A multi-objective collaborative optimization method

Abstract

This application discloses a multi-objective collaborative optimization method. First, a multi-objective collaborative optimization model for a collaborative energy system is established. The decision variables of this model comprehensively cover unit output, energy storage charging and discharging power and state of charge, and carbon capture energy consumption and capture rate. The objective function simultaneously minimizes system operating costs, total carbon emissions, energy loss rate, and load response deviation. The constraints finely characterize power balance, equipment operating limits, and inter-system coupling relationships. Second, a multi-objective optimization problem is constructed based on this model and solved using a multi-objective evolutionary algorithm based on K-means clustering fitness strategy to generate a high-quality Pareto solution set. Finally, a scheduling scheme is selected from the Pareto solution set, and scheduling instructions are output to the source, storage, and carbon physical equipment. This achieves comprehensive coordinated optimization of the energy system among multiple objectives such as economy, low carbon emissions, energy efficiency, and response accuracy.

Description

Technical Field

[0001] This application relates to the field of multi-objective optimization technology, and more specifically, to a multi-objective collaborative optimization method. Background Technology

[0002] Driven by the dual carbon goals, building a clean, low-carbon, safe, and efficient new power system is a crucial direction for the current energy sector transformation. Integrating traditional coal-fired and gas-fired power units, fast-response electrochemical energy storage arrays, and carbon capture systems into a unified system can synergistically leverage the supporting role of traditional power sources, the flexible regulation capabilities of energy storage, and the emission reduction benefits of carbon capture. This is considered an important technological path that balances energy security and green transformation. Through deep interaction of energy, materials, and control, this coupled system is expected to significantly reduce carbon emissions while ensuring a stable power supply, demonstrating significant application potential.

[0003] Currently, research on modeling and operational optimization of this coupled system still has many shortcomings. First, existing methods typically treat each subsystem as an independent entity for separate modeling and optimization, neglecting the close physical coupling and operational constraints between them. For example, they fail to fully consider the impact of carbon capture system energy consumption on the unit's net output, the coordination between energy storage charging and discharging and unit regulation, and the chain effects caused by dynamic changes in carbon capture rate. This can lead to the optimization scheme facing the risk of power imbalance or equipment overload during actual coordinated operation. Second, optimization objectives are often too simplistic, focusing primarily on minimizing economic costs or simply weighting multiple objectives such as economy, environmental protection, and reliability. This makes it difficult to truly reflect and balance the complex trade-offs between multiple dimensions of indicators, resulting in poor overall performance of the obtained schemes under strict carbon constraints. Furthermore, existing models are relatively simplified in setting decision variables and characterizing constraints, often neglecting key control variables such as the dynamic capture rate of carbon capture systems and the fine state of charge of energy storage systems. They also lack precise descriptions of the underlying mechanisms such as the dynamic process of energy storage and the coupling relationship between carbon capture energy consumption and capture rate, resulting in insufficient model degrees of freedom and limited feasibility and potential for optimization results.

[0004] Furthermore, at the algorithmic level, traditional single-objective optimization algorithms cannot handle multi-objective conflicts, while conventional multi-objective evolutionary algorithms often suffer from slow convergence, susceptibility to local optima, and poor solution set distribution when facing high-dimensional, nonlinear, and strongly constrained optimization problems in this system. These issues make it difficult to provide a series of high-quality, high-precision Pareto optimal solutions within a limited timeframe, thus hindering the practical engineering application of optimization techniques for coupled systems and the improvement of overall efficiency. Therefore, there is an urgent need to develop a new modeling and optimization method that can accurately characterize the coupling characteristics of the system, comprehensively consider multi-objective trade-offs, and possess efficient solution capabilities. Summary of the Invention

[0005] The purpose of this application is to overcome the shortcomings of existing technologies and provide a multi-objective collaborative optimization method. This method involves constructing a collaborative optimization model that includes multi-dimensional decision variables, multi-objective functions, and deeply coupled constraints between energy flow, material flow, and control flow. A K-based approach is then employed. MOEA, a multi-objective optimization algorithm for clustering means The CFS solution addresses the problems in existing technologies that neglect the deep coupling between coal-fired power plant and gas turbine clusters, electrochemical energy storage arrays and carbon capture systems, as well as the inability to achieve multi-objective synergistic optimization of economy, low carbon, energy efficiency and response due to single-objective optimization.

[0006] The objective of this application is achieved through the following technical solution: Firstly, this application proposes a multi-objective collaborative optimization method, which is applied to a smart power generation system, and the method includes: A multi-objective collaborative optimization model is established based on the collaborative energy system. The multi-objective collaborative optimization model includes decision variables, objective function and constraints. Based on the objective function and constraints, construct a multi-objective optimization problem; A multi-objective optimization algorithm is used to solve the multi-objective optimization problem and generate a Pareto solution set; Select a scheduling scheme from the Pareto solution set and output scheduling instructions to the physical devices of the collaborative energy system.

[0007] In one possible implementation, the decision variables include the output of coal-fired power and gas turbine units, the power of the electrochemical energy storage system, the energy consumption of the carbon capture system, the capture rate of the carbon capture system, and the state of charge of the electrochemical energy storage system. The objective function includes system operating costs, total system carbon emissions, system energy loss rate, and system load response deviation; The constraints include power balance constraints, upper and lower limits of unit output constraints, unit ramp rate constraints, energy storage power constraints, energy storage SOC state transition constraints, upper and lower limits of energy storage SOC constraints, upper and lower limits of carbon capture system capture rate constraints, and carbon capture system energy consumption-capture rate coupling constraints.

[0008] In one possible implementation, the total system operating cost includes unit fuel cost, energy storage operation and maintenance cost, carbon capture system operating cost, and carbon trading cost. The total carbon emissions of the system are the difference between the total carbon emissions of the generating unit and the total carbon capture of the carbon capture system. The system energy loss rate is calculated as the ratio of the sum of energy loss during the energy storage charging and discharging process and the energy consumption of the carbon capture system to the total power generation. The system load response deviation is the cumulative deviation between the system's total output power and the grid load demand.

[0009] In one possible implementation, the multi-objective optimization algorithm is the MOEA-CFS algorithm based on the K-means clustering algorithm. The MOEA-CFS algorithm constructs a mating pool through a clustering fitness strategy and adopts a multi-strategy mating mechanism.

[0010] In one possible implementation, the clustering fitness strategy is as follows: The K-means algorithm is used to divide the population into K clusters in the target space; The objective function value of each individual in the population is normalized. Calculate the fitness of each individual.

[0011] In one possible implementation, the fitness calculation formula is: , and For the number of iterations Dynamically adjusted weighting coefficients As the reference vector, Represents an individual With reference vector The included angle, Indicates and The angle between the reference vectors with the smallest included angle. Indicates the first Individuals at the number of iterations The normalized objective function value vector at time.

[0012] In one possible implementation, the multi-strategy mating mechanism integrates the SBX crossover operator, the BLX-α crossover operator, and the non-uniform mutation operator.

[0013] In one possible implementation, the step of selecting a scheduling scheme from the Pareto solution set includes: A compromise solution is selected from the Pareto solution set using the fuzzy membership method, which is then used as the final scheduling scheme.

[0014] In one possible implementation, the smart power generation system includes: The physical equipment layer includes coal-fired and gas-fired power units, electrochemical energy storage arrays, and carbon capture systems; The sensor network layer is used to collect the operating status parameters of the physical device layer in real time. The communication network layer is used to enable data transmission between the sensor network layer and the upper layers. The data middle platform layer is used to store real-time data, historical data, and model parameters; The algorithm engine layer is used to manage the Pareto front solution set; The application interaction layer is used to output scheduling instructions, perform visual monitoring, and interact with external systems.

[0015] In one possible implementation, the step of outputting scheduling instructions to the physical devices of the coordinated energy system includes: It outputs the unit output setpoint to the coal-fired power plant and gas-fired power plant, the charge and discharge power setpoint to the electrochemical energy storage array, and the capture rate setpoint to the carbon capture system.

[0016] The main solution and its various further alternatives described above can be freely combined to form multiple solutions, all of which are solutions that can be adopted and are claimed in this application; furthermore, the (non-conflicting alternatives) can also be freely combined with each other and with other alternatives. Those skilled in the art, after understanding the solution of this application, will realize from the prior art and common general knowledge that there are many combinations, all of which are technical solutions to be protected in this application, and will not be exhaustively listed here.

[0017] This application discloses a multi-objective collaborative optimization method. First, a multi-objective collaborative optimization model for a collaborative energy system is established. The decision variables of this model comprehensively cover unit output, energy storage charging and discharging power and state of charge, and carbon capture energy consumption and capture rate. The objective function simultaneously minimizes system operating costs, total carbon emissions, energy loss rate, and load response deviation. The constraints finely characterize power balance, equipment operating limits, and inter-system coupling relationships. Second, a multi-objective optimization problem is constructed based on this model and solved using a multi-objective evolutionary algorithm based on K-means clustering fitness strategy to generate a high-quality Pareto solution set. Finally, a scheduling scheme is selected from the Pareto solution set, and scheduling instructions are output to the source, storage, and carbon physical equipment. This achieves comprehensive coordinated optimization of the energy system among multiple objectives such as economy, low carbon emissions, energy efficiency, and response accuracy. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A schematic diagram of a multi-objective collaborative optimization method proposed in an embodiment of this application is shown.

[0020] Figure 2 The flowchart illustrating the collaborative modeling and optimization process for a coal-fired power plant-gas turbine cluster-electrochemical energy storage array-carbon capture energy system proposed in this application embodiment is shown.

[0021] Figure 3 The flowchart of the multi-objective optimization algorithm MOEA-CFS is shown.

[0022] Figure 4 The core flowchart of constructing a mating pool using the clustering fitness strategy proposed in this application embodiment is shown. Detailed Implementation

[0023] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0024] Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this application.

[0025] In existing technologies, the current coal-fired power plant and gas turbine clusters, electrochemical energy storage arrays, and carbon capture systems suffer from serious modeling defects in actual modeling and optimization. On the one hand, existing methods often treat the three as independent subsystems for decoupled modeling and optimization, ignoring the deep coupling relationship between energy flow, material flow, and control flow. This fails to accurately reflect the chain reaction of unit output changes, carbon capture energy consumption fluctuations, and energy storage regulation demands, easily leading to power imbalance or equipment overload. On the other hand, the optimization objectives are too singular, often using single objectives such as minimum cost or weighted summation methods, failing to consider economy, low carbon emissions, efficiency, and response accuracy. Furthermore, the decision variable dimension is lacking, often ignoring key variables such as dynamic carbon capture rate and energy storage state of charge. The constraints also lack a detailed characterization of energy storage SOC state transition, SOC limits, and the coupling relationship between carbon capture energy consumption and capture rate, causing infeasibility issues such as SOC exceeding limits and mismatch between energy consumption and capture rate in actual operation.

[0026] At the algorithmic level, traditional single-objective algorithms are inherently unable to handle multi-dimensional conflicting objectives. Forced weighted summation will lose crucial Pareto front information and cannot provide diverse trade-off solutions. On the other hand, conventional multi-objective evolutionary algorithms often have slow convergence speeds, are prone to getting trapped in local optima, and have uneven solution set distributions when facing the strong nonlinearity, high-dimensional decision space, and complex coupling constraints of power generation-storage-carbon capture systems. They are difficult to obtain high-quality, high-precision non-dominated solution sets within a limited computation time and cannot meet the dual high standards of efficiency and accuracy required by engineering practice.

[0027] To address the aforementioned technical issues, this application proposes a multi-objective collaborative optimization method. This method establishes a mathematical model that deeply couples energy flow, material flow, and control flow among the three entities, along with a high-performance multi-objective optimization algorithm. It constructs a scheduling framework for multi-objective collaborative optimization of source-storage-carbon, achieving comprehensive coordinated optimization of system economy, low carbon emissions, energy efficiency, and response accuracy. This provides decision-makers with diverse optimization schemes and effectively improves the system's operational flexibility, low carbon emissions, and reliability.

[0028] Please refer to Figure 1 , Figure 1 The diagram illustrates a multi-objective collaborative optimization method proposed in an embodiment of this application. This method is applied to a smart power generation system and includes: Step S1: Establish a multi-objective collaborative optimization model based on the collaborative energy system. The multi-objective collaborative optimization model includes decision variables, objective function, and constraints. Step S1 constructs a comprehensive, detailed, and tightly coupled mathematical model. This model first establishes decision variables encompassing core system operation control parameters, including the active power output of coal-fired and gas-fired power units, the charging and discharging power of the electrochemical energy storage system, the operating energy consumption and dynamic capture rate of the carbon capture system, and the state of charge of the electrochemical energy storage system. Secondly, the model defines a multi-dimensional objective function that considers economy, environmental protection, energy efficiency, and reliability, specifically the total system operating cost, total system carbon emissions, system energy loss rate, and system load response deviation. Finally, the model constructs a detailed and physically consistent constraint system, including power balance, unit output and ramp-up limits, dynamic constraints on energy storage power and state of charge, carbon capture rate limits, and crucial carbon capture system energy consumption-capture rate coupling constraints, thereby ensuring the feasibility and optimality of the optimization scheme in actual operation.

[0029] The decision variables include the output of coal-fired and gas-fired power units, the power of electrochemical energy storage systems, the energy consumption of carbon capture systems, the capture rate of carbon capture systems, and the state of charge of electrochemical energy storage systems.

[0030] The output of the coal-fired and gas-fired power units is , Indicates the first Taiwan's coal-fired or gas-fired power units in the first Active power output over a given time period, measured in megawatts (MW), refers to the real-time active power output of a coal-fired and gas-fired power generation unit group during power generation. It is a core parameter of the system's energy supply. Its dynamic changes directly determine the net output level and regulation rate of the unit group, while also providing necessary power supply and high-temperature thermal energy support for the carbon capture system. It is a key variable for coordinating the coupled operation of the unit group and the carbon capture system, directly affecting the overall operating efficiency and carbon emission intensity of the system.

[0031] The power of the electrochemical energy storage system is ,in Indicates the first The energy storage unit in the first The charging and discharging power during a given period, measured in megawatts, is a continuous and directional variable: positive values ​​indicate that the energy storage system is discharging into the grid, while negative values ​​indicate that it is charging from the grid. This power is used to smooth out instantaneous power disturbances caused by grid load fluctuations and changes in energy consumption of the carbon capture system. Its millisecond-level response characteristic allows the system to quickly adjust its output, effectively alleviating peak-shaving pressure on coal-fired and gas-fired power units, reducing the frequency of unit start-ups and shutdowns, and improving the accuracy of grid dispatching and the flexibility of system operation.

[0032] The energy consumption of the carbon capture system is ,in Indicates the first The carbon capture device in the first The operating energy consumption during a given period is measured in megawatts (MW). During operation, the carbon capture system consumes electrical energy and high-temperature heat for normal system operation and solvent regeneration. This energy consumption is deeply coupled with the output of coal-fired and gas-fired power units, significantly affecting the net output and regulation capability of the unit group. It is the core constraint of the system's energy coupling, and its dynamic changes directly drive the regulation needs of the energy storage system to maintain power balance.

[0033] The carbon capture system has a capture rate of ,in Indicates the first The carbon capture device in the first The capture rate over a given period is a dimensionless parameter. The capture rate of a carbon capture system refers to the efficiency with which the system captures CO2 from the flue gas emitted by a group of generating units, directly impacting the system's net carbon emissions. This invention introduces it as a dynamic decision variable, breaking through the limitations of traditional fixed capture rates. This allows the optimization process to adaptively adjust the capture intensity based on the unit's output and energy consumption status, thereby accurately quantifying the carbon reduction effect and optimizing the system's low-carbon objectives.

[0034] The state of charge of the electrochemical energy storage system is ,in Indicates the first The energy storage unit in the first The state of charge (SOC) at the end of the time period is a dimensionless parameter. This parameter refers to the SOC of the electrochemical energy storage system, i.e., the current percentage of remaining charge in the energy storage unit. As a core decision variable, this state directly characterizes the energy storage and release capacity of the energy storage system. Its dynamic changes affect the energy throughput and available capacity throughout the entire energy storage cycle, and it is a key parameter for achieving fine-grained control of the energy storage system, avoiding SOC exceeding limits, and improving energy time-shift efficiency.

[0035] The objective function includes system operating cost, total system carbon emissions, system energy loss rate, and system load response deviation.

[0036] The total operating cost of the system includes the unit's fuel cost, energy storage operation and maintenance cost, carbon capture system operation cost, and carbon trading cost; The system operating cost is: ; ; ; ; ; in This represents the total system operating cost within the scheduling period, expressed in yuan, including unit fuel costs. (A quadratic function of unit fuel cost, where) , , (Cost coefficient), energy storage operation and maintenance costs (Proportional to the absolute value of charging and discharging power) (Per unit power operation and maintenance cost coefficient), operating cost of carbon capture system (The first item is the cost of capture, which is directly proportional to the amount captured.) This is the cost coefficient. flue gas to be treated Total amount; the second item is energy consumption cost. (Cost coefficient), carbon trading cost ( For carbon prices, (The four items are: unit carbon emission intensity (net carbon emissions in parentheses)). It is a core indicator for measuring the economic efficiency of a system.

[0037] Total carbon emissions of the system are the difference between the total carbon emissions of the generating unit and the total carbon capture of the carbon capture system.

[0038] The total carbon emissions of the system are ,in The total net carbon emissions of the system during the scheduling period, in units of It reflects the low-carbon nature of the system and is a direct quantitative indicator for evaluating the low-carbon performance of the system.

[0039] The system energy loss rate is calculated as the ratio of the sum of energy loss during the energy storage charging and discharging process and the energy consumption of the carbon capture system to the total power generation.

[0040] The system energy loss rate is in This represents the proportion of energy lost during energy conversion and transmission in the system. It primarily covers energy losses during the charging and discharging of the electrochemical energy storage system, as well as additional energy losses due to the impact of carbon capture system operating energy consumption on unit processing. It is a key indicator for measuring system energy efficiency. Its numerator is the sum of energy storage charging and discharging losses and carbon capture system energy consumption, and the denominator is the total power generation. The smaller the value, the higher the system's energy efficiency.

[0041] The system load response deviation is the cumulative deviation between the system's total output power and the grid load demand.

[0042] System load response deviation is ,in This indicates the system's output power and grid load demand. The cumulative value of the instantaneous deviation between the two, in megawatts, reflects the accuracy and stability of the system's response to power grid dispatch commands.

[0043] The constraints include power balance constraints, upper and lower limits of unit output constraints, unit ramp rate constraints, energy storage power constraints, energy storage SOC state transition constraints, upper and lower limits of energy storage SOC constraints, upper and lower limits of carbon capture system capture rate constraints, and carbon capture system energy consumption-capture rate coupling constraints.

[0044] Power balance constraint is The power balance constraint requires that, within any given time period, the sum of the output of all coal-fired and gas-fired power plants and the charging and discharging power of the electrochemical energy storage system must equal the sum of the grid load demand and the energy consumption of the carbon capture system during that time period. Ensuring the power balance between power generation, energy storage, load, and the carbon capture system is the primary prerequisite for the feasibility of the dispatch optimization scheme.

[0045] Unit output upper and lower limit constraints are The upper and lower limit constraints of the unit stipulate that the active power output of each unit must be between its technically permissible minimum stable output and rated maximum output. This is a rigid constraint determined by the physical characteristics of the unit, in order to ensure that the unit operates stably within a safe and efficient operating range.

[0046] The unit's ramp rate constraint is The unit ramp-up rate constraint limits the maximum change in unit output within adjacent time periods, reflecting the unit's dynamic adjustment capability in response to load changes. This prevents equipment wear or control instability caused by over-speed adjustment and ensures the feasibility of scheduling optimization results in actual implementation.

[0047] Energy storage power constraint is The energy storage power constraint stipulates that the charging and discharging power of the energy storage system at each time period shall not exceed its technically permissible maximum charging power and maximum discharging power. This is to protect the physical limits of the energy storage system, ensure that it operates within a safe range, and avoid the risks of overcharging and over-discharging.

[0048] Energy storage SOC state transition constraints are in Indicates the rated capacity of energy storage. The time period is defined as . The energy storage SOC state transition constraint establishes the mathematical relationship between the energy storage state of charge (SOC) and the charging / discharging power within adjacent time periods through the state transition equation, describing the dynamic change of the state of charge over time.

[0049] Energy storage SOC upper and lower limits constraints are The upper and lower limits of the energy storage SOC require that the energy storage system maintain its state of charge between the minimum and maximum values ​​at the end of each period. This is to protect the operating depth of the energy storage battery, prevent overcharging and discharging from damaging the battery life, and reserve the necessary adjustment margin to cope with the uncertainties in subsequent periods.

[0050] The upper and lower limits of the capture rate of the carbon capture system are constrained. .

[0051] The upper and lower limits of the capture rate of a carbon capture system stipulate that the CO2 capture rate of the carbon capture device at each time period must be between its technically permissible minimum and maximum capture rate. It reflects the actual adjustment range of the carbon capture process, ensuring that the capture system operates within its design conditions and avoiding exceeding the equipment's performance limits.

[0052] The energy consumption-capture rate coupling constraint of the carbon capture system is ,in This represents the energy consumption required per unit of carbon capture. The energy consumption-capture rate coupling constraint of the carbon capture system establishes a quantitative relationship between the energy consumption and capture rate of the carbon capture system, reflects the physical nature of the carbon capture process, ensures that energy consumption matches the capture effect, and prevents unreasonable operating conditions of high capture rate but low energy consumption.

[0053] Step S2: Based on the objective function and constraints, construct a multi-objective optimization problem.

[0054] Based on the refined mathematical model established in step S1, the four objective functions (total system operating cost f1, total carbon emissions f2, energy loss rate f3, and load response deviation f4) covering economy, low carbon emissions, efficiency, and responsiveness are integrated into a multi-objective vector MinF(x)=[f1,f2,f3,f4] that needs to be minimized simultaneously. Eight types of constraints, including power balance, unit operation limitations, energy storage dynamics, and physical coupling of the carbon capture system, are also included as the constraint set of the optimization problem. Thus, a practical "source-storage-carbon" coordinated scheduling engineering problem is transformed into a multi-objective optimization problem with a clear mathematical form, containing high-dimensional decision variables and complex constraints.

[0055] Step S3: Use a multi-objective optimization algorithm to solve the multi-objective optimization problem and generate a Pareto solution set.

[0056] The multi-objective optimization problem constructed in step S2 is solved using the multi-objective optimization algorithm MOEA-CFS based on K-means clustering proposed in this invention. This algorithm constructs a mating pool through a clustering fitness strategy to maintain population diversity and convergence performance, and integrates multiple mating strategies such as SBX crossover, BLX-α crossover, and non-uniform mutation to improve search efficiency. During the iterative evolution process, the algorithm continuously generates non-dominated solutions and maintains an external archive through non-dominated sorting to manage the Pareto front. Finally, after satisfying the termination condition, a high-quality Pareto front solution set is output.

[0057] The multi-objective optimization algorithm is the MOEA-CFS algorithm based on the K-means clustering algorithm. The MOEA-CFS algorithm constructs a mating pool through a clustering fitness strategy and adopts a multi-strategy mating mechanism.

[0058] Figure 2 The flowchart illustrating the collaborative modeling and optimization process for a coal-fired power plant-gas turbine cluster-electrochemical energy storage array-carbon capture energy system proposed in this application embodiment is shown.

[0059] Figure 3The flowchart of the multi-objective optimization algorithm MOEA-CFS is shown. The algorithm starts by generating an initial population P_it and setting parameters, then enters the main loop. In each iteration, the algorithm first applies BLX-α crossover and non-uniform mutation operators to the current population P_it to generate offspring Q1, enhancing global exploration capabilities. Next, P_it and Q1 are merged, and the core clustering fitness strategy (CFS) is used to select high-quality individuals to form a mating pool M. Then, the algorithm merges P_it and M again, and applies SBX crossover and non-uniform mutation to generate offspring Q2, which focuses on local exploitation. Then, P_it and Q2 are merged to form a temporary population P, and N individuals are selected from it based on natural selection principles (such as non-dominated ranking and crowding calculation) to form a new generation population P_it. This process is repeated until the maximum number of iterations It_max is reached, and finally, the optimized Pareto front solution set is output. The entire process demonstrates the algorithm's innovative mechanism of effectively balancing global exploration and local exploitation through multi-stage, multi-operator collaborative search and the targeted selection of the CFS strategy.

[0060] The algorithm constructs a mating pool through a clustering fitness strategy. This strategy innovatively applies the K-means clustering algorithm to the population, dividing individuals into multiple clusters in the target space to maintain distribution diversity. It also combines a novel fitness index that integrates convergence and distribution to evaluate and select individuals, thus effectively guiding the search direction while ensuring population diversity. Secondly, the algorithm employs a multi-strategy mating mechanism that integrates the advantages of three genetic operations: SBX crossover, BLX-α crossover, and non-uniform mutation. By flexibly applying these strategies under different conditions or at different evolutionary stages, the algorithm enhances its global exploration and local exploitation capabilities in complex multi-objective optimization problems.

[0061] The clustering fitness strategy is as follows: The K-means algorithm is used to divide the population into K clusters in the target space; The objective function value of each individual in the population is normalized. Calculate the fitness of each individual.

[0062] The formula for calculating fitness is: , and For the number of iterations Dynamically adjusted weighting coefficients As the reference vector, Represents an individual With reference vector The included angle, Indicates and The angle between the reference vectors with the smallest included angle. Indicates the first Individuals at the number of iterations The normalized objective function value vector at time.

[0063] First, the K-means clustering algorithm is used to divide the current population into K clusters in the target space. The K-means algorithm uses the Euclidean distance between individuals in the target space as the classification criterion; the distance between any two individuals is calculated using the following formula: ,in The target number is [number of elements]. Clustering allows groups of individuals in similar locations within the target space to be grouped into the same cluster.

[0064] Secondly, to eliminate the influence of different objective function units and magnitudes, the objective function value of each individual in the population is normalized. The normalization formula is as follows: ,in and The current population is at the th The maximum and minimum values ​​of each objective function are determined. After normalization, all objective function values ​​are mapped to the interval [0,1], making subsequent fitness calculations more fair and reasonable.

[0065] Finally, the fitness of each individual is calculated. This fitness score comprehensively reflects an individual's convergence performance and diversity contribution. The calculation formula is: .in The magnitude of the individual normalized target vector represents its convergence (the smaller the magnitude, the closer it is to the ideal point). Represents an individual With reference vector The angle between them Indicates and The ratio of the smallest included angle between reference vectors with the smallest included angle. Used to measure the uniformity of individual distribution along the direction of the reference vector. Weighting coefficients. and With the number of iterations Dynamic adjustment, defined as follows: That is, the early stage of evolution. Larger, the algorithm focuses on convergence; later stages of evolution The population is relatively large, and the algorithm places greater emphasis on diversity. Through the above design, the CFS strategy can effectively guide the population to approach the Pareto front and distribute it evenly on the front.

[0066] Step S4: Select a scheduling scheme from the Pareto solution set and output scheduling instructions to the physical devices of the collaborative energy system.

[0067] The multi-strategy mating mechanism integrates the SBX crossover operator, the BLX-α crossover operator, and the non-uniform mutation operator.

[0068] The multi-strategy mating mechanism integrates the advantages of three genetic operation methods: SBX crossover operator, BLX-α crossover operator, and non-uniform mutation operator. The strategy for combining multiple genetic operators is as follows: SBX Crossover and BLX- Cross-strategy: Assumption and There are two The parent individual of wei, , The offspring calculated by the SBX crossover strategy are shown in the following formula: ,in It is determined by the distribution factor The dynamic parameters are determined, and the calculation formula is as follows: .

[0069] By BLX- The offspring calculated using the crossover strategy are shown in the following formula: .

[0070] Non-uniform mutation strategy: assumption and There are two The parent individual of wei, , The offspring obtained by the non-uniform mutation operator can be calculated using the following formula:

[0071] Where the function The calculation formula is as follows: .

[0072] The steps for selecting a scheduling scheme from the Pareto solution set include: A compromise solution is selected from the Pareto solution set using the fuzzy membership method, which is then used as the final scheduling scheme.

[0073] For each non-dominated solution in the Pareto solution set, its membership degree value on each objective function is calculated. Then, for each solution, its membership degree on each objective is weighted and summed or averaged to obtain the comprehensive satisfaction index of the solution. Finally, the solution with the highest comprehensive satisfaction is selected as the compromise solution output. This compromise solution can achieve the best balance among the four conflicting objectives of economy, low carbon emissions, efficiency and response accuracy.

[0074] Intelligent power generation systems include: The physical equipment layer includes coal-fired and gas-fired power units, electrochemical energy storage arrays, and carbon capture systems; The sensor network layer is used to collect the operating status parameters of the physical device layer in real time. The communication network layer is used to enable data transmission between the sensor network layer and the upper layers. The data middle platform layer is used to store real-time data, historical data, and model parameters; The algorithm engine layer is used to manage the Pareto front solution set; The application interaction layer is used to output scheduling instructions, perform visual monitoring, and interact with external systems.

[0075] The physical equipment layer includes power generation equipment such as coal-fired power units and gas-fired power units, electrochemical energy storage arrays such as lithium battery energy storage units, and carbon capture systems such as CCUS devices and CO2 compression and purification systems. This layer is the physical carrier of energy flow, material flow, and control flow. Coal-fired and gas-fired power units undertake baseload and peak-shaving tasks and provide the energy required for carbon capture. Electrochemical energy storage arrays are responsible for smoothing power fluctuations, and carbon capture systems realize the separation and capture of CO2 in flue gas.

[0076] The sensor network layer is used to collect the operating status parameters of the physical device layer in real time. This layer deploys an intelligent sensor array and is configured with a remote terminal unit (RTU) and a programmable logic controller (PLC) for data acquisition and instruction parsing. It also integrates data acquisition devices such as smart meters and carbon meters to achieve real-time monitoring of key parameters such as unit output, state of charge (SOC) of energy storage units, carbon capture rate of carbon capture devices, and carbon emissions.

[0077] The communication network layer is used to achieve low-latency, high-reliability data transmission between the sensor network layer and the upper layers. This layer uses industrial Ethernet, 5G private networks, OPC UA protocol conversion, and edge computing gateways to ensure that collected data can be uploaded to the data platform layer quickly and accurately, while reliably sending upper-layer scheduling commands to physical devices.

[0078] The data platform layer stores real-time data, historical data, and model parameters. This layer includes a real-time database, a historical database, and a model parameter library; the model parameter library records cost coefficients, constraint boundaries, and algorithm hyperparameters.

[0079] The algorithm engine layer, as the core optimization module of the system, is used to manage the Pareto front solution set and realize intelligent scheduling optimization. This layer first completes the collaborative optimization modeling of the intelligent power generation system, clarifying the decision variables, objective functions, and constraints; then it carries the MOEA-CFS multi-objective optimization algorithm, which maintains the external archive through non-dominated sorting to manage the Pareto front, and uses the fuzzy membership method to select compromise solutions from the Pareto solution set, thereby generating a high-quality non-dominated solution set within a limited computation time for the application interaction layer to call.

[0080] The application interaction layer is used to output dispatch commands, perform visual monitoring, and interact with external systems. It outputs dispatch commands such as unit output setpoints, energy storage charging and discharging power setpoints, and CCUS capture rate setpoints to physical equipment; provides visual monitoring functions such as a large-screen display of operating status, Pareto dynamic displays, and carbon emission reports; and connects to the power grid dispatch center, carbon trading platform, and electricity market system through external interfaces to access external information such as load forecast curves, carbon trading prices, electricity market prices, fuel prices, and weather forecasts in real time.

[0081] The steps for outputting scheduling instructions to the physical devices of the coordinated energy system include: It outputs the unit output setpoint to the coal-fired power plant and gas-fired power plant, the charge and discharge power setpoint to the electrochemical energy storage array, and the capture rate setpoint to the carbon capture system.

[0082] First, the power output setpoints of the coal-fired and gas-fired power units are output to precisely control the active power output of each unit during each period of the scheduling cycle, thereby meeting the grid load demand and providing the necessary energy support for the carbon capture system. Second, the charge and discharge power setpoints are output to the electrochemical energy storage array to smooth out instantaneous power disturbances caused by load fluctuations and changes in carbon capture energy consumption, while realizing time-shifted energy scheduling. Third, the capture rate setpoints are output to the carbon capture system to dynamically adjust the CO2 capture ratio, thereby achieving the optimal balance between carbon emission reduction and system energy consumption.

[0083] Figure 4 The diagram illustrates the core flowchart of the clustering fitness strategy proposed in this application for constructing a mating pool. Starting with input parameters such as the initial population P, size N, number of clusters K, and target number m, the strategy first initializes an empty mating pool M and an intermediate population FI for processing. Subsequently, the individuals in FI are normalized to their target values, and the comprehensive fitness value of each individual is calculated. Next, the K-means algorithm is used to divide the entire population into K different clusters based on the position of the individuals in the target space. Then, a cyclic selection process is entered—for each cluster, the number of individuals it contains is determined. If there are more than one, the individual with the smallest LCD value in the cluster is selected. If there is exactly one, the individual is directly selected and added to the mating pool M. After traversing all clusters, a mating pool M with high quality and good distribution, composed of representative individuals from each cluster, is finally output.

[0084] Compared with the prior art, the embodiments of this application have the following beneficial effects: First, this scheme fully depicts the coupling relationship of energy flow, material flow, and control flow among coal-fired and gas-fired power units, electrochemical energy storage arrays, and carbon capture systems. It can accurately reflect the chain reaction of unit output changes, carbon capture energy consumption fluctuations, and energy storage regulation needs. This avoids the risks of power imbalance or equipment overload caused by traditional decoupling modeling, and truly achieves source-storage-carbon synergistic optimization.

[0085] Secondly, by constructing a multi-objective optimization function encompassing four dimensions—economic efficiency, low carbon emissions, efficiency, and responsiveness—this effectively fills the technical gap in existing models that only focus on a single objective or weighted summation. This design avoids high-carbon emission strategies caused by a single economic objective, while quantifying the impact of energy loss during energy storage charging and discharging on load tracking deviation.

[0086] Third, by introducing multi-dimensional decision variables such as the dynamic output of coal-fired and gas-fired power units, the state of charge of energy storage, and the dynamic capture rate of carbon capture, the system's degree of flexibility in regulation is significantly improved. This ensures that carbon capture energy consumption and the time-shifting capability of energy storage can be precisely controlled, further unlocking economic and environmental benefits.

[0087] Fourth, the optimization model of this scheme comprehensively covers the dynamic process of energy storage and the internal coupling relationship of the carbon capture system, accurately characterizing various constraints in actual engineering. This effectively avoids operational infeasibility issues such as exceeding the energy storage SOC limit and the mismatch between carbon capture energy consumption and capture rate, ensuring the physical feasibility of the optimization results.

[0088] Fifth, a multi-objective optimization algorithm, MOEA-CFS, based on K-means clustering, is proposed. By leveraging clustering fitness strategies and a multi-strategy matching mechanism, the algorithm significantly improves convergence speed, distribution uniformity, and solution accuracy. This algorithm can generate high-quality Pareto front solution sets within a finite computation time, providing decision-makers with diverse trade-off options and achieving comprehensive coordinated optimization that is economical, low-carbon, efficient, and reliable.

[0089] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A multi-objective collaborative optimization method, characterized in that, The method is applied to a smart power generation system, and the method includes: A multi-objective collaborative optimization model is established based on the collaborative energy system. The multi-objective collaborative optimization model includes decision variables, objective function and constraints. Based on the objective function and constraints, construct a multi-objective optimization problem; A multi-objective optimization algorithm is used to solve the multi-objective optimization problem and generate a Pareto solution set; Select a scheduling scheme from the Pareto solution set and output scheduling instructions to the physical devices of the collaborative energy system.

2. The multi-objective collaborative optimization method as described in claim 1, characterized in that, Decision variables include the output of coal-fired and gas-fired power units, the power of electrochemical energy storage systems, the energy consumption of carbon capture systems, the capture rate of carbon capture systems, and the state of charge of electrochemical energy storage systems. The objective function includes system operating costs, total system carbon emissions, system energy loss rate, and system load response deviation; The constraints include power balance constraints, upper and lower limits of unit output constraints, unit ramp rate constraints, energy storage power constraints, energy storage SOC state transition constraints, upper and lower limits of energy storage SOC constraints, upper and lower limits of carbon capture system capture rate constraints, and carbon capture system energy consumption-capture rate coupling constraints.

3. The multi-objective collaborative optimization method as described in claim 2, characterized in that, The total operating cost of the system includes the unit's fuel cost, energy storage operation and maintenance cost, carbon capture system operation cost, and carbon trading cost; The total carbon emissions of the system are the difference between the total carbon emissions of the generating unit and the total carbon capture of the carbon capture system. The system energy loss rate is calculated as the ratio of the sum of energy loss during the energy storage charging and discharging process and the energy consumption of the carbon capture system to the total power generation. The system load response deviation is the cumulative deviation between the system's total output power and the grid load demand.

4. The multi-objective collaborative optimization method as described in claim 1, characterized in that, The multi-objective optimization algorithm is the MOEA-CFS algorithm based on the K-means clustering algorithm. The MOEA-CFS algorithm constructs a mating pool through a clustering fitness strategy and adopts a multi-strategy mating mechanism.

5. The multi-objective cooperative optimization method as described in claim 4, characterized in that, The clustering fitness strategy is as follows: The K-means algorithm is used to divide the population into K clusters in the target space; The objective function value of each individual in the population is normalized. Calculate the fitness of each individual.

6. The multi-objective cooperative optimization method as described in claim 5, characterized in that, The formula for calculating fitness is: , and For the number of iterations Dynamically adjusted weighting coefficients As the reference vector, Represents an individual With reference vector The included angle, Indicates and The angle between the reference vectors with the smallest included angle. Indicates the first Individuals at the number of iterations The normalized objective function value vector at time.

7. The multi-objective cooperative optimization method as described in claim 4, characterized in that, The multi-strategy mating mechanism integrates the SBX crossover operator, the BLX-α crossover operator, and the non-uniform mutation operator.

8. The multi-objective collaborative optimization method as described in claim 1, characterized in that, The steps for selecting a scheduling scheme from the Pareto solution set include: A compromise solution is selected from the Pareto solution set using the fuzzy membership method, which is then used as the final scheduling scheme.

9. The multi-objective collaborative optimization method as described in claim 1, characterized in that, Intelligent power generation systems include: The physical equipment layer includes coal-fired and gas-fired power units, electrochemical energy storage arrays, and carbon capture systems; The sensor network layer is used to collect the operating status parameters of the physical device layer in real time. The communication network layer is used to enable data transmission between the sensor network layer and the upper layers. The data middle platform layer is used to store real-time data, historical data, and model parameters; The algorithm engine layer is used to manage the Pareto front solution set; The application interaction layer is used to output scheduling instructions, perform visual monitoring, and interact with external systems.

10. The multi-objective cooperative optimization method as described in claim 1, characterized in that, The steps for outputting scheduling instructions to the physical devices of the coordinated energy system include: It outputs the unit output setpoint to the coal-fired power plant and gas-fired power plant, the charge and discharge power setpoint to the electrochemical energy storage array, and the capture rate setpoint to the carbon capture system.

Images