this embodiment, and a multi-time scale modeling framework is established. Exemplarily, the above-mentioned supply, demand, energy conversion and storage equipment includes supply, demand, energy conversion and storage, etc.

[0138] In this embodiment, the supply and demand dynamic model is described by the following equation:

[0139]

[0140] in, represents the state change of the system at time t, P d (t) is the required power, P s (t) is the supplied power, P st (t) is the power output of the energy storage system. Through dynamic adjustment at multiple time scales, the supply and demand balance of the system and the optimal regulation of energy storage and release are ensured in different time periods.

[0141] In the green hydrogen metallurgical integrated energy system, each link not only has its own regulation characteristics, but also needs to consider the mutual conversion and interaction between different energy qualities (electricity, hydrogen, heat) on the overall scheduling. For example, in the process of electrolytic hydrogen production, the conversion efficiency of electricity to hydrogen energy is not only subject to the electrolyzer itself, but also affected by the fluctuation of upstream power grid energy supply.

[0142] Furthermore, in step X1 of this embodiment, the method for analyzing the cross-link regulation characteristics in the multi-energy interaction is specifically as follows:

[0143] By constructing an energy flow coupling model and analyzing the delay effect in the cross-link energy interaction process, the impact of the mutual conversion and interaction between different energy qualities on the overall scheduling of the green hydrogen metallurgical integrated energy system is analyzed; among them, the delay effect includes the cross-link energy transmission delay and conversion delay.

[0144] After analyzing the cross-link regulation characteristics in multi-energy interactions, it is necessary to introduce deep learning technology to assist in modeling because the nonlinearity and uncertainty of various energy devices in the actual system are difficult to fully describe through traditional physical models. The key to this step is to build a high-precision equivalent model of the external characteristics of regulation through the analysis of historical data.

[0145] Specifically, the method for constructing a high-precision equivalent model of the external regulation characteristics is as follows: for different regulation characteristic data of various types of equipment, the corresponding deep neural network (such as a feedforward neural network (FFNN) or a long short-term memory network) is selected, the device control variable is set as the network input (such as input electric power), and the regulation result is set as the network output (such as hydrogen production or energy efficiency). The corresponding deep neural network is trained according to the historical operation data and dynamically updated to obtain the external regulation characteristic models corresponding to the different regulation characteristics of various types of equipment.

[0146] Step X2 of this embodiment includes the following sub-steps:

[0147] X21. Classify the production tasks in the green hydrogen metallurgical integrated energy system according to production demand, determine the demand characteristics of various production tasks for energy and mass flow in different periods, fit their historical data, establish dynamic demand curves for various production tasks, and then produce multiple association models for different types of production tasks;

[0148] For example, in the green hydrogen metallurgical integrated energy system, the energy and mass flow demand types and intensities of tasks such as hydrogen preparation, electric energy storage, hydrogen storage, and vertical furnace steelmaking are different. By classifying the tasks, the demand characteristics of energy and mass flow for various tasks at different times can be clarified.

[0149] In the above process, the distribution of energy and mass flow can be regarded as a function of time and task progress. By fitting historical data, a dynamic demand curve for tasks is established, and multiple association models are generated for different task types to ensure that the distribution of energy and mass flow meets real-time production needs.

[0150] X22. Model the energy demand characteristics of each production process of each production task in the green hydrogen metallurgical integrated energy system through dynamic programming or optimization algorithm, adjust the execution order and energy flow distribution of the production process of the production task in real time, so that the coordinated scheduling between production processes is optimal, and construct a coordinated regulation model of energy flow between processes; the coordinated regulation model of energy flow between processes takes the time scheduling cost between production processes, the dependency relationship between production processes and resource constraints as constraints;

[0151] For example, the hydrogen and electricity requirements of a shaft furnace during startup and operation are nonlinear, and may consume a large amount of energy during the startup phase, but require less during stable operation; similarly, the energy requirements of an electrolyzer vary at different stages;

[0152] X23. Based on the constructed production task dynamic association model and the energy-mass flow coordinated control model between processes, and by introducing the dynamic characteristics of time and space energy-mass flow, an extended resource-task network model is constructed;

[0153] The extended resource-task network model is a model for dynamically adjusting resource supply and task demand in the green hydrogen metallurgical integrated energy system according to time series data.

[0154] In this embodiment, the traditional resource-task network model usually assumes that resources are static and independent, but in a multi-energy flow system, the interaction between resources and tasks is dynamic and interdependent. To adapt to this complexity, it is necessary to expand the traditional resource-task network model. In this embodiment, the dynamic characteristics of time, space and energy flow are introduced to construct an extended resource-task network model. Based on this, the extended resource-task network model in this embodiment is a model that dynamically adjusts the supply of resources and the demand for tasks in the green hydrogen metallurgical integrated energy system according to time series data.

[0155] In this embodiment, the temporal and spatial distribution of different energy flows (such as electricity, hydrogen, and heat) is crucial to the overall operation of the system; in a specific example of this embodiment, in the process of hydrogen production by electrolysis, the supply of electricity is closely related to the demand for hydrogen production, and it is necessary to define the "supply and demand" node in the network model and introduce a time-varying function to describe the relationship between the two. elec 、Hydrogen E H2 ) is distributed in time and space by functions h(t,x) and k(t,x) respectively, and the extended resource-task network model can be defined as:

[0156] E elec (t,x)=h(t,x)(10)

[0157] E H2 (t,x)=k(t,x)(11)

[0158] Among them, t is time, x is spatial position, and functions h and k describe the dynamic distribution of different energy flows in the space-time dimension.

[0159] In this embodiment, when the extended resource-task network model performs collaborative regulation on resources involved in multiple production tasks, it performs multi-correlation processing on them, establishes an optimization model with constraints, and realizes collaborative matching of multiple energy and quality supplies to production task requirements; wherein the constraints include resource supply capacity, dynamic allocation of energy and quality flows, and timing dependencies of production tasks.

[0160] In this embodiment, in the extended resource-task network model, each task node not only represents the execution progress of the task, but also reflects the consumption of energy and mass flow; based on this, when the extended resource-task network model allocates resources-tasks, a dynamic optimization algorithm is used to dynamically adjust the execution order of production tasks and the allocation strategy of energy and mass flow according to the current energy and mass flow distribution in the green hydrogen metallurgical integrated energy system to ensure the optimal efficiency and stability of the system.

[0161] In this embodiment, the dynamic optimization algorithm in the above process may be a heuristic algorithm or a large-scale linear programming method; illustratively, the above heuristic algorithm includes a genetic algorithm and a particle swarm optimization algorithm).

[0162] In this embodiment, in a complex multi-energy and mass flow system, due to the volatility of energy and mass flow and the uncertainty of system requirements, traditional control strategies are difficult to achieve efficient production control. Therefore, it is necessary to use a multi-model predictive control (MPC) strategy to achieve real-time control of system operation. In this embodiment, a prediction model of multiple random variables is designed, and production control is performed by combining multiple models to effectively deal with system complexity and uncertainty.

[0163] Specifically, in this embodiment, the method for outputting the energy and mass flow control strategy is specifically as follows:

[0164] Based on the task-resource allocation results obtained by the extended resource-task network model, the random variables involved in the production process at future moments are predicted and input into the multi-model predictive control mechanism to output the energy and quality flow control strategy for the current working conditions; the multi-model predictive control mechanism includes a multi-model prediction mechanism based on a combination of local models and a multi-model prediction mechanism based on a combination of multiple controllers. Among them, when predicting random variables, a time series analysis is performed on each random variable to capture the volatility characteristics in its historical data. Classical time series models such as the autoregressive integrated moving average model (ARIMA) and the generalized autoregressive conditional heteroskedasticity model (GARCH) can be used to fit the volatility law of each random variable on different time scales.

[0165] In this embodiment, the multi-model prediction mechanism based on local model combination refers to:

[0166] A local model is constructed to realize energy and mass flow distribution prediction under different working conditions, and based on the conditional model switching mechanism, the optimal local model is switched according to the operating status of the green hydrogen metallurgical integrated energy system to perform corresponding energy and mass flow distribution prediction.

[0167] Specifically, in this embodiment, local models are constructed for different operating stages of the system. For example, under low load and high load conditions, the energy and mass flow distribution rules and response characteristics of the system are different. For these different working conditions, local models based on linear systems, nonlinear models or machine learning models (such as long short-term memory networks LSTM or support vector machines SVM, etc.) can be established respectively.

[0168] In this embodiment, a conditional model switching mechanism is introduced into the above-mentioned multi-model prediction mechanism based on local models, so that an appropriate local model can be selected under different system operating conditions. For example, when the system load fluctuates greatly, a local model suitable for high-fluctuation conditions can be switched, while a simple linear model can be selected when the system operates smoothly. The model switching mechanism can be implemented by setting a threshold, a state observer, or a dynamic discrimination method based on real-time data.

[0169] In this embodiment, the multi-model prediction mechanism based on multi-controller combination refers to:

[0170] On the basis of the energy and mass flow allocation strategy predicted by the multi-model prediction mechanism of local model combination, corresponding independent controllers are designed according to different energy and mass flow characteristics. In view of different working conditions and energy and mass flow control requirements, a collaborative mechanism based on constraint optimization is adopted, and a global objective function is introduced to optimize the control target of each independent controller to obtain the optimal energy and mass flow control strategy for the system as a whole. Among them, the outputs of each independent controller are interrelated.

[0171] In this embodiment, a conditional model switching mechanism is introduced into the above-mentioned multi-model prediction mechanism based on local models, so that an appropriate local model can be selected under different system operating conditions. For example, when the system load fluctuates greatly, a local model suitable for high-fluctuation conditions can be switched, while a simple linear model can be selected when the system operates smoothly. The model switching mechanism can be implemented by setting a threshold, a state observer, or a dynamic discrimination method based on real-time data.

[0172] In this embodiment, the basic model of the above multi-model predictive control mechanism is expressed as:

[0173]

[0174] In the formula, x(t+k|t) represents the predicted value of state x at time t+k, u(t+k|t) represents the predicted value of control input u at time t+k, Q and R are the weight matrices of state and control respectively. u(t) represents the control input function and N represents the number of predictions or time steps of the optimization process.

[0175] The above model makes rolling predictions of future states and dynamically adjusts the production plan, thereby improving the system's control accuracy and response speed.

[0176] In the embodiment of the present invention, based on the uncertainty variables in the green hydrogen metallurgical integrated energy system, a multi-objective optimization operation model is constructed by a robust optimization method to randomly generate massive operation scenarios. The method for performing energy efficiency-flexibility-risk dynamic quantitative analysis on the operation scenario data through an oblique decision tree model is specifically as follows:

[0177] T1. Characterize the uncertainty variables in the green hydrogen metallurgical integrated energy system as uncertainty sets through probability distribution fitting. Use robust optimization methods based on the uncertainty sets to build a multi-objective optimization operation model for random production simulation. Use continuous convex envelope approximation strategy to transform the multi-objective optimization operation model with nonlinear constraints into a linear model.

[0178] The uncertain variables include power supply fluctuations and market price fluctuations; the constraints of the multi-objective optimization operation model are an extended resource-task network model of the green hydrogen metallurgical integrated energy system and multi-energy and mass flow distribution;

[0179] T2. Randomly generate uncertainty variables contained in different operation scenarios, and take maximizing the system's operating efficiency and minimizing risks as the optimization goal of the operation scenario, thereby generating a large number of operation scenarios;

[0180] T3. Analyze the generated operation scenarios, extract the main features that affect the key indicators of system operation, construct an oblique decision tree based on multi-indicator sparse weights, and regularize its sparse weights to obtain an oblique decision tree model;

[0181] The key indicators mentioned include energy efficiency, flexibility and risk;

[0182] T4. Use the oblique decision tree model to extract the association rules between energy efficiency, flexibility and risk in the operating scenario data to achieve dynamic quantitative analysis of energy efficiency-flexibility-risk.

[0183] In step T1 of this embodiment, the method of characterizing the uncertainty variables in the green hydrogen metallurgical integrated energy system as uncertainty sets through probability distribution fitting is specifically as follows:

[0184] The historical operation data in the green hydrogen metallurgical integrated energy system is statistically analyzed, and the uncertainty variables are fitted with probability distributions, and the uncertainty variables are characterized as corresponding uncertainty sets based on the distances between different probability distributions. For example, the probability distribution fitting technology is used to generate probability intervals for multiple uncertainty variables in power supply and market, which can effectively characterize a variety of possible operating states and provide a data basis for subsequent optimization scheduling; further, the Wasserstein distance is used to measure the distances between different probability distributions.

[0185] In step T1 of this embodiment, in the process of realizing the multi-objective optimization operation model of random production simulation, taking into account the impact of power supply and market uncertainties on system operation, the distributionally robust optimization (DRO) method is introduced to achieve the optimization of random production simulation. In this process, robustness parameters are introduced through the optimization model, so that the system can balance the expectancy and conservatism of operation under different power supply and market environments. Specifically, the optimization objectives cover the safety, economy and flexibility of system operation. In a random environment, distributionally robust optimization can not only provide a stable and reliable production strategy for the green hydrogen metallurgical system, but also minimize operational risks.

[0186] Specifically, in this embodiment, under multiple uncertainties, the distributionally robust optimization (DRO) method is used. Considering the stochastic production optimization problem of the green hydrogen metallurgical system, the objective function can be expressed as the worst-case expectation under the uncertainty set:

[0187]

[0188] Where x represents the decision variable of the system (for example, power generation scheduling, market transaction volume, etc.), ξ represents the uncertainty variable (such as power supply fluctuation, market price, etc.), and f(x,ξ) is the profit and loss function of the system (including indicators such as economy, safety, and flexibility).

[0189] In order to further consider the safety and flexibility of system operation, the extended resource-task network of the green hydrogen metallurgical integrated energy system is introduced as a model constraint, so that the system can achieve the optimization effect under a certain robustness level, and the robustness parameter θ is introduced to express the multi-objective optimization operation model for random production simulation as follows:

[0190]

[0191] In the formula, x is the decision variable of the system, ξ is the uncertainty variable, f(x,ξ) is the profit and loss function of the system corresponding to the optimization target, and θ represents the robustness parameter, which is used to control the trade-off between the desirability and conservatism of the system. Indicates based on The expected function of represents the uncertainty set, P ξ represents the probability distribution of the uncertainty variable ξ, represents the reference distribution obtained by fitting the operational history data;

[0192] On this basis, the continuous convex envelope approximation method is used to optimize the nonlinear constraint problem in the system. Assuming that the objective function f(x) in the original problem is nonlinear and difficult to solve directly, the problem can be simplified by finding its convex hull approximation.

[0193] Assume f(x) is a nonlinear function whose k The linear approximation at is:

[0194]

[0195] Through this linearization method, the original problem is approximated as a linear programming (LP) problem, so that high-quality solutions can be obtained under limited computing resources. In the optimization solution process, this strategy can be used to improve the computational efficiency and ensure that the accuracy of the solution meets the system's operating requirements.

[0196] In step T2 of this embodiment, in order to comprehensively evaluate the operating performance of the green hydrogen metallurgical integrated energy system under asynchronous deterministic conditions, a large number of different operating scenarios are generated through random production simulation. By simulating a large number of possible power supply and market environments, the energy efficiency, flexibility and risk characteristics of the system under various operating conditions can be captured. These scenario data not only provide a solid foundation for subsequent optimization and decision-making, but also provide a comprehensive risk assessment and flexibility analysis tool for system operation.

[0197] Specifically, power supply fluctuation ξ and market price fluctuation η are set as uncertainty variables, and different scenario sets S are constructed. Each scenario s∈S contains variables (ξ s ,η s) is generated by random sampling or Monte Carlo simulation. For each scenario, the optimization goal is to maximize the system's operating efficiency while minimizing the risk; based on this, in step T2, the optimization goal of the operating scenario is expressed as:

[0198] In the formula, x s represents the decision variable of running scenario s, f(x s ,ξ s ,η s ) represents the system’s operating performance under operating scenario s, R(x s ,ξ s ,η s ) represents the system risk, η s and s They represent the power supply fluctuation and market price fluctuation in the operating scenario s respectively, λ represents the weight coefficient for balancing system performance and risk, s∈S, and S represents the operating scenario set.

[0199] In step T3 of this embodiment, based on a large number of operation scenarios, an oblique decision tree based on multi-indicator sparse weights is proposed to dynamically quantify the energy efficiency, flexibility and risk of the system. The oblique decision tree extracts the main features that affect the key indicators of system operation by introducing sparse weights, and controls the complexity of the model through regularization methods to avoid overfitting. The model can not only efficiently process multidimensional data, but also dynamically identify the intrinsic connection between various indicators in system operation, and provide high-precision quantitative analysis for multi-objective optimization problems in uncertain environments.

[0200] Specifically, the oblique decision tree is a decision tree based on linear segmentation, which is used to process multi-dimensional, continuous data. Assume that the linear segmentation function of the kth node in the decision tree is:

[0201]

[0202] Among them, x is the input feature vector (such as energy efficiency, flexibility, risk and other indicators), ω k is the sparse weight vector of the kth node of the oblique decision tree, b k is the bias term.

[0203] In order to ensure the sparsity of the model and avoid overfitting, regularization is introduced, and the objective function E(q) of sparse weight regularization in the oblique decision tree is expressed as:

[0204]

[0205] In the formula, represents the weight entropy, l 1 and l 2 represents the regularization parameter that controls the degree of sparsity, q and q represent the sparse regularization terms, and W i(q) is expressed as the weight function associated with sample i, H i (q) represents the indicator or function related to sample i, N represents the number of samples, and i∈I represents the sample index.

[0206] In this embodiment, through regularization processing, the model can automatically select the most important features and remove variables that have little impact on system operation, thereby avoiding overfitting of complex models.

[0207] In step T4 of this embodiment, through the analysis of the decision tree model, the significant relationship between energy efficiency and flexibility, risk and other indicators in different operating scenarios can be identified, providing a basis for system optimization and operation strategy.

[0208] In the embodiment of the present invention, in the current energy system operation, not only the optimal scheduling of energy resources such as electricity and hydrogen should be considered, but also the carbon emission trading mechanism should be integrated, especially in the context of carbon neutrality, carbon block trading has become an important factor affecting the efficiency of system operation. In order to achieve flexible and efficient system operation in a multi-market environment of energy blocks and carbon blocks, the present invention proposes a multi-market trading strategy based on a two-layer optimization model.

[0209] In the green hydrogen metallurgical integrated energy system, there is a close connection between "energy blocks" (such as electricity and hydrogen energy) and "carbon blocks" (i.e., carbon emission quotas). The consumption of energy blocks directly affects carbon emissions, which in turn are subject to limits and market prices. In order to achieve flexible scheduling of the system in a multi-market environment, the transaction process of energy blocks and carbon blocks is modeled as a two-layer optimization problem in this embodiment. Specifically, the two-layer optimization model constructed according to the results of the dynamic quantitative analysis of energy efficiency-flexibility-risk includes an upper model and a lower model;

[0210] The optimization goal of the upper model is to minimize the energy cost and carbon emission cost of the system while meeting the production demand; the optimization goal of the lower model is to purchase the required energy and carbon emission quotas at the lowest price in the energy market and carbon market according to the resource scheduling requirements of the upper model, or sell the excess energy and carbon quotas to obtain profits; wherein the upper model represents the energy scheduling and carbon emission constraints of the system, and the lower model represents the energy block and carbon block trading strategies in the market;

[0211] Specifically, the objective function of the upper model is:

[0212]

[0213] Among them, C energy (P) represents the cost of energy blocks, represents the cost of carbon blocks, P is the energy consumption of the system, is carbon emissions;

[0214] The constraints of the above upper model include energy balance constraints and carbon emission constraints, which are respectively expressed as:

[0215] P generation +P import =P demand +P storage (20)

[0216]

[0217] Where P generation represents the local power generation, P import represents the energy imported from the market, P demand and P storage represent the demand and storage capacity respectively, Carbon credits purchased for the system;

[0218] The objective function of the lower model is expressed as:

[0219]

[0220] In the formula, λ energy (t) represents the price of the energy market in the tth period, λ carbon (t) is the price of carbon market, P import (t) is the energy purchased from the market in period t, E buy (t) and E sell (t) are the carbon emission quotas purchased and sold in period t;

[0221] The constraints of the above lower-level model include market supply and demand balance constraints and carbon market rating constraints, which are expressed as:

[0222] P import (t)+P local_generation (t) = P demand (t) (23)

[0223] E buy (t)≥E CO2 (t)-Cap CO2 (t)(24)

[0224] Where P import (t) represents the energy imported from the market in period t, P local_generation (t) represents the energy generated locally by the system in the tth period, P demand (t) represents the demand in the tth period, E CO2 (t) represents the carbon dioxide emissions at time t, Cap CO2 (t) represents the capital cost or constraint on CO2 emissions at time t.

[0225] In this embodiment, in order to ensure the flexibility of the system in the face of price fluctuations in different markets, a multi-market bidding and auction coupling analysis model strategy for energy blocks and carbon blocks is also proposed. During price peaks, local power generation and energy storage are increased to reduce imported energy from the market, while profits are obtained by selling excess carbon emission quotas. During price troughs, cheap energy or carbon emission quotas are purchased from the market to optimize the cost of the entire production process.

[0226] In the embodiment of the present invention, a multi-market adaptive transaction auxiliary decision-making system based on a two-layer optimization model is constructed, and the method of outputting the scheduling and trading strategies of energy blocks and carbon blocks through the system is specifically as follows:

[0227] G1. Based on the results of the dynamic quantitative analysis of energy efficiency, flexibility and risk, the time series coupling analysis and confidence interval analysis of the energy block and the carbon block are carried out to quantitatively infer the coupling characteristics of the energy block and the carbon block;

[0228] G2. Based on the inferred coupling characteristics of energy blocks and carbon blocks, the optimal bidding strategy of energy blocks and carbon blocks in market transactions is explored, and combined with the constructed two-layer optimization model to form power trading agents, carbon emission rights trading agents and green hydrogen trading agents;

[0229] G3. Based on the power trading agent, carbon emission trading agent and green hydrogen trading agent, a multi-agent reinforcement learning adaptive trading auxiliary decision-making system is constructed to continuously learn and optimize the trading strategies of energy blocks and carbon blocks to adapt to different market changes and output the scheduling and trading strategies of energy blocks and carbon blocks;

[0230] The optimization goal of the adaptive trading auxiliary decision-making system is to maximize the cumulative reward of each intelligent agent through the scheduling and trading strategies of the output energy blocks and carbon blocks.

[0231] In step G1 of this embodiment, a time-series coupling analysis method is proposed for energy blocks and carbon blocks in multi-market transactions, and a quantitative evaluation is performed through confidence inference. The energy block and the carbon block represent the transmission of physical energy and the changes in carbon emission quotas in the integrated energy system, respectively, and their transactions in the market are highly correlated. Through the coupling analysis model, the interaction between energy and carbon is studied, especially its transmission and trading patterns in different time periods. In this embodiment, an inference method based on confidence intervals is also proposed to quantify the coupling characteristics of energy blocks and carbon blocks under different market conditions, and to provide quantitative analysis results for system optimization. Through confidence interval analysis, the coupling characteristics of energy blocks and carbon blocks can be quantitatively inferred.

[0232] In step G2 of this embodiment, the coupling characteristics of energy blocks and carbon blocks can be quantitatively inferred through confidence interval analysis; specifically, the optimal bidding strategy of market participants is described through a game theory model, and when the bids in each market meet the Nash equilibrium condition, the optimal bidding strategy of energy blocks and carbon blocks in different markets is obtained by optimizing the utility function that characterizes the total revenue of the energy market and the carbon market;

[0233] Among them, the Nash equilibrium condition is expressed as:

[0234]

[0235] Among them, U m (b m ,b -m ) represents the utility function of market m under other market bidding bm, which is expressed as:

[0236]

[0237] Where b m Indicates price, b -m Indicates other prices, E m (b m ) represents the potential field m and b m The relevant expected utility, C m (b m ) represents the relationship between market m and b m The associated costs, p m Represents the price of market m.

[0238] Through the above-mentioned optimization utility function, the bidding strategies of energy blocks and carbon blocks in different markets can be obtained, and the coupling analysis of energy and carbon markets can be realized.

[0239] In step G3 of this embodiment, in order to improve the flexibility and adaptability of energy blocks and carbon blocks in market transactions, multi-agent reinforcement learning technology is introduced to involve an adaptive trading auxiliary decision-making model; in this model, the interaction between multiple agents is continuously improved through reinforcement learning, so that the system can adapt to different market changes and make the best decision in a complex trading environment. Through continuous strategy updates and optimization, this method can achieve efficient scheduling and trading of energy blocks and carbon blocks in a multi-market environment, and enhance the market responsiveness and profitability of the system.

[0240] For example, a multi-agent reinforcement learning (MARL) model is introduced. Assume that there are N agents, and the state of each agent i is s i (t), the action is a i (t), the reward is γi (t). The goal of reinforcement learning is to use the strategy π i Make each agent maximize its cumulative reward R i :

[0241]

[0242] In this embodiment, the multi-agent Tsinghua learning method is introduced to realize the linkage optimization of the adaptive trading strategies of energy blocks and carbon blocks. Through the continuous interaction between agents, the reinforcement learning system can adaptively adjust the strategy to optimize the market transactions of energy blocks and carbon blocks. Ultimately, this multi-agent system can achieve the best scheduling and trading decisions in a complex market environment.

[0243] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0244] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for optimizing a green hydrogen metallurgical integrated energy system, characterized in that: include: Aiming at the multi-energy and mass flow characteristics in the green hydrogen metallurgical integrated energy system, a multi-energy and mass flow network matrix model based on differential dynamics equations that unifies heterogeneous energy flows and material flows is constructed, and the multi-energy and mass flow distribution calculation is performed based on it; Analyze the multi-time scale characteristics of flexible and adjustable resources in the green hydrogen metallurgical integrated energy system, construct an extended resource-task network model that takes into account the distribution of multiple energy and mass flows, and combine the multi-model predictive control mechanism to coordinate the multi-link regulation of energy and mass flows; Based on the uncertainty variables in the green hydrogen metallurgical integrated energy system, a robust optimization method is used to construct a multi-objective optimization operation model and randomly generate massive operation scenarios. The energy efficiency-flexibility-risk dynamic quantitative analysis of the operation scenario data is carried out through the oblique decision tree model. According to the results of the energy efficiency-flexibility-risk dynamic quantitative analysis, a multi-market adaptive transaction auxiliary decision-making system based on a two-layer optimization model is constructed. The scheduling and trading strategies of the energy blocks and carbon blocks are output to achieve efficient operation of the system. The energy blocks and carbon blocks represent the physical energy transmission and carbon emission quota in the green hydrogen metallurgical integrated energy system, respectively. During the system optimization process, the multi-energy mass flow distribution calculated by the multi-energy mass flow network matrix model provides the required multi-energy mass flow distribution state for constructing the extended resource-task network model; the extended resource-task network model serves as a constraint condition for constructing a multi-objective optimization operation model; the energy mass flow regulation strategy obtained by the multi-model predictive control mechanism serves as the production regulation strategy required for constructing the multi-energy mass flow network matrix model and generating massive operation scenarios; the scheduling and trading strategy of energy blocks and carbon blocks output by the multi-market adaptive transaction auxiliary decision-making system provides optimized decision information for the energy mass flow regulation strategy output by the multi-model predictive control mechanism; Through the system optimization process, multi-energy and mass flow distribution calculation in seconds, multi-link coordinated regulation of energy and mass flow in minutes, and efficient operation of the system in hours are achieved.

2. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 1 is characterized in that: The method of constructing a multi-energy and material flow network matrix model based on differential dynamics equations to unify heterogeneous energy flows and material flows is as follows: S1. Analyze the characteristics of multi-energy-mass flow in the green hydrogen metallurgical integrated energy system, classify the physical quantities involved into intensive quantities and extensive quantities, and then construct a unified dynamic equation of multi-energy-mass flow based on differential dynamic equations; The multi-energy-mass-flow unified dynamic equation includes a multi-energy-mass-flow transmission unified dynamic equation and a multi-energy-mass-flow coupling transformation unified dynamic equation; S2. The constructed unified dynamic equations of multi-energy mass flow are subjected to static algebraic analysis based on Laplace transform and external port equivalence based on distributed parameter approximation in turn, and a unified model of multi-energy mass flow is constructed; The multi-energy-mass-flow unified model includes a multi-energy-mass-flow transmission unified model obtained by sequentially performing transmission static algebraic analysis and branch external port equivalence, and a multi-energy-mass-flow coupling transformation unified model obtained by sequentially performing energy-mass transformation static algebraic analysis and coupling node external port equivalence; S3. Based on the unified model of multi-energy mass flow, a dynamic characteristic model of the whole network of multi-energy mass flow network is constructed; S4. Reduce the dimension of the dynamic characteristic model of the multi-energy mass flow network to a boundary equivalent model that only contains key boundary nodes, and then construct its corresponding unified Jacobian matrix as the matrix model of the multi-energy mass flow network.

3. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 2 is characterized in that: In step S1, in the constructed unified dynamics equation of multi-energy-mass flow, the spatial differential of the intensity quantity and the extension quantity is used to represent the left side of the equal sign, and the time differential of the intensity quantity and the extension quantity and related terms are used to represent the right side of the equal sign, which is expressed as follows: Where ψ represents the generalized intensity quantity, ζ represents the extension quantity, and α1, α2, β1 and β2 all represent branch parameters; The step S4 is specifically as follows: According to the importance and physical properties of network nodes, they are divided into key boundary nodes and non-key nodes. The intensity and extension injection values ​​of the nodes involved in the dynamic characteristics model of the multi-energy mass flow network are subjected to matrix operations according to the divided node types, and a block matrix is ​​constructed. The constructed block matrix is ​​subjected to Gaussian elimination to obtain a boundary equivalent model containing only key boundary nodes, and then the corresponding unified Jacobian matrix is ​​constructed as the matrix model of the multi-energy mass flow network. Among them, the constructed block matrix is ​​expressed as: Where Y BB , Y BI , Y IB , Y II represents the block matrix of Y, Y represents the energy node admittance matrix, ψ B (s) and ψ I (s) represents two sets of node strength quantities divided according to node types, ζ B (s) and ζ I (s) represents two sets of values ​​injected into the extensive quantity divided according to the node type; The constructed boundary equivalent model is expressed as: The constructed unified Jacobian matrix is ​​expressed as: In the formula, It represents the extensive and intensive quantity characterization of process J in the green hydrogen metallurgical integrated energy system. The subscripts m and n represent the parameter index in the green hydrogen metallurgical integrated energy system.

4. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 1 is characterized in that: The method of constructing an extended resource-task network model taking into account the distribution of multiple energy and mass flows is as follows: X1. Analyze the regulation response characteristics of energy supply, demand, energy conversion and storage links in the green hydrogen metallurgical integrated energy system at different time scales, construct a regulation response time scale matrix, and analyze the cross-link regulation characteristics in multi-energy interaction based on it, and construct an equivalent model of regulation external characteristics based on heterogeneous links; X2. Based on the equivalent model of external regulation characteristics corresponding to different regulation characteristics of various types of equipment, dynamic correlation analysis is carried out to analyze the impact of different production tasks on the distribution of energy and mass flow, and an extended resource-task network model is established to characterize the relationship between production tasks and the distribution of supply and demand of multiple energy and mass flows.

5. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 4 is characterized in that: The step X2 comprises the following sub-steps: X21. Classify the production tasks in the green hydrogen metallurgical integrated energy system according to production demand, determine the demand characteristics of various production tasks for energy and mass flow in different periods, fit their historical data, establish dynamic demand curves for various production tasks, and then produce multiple association models for different types of production tasks; X22. Model the energy quality demand characteristics of each production process of each production task in the green hydrogen metallurgical integrated energy system through dynamic programming or optimization algorithm, adjust the execution order and energy quality flow distribution of the production process of the production task in real time, so that the coordinated scheduling between production processes is optimal, and construct a coordinated regulation model of energy quality flow between processes; The inter-process energy and mass flow collaborative control model takes the time scheduling cost between production processes, the dependency between production processes and resource limitations as constraints; X23. Based on the constructed production task dynamic association model and the energy-mass flow coordinated control model between processes, and by introducing the dynamic characteristics of time and space energy-mass flow, an extended resource-task network model is constructed; The extended resource-task network model is a model for dynamically adjusting resource supply and task demand in the green hydrogen metallurgical integrated energy system according to time series data.

6. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 4 is characterized in that: Based on the multi-model predictive control mechanism, the method of coordinating the multi-link energy and mass flow and outputting the energy and mass flow control strategy is as follows: Based on the task-resource allocation results obtained by the extended resource-task network model, the random variables involved in the production process at future moments are predicted and input into the multi-model predictive control mechanism to output the energy and mass flow control strategy for the current working conditions; The multi-model predictive control mechanism includes a multi-model predictive mechanism based on a combination of local models and a multi-model predictive mechanism based on a combination of multiple controllers; The multi-model prediction mechanism based on local model combination refers to: Construct a local model for predicting energy and mass flow distribution under different working conditions, and switch the optimal local model to perform corresponding energy and mass flow distribution prediction according to the operating status of the green hydrogen metallurgical integrated energy system based on the conditional model switching mechanism; The multi-model prediction mechanism based on multi-controller combination refers to: On the basis of the energy flow allocation strategy predicted by the multi-model prediction mechanism of the local model combination output, the corresponding independent controllers are designed according to different energy flow characteristics, and according to different working conditions and energy flow control requirements, a collaborative mechanism based on constraint optimization is adopted, and the global objective function is introduced to optimize the control target of each independent controller to obtain the optimal energy flow control strategy for the system as a whole; wherein the outputs of each independent controller are interrelated; The basic model of the multi-model predictive control mechanism is expressed as: In the formula, x(t+k|t) represents the predicted value of state x at time t+k, u(t+k|t) represents the predicted value of control input u at time t+k, Q and R are the weight matrices of state and control respectively. u(t) represents the control input function and N represents the number of predictions or time steps of the optimization process.

7. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 1, characterized in that: Based on the uncertainty variables in the green hydrogen metallurgical integrated energy system, a multi-objective optimization operation model is constructed using a robust optimization method to randomly generate massive operation scenarios. The method for dynamic quantitative analysis of energy efficiency, flexibility and risk of the operation scenario data is carried out through an oblique decision tree model as follows: T1. Characterize the uncertainty variables in the green hydrogen metallurgical integrated energy system as uncertainty sets through probability distribution fitting. Use robust optimization methods based on the uncertainty sets to build a multi-objective optimization operation model for random production simulation. Use continuous convex envelope approximation strategy to transform the multi-objective optimization operation model with nonlinear constraints into a linear model. The uncertain variables include power supply fluctuations and market price fluctuations; the constraints of the multi-objective optimization operation model are an extended resource-task network model of the green hydrogen metallurgical integrated energy system and multi-energy and mass flow distribution; T2. Randomly generate uncertainty variables contained in different operation scenarios, and take maximizing the system's operating efficiency and minimizing risks as the optimization goal of the operation scenario, thereby generating a large number of operation scenarios; T3. Analyze the generated operation scenarios, extract the main features that affect the key indicators of system operation, construct an oblique decision tree based on multi-indicator sparse weights, and regularize its sparse weights to obtain an oblique decision tree model; The key indicators mentioned include energy efficiency, flexibility and risk; T4. Use the oblique decision tree model to extract the association rules between energy efficiency, flexibility and risk in the operating scenario data to achieve dynamic quantitative analysis of energy efficiency-flexibility-risk.

8. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 7, characterized in that: In step T1, the multi-objective optimization operation model for realizing random production simulation is expressed as: In the formula, x is the decision variable of the system, ξ is the uncertainty variable, f(x,ξ) is the profit and loss function of the system corresponding to the optimization target, and θ represents the robustness parameter, which is used to control the trade-off between the desirability and conservatism of the system. Indicates based on The expected function of represents the uncertainty set, P ξ represents the probability distribution of the uncertainty variable ξ, represents the reference distribution obtained by fitting the operational history data; In step T2, the optimization target of the running scenario is expressed as: In the formula, x s represents the decision variable of running scenario s, f(x s ,ξ s ,η s ) represents the system’s operating performance under operating scenario s, R(x s ,ξ s ,η s ) represents the system risk, η s and s They represent the power supply fluctuation and market price fluctuation in the operating scenario s respectively, λ represents the weight coefficient for balancing system performance and risk, s∈S, and S represents the operating scenario set.

9. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 1, characterized in that: The two-layer optimization model constructed based on the results of the dynamic quantitative analysis of energy efficiency-flexibility-risk includes an upper model and a lower model; The optimization goal of the upper model is to minimize the energy cost and carbon emission cost of the system while meeting the production demand; the optimization goal of the lower model is to purchase the required energy and carbon emission quotas at the lowest price in the energy market and carbon market according to the resource scheduling requirements of the upper model, or sell the excess energy and carbon quotas to obtain profits. The objective function of the upper model is: Among them, C energy (P) represents the cost of energy blocks, represents the cost of carbon blocks, P is the energy consumption of the system, is carbon emissions; The constraints of the upper model include energy balance constraints and carbon emission constraints, which are respectively expressed as: P generation +P import =P demand +P storage Where P generation represents the local power generation, P import represents the energy imported from the market, P demand and P storage represent the demand and storage capacity respectively, Carbon credits purchased for the system; The objective function of the lower model is expressed as: In the formula, λ energy (t) represents the price of the energy market in the tth period, λ carbon (t) is the price of carbon market, P import (t) is the energy purchased from the market in period t, E buy (t) and E sell (t) are the carbon emission quotas purchased and sold in period t; The constraints of the lower model include market supply and demand balance constraints and carbon market rating constraints, which are respectively expressed as: P import (t)+P local_generation (t)=P demand (t) Where P import (t) represents the energy imported from the market in period t, P local_generation (t) represents the energy generated locally by the system in the tth period, P demand (t) represents the demand in the tth period, represents the carbon dioxide emissions at time t, represents the capital cost or constraint on CO2 emissions at time t.

10. The optimization method of the green hydrogen metallurgical integrated energy system according to claim 9, characterized in that: A multi-market adaptive trading decision-making auxiliary system based on a two-layer optimization model is constructed, and the method of outputting the scheduling and trading strategies of energy blocks and carbon blocks is as follows: G1. Based on the results of the dynamic quantitative analysis of energy efficiency, flexibility and risk, the time series coupling analysis and confidence interval analysis of the energy block and the carbon block are carried out to quantitatively infer the coupling characteristics of the energy block and the carbon block; G2. Based on the inferred coupling characteristics of energy blocks and carbon blocks, the optimal bidding strategy of energy blocks and carbon blocks in market transactions is explored, and combined with the constructed two-layer optimization model to form power trading agents, carbon emission rights trading agents and green hydrogen trading agents; G3. Based on the power trading agent, carbon emission trading agent and green hydrogen trading agent, a multi-agent reinforcement learning adaptive trading auxiliary decision-making system is constructed to continuously learn and optimize the trading strategies of energy blocks and carbon blocks to adapt to different market changes and output the scheduling and trading strategies of energy blocks and carbon blocks; The optimization goal of the adaptive trading auxiliary decision-making system is to maximize the cumulative reward of each intelligent agent through the scheduling and trading strategies of the output energy blocks and carbon blocks.

Images