Multi-objective optimization method and system for ocean buoy power generation system based on energy link mismatch diagnosis guidance

Abstract

The present application relates to a marine buoy power generation system multi-objective optimization method and system based on energy link mismatch diagnosis guidance, through the closed-loop optimization framework of "modeling-mismatch diagnosis guidance-adaptive intelligent search-engineering constraint repair", the global optimal parameter configuration considering the maximum power generation, the maximum power generation efficiency and the minimum weight of the energy accumulator can be quickly and accurately obtained under the premise of ensuring physical consistency and engineering feasibility, effectively solving the problems of difficult collaborative optimization of multiple parameters, low optimization efficiency and poor engineering applicability of the results in the prior art, realizing lightweight design of the system while ensuring power generation performance, and having good engineering application prospect and popularization value. Compared with the traditional method relying on experience adjustment, single-objective optimization or general intelligent optimization algorithm, the present application can improve the physical consistency, optimization convergence efficiency and engineering implementability of the optimization solution under the condition of complex nonlinearity and strong coupling of multiple parameters.

Description

Technical Field

[0001] This invention relates to a multi-objective optimization method and system for marine buoy power generation systems based on energy link mismatch diagnosis guidance, belonging to the field of marine energy development and power generation system design technology. Background Technology

[0002] As ocean observation missions develop towards long-term, deep-sea, and autonomous operations, oceanographic buoys play a crucial role in tasks such as marine environmental monitoring, marine resource surveys, meteorological early warning, and marine scientific research. Oceanographic buoys typically need to be deployed for extended periods in complex sea conditions to continuously monitor parameters such as seawater temperature, salinity, pressure, current velocity, waves, meteorology, and the ecological environment, and transmit data back via communication systems. Due to their long operating cycles, maintenance difficulties, and deployment locations far from shore-based energy supply conditions, buoy systems require a high level of continuous and stable power supply capability. Currently, ocean buoys mostly rely on batteries, solar panels, or other external energy devices for power. However, relying solely on battery power has limitations such as limited capacity, difficulties in replacement and maintenance, and insufficient endurance; solar power is easily affected by diurnal variations, weather conditions, wave obstruction, and insufficient sunlight in high-latitude sea areas, making it difficult to guarantee long-term, stable, and self-sustaining operation of buoys in complex ocean environments. Therefore, how to improve the autonomous power supply capability of ocean buoys and extend their continuous observation cycle is an important issue in the design of oceanographic equipment.

[0003] Ocean thermal energy conversion (OTEC), as a sustainable form of marine energy, can utilize seawater temperature differences to drive power generation systems, providing auxiliary or continuous power to ocean buoys. Among these, integrated power generation systems based on accumulators, hydraulic motors, and permanent magnet synchronous generators can further convert the energy conversion process caused by temperature differences into hydraulic energy, mechanical energy, and electrical energy, providing a new technological approach for the long-term self-sustaining operation of buoys. However, due to limited ocean temperature differences, long energy conversion chains, and constraints on device size and weight, such buoy power generation systems often suffer from problems such as low power generation, unstable output, insufficient energy utilization efficiency, and difficulties in lightweight design, making it difficult to directly meet the stable power supply requirements of long-term continuous observation missions. For the aforementioned ocean buoy power generation systems, their power generation performance depends not only on temperature differences but also on key parameters such as the nominal capacity of the accumulator, initial gas pressure, gas polytropic index, hydraulic motor displacement, number of pole pairs of the permanent magnet synchronous generator, and set rotation speed. Accumulator parameters affect the system's energy storage and release capabilities, hydraulic motor parameters affect the conversion efficiency from hydraulic energy to mechanical energy, and the number of pole pairs and rotation speed of the generator affect the matching relationship between mechanical energy and electrical energy conversion. The aforementioned parameters exhibit complex nonlinear coupling relationships, significantly impacting system performance indicators such as power generation, power generation efficiency, and energy storage weight. Therefore, relying solely on single-factor experiments or empirical adjustments is insufficient to obtain the optimal combination of system parameters that balances power generation performance and lightweight requirements.

[0004] Current research on ocean thermal energy conversion (OTEC) power generation systems largely focuses on thermodynamic cycles, single component performance analysis, or single-objective parameter optimization. Research on system-level multi-objective optimization of the integrated power generation link in an ocean buoy—from accumulator to hydraulic motor to permanent magnet synchronous generator—remains insufficient. While some optimization methods can improve computational efficiency using surrogate models or intelligent algorithms, they typically only focus on the data fitting relationship between design variables and objective functions such as power generation efficiency and output power, lacking an effective integration of the system's energy transfer link and engineering constraints. This easily leads to optimization parameter combinations with insufficient physical feasibility. Furthermore, existing intelligent optimization methods such as multi-island genetic algorithms, when applied to parameter optimization of ocean buoy power generation systems, often rely on random crossover and random mutation for global search, making it difficult to identify the specific reasons for the inadequate performance of optimized parameter combinations. For example, the algorithm cannot determine whether the current optimal solution is primarily due to insufficient accumulator power supply, hydraulic motor drive mismatch, or unreasonable generator frequency matching leading to performance degradation. Therefore, it tends to repeatedly search in physically infeasible regions or inefficient parameter regions, resulting in low convergence efficiency and insufficient engineering feasibility of the obtained Pareto non-dominated solutions. Therefore, it is of great significance to study a novel multi-objective optimization method for ocean thermal energy conversion power generation systems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multi-objective optimization method and system for marine buoy power generation systems based on Isight co-simulation and energy link mismatch diagnosis. This method leverages the multidisciplinary integration, parameter-driven, and automated iterative capabilities of the Isight platform, integrating the Amesim system simulation model, Calculator module, surrogate evaluation model, and MATLAB data processing module to construct a co-simulation optimization process for an integrated power generation link of accumulator-hydraulic motor-permanent magnet synchronous generator, achieving coordinated optimization of key parameters of the marine buoy power generation system. This invention uses the accumulator's nominal capacity, initial gas pressure, gas polyvariance index, motor displacement, number of pole pairs of the permanent magnet synchronous generator, and set rotational speed as optimization design variables, with the optimization objectives being maximizing power generation efficiency, maximizing power output, and minimizing accumulator weight. Simultaneously, this invention constructs an energy link mismatch diagnosis model, identifying the dominant mismatch type based on the mismatch status of the optimized parameter combination in terms of accumulator power supply, release speed, hydraulic motor drive, generator matching, and lightweight benefits. For different dominant mismatch types, this invention adopts a rule-based directional mutation mechanism to prioritize the adjustment of variable groups related to the cause of mismatch. At the same time, it maintains the local good matching relationship between the accumulator, hydraulic motor and generator subsystems through modular cross-linking, and performs real-time correction of engineering constraints such as electrical frequency, flow demand and weight gain through a physical feasibility repair mechanism, thereby reducing invalid search individuals and physically infeasible optimization solutions.

[0006] This invention improves the traditional indiscriminate random search process into a directional search process guided by energy link mismatch diagnosis through multidisciplinary model integration and automated iterative computation. This allows the optimization algorithm to dynamically adjust the search direction based on the specific causes of mismatch in the system's energy transfer links. Compared to traditional methods that rely on experience-based adjustment, single-objective optimization, or general intelligent optimization algorithms, this invention can improve the physical consistency, convergence efficiency, and engineering feasibility of the optimization solution under complex nonlinear and strongly coupled multi-parameter conditions.

[0007] In summary, this invention, through a closed-loop optimization framework of "modeling-mismatch diagnosis guidance-adaptive intelligent search-engineering constraint repair," can quickly and accurately obtain the globally optimal parameter configuration that balances maximizing power generation, maximizing power generation efficiency, and minimizing energy storage weight, while ensuring physical consistency and engineering feasibility. It effectively solves the problems of difficulty in coordinating the optimization of multiple parameters, low optimization efficiency, and poor engineering applicability in existing technologies. It achieves lightweight system design while ensuring power generation performance, and has good engineering application prospects and promotion value.

[0008] Terminology Explanation:

[0009] 1. Isight: Process integration, optimization, and robustness design software produced by Engineeringious.

[0010] 2. Amesim: An advanced modeling and simulation platform for engineering systems.

[0011] 3. Orthogonal Array: This is an experimental design method that aims to reduce the number of experiments by selecting a subset of representative experimental combinations, thereby improving experimental efficiency and reducing costs. Its core tool is the orthogonal array, used to scientifically arrange multi-factor, multi-level combinations.

[0012] 4. Coefficient of Determination: In data analysis and machine learning, regression models are often used to predict the value of a variable. R² (Coefficient of Determination) is a key indicator for measuring the goodness of a regression model. Essentially, it measures how well the regression model fits the data, that is, the proportion of variation in the dependent variable (target variable) that the model can explain. The closer R² is to 1, the better the model fits the data.

[0013] 5. Multi-Island Genetic Algorithm (MAGA): This is an improved genetic algorithm that has wide applications in fields such as engineering optimization and machine learning parameter tuning. It is especially suitable for handling complex problems with high dimensionality and multiple constraints.

[0014] The technical solution of the present invention is as follows:

[0015] The first aspect of this invention provides a multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance, comprising:

[0016] Step 1: Establish a simulation model of the ocean thermal energy conversion system and conduct simulation analysis to determine the optimal design variables and optimization target parameters; perform joint simulation on the ocean thermal energy conversion system simulation model to obtain the combination of optimal design variables and the objective function value;

[0017] Step 2: Obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to the preset accuracy threshold to obtain the adjusted RBF approximation model;

[0018] Step 3: Use the adaptive multi-island genetic algorithm to perform multi-objective optimization, and obtain the optimal objective function value and the corresponding optimization design variables;

[0019] Adaptive multi-island genetic algorithms include:

[0020] Based on the optimized design variable combination, the populations of multiple islands are set, the adjusted RBF approximation model is used to evaluate the populations of each island, the predicted values ​​are obtained, and an external non-dominated solution archive is constructed.

[0021] Based on the island's population and predicted values, calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index, and determine the dominant mismatch type and severity.

[0022] The mutation probability is calculated based on the severity of the dominant mismatch, and the decision is made on whether to perform mutation based on the mutation probability. If mutation is to be performed, the priority variable group to be adjusted is determined based on the type of dominant mismatch and mutation is then performed.

[0023] Calculate the non-dominated solution contribution and population diversity of the island population, and perform population migration based on the non-dominated solution contribution and population diversity;

[0024] Repeat the adaptive multi-island genetic algorithm until the loop termination condition is met to obtain the optimal objective function;

[0025] Step 4: Substitute the optimization design variables corresponding to the optimal objective function value into the simulation model of the ocean thermal energy conversion system to complete the multi-objective optimization.

[0026] According to a preferred embodiment of the present invention, the specific implementation process of step 1 includes:

[0027] A simulation model of an ocean thermal energy conversion system was built in Amesim software, including an energy storage device, a directional valve, a motor, a torque sensor, a speed sensor, a PID module, a load, a rectifier module, a current sensor, a power generation acquisition module, a power generation efficiency acquisition module, and a permanent magnet synchronous generator.

[0028] The accumulator simulates the volume change of the heat exchange medium by varying its internal pressure, enabling charge and discharge control of the ocean thermal energy conversion system simulation model. A reversing valve regulates whether the pressure inside the accumulator drops. A motor converts the hydraulic energy of the oil into mechanical energy. A torque sensor detects the motor torque, with the motor and speed sensor connected to its two sides respectively. The speed sensor connects to a permanent magnet synchronous generator (PMSG) to collect the actual speed of the PMSG. The PMSG converts mechanical energy into electrical energy.

[0029] The PID module is used to calculate the error between the actual speed and the target speed of the permanent magnet synchronous generator, and adjust the speed of the permanent magnet synchronous generator to the target speed.

[0030] The rectifier module is used to convert electrical energy from AC to DC and supply power to the load;

[0031] The current sensor, power generation acquisition module, and power generation efficiency acquisition module collect the current, power generation, and power generation efficiency of the ocean thermal energy conversion power generation system simulation model in real time.

[0032] Set the pre-charge pressure of the energy storage device and set the optimization target parameter as power generation efficiency. The parameters in the simulation model of the ocean thermal energy conversion system are changed and analyzed by measuring power generation and energy storage weight. The optimization design variables that affect the optimization target parameters of the simulation model are obtained, including: nominal capacity of energy storage V, initial gas pressure P1, gas polyvariance index r, motor displacement L1, number of pole pairs of permanent magnet synchronous generator P, and set speed N.

[0033] The co-simulation model is driven by the DOE module, the Amesim simulation module, the Calculator module, and the MATLAB module.

[0034] In the DOE module, the DOE component in the Isight software is used to perform a parametric scan of the optimization design variables. Specifically, the orthogonal experimental design method is selected in the DOE component, and the range of variation of the optimization design variables is set. Orthogonal experimental design is then performed on the optimization design variables. The Array type is specified as L64 in the DOE component to generate the combination of optimization design variables X_i = [L1, P1, r, P, V, N], where i=1,2,...,64.

[0035] The Amesim simulation module calls the established ocean thermal energy conversion power generation system simulation model and receives i sets of optimized design variable combinations from the DOE module. It then performs dynamic simulation calculations in sequence and outputs the ocean thermal energy conversion power generation efficiency curve η(t), power generation curve power(t), and energy storage nominal capacity V.

[0036] The Calculator module calculates the weight of the accumulator by establishing a mapping relationship between the nominal capacity and weight of the accumulator. Specifically, the Calculator module reads the nominal capacity V value of the accumulator output by the Amesim simulation module and calculates the weight weight_i of the accumulator through the mapping relationship between the nominal capacity and weight.

[0037] The MATLAB module is used to extract the maximum value of power generation efficiency as the power generation efficiency max_eff_i and the maximum value of power generation as the power generation max_pwr_i from the time history curves output by the Amesim simulation module, namely the power generation efficiency curve η(t) and the power generation curve power(t).

[0038] The objective function values ​​[max_eff_i, max_pwr_i, weight_i] of the i sets of optimized design variable combinations are stored in the joint simulation sample database to obtain the final joint simulation sample database.

[0039] Further preferably, the nominal capacity V of the energy storage device varies in the range of [0.4 0.63 1.0 1.6], the initial gas pressure P1 varies in the range of [100 150 200], the gas polytropic index r varies in the range of [1.0 1.2 1.4], the motor displacement L1 varies in the range of [0.13 0.16 0.25], the number of pole pairs P of the permanent magnet synchronous generator varies in the range of [2 4 5 10], and the set speed N varies in the range of [1000 2000 2500 3000].

[0040] According to a preferred embodiment of the present invention, step 2 includes the following specific implementation process:

[0041] In Isight, the optimal design variable combination X_i = [L1, P1, r, P, V, N] is used as input, and Y_i = [max_eff_i, max_pwr_i, weight_i] is used as output. The corresponding output targets, i.e., the predicted power generation efficiency values, are obtained through the RBF approximation model. Forecasted power generation Predicted weight of accumulator ;

[0042] Based on the predicted values ​​of power generation efficiency, power generation, and energy storage weight, the prediction accuracy evaluation index, i.e., the coefficient of determination R^2, is calculated. It is then determined whether the prediction accuracy evaluation index meets the preset accuracy threshold: R^2 > R_lim^2; where R_lim^2 is the preset accuracy threshold.

[0043] If the preset accuracy threshold is not met, the radial basis function type, shape parameter, or smoothing parameter of the RBF approximation model is adjusted, and multiple optimized design variable combinations are added to the regions with large prediction errors until the predicted value of the RBF approximation model meets the preset accuracy threshold R_lim^2, thus obtaining the adjusted RBF approximation model.

[0044] Further preferred, the precision threshold R_lim^2 is 0.9.

[0045] According to a preferred embodiment of the present invention, multiple island populations are defined based on optimized design variable combinations, and the populations of each island are evaluated using an adjusted RBF approximation model to obtain predicted values ​​and construct an external non-dominated solution profile; including:

[0046] Let the population of the s-th island in the k-th iteration be... for:

[0047] ;

[0048] Where n_s represents the number of individuals in the s-th island, i.e., the number of combinations of candidate design variables. Let be the i-th candidate design variable combination in the s-th island of the k-th iteration, where s∈S and S represents the number of islands; where, in the 1st iteration, the candidate design variable combination is the optimal design variable combination that drives the co-simulation model to generate.

[0049] The adjusted RBF approximation model was used to evaluate the populations on each island, and the target predicted values ​​corresponding to the candidate design variable combinations were obtained, i.e., the predicted power generation efficiency values. Forecasted power generation And the predicted weight of the accumulator ,in Indicates the combination of candidate design variables;

[0050] The target prediction value is normalized as follows:

[0051] ;

[0052] ;

[0053] ;

[0054] in, , , These represent the normalized predicted power generation efficiency, the normalized predicted power generation, and the normalized predicted energy storage weight, respectively. This represents the minimum predicted value of power generation efficiency among the candidate design variable combinations. This represents the maximum predicted power generation efficiency value among the candidate design variable combinations. This represents the maximum predicted power generation value among the candidate design variable combinations. This represents the minimum predicted power generation value among the candidate design variable combinations. This represents the maximum predicted value of the accumulator weight among the candidate design variable combinations. This represents the minimum predicted value of the accumulator weight among the candidate design variable combinations;

[0055] During the k-th iteration, the population of each island is... Merge with the current external non-dominated solution archive, as shown below:

[0056] ;

[0057] in, Let be the candidate set for the k-th iteration. This represents the external non-dominated solution archive, which is empty during the first iteration.

[0058] Traverse the candidate set Then, the external non-dominated solution archives of the (k+1)th generation are generated through screening. As shown below:

[0059] ;

[0060] in, This indicates the Pareto non-dominated selection operation, i.e., for the candidate set... Any two optimal solutions are combinations of candidate design variables. and If the following conditions are met: and and And at least one objective is strictly superior to ,Right now or or Then determine Dominate Remove from external non-inferior solution archives For the candidate set After performing Pareto nondominated screening on all candidate design variable combinations, the (k+1)th generation external nondominated solution archive is formed. According to a preferred embodiment of the present invention, based on the island's population and predicted values, an energy supply mismatch index, a motor drive mismatch index, and a generator matching mismatch index are calculated, and the dominant mismatch type and severity are determined; including:

[0061] Calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index;

[0062] Among them, the calculation of energy supply insufficiency and mismatch index This includes: calculating the hydraulic energy that the accumulator can release based on the combination of candidate design variables, as shown below:

[0063] ;

[0064] When r=1, the hydraulic energy that the accumulator can release is expressed as:

[0065] ;

[0066] in, This indicates that the accumulator can release hydraulic energy. This represents the initial gas pressure of the candidate design variable combination. This represents the nominal capacity of the energy storage device, representing the combination of candidate design variables. The gas variability index represents the combination of candidate design variables. This is the minimum operating pressure of the accumulator, which is slightly higher than the pre-charge pressure of the accumulator. This indicates the highest operating pressure of the accumulator, i.e., the initial gas pressure P1.

[0067] Introducing an energy storage energy release correction factor To obtain effective energy supply capacity As shown below:

[0068] ;

[0069] Power generation prediction values ​​obtained from the RBF approximation model and power generation efficiency forecast Calculate the equivalent input energy demand As shown below:

[0070] ;

[0071] Construct an energy supply mismatch index:

[0072] ;

[0073] in, Indicators of insufficient or mismatched energy supply;

[0074] Calculate motor drive mismatch index This includes: selecting the top 20%–30% of optimal design variable combinations with max_eff_i as the reference sample set from the joint simulation sample database; calculating the motor drive capability index in the reference sample set, as shown below:

[0075] ;

[0076] in, This represents the motor drive capability index of the reference sample set. This represents the initial gas pressure of the reference sample set. This represents the motor displacement of the reference sample set. This indicates the set rotational speed of the reference sample set;

[0077] The range of values ​​for calculating the motor drive capability index:

[0078] ;

[0079] in, and These are the minimum and maximum values ​​of the motor drive capability index, respectively, serving as the lower and upper limits;

[0080] For candidate design variable combinations in the island population, calculate the motor drive capability index for the candidate design variable combinations and construct a motor drive mismatch index:

[0081] ;

[0082] in, These are the weighting coefficients. This indicates a motor drive mismatch index. This represents the normalized predicted power generation efficiency. A motor-driving capability index representing the combination of candidate design variables in an island population;

[0083] Calculate generator matching mismatch index This includes: calculating the electrical frequency based on the number of permanent magnet synchronous generator pole pairs P and the set rotational speed N in the candidate design variable combinations for the island, as shown below:

[0084] ;

[0085] Assume the permissible frequency range of the generator is:

[0086] ;

[0087] in, The generator's electrical frequency, and These are the lower and upper limits of the generator's effective frequency operating range, respectively. That is, the corresponding model of the permanent magnet synchronous generator is found based on P, and the upper and lower limits are determined based on the corresponding model.

[0088] Construct generator frequency mismatch index :

[0089] ;

[0090] Calculate the power generation revenue per unit rotational speed based on the predicted power generation value and the set rotational speed. :

[0091] ;

[0092] Constructing a power generation revenue mismatch index :

[0093] ;

[0094] in, This represents the lower limit of the revenue generated per unit speed of electricity.

[0095] The generator matching mismatch index was obtained by comprehensive analysis. :

[0096] ;

[0097] in, and These are the weighting coefficients;

[0098] Three types of mismatch indicators, namely, insufficient energy supply mismatch indicators Motor drive mismatch index Generator matching mismatch index Normalization is performed as follows:

[0099] ;

[0100] in, This represents the normalized mismatch index, j=1,2,3. This represents the maximum value of the j-th type of mismatch index in the current population;

[0101] The type corresponding to the maximum value in the normalized mismatch index is taken as the dominant mismatch type of the candidate design variable combination, as shown below:

[0102] ;

[0103] in, The dominant mismatch type for candidate design variable combinations. Indicates selection The number corresponding to the maximum value; the severity of the dominant mismatch is defined as:

[0104] ;

[0105] in, To determine the severity of the mismatch, Indicates selection The maximum value.

[0106] Further optimized, the lower limit of power generation revenue per unit speed Determined based on the joint simulation sample database; including:

[0107] The performance scores of the objective function values ​​in the co-simulation sample database are calculated as follows:

[0108] );

[0109] in, Indicates the performance score. This represents the i-th combination of optimal design variables in the joint simulation sample database. , , These represent the normalized power generation efficiency, normalized power generation, and normalized energy storage weight, respectively. , , Represents the weight parameters. + + =1;

[0110] Will Sort by high to low, select the top 20% to 30% of the optimized design variable combinations in the joint simulation sample database as high-performance reference samples;

[0111] Calculate the power generation revenue per unit rotational speed for each sample in the high-performance reference sample. The minimum value among them is taken as the lower limit of power generation revenue per unit speed. ;in, This represents the i-th combination of optimized design variables in the high-performance reference sample.

[0112] According to a preferred embodiment of the present invention, the mutation probability is calculated based on the severity of the dominant mismatch, and a determination is made based on the mutation probability whether to perform mutation; if mutation is to be performed, a priority adjustment variable group is determined based on the type of dominant mismatch and mutation is performed; including:

[0113] The probability of variation is calculated based on the severity of the dominant mismatch, and the base probability of variation is set as follows: Then the mutation probability for:

[0114] ;

[0115] in, This is the variation intensity adjustment coefficient; () represents the basic mutation probability;

[0116] The decision to perform mutation is based on the mutation probability. If mutation is deemed necessary, the priority variable group to be adjusted and mutated is determined according to the dominant mismatch type, including:

[0117] when When =1, it indicates that the dominant mismatch type of the candidate design variable combination is insufficient energy supply mismatch; the priority adjustment variable group is: increase or decrease the nominal capacity V of the accumulator, increase the initial gas pressure P1, and increase or decrease the gas polyvariance index r within the range of variation.

[0118] when When = 2, it indicates that the dominant mismatch type of the candidate design variable combination is motor drive mismatch; when the drive capability is insufficient, i.e. When the variable group is adjusted, the priority is to increase or decrease P1 or increase L1 within the set range of change; when L1 or N exceeds the upper limit of the range of change and the predicted power generation efficiency is lower than the current population average level, it is determined that the motor displacement or speed is too high, resulting in a decrease in efficiency, and the priority is to decrease L1 or N within the set range of change.

[0119] when When =3, the dominant mismatch type of the candidate design variable combination is generator matching mismatch; when the electric frequency f is higher than the allowable electric frequency range of the generator. Then, the priority adjustment of the variable group is: within the set range of change, increase or decrease P, or decrease N; when the electrical frequency f is lower than the allowable range. If so, the variable group should be adjusted to increase or decrease P or increase N within the set range of change;

[0120] Further optimization prioritizes adjusting the variable set, i.e., the dominant mismatch type. =1 or =2 or =3 Variables in the candidate design variable combinations that need to be increased or decreased When increasing or decreasing, adjustments are made based on the directional disturbance, as shown below:

[0121] ;

[0122] in, The mutated variable value; and They are respectively The upper and lower limits; A random number within the interval [0,1]; The direction coefficient is used to adjust the variable, indicating the direction of increase or decrease of the variable. It is used when it is necessary to increase the variable. hour, =1, when it is necessary to reduce the variable hour, =-1; The adaptive perturbation coefficients for the k-th iteration are shown below:

[0123] ;

[0124] in, The initial disturbance coefficients are... The minimum disturbance coefficient, The maximum number of iterations;

[0125] After completing the directional perturbation, the mutated variable values ​​are subjected to boundary processing, as shown below:

[0126] ;

[0127] in, and They are respectively The upper and lower limits;

[0128] Further preferred methods include calculating the non-dominated solution contribution and population diversity of the island population, and performing population migration based on the non-dominated solution contribution and population diversity; including:

[0129] The contribution of the s-th island to the non-dominated solution in the k-th iteration is calculated as follows:

[0130] ;

[0131] in, This represents the contribution of the s-th island to the external non-dominated solution archive in the k-th iteration, i.e., the non-dominated solution contribution. This represents the number of individuals added to the external non-dominated solution file for the s-th island in the k-th iteration, which is the number of candidate design variable combinations. This represents the population size, or number of individuals, of the s-th island;

[0132] The population diversity of the s-th island in the k-th iteration is calculated as follows:

[0133] ;

[0134] in, This represents the population diversity of the s-th island; Let represent the normalized vector of the i-th individual in the s-th island; This represents the population center of the s-th island after normalization. Indicates Euclidean distance;

[0135] according to and Population migration includes:

[0136] When the non-dominated solution contribution of a certain island When the contribution exceeds the preset threshold, the island is designated as the elite output island, and individuals in the external non-dominated solution archive of the elite output island are migrated to other islands. Alternatively, the Euclidean distance between individuals in the current population of the elite output island and each individual in the external non-dominated solution archive of the elite output island is calculated, and individuals whose Euclidean distance is less than or equal to the preset threshold are migrated to other islands.

[0137] When the population diversity of a certain island When the diversity threshold is lower than the preset threshold, the island is used as the differential input island; calculate the Euclidean distance between other islands and the differential input island, and introduce individuals from other islands whose Euclidean distance is greater than or equal to the preset distance threshold into the differential input island;

[0138] When an island simultaneously satisfies both high contribution and high diversity, that is... >0.25 and When the value is greater than 0.25, some individuals in the elite output island will be retained and not replaced;

[0139] Further preferred, when individuals migrate to other islands or are introduced into differentiated input islands, the migration or introduction ratio is 5%; when an island fails to produce a new non-dominated solution in multiple consecutive iterations, i.e., the number of newly added individuals in the external non-dominated solution archive. When the value is 0, increase the island migration or introduction ratio to 15%, or inject resources into the island to meet the set variation range and generator frequency constraints. And the energy storage capacity constraint, i.e. ≤ Random individuals.

[0140] According to a preferred embodiment of the present invention, the adaptive multi-island genetic algorithm is repeatedly executed until the loop termination condition is met to obtain the optimal objective function; including:

[0141] Define the proportion of newly added non-dominated solutions in the external non-dominated solution archive during the k-th iteration as:

[0142] ;

[0143] in, This represents the proportion of newly added non-dominated solutions; if the proportion of newly added non-dominated solutions in the external non-dominated solution archive is G consecutive iterations... If all values ​​are less than the set threshold, the external non-dominated solution archive is considered to be stable.

[0144] After the external non-dominated solution portfolio stabilizes, the iterative loop stops. Using MATLAB software, Pareto plots are generated for the objective function values ​​corresponding to the Pareto non-dominated solutions (i.e., combinations of candidate design variables) in the stabilized external non-dominated solution portfolio. A decision mechanism is then used to calculate the comprehensive evaluation value of the Pareto non-dominated solutions, as shown below:

[0145] ;

[0146] in, This is expressed as a comprehensive evaluation value. , , These represent the normalized predicted power generation efficiency, the normalized predicted power generation, and the normalized predicted energy storage weight, respectively. , , Let these be the weighting coefficients for power generation efficiency, power generation, and weight, respectively, and satisfy the following conditions: + + =1;

[0147] Comprehensive evaluation value The highest Pareto nondominated solution is taken as the optimal objective function. The optimal design variables corresponding to the optimal objective function value, namely the nominal capacity V of the energy storage device, the initial gas pressure P1, the gas polyvariance index r, the motor displacement L1, the number of pole pairs P of the permanent magnet synchronous generator, and the set speed N, are substituted into the simulation model of the ocean thermal energy conversion system for simulation analysis. The optimized effects, namely the power generation and power generation efficiency, are obtained, and the multi-objective optimization of the power generation system is completed.

[0148] Further optimized, weighting coefficients , , The values ​​are 0.4, 0.4, and 0.2 respectively.

[0149] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of a multi-objective optimization method for an ocean buoy power generation system guided by energy link mismatch diagnosis.

[0150] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a multi-objective optimization method for an ocean buoy power generation system guided by energy link mismatch diagnosis.

[0151] A second aspect of the present invention provides a multi-objective optimization system for marine buoy power generation systems based on energy link mismatch diagnosis guidance, comprising:

[0152] The simulation analysis module is configured to: establish a simulation model of the ocean thermal energy conversion power generation system and perform simulation analysis to determine the optimal design variables and optimization target parameters; and perform joint simulation of the ocean thermal energy conversion power generation system simulation model to obtain the combination of optimal design variables and the objective function value.

[0153] The co-simulation module is configured to: obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to a preset accuracy threshold to obtain the adjusted RBF approximation model;

[0154] The RBF approximation model optimization module is configured to: obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to the preset accuracy threshold to obtain the adjusted RBF approximation model;

[0155] The multi-objective optimization module is configured to use an adaptive multi-island genetic algorithm to perform multi-objective optimization and obtain the optimal objective function value and the corresponding optimization design variables.

[0156] Adaptive multi-island genetic algorithms include:

[0157] Based on the optimized design variable combination, the populations of multiple islands are set, the adjusted RBF approximation model is used to evaluate the populations of each island, the predicted values ​​are obtained, and an external non-dominated solution archive is constructed.

[0158] Based on the island's population and predicted values, calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index, and determine the dominant mismatch type and severity.

[0159] The mutation probability is calculated based on the severity of the dominant mismatch, and the decision is made on whether to perform mutation based on the mutation probability. If mutation is to be performed, the priority variable group to be adjusted is determined based on the type of dominant mismatch and mutation is then performed.

[0160] Calculate the non-dominated solution contribution and population diversity of the island population, and perform population migration based on the non-dominated solution contribution and population diversity;

[0161] Repeat the adaptive multi-island genetic algorithm until the loop termination condition is met to obtain the optimal objective function;

[0162] The simulation module is configured as follows: Step 4: Substitute the optimization design variables corresponding to the optimal objective function value into the simulation model of the ocean thermal energy conversion system to complete the multi-objective optimization.

[0163] The beneficial effects of this invention are as follows:

[0164] 1. This invention integrates the Amesim simulation model, Calculator calculation module, and MATLAB data processing module through the Isight platform to achieve unified modeling and automated iterative optimization of six key parameters, including the nominal capacity of the energy storage device, gas pressure, motor displacement, motor speed, and number of pole pairs. It establishes a mapping relationship between input parameters and power generation efficiency, power generation, and energy storage device weight, which significantly improves optimization efficiency and result reliability.

[0165] 2. This invention is based on multi-objective optimization of marine buoy power generation systems guided by energy link mismatch diagnosis. It performs rule-based directed mutation based on the dominant mismatch types of the optimized individual, such as insufficient power supply, motor drive mismatch, and generator matching mismatch. Compared to traditional random crossover and random mutation methods, this invention can prioritize adjusting variable groups related to the cause of mismatch, reducing invalid searches and blind perturbations to irrelevant variables.

[0166] 3. This invention introduces a physical feasibility repair mechanism after optimizing individual solutions. This mechanism corrects engineering constraints such as electrical frequency, flow requirements, design variable boundaries, and weight gains, ensuring that the newly generated optimized parameter combinations remain within the physically feasible region as much as possible. Compared to methods that only use penalty functions or directly eliminate infeasible solutions, this invention reduces physically infeasible solutions and improves the quality of the optimized Pareto non-dominated solution archive.

[0167] 4. This invention adaptively adjusts the island migration strategy based on the non-dominated solution contribution and population diversity of each island. This allows islands with higher contributions to spread superior search directions to other islands, while islands with insufficient diversity or that have not produced non-dominated solutions for a long time can introduce differentiated individuals or feasible random individuals. Therefore, this invention can reduce the risk of premature convergence while maintaining global search capabilities, and improve the distribution and stability of the Pareto solution set in multi-objective optimization. Attached Figure Description

[0168] Figure 1 This is a schematic diagram of a simulation model of an ocean thermal energy conversion power generation system based on the multi-objective optimization method for ocean buoy power generation systems guided by energy link mismatch diagnosis, as described in this invention.

[0169] Figure 2 This is a flowchart of the multi-objective optimization of the marine buoy power generation system based on energy link mismatch diagnosis guided by the present invention;

[0170] Figure 3 The simulation model diagram is shown for the co-simulation model driving this invention;

[0171] Figure 4 This is a Pareto diagram of the candidate design variable combinations for this invention;

[0172] Figure 5 This is a comparison chart of the power generation efficiency of the present invention;

[0173] Figure 6 This is a comparison chart of the power generation of the present invention;

[0174] The components include: 1. Energy accumulator; 2. Reversing valve; 3. Motor; 4. Torque sensor; 5. Speed ​​sensor; 6. Target speed; 7. PID module; 8. Load; 9. Rectifier module; 10. Current sensor; 11. Power generation acquisition module; 12. Power generation efficiency acquisition module; and 13. Permanent magnet synchronous generator. Detailed Implementation

[0175] The present invention will be further described below with reference to the embodiments and accompanying drawings, but is not limited thereto.

[0176] Example 1

[0177] A multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance, such as Figure 2 As shown, it includes:

[0178] Step 1: Establish a simulation model of the ocean thermal energy conversion system and conduct simulation analysis to determine the optimal design variables and optimization target parameters; perform joint simulation on the ocean thermal energy conversion system simulation model to obtain the combination of optimal design variables and the objective function value;

[0179] Step 2: Obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to the preset accuracy threshold to obtain the adjusted RBF approximation model;

[0180] Step 3: Use the adaptive multi-island genetic algorithm to perform multi-objective optimization, and obtain the optimal objective function value and the corresponding optimization design variables;

[0181] Adaptive multi-island genetic algorithms include:

[0182] Based on the optimized design variable combination, the populations of multiple islands are set, the adjusted RBF approximation model is used to evaluate the populations of each island, the predicted values ​​are obtained, and an external non-dominated solution archive is constructed.

[0183] Based on the island's population and predicted values, calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index, and determine the dominant mismatch type and severity.

[0184] The mutation probability is calculated based on the severity of the dominant mismatch, and the decision is made on whether to perform mutation based on the mutation probability. If mutation is to be performed, the priority variable group to be adjusted is determined based on the type of dominant mismatch and mutation is then performed.

[0185] Calculate the non-dominated solution contribution and population diversity of the island population, and perform population migration based on the non-dominated solution contribution and population diversity;

[0186] Repeat the adaptive multi-island genetic algorithm until the loop termination condition is met to obtain the optimal objective function;

[0187] Step 4: Substitute the optimization design variables corresponding to the optimal objective function value into the simulation model of the ocean thermal energy conversion system to complete the multi-objective optimization.

[0188] Example 2

[0189] The multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guided by the method described in Example 1 differs in that it includes:

[0190] The specific implementation process of step 1 includes:

[0191] like Figure 1 As shown, a simulation model of an ocean thermal energy conversion power generation system is established in Amesim software, including an energy storage device 1, a reversing valve 2, a motor 3, a torque sensor 4, a speed sensor 5, a PID module 7, a load 8, a rectifier module 9, a current sensor 10, a power generation acquisition module 11, a power generation efficiency acquisition module 12, and a permanent magnet synchronous generator 13.

[0192] Accumulator 1 simulates the volume change of the heat exchange medium by changing its internal pressure, thus realizing the charging and discharging control of the ocean thermal energy conversion system simulation model. Reversing valve 2 is used to regulate whether the pressure inside the accumulator drops. When the reversing valve is energized, the valve port opens, the oil inside the accumulator begins to flow out, and the accumulator pressure drops. Motor 3 is used to convert the hydraulic energy of the oil into mechanical energy. Torque sensor 4 is used to detect the motor torque, and is connected to motor 3 and speed sensor 5 on both sides respectively. Speed ​​sensor 5 is connected to permanent magnet synchronous generator 13 to collect the actual speed of the permanent magnet synchronous generator. The permanent magnet synchronous generator is used to convert mechanical energy into electrical energy.

[0193] PID module 7 is used to calculate the error between the actual speed of the permanent magnet synchronous generator and the target speed 6, and adjust the speed of the permanent magnet synchronous generator to the target speed (by changing the electromagnetic load torque of the generator through "adjusting the resistor", and using the balance relationship of input mechanical torque = electromagnetic braking torque + loss, the speed can automatically track the target value).

[0194] The rectifier module is used to convert electrical energy from AC to DC and supply power to the load;

[0195] The current sensor 10, power generation acquisition module 11, and power generation efficiency acquisition module 12 collect the current, power generation, and power generation efficiency of the ocean thermal energy conversion power generation system simulation model in real time.

[0196] Set the pre-charge pressure of the energy storage device (10MPa) and set the optimization target parameter to power generation efficiency. The parameters in the simulation model of the ocean thermal energy conversion system are changed and analyzed by measuring power generation and energy storage weight. The optimization design variables that affect the optimization target parameters of the simulation model are obtained, including: nominal capacity of energy storage V, initial gas pressure P1, gas polyvariance index r, motor displacement L1, number of pole pairs of permanent magnet synchronous generator P, and set speed N.

[0197] like Figure 3 As shown, the driving co-simulation model includes the DOE module, Amesim simulation module, Calculator module, and MATLAB module;

[0198] In the DOE module, the DOE component in the Isight software is used to perform a parametric scan of the optimization design variables. Specifically, the orthogonal experimental design method is selected within the DOE component (utilizing the mathematical and statistical properties of "balanced distribution" and "uniform comparability" of the "Orthogonal Array" method to scientifically select representative sample points from a comprehensive experiment. This method not only ensures the uniform distribution of sample points in the multidimensional design space, avoiding sample clustering and achieving effective coverage of the parameter space, but also guarantees the balanced frequency of occurrence of different levels of each parameter and any combination of parameters in the experiment, thus effectively decoupling the interaction effects between multiple parameters and providing a solid mathematical foundation for subsequent independent and accurate evaluation of the effects of each parameter). Simultaneously, the range of values ​​for the optimization design variables is set, and orthogonal experimental design is performed on the optimization design variables. The Array type is specified as L64 in the DOE component, generating the optimization design variable combination X_i = [L1, P1, r, P, V, N], where i = 1, 2, ..., 64.

[0199] The L64 orthogonal array was selected to achieve the optimal balance between computational efficiency and experimental accuracy. This method significantly reduced the number of full factorial experiments that originally required thousands of trials to 64, improving computational efficiency by more than 90% and significantly reducing simulation costs. At the same time, the high capacity of the L64 orthogonal array fully covered the variable requirements of this scheme and had high resolution, which could effectively identify main effects and some interaction effects, providing sufficient and high-quality data support for building a high-precision surrogate model.

[0200] The Amesim simulation module calls the established ocean thermal energy conversion power generation system simulation model and receives i sets of optimized design variable combinations from the DOE module. It then performs dynamic simulation calculations in sequence and outputs the ocean thermal energy conversion power generation efficiency curve η(t), power generation curve power(t), and energy storage nominal capacity V.

[0201] The Calculator module calculates the weight of the accumulator by establishing a mapping relationship between the accumulator's nominal capacity and weight. Specifically, the Calculator module reads the nominal capacity V value of the accumulator output by the Amesim simulation module and calculates the accumulator weight_i through the mapping relationship between nominal capacity and weight (different volumes will have certain weights, which are determined by the accumulator manufacturer, such as the following specific mapping relationships: 0.4L-3kg, 0.63L-3.5kg, 1L-5.5kg, 1.6L-12.5kg).

[0202] The MATLAB module is used to extract the maximum value of power generation efficiency as the power generation efficiency max_eff_i and the maximum value of power generation as the power generation max_pwr_i from the time history curves output by the Amesim simulation module, namely the power generation efficiency curve η(t) and the power generation curve power(t).

[0203] The objective function values ​​[max_eff_i, max_pwr_i, weight_i] of the i sets of optimized design variable combinations are stored in the joint simulation sample database to obtain the final joint simulation sample database.

[0204] The nominal capacity V of the accumulator varies from [0.4 0.63 1.0 1.6], the initial gas pressure P1 varies from [100 150 200], the gas polytropic index r varies from [1.0 1.2 1.4], the motor displacement L1 varies from [0.13 0.16 0.25], the number of pole pairs P of the permanent magnet synchronous generator varies from [2 4 5 10], and the set speed N varies from [1000 2000 2500 3000].

[0205] The specific implementation process of step 2 includes:

[0206] In Isight, the optimal design variable combination X_i = [L1, P1, r, P, V, N] is used as input, and Y_i = [max_eff_i, max_pwr_i, weight_i] is used as output. The corresponding output target, i.e., the predicted value of power generation efficiency, is obtained by using the RBF approximation model (select the RBF approximation model as the approximation technique in the Approximation module of Isight software). Forecasted power generation Predicted weight of accumulator ;

[0207] The generated approximate model is verified and corrected to determine whether the RBF approximate model meets the rapid evaluation requirements in the subsequent multi-objective optimization main loop. Specifically, based on the predicted power generation efficiency, predicted power generation, and predicted energy storage weight, the prediction accuracy evaluation index, i.e., the coefficient of determination R^2, is calculated, and it is determined whether the prediction accuracy evaluation index meets the preset accuracy threshold: R^2 > R_lim^2; where R_lim^2 is the preset accuracy threshold.

[0208] If the preset accuracy threshold is not met, the radial basis function type, shape parameter, or smoothing parameter of the RBF approximation model is adjusted, and multiple optimized design variable combinations are added to the regions with large prediction errors to make the samples more evenly distributed within the range of values ​​of each design variable, until the generated approximation model meets the prediction accuracy evaluation index, until the predicted value of the RBF approximation model meets the preset accuracy threshold R_lim^2, and the adjusted RBF approximation model is obtained.

[0209] The precision threshold R_lim^2 is 0.9.

[0210] Based on the optimized design variable combination, populations for multiple islands are defined. An adjusted RBF approximation model is used to evaluate the populations of each island, obtain predicted values, and construct an external non-dominated solution archive; including:

[0211] An adaptive multi-island genetic algorithm is used for multi-objective optimization. The main loop of the adaptive multi-island genetic algorithm is based on the multi-population parallel search and island migration mechanism of the multi-island genetic algorithm, and introduces external non-dominated solution file update, energy link mismatch diagnosis, rule-based directional mutation, modular crossover, physical feasibility repair and adaptive island migration mechanism.

[0212] Let the population of the s-th island in the k-th iteration be... for:

[0213] ;

[0214] Where n_s represents the number of individuals in the s-th island, i.e., the number of combinations of candidate design variables. Let be the i-th candidate design variable combination in the s-th island of the k-th iteration, where s∈S and S represents the number of islands (S=4); where, in the 1st iteration, the candidate design variable combination is the optimal design variable combination that drives the co-simulation model to generate.

[0215] The adjusted RBF approximation model was used to evaluate the populations on each island, and the target predicted values ​​corresponding to the candidate design variable combinations were obtained, i.e., the predicted power generation efficiency values. Forecasted power generation And the predicted weight of the accumulator The multi-objective optimization direction is to maximize power generation efficiency, maximize power generation, and minimize the weight of the energy storage device. Indicates the combination of candidate design variables;

[0216] To facilitate subsequent evaluation, the target predicted value is normalized as follows:

[0217] ;

[0218] ;

[0219] ;

[0220] in, , , These represent the normalized predicted power generation efficiency, the normalized predicted power generation, and the normalized predicted energy storage weight, respectively. This represents the minimum predicted value of power generation efficiency among the candidate design variable combinations. This represents the maximum predicted power generation efficiency value among the candidate design variable combinations. This represents the maximum predicted power generation value among the candidate design variable combinations. This represents the minimum predicted power generation value among the candidate design variable combinations. This represents the maximum predicted value of the accumulator weight among the candidate design variable combinations. This represents the minimum predicted value of the accumulator weight among the candidate design variable combinations;

[0221] During the k-th iteration, the population of each island is... Merge with the current external non-dominated solution archive, as shown below:

[0222] ;

[0223] in, Let be the candidate set for the k-th iteration. This represents the external non-dominated solution archive. In the first iteration, the external non-dominated solution archive is empty. Using this formula, the current external non-dominated solution archive is merged with the newly generated optimized individuals from each island to form a unified candidate set. This is used for subsequent Pareto dominance determination and external non-dominated solution file updates;

[0224] Traverse the candidate set Then, the external non-dominated solution archives of the (k+1)th generation are generated through screening. As shown below:

[0225] ;

[0226] in, This indicates the Pareto non-dominated selection operation, i.e., for the candidate set... Any two optimal solutions are combinations of candidate design variables. and If the following conditions are met: and and And at least one objective is strictly superior to ,Right now or or Then determine Dominate Remove from external non-inferior solution archives For the candidate set After performing Pareto nondominated screening on all candidate design variable combinations, the (k+1)th generation external nondominated solution archive is formed. .

[0227] Based on the island's population and predicted values, calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index, and determine the dominant mismatch type and severity; including:

[0228] Energy link mismatch diagnosis: For each candidate design variable combination, an energy link mismatch diagnosis index is constructed based on the design variables, RBF predicted target value, and energy link physical rules to determine the dominant mismatch type of the optimization individual;

[0229] Calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index;

[0230] Among them, the calculation of energy supply insufficiency and mismatch index This includes: calculating the releaseable hydraulic energy of the accumulator based on the polytropic relationship of the accumulator gas and the combination of candidate design variables, as shown below:

[0231] ;

[0232] When r=1, the hydraulic energy that the accumulator can release is expressed as:

[0233] ;

[0234] in, This indicates that the accumulator can release hydraulic energy. This represents the initial gas pressure of the candidate design variable combination. This represents the nominal capacity of the energy storage device, representing the combination of candidate design variables. The gas variability index represents the combination of candidate design variables. This is the minimum operating pressure of the accumulator, which is slightly greater than the pre-charge pressure of the accumulator (11MPa). This indicates the highest operating pressure of the accumulator, i.e., the initial gas pressure P1.

[0235] Introducing an energy storage energy release correction factor (0.8-0.9) to obtain effective energy supply capacity As shown below:

[0236] ;

[0237] Power generation prediction values ​​obtained from the RBF approximation model and power generation efficiency forecast Calculate the equivalent input energy demand As shown below:

[0238] ;

[0239] Construct an energy supply mismatch index:

[0240] ;

[0241] in, Indicators of insufficient or mismatched energy supply; When =0, it means that the combination of candidate design variables meets the input energy demand corresponding to the predicted power generation in terms of the energy storage capacity, and does not show the risk of insufficient energy supply. >0 indicates that there is a risk of insufficient energy supply in the combination of candidate design variables;

[0242] Calculate motor drive mismatch index This includes: To determine a reasonable range of motor drive capability, in the joint simulation sample database, selecting the optimal design variable combinations where max_eff_i is in the top 20% to 30% (preferably the top 25%) as the reference sample set; calculating the motor drive capability index in the reference sample set, as shown below:

[0243] ;

[0244] in, This represents the motor drive capability index of the reference sample set. This represents the initial gas pressure of the reference sample set. This represents the motor displacement of the reference sample set. This indicates the set rotational speed of the reference sample set;

[0245] The range of values ​​for calculating the motor drive capability index:

[0246] ;

[0247] in, and These are the minimum and maximum values ​​of the motor drive capability index, respectively, serving as the lower and upper limits;

[0248] For candidate design variable combinations in the island population, calculate the motor drive capability index for the candidate design variable combinations and construct a motor drive mismatch index:

[0249] ;

[0250] in, This is the weighting coefficient (with values ​​in the range [0.1, 1], preferably 0.5). This indicates a motor drive mismatch index. This represents the normalized predicted power generation efficiency. This represents the motor-driving capability index of candidate design variable combinations in an island population; when When =0, it indicates that the motor driving capability of the candidate design variable combination is within the reference matching range; when A value greater than 0 indicates a risk of motor drive mismatch in the combination of candidate design variables;

[0251] Calculate generator matching mismatch index This includes: Generator matching mismatch indices are constructed based on the number of permanent magnet synchronous generator pole pairs, set speed, allowable electrical frequency range, and power generation revenue per unit speed; the electrical frequency is calculated based on the number of permanent magnet synchronous generator pole pairs P and set speed N in the candidate design variable combinations for the island, as shown below:

[0252] ;

[0253] Assume the permissible frequency range of the generator is:

[0254] ;

[0255] in, The generator's electrical frequency, and These are the lower and upper limits of the generator's effective frequency operating range, respectively. That is, the corresponding model of the permanent magnet synchronous generator is found based on P, and the upper and lower limits are determined based on the corresponding model.

[0256] Construct generator frequency mismatch index :

[0257] ;

[0258] Calculate the power generation revenue per unit rotational speed based on the predicted power generation value and the set rotational speed. :

[0259] ;

[0260] Constructing a power generation revenue mismatch index :

[0261] ;

[0262] in, This represents the lower limit of the revenue generated per unit speed of electricity.

[0263] The generator matching mismatch index was obtained by comprehensive analysis. :

[0264] ;

[0265] in, and Weighting coefficients ( =[0.6,0.8], =[0.2,0.4]); >0 indicates that the selected design variable combination has a mismatch between the number of generator pole pairs, the set speed, or the utilization effect of mechanical input; =0 indicates that the generator frequency is within the effective operating range and the power generation revenue per unit speed meets the lower limit requirement determined by the reference sample, and there is no risk of generator mismatch.

[0266] Mismatch type determination: To eliminate the dimensional differences between different mismatch indicators, three types of mismatch indicators, namely, insufficient energy supply mismatch indicators, are classified. Motor drive mismatch index Generator matching mismatch index Normalization is performed as follows:

[0267] ;

[0268] in, This represents the normalized mismatch index, j=1,2,3. This represents the maximum value of the j-th type of mismatch index in the current population;

[0269] The type corresponding to the maximum value in the normalized mismatch index is taken as the dominant mismatch type of the candidate design variable combination, as shown below:

[0270] ;

[0271] in, The dominant mismatch type for candidate design variable combinations. Indicates selection The maximum value corresponds to the index (j=1,2,3); the severity of the dominant mismatch is defined as:

[0272] ;

[0273] in, To determine the severity of the mismatch, Indicates selection The maximum value.

[0274] Lower limit of power generation revenue per unit speed Determined based on the joint simulation sample database; including:

[0275] The performance scores of the objective function values ​​in the co-simulation sample database are calculated as follows:

[0276] );

[0277] in, Indicates the performance score. This represents the i-th combination of optimal design variables in the joint simulation sample database. , , These represent the normalized power generation efficiency, normalized power generation, and normalized energy storage weight, respectively. , , Represents the weight parameters. + + =1;

[0278] Will Sort by high to low, select the top 20% to 30% (preferably the top 25%) of the joint simulation sample database of optimized design variable combinations as high-performance reference samples;

[0279] Calculate the power generation revenue per unit rotational speed for each sample in the high-performance reference sample. The minimum value among them is taken as the lower limit of power generation revenue per unit speed. ;in, This represents the i-th combination of optimization design variables in the high-performance reference sample.

[0280] The mutation probability is calculated based on the severity of the dominant mismatch, and a decision is made on whether to perform mutation based on the mutation probability. If mutation is performed, the priority variable group to be adjusted is determined based on the type of dominant mismatch, and mutation is then performed. This includes:

[0281] The probability of variation is calculated based on the severity of the dominant mismatch, and the base probability of variation is set as follows: Then the mutation probability for:

[0282] ;

[0283] in, The variation intensity adjustment coefficient is [0.5,2]; () represents the basic probability of variation (which can be determined based on the number of candidate design variables, with the preferred option being...). ,in The number of candidate design variables; in this invention ,therefore (1 / 6 can be taken).

[0284] The decision to perform mutation is based on the mutation probability. If mutation is deemed necessary, the priority variable group to be adjusted and mutated is determined according to the dominant mismatch type, including:

[0285] Rule-based directional optimization: based on the dominant mismatch type of candidate design variable combinations. Determine which group of variables to adjust first;

[0286] when When =1, it indicates that the dominant mismatch type of the candidate design variable combination is the insufficient energy supply mismatch type; the priority adjustment variable group is: within the range of change (the range of change of the optimization design variables set in the driving co-simulation model, i.e. the value range) increase or decrease the nominal capacity V of the accumulator, increase the initial gas pressure P1, and increase or decrease the gas polyvariance index r to improve the energy release process of the accumulator.

[0287] when When = 2, it indicates that the dominant mismatch type of the candidate design variable combination is motor drive mismatch; when the drive capability is insufficient, i.e. When the variable group is adjusted, the priority is to increase or decrease P1 or increase L1 within the set range of change; when L1 or N exceeds the upper limit of the range of change and the predicted power generation efficiency is lower than the current population average level, it is determined that the motor displacement or speed is too high, resulting in a decrease in efficiency, and the priority is to decrease L1 or N within the set range of change.

[0288] when When the value is 3, the dominant mismatch type of the candidate design variable combination is generator matching mismatch; when the electric frequency f is higher than the allowable electric frequency range of the generator... Then, the priority adjustment of the variable group is: within the set range of change, increase or decrease P, or decrease N; when the electrical frequency f is below the allowable range. If so, the variable group should be adjusted to increase or decrease P or increase N within the set range of change;

[0289] Prioritize adjusting the variable group, i.e., the dominant mismatch type. =1 or =2 or =3 Variables in the candidate design variable combinations that need to be increased or decreased When increasing or decreasing, adjustments are made based on the directional disturbance, as shown below:

[0290] ;

[0291] in, The mutated variable value; and They are respectively The upper and lower limits (the upper and lower limits of the range of values). A random number within the interval [0,1]; The direction coefficient is used to adjust the variable, indicating the direction of increase or decrease of the variable. It is used when it is necessary to increase the variable. hour, =1, when it is necessary to reduce the variable hour, =-1; The adaptive perturbation coefficients for the k-th iteration are shown below:

[0292] ;

[0293] in, The initial perturbation coefficients are [0.05, 0.2], The minimum perturbation coefficient is [0.005, 0.05]. The maximum number of iterations is (200). Therefore, a larger perturbation amplitude is maintained in the early stage of optimization to enhance the global search capability, and the perturbation amplitude is gradually reduced in the later stage of optimization to improve the local convergence accuracy.

[0294] After completing the directional perturbation, the mutated variable values ​​are subjected to boundary processing, as shown below:

[0295] ;

[0296] in, and They are respectively The upper and lower limits; to ensure that the combination of candidate design variables after mutation is within the allowable range of values;

[0297] The design variables are then divided into three functional modules: accumulator, hydraulic motor and generator. The modules are then cross-checked to ensure that the offspring can inherit the better local matching structure from the parent. After the cross-check, the coupling variable N is checked in a unified manner to avoid conflicts between the motor module and the generator module.

[0298] Calculate the non-dominated solution contribution and population diversity of the island population, and perform population migration based on the non-dominated solution contribution and population diversity; including:

[0299] Adaptive Island Migration: To avoid premature convergence of island populations during evolution and to improve the exchange efficiency of superior parameter combinations between different islands, this invention adaptively adjusts the island migration strategy based on the non-dominated solution contribution and population diversity of each island.

[0300] The contribution of the s-th island to the non-dominated solution in the k-th iteration is calculated as follows:

[0301] ;

[0302] in, This represents the contribution of the s-th island to the external non-dominated solution archive in the k-th iteration, i.e., the non-dominated solution contribution. This represents the number of individuals added to the external non-dominated solution file for the s-th island in the k-th iteration, which is the number of candidate design variable combinations. This represents the population size, or number of individuals, of the s-th island; The larger the value, the better the current search direction for the island and the stronger its ability to generate non-dominated solutions;

[0303] The population diversity of the s-th island in the k-th iteration is calculated as follows:

[0304] ;

[0305] in, This represents the population diversity of the s-th island; Let represent the normalized vector of the i-th individual in the s-th island; This represents the normalized population center of the s-th island (the mean of all normalized individual vectors in the s-th island, which is a common calculation method in multi-island genetic algorithms or population diversity assessment). Indicates Euclidean distance;

[0306] according to and Population migration includes:

[0307] When the non-dominated solution contribution of a certain island When the contribution rate is higher than the preset contribution threshold (20%), the island is designated as the elite output island, and individuals in the external non-dominated solution archive of the elite output island are migrated to other islands. Alternatively, the Euclidean distance between individuals in the current population of the elite output island and each individual in the external non-dominated solution archive of the elite output island is calculated, and individuals with an Euclidean distance less than or equal to the preset threshold (0.1) are migrated to other islands to accelerate the diffusion of superior search directions.

[0308] When the population diversity of a certain island When the diversity threshold is lower than the preset threshold (0.15), the island is used as the differential input island; calculate the Euclidean distance between other islands and the differential input island, and introduce individuals from other islands with an Euclidean distance greater than or equal to the preset distance threshold (0.3) into the differential input island to reduce the risk of premature convergence.

[0309] When an island simultaneously satisfies both high contribution and high diversity, that is... >0.25 and When the value is greater than 0.25, a portion (20%) of the individuals in the elite output island will not be replaced in order to maintain the effective search direction of the island;

[0310] When individuals migrate to other islands or are introduced into differentiated input islands, the migration or introduction ratio is 5%; when an island has not produced a new non-dominated solution for multiple consecutive iterations (10 generations), the number of newly added individuals in the external non-dominated solution archive is [not specified]. When the value is 0, increase the island migration or introduction ratio to 15%, or inject resources into the island to meet the set variation range and generator frequency constraints. And the energy storage capacity constraint, i.e. ≤ Random individuals.

[0311] Repeatedly execute the adaptive multi-island genetic algorithm until the loop termination condition is met to obtain the optimal objective function; including:

[0312] Define the proportion of newly added non-dominated solutions in the external non-dominated solution archive during the k-th iteration as:

[0313] ;

[0314] in, This indicates the proportion of newly added non-dominated solutions; if it is the proportion of newly added non-dominated solutions in the external non-dominated solution archive after G consecutive iterations (e.g., G=10). If all values ​​are less than the set threshold (0.02), then the external non-dominated solution archive is considered to be stable.

[0315] After the external non-dominated solution portfolio stabilizes, the iterative loop stops. Using MATLAB software, Pareto plots are generated for the Pareto non-dominated solutions (i.e., the objective function values ​​corresponding to the candidate design variable combinations) within the stabilized external non-dominated solution portfolio. Figure 4 As shown, the comprehensive evaluation value of the Pareto non-dominated solution is calculated using a decision mechanism, as shown below:

[0316] ;

[0317] in, This is expressed as a comprehensive evaluation value. , , These represent the normalized predicted power generation efficiency, the normalized predicted power generation, and the normalized predicted energy storage weight, respectively. , , Let these be the weighting coefficients for power generation efficiency, power generation, and weight, respectively, and satisfy the following conditions: + + =1;

[0318] Comprehensive evaluation value The highest Pareto nondominated solution is taken as the optimal objective function. The optimal design variables corresponding to the optimal objective function value, namely the nominal capacity V of the energy storage device, the initial gas pressure P1, the gas polyvariance index r, the motor displacement L1, the number of pole pairs P of the permanent magnet synchronous generator, and the set speed N, are substituted into the simulation model of the ocean thermal energy conversion system for simulation analysis. The optimized effects, namely the power generation and power generation efficiency, are obtained, and the multi-objective optimization of the power generation system is completed.

[0319] Weighting coefficient , , The values ​​are 0.4, 0.4, and 0.2 respectively.

[0320] like Figure 5 , Figure 6 As shown in the graphs, the red line represents the result after optimization, and the blue line represents the result before optimization. Both graphs demonstrate a significant improvement in both power generation efficiency and power output (in Joules) after optimization. The nominal capacity of the energy storage device remains at 1L both before and after optimization, so the weight of the energy storage device remains unchanged. The overall optimization result of "increased power generation efficiency, increased power output, and unchanged weight" is essentially a successful application of a multi-objective optimization algorithm under engineering constraints: it focuses on core performance requirements through a weighting strategy while achieving multi-parameter collaborative optimization without increasing weight. This proves the engineering value of "increasing efficiency without increasing weight, and achieving balanced improvement across multiple objectives," fully meeting the practical needs of ocean thermal energy conversion systems for "lightweight design and high performance."

[0321] Example 3

[0322] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the multi-objective optimization method for marine buoy power generation system based on energy link mismatch diagnosis guidance as described in Embodiment 1 or 2.

[0323] Example 4

[0324] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the multi-objective optimization method for marine buoy power generation system based on energy link mismatch diagnosis guidance as described in Embodiment 1 or 2.

[0325] Example 5

[0326] A multi-objective optimization system for marine buoy power generation systems based on energy link mismatch diagnosis guidance includes:

[0327] The simulation analysis module is configured to: establish a simulation model of the ocean thermal energy conversion power generation system and perform simulation analysis to determine the optimal design variables and optimization target parameters; and perform joint simulation of the ocean thermal energy conversion power generation system simulation model to obtain the combination of optimal design variables and the objective function value.

[0328] The co-simulation module is configured to: obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to a preset accuracy threshold to obtain the adjusted RBF approximation model;

[0329] The RBF approximation model optimization module is configured to: obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to the preset accuracy threshold to obtain the adjusted RBF approximation model;

[0330] The multi-objective optimization module is configured to use an adaptive multi-island genetic algorithm to perform multi-objective optimization and obtain the optimal objective function value and the corresponding optimization design variables.

[0331] Adaptive multi-island genetic algorithms include:

[0332] Based on the optimized design variable combination, the populations of multiple islands are set, the adjusted RBF approximation model is used to evaluate the populations of each island, the predicted values ​​are obtained, and an external non-dominated solution archive is constructed.

[0333] Based on the island's population and predicted values, calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index, and determine the dominant mismatch type and severity.

[0334] The mutation probability is calculated based on the severity of the dominant mismatch, and the decision is made on whether to perform mutation based on the mutation probability. If mutation is to be performed, the priority variable group to be adjusted is determined based on the type of dominant mismatch and mutation is then performed.

[0335] Calculate the non-dominated solution contribution and population diversity of the island population, and perform population migration based on the non-dominated solution contribution and population diversity;

[0336] Repeat the adaptive multi-island genetic algorithm until the loop termination condition is met to obtain the optimal objective function;

[0337] The simulation module is configured as follows: Step 4: Substitute the optimization design variables corresponding to the optimal objective function value into the simulation model of the ocean thermal energy conversion system to complete the multi-objective optimization.

Claims

1. A multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance, characterized in that, include: Step 1: Establish a simulation model of the ocean thermal energy conversion system and conduct simulation analysis to determine the optimization design variables and optimization target parameters; A joint simulation was performed on a simulation model of an ocean thermal energy conversion power generation system to obtain the optimal combination of design variables and the objective function value. Step 2: Obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to the preset accuracy threshold to obtain the adjusted RBF approximation model; Step 3: Use the adaptive multi-island genetic algorithm to perform multi-objective optimization, and obtain the optimal objective function value and the corresponding optimization design variables; Adaptive multi-island genetic algorithms include: Based on the optimized design variable combination, the populations of multiple islands are set, the adjusted RBF approximation model is used to evaluate the populations of each island, the predicted values ​​are obtained, and an external non-dominated solution archive is constructed. Based on the island's population and predicted values, calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index, and determine the dominant mismatch type and severity. The mutation probability is calculated based on the severity of the dominant mismatch, and the decision is made on whether to perform mutation based on the mutation probability. If mutation is to be performed, the priority variable group to be adjusted is determined based on the type of dominant mismatch and mutation is then performed. Calculate the non-dominated solution contribution and population diversity of the island population, and perform population migration based on the non-dominated solution contribution and population diversity; Repeat the adaptive multi-island genetic algorithm until the loop termination condition is met to obtain the optimal objective function; Step 4: Substitute the optimization design variables corresponding to the optimal objective function value into the simulation model of the ocean thermal energy conversion system to complete the multi-objective optimization.

2. The multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance as described in claim 1, characterized in that, The specific implementation process of step 1 includes: A simulation model of an ocean thermal energy conversion system was built in Amesim software, including an energy storage device, a directional valve, a motor, a torque sensor, a speed sensor, a PID module, a load, a rectifier module, a current sensor, a power generation acquisition module, a power generation efficiency acquisition module, and a permanent magnet synchronous generator. The accumulator simulates the volume change of the heat exchange medium by varying its internal pressure, enabling charge and discharge control of the ocean thermal energy conversion system simulation model. A reversing valve regulates whether the pressure inside the accumulator drops. A motor converts the hydraulic energy of the oil into mechanical energy. A torque sensor detects the motor torque, with the motor and speed sensor connected to its two sides respectively. The speed sensor connects to a permanent magnet synchronous generator (PMSG) to collect the actual speed of the PMSG. The PMSG converts mechanical energy into electrical energy. The PID module is used to calculate the error between the actual speed and the target speed of the permanent magnet synchronous generator, and adjust the speed of the permanent magnet synchronous generator to the target speed. The rectifier module is used to convert electrical energy from AC to DC and supply power to the load; The current sensor, power generation acquisition module, and power generation efficiency acquisition module collect the current, power generation, and power generation efficiency of the ocean thermal energy conversion power generation system simulation model in real time. Set the pre-charge pressure of the energy storage device and set the optimization target parameter as power generation efficiency. The parameters in the simulation model of the ocean thermal energy conversion system are changed and analyzed by measuring power generation and weight of the energy storage device. The optimization design variables that affect the optimization target parameters of the simulation model are obtained, including: nominal capacity of energy storage device V, initial gas pressure P1, gas polyvariance index r, motor displacement L1, number of pole pairs of permanent magnet synchronous generator P, and set speed N. The co-simulation model is driven by the DOE module, the Amesim simulation module, the Calculator module, and the MATLAB module. In the DOE module, the DOE component in the Isight software is used to perform a parametric scan of the optimization design variables. Specifically, the orthogonal experimental design method is selected in the DOE component, and the range of variation of the optimization design variables is set. Orthogonal experimental design is then performed on the optimization design variables. The Array type is specified as L64 in the DOE component to generate the combination of optimization design variables X_i = [L1, P1, r, P, V, N], where i = 1, 2, ..., 64. The Amesim simulation module calls the established ocean thermal energy conversion power generation system simulation model and receives i sets of optimized design variable combinations from the DOE module. It then performs dynamic simulation calculations in sequence and outputs the ocean thermal energy conversion power generation efficiency curve η(t), power generation curve power(t), and energy storage nominal capacity V. The Calculator module calculates the weight of the accumulator by establishing a mapping relationship between the nominal capacity and weight of the accumulator. Specifically, the Calculator module reads the nominal capacity V value of the accumulator output by the Amesim simulation module and calculates the weight weight_i of the accumulator through the mapping relationship between the nominal capacity and weight. The MATLAB module is used to extract the maximum value of power generation efficiency as the power generation efficiency max_eff_i and the maximum value of power generation as the power generation max_pwr_i from the time history curves output by the Amesim simulation module, namely the power generation efficiency curve η(t) and the power generation curve power(t). The objective function values ​​[max_eff_i, max_pwr_i, weight_i] of the i sets of optimized design variable combinations are stored in the joint simulation sample database to obtain the final joint simulation sample database. The nominal capacity V of the accumulator varies from [0.4 0.63 1.0 1.6], the initial gas pressure P1 varies from [100 150 200], the gas polytropic index r varies from [1.0 1.2 1.4], the motor displacement L1 varies from [0.13 0.16 0.25], the number of pole pairs P of the permanent magnet synchronous generator varies from [2 4 5 10], and the set speed N varies from [1000 2000 2500 3000].

3. The multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance as described in claim 2, characterized in that, The specific implementation process of step 2 includes: In Isight, the optimal design variable combination X_i = [L1, P1, r, P, V, N] is used as input, and Y_i = [max_eff_i, max_pwr_i, weight_i] is used as output. The corresponding output targets, i.e., the predicted power generation efficiency values, are obtained through the RBF approximation model. Forecasted power generation Predicted weight of accumulator ; Based on the predicted values ​​of power generation efficiency, power generation, and energy storage weight, the prediction accuracy evaluation index, i.e., the coefficient of determination R^2, is calculated. It is then determined whether the prediction accuracy evaluation index meets the preset accuracy threshold: R^2 > R_lim^2; where R_lim^2 is the preset accuracy threshold. If the preset accuracy threshold is not met, the radial basis function type, shape parameter or smoothing parameter of the RBF approximation model is adjusted, and multiple optimized design variable combinations are added to the region with large prediction error until the prediction value of the RBF approximation model meets the preset accuracy threshold R_lim^2, and the adjusted RBF approximation model is obtained. The precision threshold R_lim^2 is 0.

9.

4. The multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance according to claim 3, characterized in that, Based on the optimized design variable combination, populations for multiple islands are defined. An adjusted RBF approximation model is used to evaluate the populations of each island, obtain predicted values, and construct an external non-dominated solution archive; including: Let the population of the s-th island in the k-th iteration be... for: ； Where n_s represents the number of individuals in the s-th island, i.e., the number of combinations of candidate design variables. Let be the i-th candidate design variable combination in the s-th island of the k-th iteration, where s∈S and S represents the number of islands; where, in the 1st iteration, the candidate design variable combination is the optimal design variable combination that drives the co-simulation model to generate. The adjusted RBF approximation model was used to evaluate the populations on each island, and the target predicted values ​​corresponding to the candidate design variable combinations were obtained, i.e., the predicted power generation efficiency values. Forecasted power generation And the predicted weight of the accumulator ,in Indicates the combination of candidate design variables; The target prediction value is normalized as follows: ； ； ； in, , , These represent the normalized predicted power generation efficiency, the normalized predicted power generation, and the normalized predicted energy storage weight, respectively. This represents the minimum predicted value of power generation efficiency among the candidate design variable combinations. This represents the maximum predicted power generation efficiency value among the candidate design variable combinations. This represents the maximum predicted power generation value among the candidate design variable combinations. This represents the minimum predicted power generation value among the candidate design variable combinations. This represents the maximum predicted value of the accumulator weight among the candidate design variable combinations. This represents the minimum predicted value of the accumulator weight among the candidate design variable combinations; During the k-th iteration, the population of each island is... Merge with the current external non-dominated solution archive, as shown below: ； in, Let be the candidate set for the k-th iteration. This represents the external non-dominated solution archive, which is empty during the first iteration. Traverse the candidate set Then, the external non-dominated solution archives of the (k+1)th generation are generated through screening. As shown below: ； in, This indicates the Pareto non-dominated selection operation, i.e., for the candidate set... Any two optimal solutions are combinations of candidate design variables. and If the following conditions are met: and and And at least one objective is strictly superior to ,Right now or or Then determine Dominate Remove from external non-inferior solution archives For the candidate set After performing Pareto nondominated screening on all candidate design variable combinations, the (k+1)th generation external nondominated solution archive is formed. .

5. The multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance according to claim 4, characterized in that, Based on the island's population and predicted values, calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index, and determine the dominant mismatch type and severity; including: Calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index; Among them, the calculation of energy supply insufficiency and mismatch index This includes: calculating the hydraulic energy that the accumulator can release based on the combination of candidate design variables, as shown below: ； When r=1, the hydraulic energy that the accumulator can release is expressed as: ； in, This indicates that the accumulator can release hydraulic energy. This represents the initial gas pressure of the candidate design variable combination. This represents the nominal capacity of the energy storage device, representing the combination of candidate design variables. The gas variability index represents the combination of candidate design variables. This is the minimum operating pressure of the accumulator, which is slightly higher than the pre-charge pressure of the accumulator. This indicates the highest operating pressure of the accumulator, i.e., the initial gas pressure P1. Introducing an energy storage energy release correction factor To obtain effective energy supply capacity As shown below: ； Power generation prediction values ​​obtained from the RBF approximation model and power generation efficiency forecast Calculate the equivalent input energy demand As shown below: ； Construct an energy supply mismatch index: ； in, Indicators of insufficient or mismatched energy supply; Calculate motor drive mismatch index This includes: selecting the top 20%–30% of optimal design variable combinations with max_eff_i as the reference sample set from the joint simulation sample database; calculating the motor drive capability index in the reference sample set, as shown below: ； in, This represents the motor drive capability index of the reference sample set. This represents the initial gas pressure of the reference sample set. This represents the motor displacement of the reference sample set. This indicates the set rotational speed of the reference sample set; The range of values ​​for calculating the motor drive capability index: ； in, and These are the minimum and maximum values ​​of the motor drive capability index, respectively, serving as the lower and upper limits; For candidate design variable combinations in the island population, calculate the motor drive capability index for the candidate design variable combinations and construct a motor drive mismatch index: ； in, These are the weighting coefficients. This indicates a motor drive mismatch index. This represents the normalized predicted power generation efficiency. A motor-driving capability index representing the combination of candidate design variables in an island population; Calculate generator matching mismatch index This includes: calculating the electrical frequency based on the number of permanent magnet synchronous generator pole pairs P and the set rotational speed N in the candidate design variable combinations for the island, as shown below: ； Assume the permissible frequency range of the generator is: ； in, The generator's electrical frequency, and These are the lower and upper limits of the generator's effective frequency operating range, respectively. That is, the corresponding model of the permanent magnet synchronous generator is found based on P, and the upper and lower limits are determined based on the corresponding model. Construct generator frequency mismatch index : ； Calculate the power generation revenue per unit rotational speed based on the predicted power generation value and the set rotational speed. : ； Constructing a power generation revenue mismatch index : ； in, This represents the lower limit of the revenue generated per unit speed of electricity. The generator matching mismatch index was obtained by comprehensive analysis. : ； in, and These are the weighting coefficients; Three types of mismatch indicators, namely, insufficient energy supply mismatch indicators Motor drive mismatch index Generator matching mismatch index Normalization is performed as follows: ； in, This represents the normalized mismatch index, j=1,2,3. This represents the maximum value of the j-th type of mismatch index in the current population; The type corresponding to the maximum value in the normalized mismatch index is taken as the dominant mismatch type of the candidate design variable combination, as shown below: ； in, The dominant mismatch type for candidate design variable combinations. Indicates selection The number corresponding to the maximum value; the severity of the dominant mismatch is defined as: ； in, To determine the severity of the mismatch, Indicates selection The maximum value; Lower limit of power generation revenue per unit speed Determined based on the joint simulation sample database; including: The performance scores of the objective function values ​​in the co-simulation sample database are calculated as follows: ）； in, Indicates the performance score. This represents the i-th combination of optimal design variables in the joint simulation sample database. , , These represent the normalized power generation efficiency, normalized power generation, and normalized energy storage weight, respectively. , , Represents the weight parameters. + + =1; Will Sort by high to low, select the top 20% to 30% of the optimized design variable combinations in the joint simulation sample database as high-performance reference samples; Calculate the power generation revenue per unit rotational speed for each sample in the high-performance reference sample. The minimum value among them is taken as the lower limit of power generation revenue per unit speed. ;in, This represents the i-th combination of optimization design variables in the high-performance reference sample.

6. The multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance according to claim 5, characterized in that, Calculate the mutation probability based on the severity of the dominant mismatch, and determine whether to perform mutation based on the mutation probability. If mutation is performed, the priority variable group to be adjusted and mutated is determined based on the dominant mismatch type; including: The probability of variation is calculated based on the severity of the dominant mismatch, and the base probability of variation is set as follows: Then the mutation probability for: ； in, This is the variation intensity adjustment coefficient; () represents the basic mutation probability; The decision to perform mutation is based on the mutation probability. If mutation is deemed necessary, the priority variable group to be adjusted and mutated is determined according to the dominant mismatch type, including: when When =1, it indicates that the dominant mismatch type of the candidate design variable combination is insufficient energy supply mismatch; the priority adjustment variable group is: increase or decrease the nominal capacity V of the accumulator, increase the initial gas pressure P1, and increase or decrease the gas polyvariance index r within the range of variation. when When = 2, it indicates that the dominant mismatch type of the candidate design variable combination is motor drive mismatch; when the drive capability is insufficient, i.e. When the variable group is adjusted, the priority is to increase or decrease P1 or increase L1 within the set range of change; when L1 or N exceeds the upper limit of the range of change and the predicted power generation efficiency is lower than the current population average level, it is determined that the motor displacement or speed is too high, resulting in a decrease in efficiency, and the priority is to decrease L1 or N within the set range of change. when When the value is 3, the dominant mismatch type of the candidate design variable combination is generator matching mismatch; when the electric frequency f is higher than the allowable electric frequency range of the generator... Then, the priority adjustment of the variable group is: within the set range of change, increase or decrease P, or decrease N; when the electrical frequency f is below the allowable range. If so, the variable group should be adjusted to increase or decrease P or increase N within the set range of change; Prioritize adjusting the variable group, i.e., the dominant mismatch type. =1 or =2 or =3 Variables in the candidate design variable combinations that need to be increased or decreased When increasing or decreasing, adjustments are made based on the directional disturbance, as shown below: ； in, The mutated variable value; and They are respectively The upper and lower limits; A random number within the interval [0,1]; The direction coefficient is used to adjust the variable, indicating the direction of increase or decrease of the variable. It is used when it is necessary to increase the variable. hour, =1, when it is necessary to reduce the variable hour, =-1; The adaptive perturbation coefficients for the k-th iteration are shown below: ； in, The initial disturbance coefficients are... The minimum disturbance coefficient, The maximum number of iterations; After completing the directional perturbation, the mutated variable values ​​are subjected to boundary processing, as shown below: ； in, and They are respectively The upper and lower limits; Calculate the non-dominated solution contribution and population diversity of the island population, and perform population migration based on the non-dominated solution contribution and population diversity; including: The contribution of the s-th island to the non-dominated solution in the k-th iteration is calculated as follows: ； in, This represents the contribution of the s-th island to the external non-dominated solution archive in the k-th iteration, i.e., the non-dominated solution contribution. This represents the number of individuals added to the external non-dominated solution file for the s-th island in the k-th iteration, which is the number of candidate design variable combinations. This represents the population size, or number of individuals, of the s-th island; The population diversity of the s-th island in the k-th iteration is calculated as follows: ； in, This represents the population diversity of the s-th island; Let represent the normalized vector of the i-th individual in the s-th island; This represents the population center of the s-th island after normalization. Indicates Euclidean distance; according to and Population migration includes: When the non-dominated solution contribution of a certain island When the contribution exceeds the preset threshold, the island is designated as the elite output island, and individuals in the external non-dominated solution archive of the elite output island are migrated to other islands. Alternatively, the Euclidean distance between individuals in the current population of the elite output island and each individual in the external non-dominated solution archive of the elite output island is calculated, and individuals whose Euclidean distance is less than or equal to the preset threshold are migrated to other islands. When the population diversity of a certain island When the diversity threshold is lower than the preset threshold, the island is used as the differential input island; calculate the Euclidean distance between other islands and the differential input island, and introduce individuals from other islands whose Euclidean distance is greater than or equal to the preset distance threshold into the differential input island; When an island simultaneously satisfies both high contribution and high diversity, that is... >0.25 and When the value is greater than 0.25, some individuals in the elite output island will be retained and not replaced; When individuals migrate to other islands or are introduced into differentiated input islands, the migration or introduction ratio is 5%; when an island fails to produce a new non-dominated solution (i.e., the number of newly added individuals from external non-dominated solution archives) for multiple consecutive iterations. When the value is 0, increase the island migration or introduction ratio to 15%, or inject resources into the island to meet the set variation range and generator frequency constraints. And the energy storage capacity constraint, i.e. ≤ Random individuals.

7. The multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance according to claim 6, characterized in that, Repeat the adaptive multi-island genetic algorithm until the loop termination condition is met to obtain the optimal objective function; include: Define the proportion of newly added non-dominated solutions in the external non-dominated solution archive during the k-th iteration as: ； in, This represents the proportion of newly added non-dominated solutions; if the proportion of newly added non-dominated solutions in the external non-dominated solution archive is G consecutive iterations... If all values ​​are less than the set threshold, the external non-dominated solution archive is considered to be stable. After the external non-dominated solution portfolio stabilizes, the iterative loop stops. Using MATLAB software, Pareto plots are generated for the objective function values ​​corresponding to the Pareto non-dominated solutions (i.e., combinations of candidate design variables) in the stabilized external non-dominated solution portfolio. A decision mechanism is then used to calculate the comprehensive evaluation value of the Pareto non-dominated solutions, as shown below: ； in, This is expressed as a comprehensive evaluation value. , , These represent the normalized predicted power generation efficiency, the normalized predicted power generation, and the normalized predicted energy storage weight, respectively. , , Let these be the weighting coefficients for power generation efficiency, power generation, and weight, respectively, and satisfy the following conditions: + + =1; Comprehensive evaluation value The highest Pareto nondominated solution is taken as the optimal objective function. The optimal design variables corresponding to the optimal objective function value, namely the nominal capacity of the energy storage device V, the initial gas pressure P1, the gas polyvariance index r, the motor displacement L1, the number of pole pairs of the permanent magnet synchronous generator P, and the set speed N, are substituted into the simulation model of the ocean thermal energy conversion system for simulation analysis. The optimized effect, namely the power generation and power generation efficiency, is obtained, and the multi-objective optimization of the power generation system is completed. Weighting coefficient , , The values ​​are 0.4, 0.4, and 0.2 respectively.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of a multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of a multi-objective optimization method for marine buoy power generation systems based on energy link mismatch diagnosis guidance.

10. A multi-objective optimization system for marine buoy power generation systems based on energy link mismatch diagnosis guidance, characterized in that: include: The simulation analysis module is configured to: establish a simulation model of the ocean thermal energy conversion power generation system and perform simulation analysis to determine the optimization design variables and optimization target parameters; A joint simulation was performed on a simulation model of an ocean thermal energy conversion power generation system to obtain the optimal combination of design variables and the objective function value. The co-simulation module is configured to: obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to a preset accuracy threshold to obtain the adjusted RBF approximation model; The RBF approximation model optimization module is configured to: obtain the predicted value of the objective function through the RBF approximation model, and adjust the RBF approximation model according to the preset accuracy threshold to obtain the adjusted RBF approximation model; The multi-objective optimization module is configured to use an adaptive multi-island genetic algorithm to perform multi-objective optimization and obtain the optimal objective function value and the corresponding optimization design variables. Adaptive multi-island genetic algorithms include: Based on the optimized design variable combination, the populations of multiple islands are set, the adjusted RBF approximation model is used to evaluate the populations of each island, the predicted values ​​are obtained, and an external non-dominated solution archive is constructed. Based on the island's population and predicted values, calculate the power supply mismatch index, motor drive mismatch index, and generator matching mismatch index, and determine the dominant mismatch type and severity. The mutation probability is calculated based on the severity of the dominant mismatch, and the decision is made on whether to perform mutation based on the mutation probability. If mutation is to be performed, the priority variable group to be adjusted is determined based on the type of dominant mismatch and mutation is then performed. Calculate the non-dominated solution contribution and population diversity of the island population, and perform population migration based on the non-dominated solution contribution and population diversity; Repeat the adaptive multi-island genetic algorithm until the loop termination condition is met to obtain the optimal objective function; The simulation module is configured as follows: Step 4: Substitute the optimization design variables corresponding to the optimal objective function value into the simulation model of the ocean thermal energy conversion system to complete the multi-objective optimization.

Images