High-dimensional multi-target multi-working-condition optimization method and system for centrifugal air compressor of fuel cell

Abstract

The invention belongs to the technical field of fuel cells, and particularly relates to a high-dimensional multi-target multi-working-condition optimization method and system for a fuel cell centrifugal air compressor, and the method comprises the steps: constructing a fuel cell air compressor performance analysis and optimization system, a comprehensive performance test platform is built, a multi-physics field coupling CFD simulation model is built, and the simulation and optimization efficiency is improved in combination with a machine learning algorithm; the pressure ratio and isentropic efficiency are defined as optimization targets based on fuel cell system requirements, and the performance of the air compressor is evaluated through experiment and simulation data; a multi-objective multi-population optimization method, an archive sharing technology, a genetic algorithm and the like are fused to establish an MMPS algorithm, and high-dimensional multi-objective optimization is carried out on related structure parameters of an impeller, a volute and a diffuser. The technical problems that the multi-target collaborative optimization requirement of an existing fuel cell centrifugal air compressor is difficult to meet, the high-dimensional variable optimization difficulty is large, and complex flow characteristic control is difficult are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of fuel cells, and in particular, relates to a high-dimensional multi-objective and multi-operating condition optimization method and system for a fuel cell centrifugal air compressor. Background Art

[0002] Fuel cell technology has attracted worldwide attention and has developed rapidly due to its high efficiency, cleanliness and low pollution emissions. As an important supporting technology for promoting green and low-carbon goals, it has shown broad application potential in transportation, stationary power generation and portable energy. The centrifugal air compressor is the core air supply unit of the fuel cell system, responsible for providing high-pressure air to the fuel cell stack. Its performance directly affects the power generation efficiency, power density, durability and overall system performance of the fuel cell. Therefore, optimizing the performance of the centrifugal air compressor has become a key research direction for improving the efficiency of the fuel cell system.

[0003] At present, the status quo of fuel cell centrifugal air compressor technology is as follows:

[0004] 1. It is difficult to meet the requirements of multi-objective collaborative optimization: Fuel cell systems often face complex working conditions in practical applications, such as instantaneous load changes, high dynamic response and long-term stability. This requires centrifugal air compressors to maintain efficient and stable performance under a variety of working conditions. However, traditional optimization methods are usually guided by a single objective or static working conditions, and it is difficult to take into account dynamic multi-objective requirements, thus limiting the performance of air compressors in complex scenarios.

[0005] 2. High-dimensional variables are difficult to optimize: The design of centrifugal air compressors involves multi-dimensional variables such as geometric parameters, fluid mechanics, speed and load, which are highly nonlinear and strongly coupled. Global optimization in multi-dimensional design space requires efficient algorithm support. However, traditional optimization methods are often limited to local optimal solutions and are difficult to meet engineering requirements in terms of accuracy and efficiency.

[0006] 3. Difficulty in controlling complex flow characteristics: Complex flow characteristics such as turbulence and boundary layer separation inside centrifugal air compressors require optimization design to consider the nonlinearity and dynamic changes of the flow field. Traditional optimization tools are difficult to fully capture these complex characteristics and cannot achieve systematic performance improvement.

[0007] For example, patent CN118013658A discloses a method for optimizing the design of a turbine air compressor based on fuel cell tail exhaust energy recovery. This scheme determines the matching points and weights corresponding to typical working conditions through cluster analysis, combines the coupling characteristics of the pressure end and the turbine end, and uses a CFD (computational fluid dynamics) model to simulate and optimize the performance of the air compressor. The specific method includes: establishing a performance prediction model based on enthalpy loss, using the Monte Carlo method to perform sensitivity analysis on turbine geometric parameters, and optimizing the particle swarm algorithm with the goal of maximizing the comprehensive efficiency of the turbine end. Although this technical solution has improved the performance of the air compressor to a certain extent, it still has the following shortcomings: 1. Limitations of single-objective optimization: The improvement of the comprehensive efficiency of the turbine end may lead to a decrease in other performance of the air compressor, and multi-objective collaborative optimization has not been achieved. 2. Insufficient optimization algorithm: The particle swarm algorithm has problems such as local convergence, poor balance between search accuracy and convergence speed, and weak algorithm adaptability and diversity retention. Compared with the advanced optimization algorithms developed in recent years, its efficiency and applicability are already insufficient. 3. High computing resource requirements: Although CFD simulation models have high accuracy, they are computationally complex and have low optimization efficiency, making them difficult to meet the high efficiency requirements in actual engineering projects.

[0008] In summary, the existing fuel cell centrifugal air compressor optimization technology still faces significant challenges in multi-objective collaborative optimization, high-dimensional variable optimization and complex flow field control. In order to further improve the overall performance of the fuel cell system, it is urgent to develop more efficient and comprehensive optimization methods to break through the current technical bottleneck. Summary of the invention

[0009] In view of the defects of the prior art, the present invention proposes a high-dimensional multi-objective and multi-operating condition optimization method and system for a fuel cell centrifugal air compressor. The method realizes high-dimensional multi-objective performance optimization of a fuel cell centrifugal air compressor under complex multi-operating conditions by integrating performance test data, CFD simulation and data-driven modeling, combining a multi-objective and multi-population optimization framework and an MMPS algorithm, and provides scientific guidance and practical basis.

[0010] In one aspect, the present invention provides a high-dimensional multi-objective multi-operating condition optimization method for a fuel cell centrifugal air compressor, the method comprising:

[0011] Step 1: Construct a fuel cell air compressor performance analysis and optimization system. By building a comprehensive performance test platform, establishing a multi-physics field coupling CFD simulation model, and verifying the consistency between the simulation results and the test data, the optimization direction of the geometric model is determined. Based on the orthogonal experimental design, batch simulation is performed to generate multi-condition data. In addition, a data-driven model is constructed and verified in combination with a machine learning algorithm to improve the simulation and optimization efficiency.

[0012] Step 2: Based on the fuel cell system requirements and multi-operating condition data analysis, the optimization objectives are defined as pressure ratio and isentropic efficiency, and the compression capacity and efficiency performance of the air compressor are evaluated through experimental and simulation data calculations;

[0013] Step 3: Integrate multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm, and K-means clustering method to establish the MMPS algorithm. Use the MMPS algorithm to perform high-dimensional multi-objective optimization on the impeller, volute, diffuser and related structural parameters to improve their performance under multiple working conditions.

[0014] Further, step 1 comprises:

[0015] Step 1.1: Build a comprehensive performance test platform for fuel cell air compressors, and use the platform to collect performance parameters of fuel cell air compressors under different operating conditions;

[0016] Step 1.2: Create a geometric model of the fuel cell air compressor, perform multi-physics simulation analysis on the geometric model based on the fluid dynamics model, thermodynamics model, and turbulence model, set the boundary conditions of each model, run multi-physics simulation to cover a variety of operating conditions, and complete the convergence and accuracy verification of the geometric model;

[0017] Step 1.3: Based on the fluid dynamics model, thermodynamics model and turbulence model, boundary conditions are set for each model to establish a multi-physics field coupled CFD simulation model of the fuel cell air compressor;

[0018] Step 1.4: Compare the simulation results with the performance parameters collected by the test platform one by one to verify whether the error of each performance indicator is within the set acceptance range;

[0019] Step 1.5: Based on the comparative analysis of simulation results and test data, determine the optimization direction of the impeller, volute and diffuser in the geometric model of the fuel cell air compressor;

[0020] Step 1.6: Based on the geometric model of the fuel cell air compressor, build a CFD integrated simulation framework for the fuel cell air compressor, perform batch simulations on different structural parameter combinations based on orthogonal experimental design, generate representative simulation data covering multiple working conditions, and pre-process the simulation data;

[0021] Step 1.7: Based on the acquired batch simulation data, use the machine learning algorithm to build a data-driven model of the fuel cell centrifugal air compressor, use the training set to complete the model construction, and use the test set to verify the model performance.

[0022] Furthermore, step 2 includes:

[0023] Step 2.1: Based on the fuel cell system gas supply demand, multi-operating conditions, thermodynamics and aerodynamic theory, and comprehensive analysis of experimental and simulation data, the optimization objectives of the fuel cell air compressor are defined as pressure ratio and isentropic efficiency;

[0024] Step 2.2: Under the condition of a fixed speed ratio, collect the inlet and outlet pressure and temperature data of the air compressor under different speed conditions, and calculate the pressure ratio and isentropic efficiency. According to the calculation results, the compression capacity and efficiency performance of the air compressor are evaluated.

[0025] Further, step 3 includes:

[0026] Step 3.1: Construct a multi-objective multi-population optimization framework, divide the optimization objectives into 10 sub-populations for parallel optimization, and combine the crossover operator and mutation operator based on the genetic algorithm to generate and improve the solution, so as to improve the diversity of the solution set and the global search capability, thereby achieving efficient solution set exploration and optimization in complex multi-objective optimization problems;

[0027] Step 3.2: The crossover operator uses a priority crossover operation to achieve gene recombination by sorting individuals with high fitness values ​​and using a double-optimal crossover strategy, thereby improving population adaptability and solution set diversity. The mutation operator uses an insertion mutation operation to enhance population diversity and global search capabilities by inserting and optimizing genes for individuals with low fitness values, thereby ensuring the stability of the optimization results and improving performance.

[0028] Step 3.3: By constructing archive sharing technology, archive A is used to store and share excellent solutions in the multi-objective multi-population optimization framework to assist information interaction and priority crossover operations between populations, and realize adaptive search and population co-evolution;

[0029] Step 3.4: Propose an archive update strategy to update the archive by screening, perturbing and preserving the diversity of the solution set of the newly generated population based on K-means clustering to improve the convergence and quality of the elite solution, thereby enhancing the co-evolutionary ability and optimization efficiency of the population in the multi-objective multi-population optimization framework;

[0030] Step 3.5: By integrating multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm and K-means clustering method, the MMPS algorithm is established to randomly initialize the population and update the population through preferential crossover operation and insertion mutation operation, and optimize the solution set in combination with the archive update strategy, and finally output the elite solution in the archive as the optimization result.

[0031] Furthermore, the fuel cell air compressor comprehensive performance test platform can simulate typical operating conditions including different flow conditions, different speed conditions, different pressure conditions, different cooling conditions, startup and steady-state conditions and comprehensive conditions.

[0032] Furthermore, in step 1.4, the simulation results are compared with the experimental data in step 1.1 one by one to verify whether the error of each performance indicator is controlled within 5%.

[0033] Furthermore, the optimization directions of the impeller include increasing the blade inlet angle to improve the inflow performance and reducing the outlet angle to improve the uniformity of the airflow; the optimization directions of the volute include optimizing the volute tongue angle to reduce turbulence and adjusting the ellipticity of the volute section to reduce flow separation; the optimization directions of the diffuser include increasing the outlet radius to improve pressure recovery and adjusting the width to improve the uniformity of the outlet flow field.

[0034] Further, the pressure ratio (σ) is determined by the following formula:

[0035] σ=p out / p in

[0036] Where: p out Indicates the outlet pressure of the air compressor; p in Indicates the inlet pressure of the air compressor;

[0037] The isentropic efficiency (η) is determined by the following formula:

[0038]

[0039] Where: T in Indicates the air compressor inlet temperature; T out Indicates the air compressor outlet temperature; k indicates the isentropic index. :

[0040] On the other hand, the present invention also provides a high-dimensional multi-objective multi-operating condition optimization system for a fuel cell centrifugal air compressor, comprising:

[0041] The performance analysis and modeling module is used to build a fuel cell air compressor performance analysis and optimization environment. By building a comprehensive performance test platform and establishing a multi-physics field coupled CFD simulation model, the consistency of simulation results and test data is verified, and multi-operating condition data is generated based on orthogonal experimental design. At the same time, a data-driven model is built and verified in combination with machine learning algorithms to improve simulation and optimization efficiency.

[0042] The optimization target definition module is used to define optimization targets, including pressure ratio and isentropic efficiency, based on fuel cell system requirements and multi-operating condition data analysis, and evaluate the compression capacity and efficiency performance of the air compressor through experimental and simulation data calculations;

[0043] The high-dimensional multi-objective optimization module is used to integrate multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm and K-means clustering method to construct the MMPS algorithm. Based on this algorithm, high-dimensional multi-objective optimization is performed on the structural parameters of the impeller, volute and diffuser to improve the performance of the fuel cell centrifugal air compressor under multiple working conditions.

[0044] The beneficial effects of the present invention are:

[0045] First, based on comprehensive performance test data, multi-physical field coupled CFD simulation model, CFD integrated simulation framework and orthogonal experimental design, the present invention obtains a large amount of high-quality test and simulation data, and establishes an efficient and accurate fuel cell centrifugal air compressor data-driven model through a data-driven algorithm, providing a solid model foundation for realizing high-dimensional, multi-objective and multi-operating condition performance optimization of the air compressor.

[0046] Second, the present invention introduces a multi-objective multi-population optimization framework, which divides the population into multiple sub-populations. Each population works independently or collaboratively to explore different objectives, thereby more effectively finding the Pareto optimal solution set in complex multi-objective problems. The information interaction mechanism is used to achieve solution sharing and migration, balance global search and local development, avoid the population from falling into the local optimum and maintain the diversity of the population, so that the algorithm can better adapt to complex, high-dimensional, multi-objective optimization environments.

[0047] Third, the present invention proposes an MMPS algorithm for the performance optimization of air compressors under different working conditions, forming a high-dimensional multi-objective optimization framework, achieving efficient solutions to complex multi-objective problems, and significantly improving the superiority of the final solution set. At the same time, it can fully consider the performance requirements of air compressors under different working conditions, provide scientific guidance and practical basis for the development of high-performance fuel cell centrifugal air compressors, and meet actual engineering needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of the high-dimensional multi-objective multi-operating condition optimization method for a fuel cell centrifugal air compressor of the present invention;

[0049] Figure 2 2 is a schematic diagram of the layout of a monitoring system for a comprehensive performance test platform for a fuel cell air compressor according to an embodiment of the present invention;

[0050] Figure 3 is a structural schematic diagram of a three-dimensional model of a fuel cell air compressor according to an embodiment of the present invention;

[0051] Figure 4 Schematic diagram of a CFD integrated simulation framework of a fuel cell air compressor according to an embodiment of the present invention;

[0052] Figure 54 is a flow chart of the MMPS algorithm of an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to enable those skilled in the art to better understand the technical solution of the present application, the present application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0054] In the description of this application, the terms "first", "second", and "third" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance; the term "plurality" refers to two or more, unless otherwise expressly defined. Terms such as "installed", "connected", "connected", and "fixed" should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; "connected" can be a direct connection or an indirect connection through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to the specific circumstances.

[0055] In the description of the present application, it should be understood that the directions or positional relationships indicated by the terms "up", "down", "left", "right", "front", "back", etc. are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or unit referred to must have a specific direction, be constructed and operated in a specific orientation, and therefore, should not be understood as a limitation on the present application.

[0056] In the description of this specification, the description of the terms "one embodiment / method", "some embodiments / methods", "specific embodiments / methods", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment / method or example are included in at least one embodiment / method or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment / method or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments / methods or examples in a suitable manner.

[0057] As the instruction manual Figure 1 The purpose of the present invention is to provide a high-dimensional multi-objective and multi-operating condition optimization method for a fuel cell centrifugal air compressor, comprising:

[0058] Step 1: Construct a fuel cell air compressor performance analysis and optimization system. By building a comprehensive performance test platform, establishing a multi-physics field coupled CFD simulation model, and verifying the consistency between the simulation results and the test data, determine the optimization direction of the geometric model, perform batch simulation based on orthogonal experimental design to generate multi-condition data, and combine machine learning algorithms to build and verify data-driven models to improve simulation and optimization efficiency.

[0059] The purpose of step 1 is to build a systematic fuel cell air compressor performance analysis and optimization process. Through comprehensive performance testing, CFD simulation modeling, data verification and optimization direction determination, combined with orthogonal experimental design and machine learning algorithms, representative simulation data under multiple working conditions are generated, and an efficient data-driven model is established to provide a scientific basis and technical support for subsequent air compressor structure optimization and performance improvement.

[0060] It should be noted that the fuel cell air compressor includes an impeller, a volute, a diffuser, a motor, a bearing and a lubrication system, an air inlet and an air outlet, a cooling system and a control system. The impeller is a core component used to convert mechanical energy into kinetic energy and pressure energy of the gas. The key parameters of its structure include the blade inlet installation angle, the outlet installation angle and the blade thickness. The volute is an outer shell surrounding the impeller, which is used to guide the gas flow and convert kinetic energy into pressure energy. The key parameters of its structure include the cross-sectional shape of the volute, the outlet diameter and the angle of the volute tongue. The diffuser is installed between the impeller and the volute for deceleration and pressure increase to improve the compression efficiency. The key parameters of its structure include the outlet radius, inlet diameter and outlet width of the diffuser. The motor is connected to the impeller to provide the mechanical energy required by the impeller. Generally, a high-speed motor is used to drive the impeller to rotate at high speed. The bearing and lubrication system are used to support the rotation of the impeller and the shaft, reduce friction and ensure stable operation of the equipment. The air inlet and the air outlet are connected to the fuel cell system to be responsible for the input and output of the gas. The cooling system includes a cooling water circuit or air cooling design for heat dissipation and maintaining the operating temperature. The control system includes sensors and control modules, which are used to monitor and adjust the operating status of the air compressor in real time, such as temperature, pressure and flow.

[0061] When the air compressor is running, the motor drives the impeller to rotate at high speed, and the blades use centrifugal force to throw the gas out from the center to the outside, increasing the kinetic energy of the gas. The high-speed gas flows into the diffuser, and the gradually expanding structure of the diffuser slows down the airflow, converting the kinetic energy into pressure energy, and significantly increasing the gas pressure. The pressurized gas enters the volute, and the curve design of the volute further optimizes the pressure and flow distribution of the gas, and finally outputs it from the outlet. The control system adjusts the motor speed and impeller parameters (such as variable frequency speed regulation or blade angle adjustment) in real time to adapt to different load requirements and working conditions. Finally, the air compressor provides high-pressure air to the fuel cell, and the oxygen in the air reacts with the hydrogen in the fuel cell to generate electricity and water.

[0062] Specifically, step 1 includes:

[0063] Step 1.1: Build a comprehensive performance test platform for fuel cell air compressors, through which the performance parameters of fuel cell air compressors under different operating conditions are collected.

[0064] As the instruction manual Figure 2The comprehensive performance test platform for fuel cell air compressor includes the air compressor and its motor, air circuit system, water cooling system, power supply and control system.

[0065] Among them, the air circuit system is the core part of the test platform, which is used to guide air through the air compressor and measure relevant performance parameters, including air filter, intake valve, intercooler and exhaust valve. The air filter is used to filter particulate matter in the air and protect the air compressor. The intake valve is used to control the gas flow entering the air compressor. The intercooler is used to reduce the temperature of the air compressed by the air compressor. The exhaust valve is used to control the output of compressed air. The air filter is connected to the air inlet of the air compressor through the intake valve and pipeline. The air inlet of the air compressor is equipped with a pressure sensor (PT) and a temperature sensor (THA) to monitor the pressure and temperature of the input air. The air compressor compresses air under the drive of the motor, and its outlet is connected to the air inlet of the intercooler through a pipeline. The air outlet of the air compressor is equipped with a pressure sensor (PT) and a temperature sensor (THA) to monitor the outlet parameters of the compressed air. The outlet of the intercooler is connected to the exhaust pipe through the exhaust valve and pipeline, and is equipped with a temperature sensor (THA) and a flow sensor (FS) to monitor the cooling effect.

[0066] The water cooling system is used to maintain the thermal management performance of the air compressor and intercooler. It dissipates heat through circulating water and includes a water tank, a water pump, a cooling water flow path, and multiple sensors. The cooling water is stored in the water tank, and the water pump is connected to the water tank to provide a circulating water flow. The water pump is connected to the intercooler and the inlet of the air compressor cooling circuit through a pipeline. The cooling water enters the intercooler through the pipeline, taking away the heat in the compressed air. The water flowing out of the intercooler flows back to the water tank through the pipeline to form a closed loop. The multiple sensors in the water system include a flow sensor (FS) for measuring the cooling water flow, a pressure sensor (PT) for monitoring the water circuit pressure, and a temperature sensor for monitoring the temperature of the cooling water.

[0067] The power supply and control system is used to drive the air compressor and collect data, including DC power supply module, controller and sensor. The DC power supply module (Q-100V adjustable power supply and 12V power supply) is connected to the motor of the air compressor through the control line, where the Q-100V adjustable power supply is used to drive the main motor of the air compressor, and the 12V power supply is used to power auxiliary equipment such as sensors and controllers. The controller is directly connected to all sensors and actuators to monitor the inlet and outlet pressure, flow, temperature data, and to adjust the motor speed, intake valve opening and exhaust valve opening in real time.

[0068] In the implementation of the present application, the fuel cell air compressor comprehensive performance test platform can simulate typical operating conditions including different flow conditions, different speed conditions, different pressure conditions, different cooling conditions, startup and steady-state conditions and comprehensive conditions.

[0069] Different flow conditions simulate low flow, medium flow and high flow conditions, and test the efficiency, pressure ratio, compression power and other performance indicators of the air compressor under different flow rates. Specifically, the intake volume is controlled by adjusting the opening of the intake valve, and the exhaust valve is used to adjust the system flow balance. Under different flow conditions, the low flow rate can be set to 10-30g / s, the medium flow rate to 30-50g / s, and the high flow rate to 50-70g / s (maximum design flow rate). The speed of the air compressor is set to a fixed value (such as 40000rpm), and the outlet pressure of the air compressor is maintained at the design pressure (such as 2bar).

[0070] Different speed conditions are to adjust the driving voltage and frequency of the motor to test the efficiency, noise, vibration and other performance indicators of the air compressor at different speeds. Specifically, use an adjustable DC power supply (Q-100V) to adjust the motor input voltage and frequency to ensure that the motor operates within a stable range. Under different speed conditions, the low speed can be set to 20000rpm, the medium speed to 40000rpm, and the high speed to 60000rpm (the design limit speed of the air compressor), maintain the medium flow rate (such as 40g / s), and the outlet pressure of the air compressor is maintained at the design pressure (such as 2bar).

[0071] Different pressure conditions are to test the compression ratio, efficiency, power consumption and other performance indicators of the air compressor at different outlet pressures by simulating low pressure, medium pressure and high pressure conditions. Specifically, the outlet pressure is controlled by adjusting the opening of the exhaust valve to ensure that the compression ratio of the air compressor during normal operation meets the working condition requirements. Under different pressure conditions, the low pressure condition can be set to 1 bar (close to atmospheric pressure), the medium pressure condition to 2 bar, and the high pressure condition to 3 bar (the upper limit of the air compressor design), maintaining the medium flow rate (such as 40g / s), and the air compressor speed is set to a fixed value (such as 40000rpm).

[0072] Different cooling conditions are used to verify the thermal management performance of the air compressor, including cooling efficiency, outlet temperature and thermal stability. Specifically, the cooling water temperature is adjusted by adjusting the output flow of the water pump and setting the water tank temperature control device. Under different cooling conditions, the low flow rate of cooling water can be set to 1L / min, the medium flow rate to 3L / min, and the high flow rate to 5L / min. The cooling water temperature is 10°C in cold water conditions, 25°C in normal temperature conditions, and 40°C in high temperature conditions. The speed of the air compressor is set to a fixed value (such as 40,000rpm), and the outlet pressure of the air compressor is maintained at the design pressure (such as 2bar).

[0073] The startup and steady-state conditions are to test the startup performance of the air compressor from a stationary state to normal operation, as well as to test the long-term performance of the air compressor under stable operating conditions. Specifically, the air compressor stops in the initial state, and by adjusting the opening of the intake valve and the exhaust valve, and using the power supply to control the dynamic change of the motor speed, as the flow and pressure gradually increase, the air compressor reaches a stable condition, and the acceleration time, stabilization time and initial vibration at startup are monitored. After the air compressor reaches a stable condition, the load change is simulated by instantaneously increasing or decreasing the flow (±20%) and instantaneously adjusting the outlet pressure (±0.5bar), so as to realize the test system response time, pressure fluctuation amplitude and flow change stability. During the whole process, the speed of the air compressor is set to change dynamically (such as 20000rpm→60000rpm), and the cooling water is kept at a constant temperature (such as 25℃).

[0074] Comprehensive working condition is a comprehensive performance indicator for evaluating air compressor under complex working conditions. Specifically, first set the parameter combination, for example: set the speed to medium speed (40000rpm), flow to medium flow (40g / s), pressure to medium pressure (2bar), cooling water temperature to 25℃, cooling water flow to 3L / min, and verify the comprehensive performance by dynamically adjusting a single parameter (such as flow ±10%, pressure ±0.2bar).

[0075] Through the above-mentioned detailed test conditions, the performance test of the fuel cell air compressor can comprehensively cover various actual operation scenarios and provide high-quality experimental data support for subsequent modeling and optimization.

[0076] Step 1.2: Create a geometric model of the fuel cell air compressor, perform multi-physics simulation analysis on the geometric model based on the fluid dynamics model, thermodynamic model and turbulence model, set the boundary conditions of each model, run multi-physics simulation to cover various operating conditions, and complete the convergence and accuracy verification of the geometric model.

[0077] It should be noted that the geometric model of the fuel cell air compressor created in step 1.2 is the same as the geometric structure of the fuel cell air compressor tested by the fuel cell air compressor comprehensive performance test platform in step 1.1. The fluid dynamics model describes the movement behavior of the fluid inside the fuel cell air compressor, helping to understand the distribution characteristics of gas flow and its impact on the performance of the air compressor. The thermodynamic model describes the temperature change and heat transfer behavior of the fluid inside the fuel cell air compressor, and is the key to analyzing the compression process and thermal management performance. The turbulence model describes the turbulent characteristics in complex flow fields, including eddies, turbulent energy dissipation, and the transfer of turbulent kinetic energy, and is the key to simulating flow losses and pressure recovery inside the air compressor.

[0078] Specifically, first use CAD software (such as SolidWorks or CATIA) to create a three-dimensional geometric model of the air compressor. Figure 3 The geometric model of the fuel cell air compressor includes an impeller, a volute, a diffuser, an inlet duct, and an outlet duct. The geometric model parameters of the impeller include the blade inlet installation angle, the outlet installation angle, and the blade thickness. The geometric model parameters of the volute include the cross-sectional ellipticity, the volute outlet diameter, and the volute tongue angle. The geometric model parameters of the diffuser include the diffuser inlet diameter, outlet radius, and outlet width. The inlet duct is used to guide the fluid into the impeller, and the outlet duct is used to guide the fluid out of the diffuser.

[0079] Then, a professional meshing tool (such as ANSYS Meshing or ICEM CFD) is used to mesh the geometric model of the fuel cell air compressor. In the implementation of the present application, an unstructured mesh (such as a tetrahedral mesh) is used in the impeller area to encrypt the blade surface; the number of mesh units is controlled within a reasonable range (such as 5 million to 10 million units). A structured mesh (such as a hexahedral mesh) is used in the volute area to optimize the calculation accuracy in the flow channel. The diffuser area encrypts the mesh in the inlet and outlet areas to ensure the accuracy of flow deceleration and pressure recovery.

[0080] Step 1.3: Based on the fluid dynamics model, thermodynamic model and turbulence model, boundary conditions are set for each model to establish a multi-physics field coupled CFD simulation model of the fuel cell air compressor.

[0081] Specifically, select appropriate fluid dynamics models, thermodynamic models and turbulence models from existing models, and set boundary conditions for each model. The boundary conditions of the fluid dynamics model include inlet conditions, outlet conditions and wall conditions, where the inlet conditions control the flow rate or velocity and provide the driving force of the flow field; the outlet conditions limit the range of the pressure field to ensure the stability of the calculation area; the wall conditions define the interaction between the fluid and the solid. The boundary conditions of the thermodynamic model include inlet temperature and wall heat exchange, where the inlet temperature controls the initial state of the compressed air; the wall heat exchange determines the thermal management effect of the air compressor. The boundary conditions of the turbulence model include turbulence parameters, which provide initial disturbance information and affect the turbulence distribution of the flow.

[0082] For example: Assume that the fuel cell air compressor has an inlet flow rate of 30 g / s, an outlet pressure of 2 bar, and an air inlet temperature of 25°C. Set boundary conditions: Fluid dynamics: inlet flow rate of 30 g / s, outlet pressure of 2 bar, wall impeller rotation speed of 40,000 rpm, no slip condition. Thermodynamic model: inlet air temperature of 25°C, wall volute insulation, diffuser convection heat exchange (cooling water temperature of 20°C). Turbulence model: turbulence intensity of 10%, turbulence viscosity ratio of 20.

[0083] Furthermore, in response to the actual operating needs of the fuel cell air compressor, the following operating conditions are set: 1. Flow range: 10g / s, 20g / s, 30g / s, 40g / s; 2. Outlet pressure range: 1.5bar, 2.0bar, 2.5bar; 3. Impeller speed range: 30000rpm, 40000rpm, 50000rpm; 4. Cooling conditions: cooling water temperature is 15℃, 20℃, 25℃.

[0084] Run the simulation using multiphysics simulation software such as ANSYS Fluent or CFX. Set the simulation convergence criteria: The residual convergence criteria is 10 -5 , key performance indicators (such as total pressure ratio, efficiency, flow) are stable. Then check the convergence, monitor whether the key performance indicators (such as total pressure ratio, flow, temperature rise, efficiency) are stable, extract data such as velocity field, pressure field, temperature field, and calculate the total pressure ratio, flow-pressure curve, isentropic efficiency, turbulence intensity distribution, and pressure recovery coefficient.

[0085] It should be noted that the velocity field is used to analyze the velocity distribution in the impeller, volute and diffuser, the pressure field is used to evaluate the pressure recovery efficiency, and the temperature field is used to observe the temperature rise during the airflow compression process. The total pressure ratio is used to reflect the compression performance, the flow-pressure curve is used to evaluate the regulation ability, the isentropic efficiency is used to quantify the energy loss of the compression process, the temperature rise is used to observe the temperature change during the compression process, the turbulence intensity distribution is used to reveal the source of vortex and flow loss, and the pressure recovery coefficient is used to evaluate the geometric design effect of the diffuser and volute.

[0086] Step 1.4: Compare the simulation results with the performance parameters collected by the test platform one by one to verify whether the error of each performance indicator is within the set acceptance range.

[0087] In the implementation of this application, the simulation results are compared with the experimental data in step 1.1 one by one to verify whether the error of each performance indicator is controlled within 5%. The error formula can be expressed as:

[0088]

[0089] Step 1.5: Based on the comparative analysis of simulation results and test data, determine the optimization direction of the impeller, volute and diffuser in the geometric model of the fuel cell air compressor.

[0090] The optimization direction of the impeller includes increasing the blade inlet angle to improve the inflow performance and reducing the outlet angle to improve the uniformity of the airflow. The optimization direction of the volute includes optimizing the volute tongue angle to reduce turbulence and adjusting the ellipticity of the volute section to reduce flow separation. The optimization direction of the diffuser includes increasing the outlet radius to improve pressure recovery and adjusting the width to improve the uniformity of the outlet flow field.

[0091] Step 1.6: Based on the geometric model of the fuel cell air compressor, construct a CFD integrated simulation framework for the fuel cell air compressor. Based on the orthogonal experimental design, perform batch simulations on different structural parameter combinations, generate representative simulation data covering multiple working conditions, and preprocess the simulation data.

[0092] It should be noted that fuel cell air compressor CFD (Computational Fluid Dynamics) refers to the simulation and analysis of the flow field inside the air compressor through computational fluid dynamics methods. Using CFD technology, the airflow distribution, pressure field, velocity field and temperature field of key components such as the impeller, volute and diffuser of the air compressor can be accurately calculated to evaluate its performance parameters such as efficiency, pressure ratio and flow characteristics.

[0093] As the instruction manual Figure 4 Specifically, the key to building a CFD integrated simulation framework for fuel cell air compressors is to achieve parameterized input, automated operation, and efficient management of result data. First, based on the geometric structure and operating conditions of the fuel cell air compressor, key design parameters (such as blade inlet angle, volute angle, diffuser outlet width) and operating parameters (such as inlet flow, outlet pressure, and speed) are tabulated to form an adjustable input parameter table, providing a basis for automated simulation. Secondly, through scripting tools (such as Python or ANSYS Workbench Automation), the full process of CFD simulation is automated, including geometric parameter input, meshing, boundary condition setting, and simulation task scheduling. In the simulation framework, boundary conditions cover typical operating conditions, such as multiple combinations of flow, pressure, and speed, to ensure a comprehensive evaluation of air compressor performance. After the simulation is completed, key performance indicators (such as total pressure ratio, efficiency, flow, and temperature rise) are automatically extracted, and these data are associated with the corresponding input parameters and stored in a unified database. The framework also integrates data verification and noise reduction functions to ensure the reliability and quality of simulation results, and supports visual analysis to intuitively display the distribution of velocity field, pressure field and temperature field, as well as the relationship between geometric parameters and performance indicators. Through this framework, efficient, flexible and scalable fuel cell air compressor CFD batch simulation is achieved, providing a solid technical foundation for optimized design.

[0094] Furthermore, the process of orthogonal experimental design includes: determining the geometric parameters to be optimized and their levels, such as: blade inlet angle (20°, 25°, 30°), volute tongue angle (40°, 45°, 50°), diffuser outlet radius (60mm, 65mm, 70mm). Use an orthogonal table (such as L9 table) to design experiments, and simplify the full factorial experiment (27 times) to 9 simulation experiments. Each set of geometric parameter combinations runs simulations under different working conditions, for example: flow rate: 10g / s, 20g / s, 30g / s, 40g / s, outlet pressure: 1.5bar, 2.0bar, 2.5bar, impeller speed: 30000rpm, 40000rpm, 50000rpm. After simulation, representative simulation data covering multiple working conditions are generated.

[0095] It should be noted that representative simulation data refers to high-quality data covering the design parameter space and operating range generated by carefully designed simulation experiments. These data can reflect the performance characteristics and variation patterns of fuel cell air compressors under typical operating conditions. They can cover the main variation ranges of air compressor geometric parameters (such as blade angle, volute shape, diffuser size) and operating parameters (such as flow, pressure, and speed). They are sufficiently extensive and accurate to guide optimization design and build data-driven models.

[0096] For example: blade inlet angle (20°, 25°, 30°), volute tongue angle (40°, 45°, 50°), diffuser width (20mm, 25mm, 30mm). Operating parameters: flow rate (20g / s, 30g / s), outlet pressure (2bar, 2.5bar). Through orthogonal experimental design, 9 groups of parameter combinations were simulated separately to generate the following data:

[0097]

[0098] Furthermore, the representative simulation data are preprocessed using multi-scale wavelet denoising, simple cross-validation and standardization methods.

[0099] Step 1.7: Based on the acquired batch simulation data, use the machine learning algorithm to build a data-driven model of the fuel cell centrifugal air compressor, use the training set to complete the model construction, and use the test set to verify the model performance.

[0100] The model construction process includes feature selection, data preprocessing, algorithm selection and parameter adjustment, and model performance evaluation to ensure that the data-driven model is efficient and accurate, thereby achieving efficient performance prediction and optimization of fuel cell centrifugal air compressors.

[0101] Specifically, key features and target variables are extracted from representative data sets generated by batch simulation to construct data sets. Geometric parameters: blade inlet angle, volute tongue angle, diffuser outlet width, etc., operating parameters: flow rate, outlet pressure, impeller speed, etc. The target variables are performance indicators: total pressure ratio, efficiency, temperature rise, pressure recovery coefficient, etc. Use correlation analysis (such as Pearson correlation coefficient or mutual information) to screen features with high correlation with target variables, remove redundant features, and reduce model complexity. Divide the data set into a training set (70%-80%) and a test set (20%-30%) to ensure that there is no overlap between the test data and the training data. Normalize or standardize all features to make their mean 0 and variance 1, and remove outliers and noise data.

[0102] Furthermore, according to the data characteristics and prediction goals, select the appropriate machine learning algorithm, such as XGBoost: a tree model based on gradient boosting, suitable for small high-dimensional data sets, with strong fitting ability; AdaBoost: improves model accuracy by weighted combination of multiple weak classifiers (such as decision trees); Random Forest: suitable for situations with large feature diversity, with strong anti-overfitting ability; Neural Network: suitable for large data sets with complex nonlinear relationships, but the training time is long. If the amount of data is small and the nonlinear relationship is weak, XGBoost is preferred; if the amount of data is large and highly nonlinear fitting is required, neural network can be selected.

[0103] Then, the selected algorithm is used to train the data-driven model on the training set; hyperparameters are set and grid search or Bayesian optimization is used to adjust the parameters to optimize model performance.

[0104] Furthermore, the model is verified using the test set, and the model performance is measured using a variety of evaluation indicators. Common evaluation indicators can be used: mean square error (MSE), determination coefficient (R 2 ), mean absolute error (MAE). Use k-fold cross validation (such as 5-fold or 10-fold cross validation) to evaluate the generalization performance of the model and ensure the model's predictive ability for unknown data. Analyze data points with large errors to identify potential model deficiencies or data quality issues; adjust feature selection, data preprocessing, or algorithm parameters to optimize the model.

[0105] After training, the model can be used to predict performance indicators under new geometric parameters and working conditions, and can be used to quickly evaluate the impact of different parameter combinations on performance when applied to the design optimization of fuel cell air compressors. The relationship curve between characteristics and performance indicators is drawn to intuitively show the impact of geometric parameters (such as blade inlet angle) on total pressure ratio, efficiency, etc.; the error distribution diagram is used to show the reliability of model prediction; and a three-dimensional scatter plot or contour map is generated to describe the impact of multiple parameters on performance.

[0106] Step 2: Based on the fuel cell system requirements and multi-operating condition data analysis, the optimization objectives are defined as pressure ratio and isentropic efficiency, and the compression capacity and efficiency performance of the air compressor are evaluated through experimental and simulation data calculations.

[0107] The purpose of step 2 is to define the optimization objectives as pressure ratio and isentropic efficiency based on the gas supply demand and multi-operating conditions of the fuel cell system, and to comprehensively evaluate the compression capacity and efficiency performance of the air compressor under a fixed growth rate ratio through experimental testing and simulation calculations, thereby providing data support and performance evaluation basis for subsequent multi-objective optimization and performance improvement.

[0108] Step 2 specifically includes:

[0109] Step 2.1: Based on the fuel cell system gas supply demand, multi-operating conditions, thermodynamics and aerodynamic theory, and comprehensive analysis of experimental and simulation data, the optimization objectives of the fuel cell air compressor are defined as pressure ratio and isentropic efficiency.

[0110] Specifically, the pressure ratio (σ) represents the ratio of the outlet pressure to the inlet pressure of the air compressor and is used to measure the compression capacity; the pressure ratio (σ) is determined by the following formula:

[0111] σ=p out / p in (1)

[0112] Where: p out Indicates the outlet pressure of the air compressor; p in Indicates the inlet pressure of the air compressor.

[0113] Isentropic efficiency (η) characterizes the thermodynamic efficiency of the air compressor and reflects the energy loss in the compression process. Isentropic efficiency (η) is determined by the following formula:

[0114]

[0115] Where: T in Indicates the air compressor inlet temperature; T out represents the air compressor outlet temperature; k represents the isentropic index.

[0116] Step 2.2: Under the condition of a fixed speed ratio, collect the inlet and outlet pressure and temperature data of the air compressor under different speed conditions, and calculate the pressure ratio and isentropic efficiency. According to the calculation results, the compression capacity and efficiency performance of the air compressor are evaluated.

[0117] It should be noted that the sources for collecting the inlet and outlet pressures, inlet and outlet temperatures and related key performance parameters of the air compressor at various speed conditions include experimental tests or CFD simulation calculations.

[0118] In the implementation of this application, the fixed speed ratio condition is: the actual speed of the fuel cell air compressor is 12.7 times the motor speed to ensure the stability of the speed condition in the test or simulation. The motor speed range is 3000, 5000, 7000, 9000, 11000 rpm, corresponding to the actual air compressor speed of 38100 ~ 139700 rpm. In the comprehensive performance test platform, the air compressor inlet pressure p is collected in real time. in , outlet pressure of air compressor p out , air compressor inlet temperature T in , air compressor outlet temperature T out , use multi-physics simulation software (such as ANSYS Fluent) to simulate the operating conditions of the air compressor. Set simulation boundary conditions (inlet pressure, outlet pressure, inlet temperature, speed, etc.) to simulate the fluid field and extract performance parameters such as flow, pressure ratio, and temperature rise. Then calculate the pressure ratio and isentropic efficiency according to formulas (1) and (2). Under different speed conditions (5 motor speeds), calculate the pressure ratio and isentropic efficiency of the air compressor respectively, and form a total of 10 optimization targets. Compare the changes in pressure ratio and isentropic efficiency under different speed conditions, analyze the stability of performance, and focus on evaluating whether the pressure ratio and efficiency meet the requirements of high-performance fuel cell systems under high speed (11000rpm) conditions. Under low speed (3000rpm) conditions, ensure the stability and efficiency of air compressor operation.

[0119] Step 3: Integrate multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm, and K-means clustering method to establish the MMPS algorithm. Use the MMPS algorithm to perform high-dimensional multi-objective optimization on the impeller, volute, diffuser and related structural parameters to improve their performance under multiple working conditions.

[0120] The purpose of step 3 is to develop an efficient multi-objective multi-population optimization algorithm (MMPS algorithm). By integrating the multi-objective multi-population optimization framework, archive sharing technology, archive update strategy, genetic algorithm and K-means clustering, the key components of the fuel cell air compressor (impeller, volute, diffuser) are optimized in high dimensions and multi-objective, which can comprehensively improve the comprehensive performance of the air compressor under multiple working conditions. By optimizing multiple structural parameters, the trade-off problem of the air compressor in performance indicators such as pressure ratio and isentropic efficiency is solved, and the complex multi-objective optimization problem is effectively solved, which ultimately improves the operating efficiency and reliability of the fuel cell system.

[0121] Step 3 specifically includes:

[0122] Step 3.1: Construct a multi-objective multi-population optimization framework. By dividing the optimization objective into 10 sub-populations for parallel optimization, the crossover operator and mutation operator based on the genetic algorithm are combined to generate and improve the solution, so as to improve the diversity of the solution set and the global search capability, thereby achieving efficient solution set exploration and optimization in complex multi-objective optimization problems.

[0123] The multi-objective multi-population optimization method is a type of swarm intelligence algorithm designed for multi-objective optimization problems. It aims to improve the diversity of solution sets and global search capabilities by searching multiple populations in parallel. The core of this method is to divide the population into multiple sub-populations, each of which works independently or collaboratively to explore different objectives, search areas or optimization strategies, so as to more effectively find the Pareto optimal solution set in complex multi-objective problems. These populations can share and migrate solutions through information interaction mechanisms, so that the algorithm can balance global exploration and local development, avoid falling into local optimal solutions and maintain population diversity. The design of multi-population methods is usually accompanied by different strategy choices, such as population splitting, differentiation of mutation crossover operations, and adaptive parameter adjustment, so that it can adapt to complex, high-dimensional, multi-objective optimization scenarios.

[0124] In the multi-objective multi-population optimization framework, 10 populations are used to optimize 10 objectives respectively. The performance of the solution in each population is evaluated by its corresponding optimization objective rather than by the Pareto optimal rule, because it is difficult to meet the Pareto optimal conditions in multi-objective multi-population optimization, which may slow down the evolution process. Genetic algorithm-based operators (i.e., crossover operator and mutation operator) are used as optimizers in each population to generate offspring.

[0125] Step 3.2: The crossover operator adopts the priority crossover operation, and realizes gene recombination by sorting individuals with high fitness values ​​and using the double-optimal crossover strategy to improve the population adaptability and solution set diversity; the mutation operator adopts the insertion mutation operation, and enhances the population diversity and global search capability by inserting and optimizing genes of individuals with low fitness values, thus ensuring the stability of the optimization results and performance improvement.

[0126] Prioritized crossover (POX) is a strategy for selecting and recombination specific individuals in the optimization process. It aims to improve the overall adaptability and convergence of the population and ensure the diversity of the solution set. Therefore, the crossover operator selects the preferred crossover operation. The core of the preferred crossover operation is to give priority to individuals with higher fitness values ​​for crossover operations, so as to retain excellent characteristic genes to the greatest extent and promote the evolution of the population towards a high-quality solution set. The specific operation steps are as follows: ① Individual priority sorting: Within the population, individuals are sorted according to their fitness values ​​(the evaluation criteria may include multi-objective comprehensive scores such as compression efficiency, energy consumption, and stability). Prioritize the top-ranked individuals to participate in the crossover to make full use of the excellent genes of individuals with high fitness values. ② Double-optimal crossover strategy: The preferred crossover operation adopts the "double-optimal crossover strategy", that is, two groups of high fitness value individuals are randomly selected from the prioritized individuals as parent individuals for crossover operations. This crossover strategy is based on probabilistic selection and local search enhancement mechanisms to improve the diversity and potential adaptability of newly generated individuals. ③ Selection of crossover mode: The crossover mode adopts a mixed gene crossover mechanism, adopts the form of multi-point crossover, and is flexibly switched according to specific goals and population status. In the crossover operation, we focus on maintaining the key gene fragments of the parent individuals, while introducing moderate gene disturbances to improve the diversity of the population. ④ Adaptive adjustment: For the new individuals generated by the crossover, they will be initially screened through a rapid adaptive evaluation mechanism, and only excellent individuals that meet the multi-objective optimization requirements will be retained for the subsequent iteration process. For individuals that do not meet the performance requirements, they will be re-marked and enter the next generation of mutation operations.

[0127] The mutation operator uses the insertion mutation operation (IM), which is designed as a key means to enhance population diversity, break local convergence and improve global search capabilities. Its purpose is to introduce new search directions and optimize the performance of centrifugal air compressors under multi-objective and multi-operating conditions through reasonable mutation strategies. The specific steps are: ① Individual selection and positioning: Select individuals with lower fitness from the current population for mutation operations. By sorting the fitness values ​​of individuals in the population, the individuals with lower rankings are preferentially selected as mutation objects, aiming to introduce new gene characteristics through mutation and improve the overall fitness of these individuals. ② Selection of insertion position: For the individuals to be mutated, one or more gene sites are randomly selected as insertion positions. In order to enhance the flexibility of multi-objective optimization, the selection of insertion positions adopts a weighted random strategy to achieve a balance between global and local, ensuring that the mutation results have good diversity and stability. ③ Generation of mutant genes: Extract some gene fragments from excellent individuals in other populations, realize information sharing and recombination through insertion mutation, and enhance optimization efficiency. ④ Insertion strategy and control mechanism: The core of the insertion mutation operation lies in how to select and insert mutant genes. Specifically, this method adopts a flexible insertion strategy, including single-point insertion, segmented insertion, and multi-point insertion, to improve the mutation effect. By controlling the mutation intensity and insertion frequency, the population diversity and global search capability are guaranteed to avoid the confusion of the solution set and the decrease of convergence caused by excessive mutation. ⑤ Adaptability evaluation and update: After the mutation operation is completed, the newly generated individuals are evaluated for their adaptability. Multi-objective comprehensive evaluation indicators are used to ensure that the performance of the mutant individuals under different working conditions and objectives is improved.

[0128] Step 3.3: By constructing archive sharing technology, archive A is used to store and share excellent solutions in the multi-objective multi-population optimization framework to assist information interaction and priority crossover operations between populations, thereby achieving adaptive search and population co-evolution.

[0129] In the multi-objective multi-population optimization framework, the archive denoted by A is used to store excellent solutions discovered during the evolution process. A is initialized to all randomly generated solutions in 10 populations without duplication. Afterwards, the archive sharing technique is developed to exchange information between populations and assist the crossover operator, thereby helping to achieve efficient population co-evolution. The archive sharing technique is designed to assist the crossover operator, namely the preferential crossover operation in Algorithm 1. Specifically, for the kth (k∈[1,10]) population, each solution in population k is selected as p1, and the solution in A is randomly selected as p2 (Lines 4-6). In each crossover operation, a random value r in the range [0,1] is generated. If r is less than or equal to the crossover probability Prc, a preferential crossover operation is performed on p1 and p2 to generate offspring c1 and c2. Among p1, c1 and c2, the one that performs best in k objectives is retained to generate a new population (Lines 8-11). Otherwise, the one with better performance among p1 and p2 will be retained (lines 12-14). Therefore, the adaptive search technology in the multi-objective multi-population optimization framework can not only retain the excellent characteristics of its corresponding optimization objective, but also effectively utilize the information of the other 9 objectives.

[0130] A specific embodiment includes the following process:

[0131]

[0132]

[0133] Step 3.4: Propose an archive update strategy to update the archive by screening, perturbing and preserving the diversity of the solution set of the newly generated population based on K-means clustering to improve the convergence and quality of the elite solution, thereby enhancing the co-evolutionary ability and optimization efficiency of the population in the multi-objective multi-population optimization framework.

[0134] In the multi-objective multi-population optimization framework, the archive update strategy is proposed to update A according to the newly generated population, thereby generating solutions with better performance. The pseudo code of the archive update strategy is given in Algorithm 2. As the number of non-dominated solutions increases during the evolution process, the size of the archive is defined by a threshold. Therefore, the maximum size of the archive is set to NA in the multi-objective multi-population optimization framework. For each generation, all solutions in the current 10 populations are added to the empty set S (lines 2-4), all elite solutions in A are also added to S (line 5), and then each elite solution in A is perturbed and then added to S (lines 6-10). The perturbation operation on A is implemented based on the insertion mutation operation (mutation operator). Based on local perturbations, the archive update strategy mainly improves the convergence of elite solutions, thereby improving the exploration ability of the multi-objective multi-population optimization framework. Since there may be duplications in the solutions collected from the 10 populations and the archive, eliminating duplicate operations can not only save storage space but also help retain more diverse solutions in the archive. If the size of S is less than or equal to the predefined size NA, all solutions in S are copied to A, otherwise, a sorting method based on K-means clustering and a diversity preservation method are performed on S to select NA elite solutions into A (lines 12-17). The archive update strategy can improve the quality of solutions in the archive, thereby helping the archive sharing technology to better guide the co-evolution process between populations.

[0135] A specific embodiment includes the following process:

[0136]

[0137]

[0138] Step 3.5: By integrating multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm and K-means clustering method, the MMPS algorithm is established to randomly initialize the population and update the population through preferential crossover operation and insertion mutation operation, and optimize the solution set in combination with the archive update strategy, and finally output the elite solution in the archive as the optimization result.

[0139] The MMPS algorithm process is as shown in the attached manual. Figure 5 As shown. First, 10 populations of equal size are randomly initialized using an operation-based encoding method. Then, the archive A is initialized to all solutions in the 10 populations. During the iteration process, the 10 populations focus on optimizing the 10 objectives respectively. The archive sharing technique based on the prioritized crossover operation (POX) and the insertion mutation operation (IM) are performed to update each population, and then A is updated by the archive update strategy. The evolutionary process is repeated until the termination condition is met. The elite solution in A is the final output of the MMPS algorithm.

[0140] A specific embodiment includes:

[0141] The optimization goal is the multi-objective performance optimization of a fuel cell centrifugal air compressor. The goals include 10 performance indicators, including maximizing efficiency, minimizing compression power, minimizing noise, minimizing vibration, etc. The constraints are to meet the design parameter range under different flow, speed and pressure conditions. The optimization framework based on the multi-objective multi-population optimization (MMPS) algorithm is applied to find the Pareto optimal solution set with high efficiency by optimizing 10 goals in parallel.

[0142] Step 3.1: Randomly generate 10 populations, each containing 50 individuals (solutions). The genetic encoding method of the population is as follows: the genetic information includes the flow rate (g / s), speed (rpm), outlet pressure (bar), cooling parameters (L / min and temperature ℃) and other control variables of the air compressor; each individual represents a design scheme, which is initially randomly distributed in the parameter space. Each population is optimized for one goal, for example, population 1 is optimized for efficiency, population 2 is optimized for compression power, and population 3 is optimized for noise. The populations are optimized in parallel, and each uses the crossover operator and mutation operator to generate new solutions. The initial solution set is stored in the archive A for subsequent sharing and improvement.

[0143] Step 3.2: Prioritized crossover operation: In population 1 (efficiency optimization), 50 individuals are sorted from high to low according to the efficiency value of the current individual, and the top 10 individuals with high fitness values ​​are selected as candidates for the parent generation of the prioritized crossover. Two groups of individuals are randomly selected from the top 10 parents, and offspring are generated through multi-point crossover. Assume that the parent genes are: Parent 1: [flow rate = 30 g / s, speed = 40000 rpm, pressure = 2 bar, cooling water flow rate = 3 L / min, cooling water temperature = 25°C], Parent 2: [flow rate = 40 g / s, speed = 45000 rpm, pressure = 2.5 bar, cooling water flow rate = 5 L / min, cooling water temperature = 30°C]; after multi-point crossover, the offspring are generated: Offspring 1: [flow rate = 30 g / s, speed = 45000 rpm, pressure = 2 bar, cooling water flow rate = 5 L / min, cooling water temperature = 25°C], Offspring 2: [flow rate = 40 g / s, speed = 40000 rpm, pressure = 2.5 bar, cooling water flow rate = 3 L / min, cooling water temperature = 30°C]. A multi-point crossover mechanism is adopted to allow the parent gene fragments to be combined alternately. For example, key genes (such as cooling water temperature = 25°C) are retained from the parent generation, and moderate disturbances (such as changes in cooling water flow) are introduced. The new individuals generated by the crossover are initially evaluated, and only solutions that meet the efficiency optimization goal are retained; individuals that do not meet the conditions will enter the next generation of mutation operations.

[0144] Insertion mutation operation: In population 2 (the optimization target is compression power), the individuals are sorted from low to high according to their fitness values ​​(compression power values). The 10 individuals with the lowest ranking are selected as mutation targets, aiming to introduce new gene characteristics through mutation and improve the overall fitness of these individuals. For example: original individual 1: [flow rate = 50g / s, speed = 40000rpm, pressure = 3bar, cooling water flow rate = 3L / min, cooling water temperature = 35℃], original individual 2: [flow rate = 45g / s, speed = 45000rpm, pressure = 2.8bar, cooling water flow rate = 4L / min, cooling water temperature = 32℃]. Through a weighted random strategy, the cooling water temperature and cooling water flow genes are given higher weights (because these parameters have a significant impact on the compression power), and other genes are given lower weights to achieve a balance between global and local optimization. For example: original individual 1: cooling water flow rate (3L / min) and cooling water temperature (35℃) are selected as mutation positions. Original individual 2: flow rate (45g / s) and speed (45000rpm) are selected as mutation positions. Gene fragments of excellent individuals are extracted from other populations (such as population 1 with optimized efficiency) as mutation sources: for example, the excellent genes extracted from population 1 are: cooling water flow rate = 4L / min, cooling water temperature = 25℃, speed = 40000rpm. After mutation, the mutation results of original individual 1 and original individual 2 are: mutant individual 1: [flow rate = 50g / s, speed = 40000rpm, pressure = 3bar, cooling water flow rate = 4L / min, cooling water temperature = 25℃], mutant individual 2: [flow rate = 45g / s, speed = 40000rpm, pressure = 2.8bar, cooling water flow rate = 4L / min, cooling water temperature = 25℃]. A mutation strategy combining single-point insertion and multi-point insertion is adopted to control the mutation intensity and frequency to ensure the optimization effect: Single-point insertion: mutate the cooling water flow rate (for example, from 3L / min to 4L / min) to improve the cooling performance. Multi-point insertion: mutate the cooling water flow rate and cooling water temperature at the same time to ensure that the impact on the compression power is significantly optimized. Control mechanism: The mutation intensity is adjusted by the predefined weight to avoid excessive global search leading to confusion in the solution set. For example, the cooling water temperature is only allowed to vary within ±5°C. After the mutation is completed, the mutant individuals are evaluated using multi-objective comprehensive evaluation indicators (such as compression power, efficiency, and stability) to ensure that their performance is improved under multiple objectives. Evaluation results: The compression power of mutant individual 1 decreased by 10%, and the efficiency increased by 5%, and it passed the evaluation; the compression power of mutant individual 2 did not change significantly, and it was marked as a low-fitness individual and entered the mutation link of the next generation.

[0145] Step 3.3: Archive A initially stores 100 solutions of all populations for information sharing. For example, in population 3 (noise optimization), each priority crossover selects the solution in A as the second parent. The current population solution p1: [flow = 35g / s, speed = 42000rpm, pressure = 2.2bar, cooling water flow = 4L / min, cooling water temperature = 27℃], select p2 from A: [flow = 30g / s, speed = 40000rpm, pressure = 2bar, cooling water flow = 3L / min, cooling water temperature = 25℃], perform priority crossover, generate offspring and retain the optimal solution.

[0146] Step 3.4: Collect all new solutions of the current population and the elite solutions in A, and store them together in set S. Through deduplication and fast non-dominated sorting, screen out the solution set that meets the multi-objective optimization requirements. For the solution sets with more duplications in S, group them with K-means clustering, select the solution closest to the center point in each group, and update the elite solution in A.

[0147] Step 3.5: Perform preferential crossover, insertion mutation and archive update in each generation, and iterate and optimize 10 populations in parallel, and repeat the iteration until convergence (such as the elite solution change is less than the threshold or the maximum number of generations is reached). Finally, the elite solution in archive A is output as the optimization result, for example: [flow rate = 40g / s, speed = 40000rpm, pressure = 2bar, cooling water flow rate = 3L / min, cooling water temperature = 25℃].

[0148] Through the above steps, the multi-objective performance of the air compressor was successfully optimized, and significant improvements were achieved in terms of efficiency, compression power, noise and vibration. At the same time, a set of Pareto optimal solutions with engineering application value was generated.

[0149] Another object of the present invention is to provide a high-dimensional multi-objective multi-operating condition optimization system for a fuel cell centrifugal air compressor, comprising:

[0150] The performance analysis and modeling module is used to build a performance analysis and optimization environment for fuel cell air compressors. By building a comprehensive performance testing platform and establishing a multi-physics field coupled CFD simulation model, the consistency of simulation results and test data is verified, and multi-operating condition data is generated based on orthogonal experimental design. At the same time, machine learning algorithms are combined to build and verify data-driven models to improve simulation and optimization efficiency.

[0151] The optimization target definition module is used to define optimization targets, including pressure ratio and isentropic efficiency, based on fuel cell system requirements and multi-operating condition data analysis, and evaluate the compression capacity and efficiency performance of the air compressor through experimental and simulation data calculations.

[0152] The high-dimensional multi-objective optimization module is used to integrate multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm and K-means clustering method to construct the MMPS algorithm. Based on this algorithm, high-dimensional multi-objective optimization is performed on the structural parameters of the impeller, volute and diffuser to improve the performance of the fuel cell centrifugal air compressor under multiple working conditions.

[0153] The above is only an embodiment of the present invention, and the common sense such as the known specific structure and characteristics in the scheme is not described in detail here. For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present invention is defined by the attached claims rather than the above description, so it is intended to include all changes within the meaning and scope of the equivalent elements of the claims. Any figure mark in the claims should not be regarded as limiting the claims involved.

Claims

1. A high-dimensional multi-objective and multi-operating-condition optimization method for a fuel cell centrifugal air compressor, characterized in that: Methods include: Step 1: Construct a fuel cell air compressor performance analysis and optimization system. By building a comprehensive performance test platform, establishing a multi-physics field coupling CFD simulation model, and verifying the consistency between the simulation results and the test data, the optimization direction of the geometric model is determined. Based on the orthogonal experimental design, batch simulation is performed to generate multi-condition data. In addition, a data-driven model is constructed and verified in combination with a machine learning algorithm to improve the simulation and optimization efficiency. Step 2: Based on the fuel cell system requirements and multi-operating condition data analysis, the optimization objectives are defined as pressure ratio and isentropic efficiency, and the compression capacity and efficiency performance of the air compressor are evaluated through experimental and simulation data calculations; Step 3: Integrate multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm, and K-means clustering method to establish the MMPS algorithm. Use the MMPS algorithm to perform high-dimensional multi-objective optimization on the impeller, volute, diffuser and related structural parameters to improve their performance under multiple working conditions.

2. The high-dimensional multi-objective multi-operating condition optimization method for a fuel cell centrifugal air compressor according to claim 1 is characterized in that: Step 1 includes: Step 1.1: Build a comprehensive performance test platform for fuel cell air compressors, and use the platform to collect performance parameters of fuel cell air compressors under different operating conditions; Step 1.2: Create a geometric model of the fuel cell air compressor, perform multi-physics simulation analysis on the geometric model based on the fluid dynamics model, thermodynamics model, and turbulence model, set the boundary conditions of each model, run multi-physics simulation to cover a variety of operating conditions, and complete the convergence and accuracy verification of the geometric model; Step 1.3: Based on the fluid dynamics model, thermodynamics model and turbulence model, boundary conditions are set for each model to establish a multi-physics field coupled CFD simulation model of the fuel cell air compressor; Step 1.4: Compare the simulation results with the performance parameters collected by the test platform one by one to verify whether the error of each performance indicator is within the set acceptance range; Step 1.5: Based on the comparative analysis of simulation results and test data, determine the optimization direction of the impeller, volute and diffuser in the geometric model of the fuel cell air compressor; Step 1.6: Based on the geometric model of the fuel cell air compressor, build a CFD integrated simulation framework for the fuel cell air compressor, perform batch simulations on different structural parameter combinations based on orthogonal experimental design, generate representative simulation data covering multiple working conditions, and pre-process the simulation data; Step 1.7: Based on the acquired batch simulation data, use the machine learning algorithm to build a data-driven model of the fuel cell centrifugal air compressor, use the training set to complete the model construction, and use the test set to verify the model performance.

3. The high-dimensional multi-objective multi-operating condition optimization method for a fuel cell centrifugal air compressor according to claim 1 is characterized in that: Step 2 includes: Step 2.1: Based on the fuel cell system gas supply demand, multi-operating conditions, thermodynamics and aerodynamic theory, and comprehensive analysis of experimental and simulation data, the optimization objectives of the fuel cell air compressor are defined as pressure ratio and isentropic efficiency; Step 2.2: Under the condition of a fixed speed ratio, collect the inlet and outlet pressure and temperature data of the air compressor under different speed conditions, and calculate the pressure ratio and isentropic efficiency. According to the calculation results, the compression capacity and efficiency performance of the air compressor are evaluated.

4. The high-dimensional multi-objective multi-operating condition optimization method for a fuel cell centrifugal air compressor according to claim 1, characterized in that: Step 3 includes: Step 3.1: Construct a multi-objective multi-population optimization framework, divide the optimization objectives into 10 sub-populations for parallel optimization, and combine the crossover operator and mutation operator based on the genetic algorithm to generate and improve the solution, so as to improve the diversity of the solution set and the global search capability, thereby achieving efficient solution set exploration and optimization in complex multi-objective optimization problems; Step 3.2: The crossover operator uses a priority crossover operation to achieve gene recombination by sorting individuals with high fitness values ​​and using a double-optimal crossover strategy, thereby improving population adaptability and solution set diversity. The mutation operator uses an insertion mutation operation to enhance population diversity and global search capabilities by inserting and optimizing genes for individuals with low fitness values, thereby ensuring the stability of the optimization results and improving performance. Step 3.3: By constructing archive sharing technology, archive A is used to store and share excellent solutions in the multi-objective multi-population optimization framework to assist information interaction and priority crossover operations between populations, and realize adaptive search and population co-evolution; Step 3.4: Propose an archive update strategy to update the archive by screening, perturbing and preserving the diversity of the solution set of the newly generated population based on K-means clustering to improve the convergence and quality of the elite solution, thereby enhancing the co-evolutionary ability and optimization efficiency of the population in the multi-objective multi-population optimization framework; Step 3.5: By integrating multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm and K-means clustering method, the MMPS algorithm is established to randomly initialize the population and update the population through preferential crossover operation and insertion mutation operation, and optimize the solution set in combination with the archive update strategy, and finally output the elite solution in the archive as the optimization result.

5. The high-dimensional multi-objective multi-operating condition optimization method for a fuel cell centrifugal air compressor according to claim 2, characterized in that: The fuel cell air compressor comprehensive performance test platform can simulate typical operating conditions including different flow conditions, different speed conditions, different pressure conditions, different cooling conditions, startup and steady-state conditions and comprehensive conditions.

6. The high-dimensional multi-objective multi-operating condition optimization method for a fuel cell centrifugal air compressor according to claim 2, characterized in that: In step 1.4, the simulation results are compared with the experimental data in step 1.1 one by one to verify whether the error of each performance indicator is controlled within 5%.

7. The high-dimensional multi-objective multi-operating condition optimization method for a fuel cell centrifugal air compressor according to claim 2, characterized in that: The optimization directions of the impeller include increasing the blade inlet angle to improve the inflow performance and reducing the outlet angle to improve the uniformity of the airflow; the optimization directions of the volute include optimizing the volute tongue angle to reduce turbulence and adjusting the ellipticity of the volute section to reduce flow separation; the optimization directions of the diffuser include increasing the outlet radius to improve pressure recovery and adjusting the width to improve the uniformity of the outlet flow field.

8. The high-dimensional multi-objective and multi-operating-condition optimization method for a fuel cell centrifugal air compressor according to claim 3 is characterized in that: The pressure ratio (σ) is determined by the following formula: σ=p out / p in Where: p out Indicates the outlet pressure of the air compressor; p in Indicates the inlet pressure of the air compressor; The isentropic efficiency (η) is determined by the following formula: Where: T in Indicates the air compressor inlet temperature; T out represents the air compressor outlet temperature; k represents the isentropic index.

9. A high-dimensional multi-objective and multi-operating-condition optimization system for a fuel cell centrifugal air compressor, characterized in that: include: The performance analysis and modeling module is used to build a fuel cell air compressor performance analysis and optimization environment. By building a comprehensive performance test platform and establishing a multi-physics field coupled CFD simulation model, the consistency of simulation results and test data is verified, and multi-operating condition data is generated based on orthogonal experimental design. At the same time, a data-driven model is built and verified in combination with machine learning algorithms to improve simulation and optimization efficiency. The optimization target definition module is used to define optimization targets, including pressure ratio and isentropic efficiency, based on fuel cell system requirements and multi-operating condition data analysis, and evaluate the compression capacity and efficiency performance of the air compressor through experimental and simulation data calculations; The high-dimensional multi-objective optimization module is used to integrate multi-objective multi-population optimization methods, archive sharing technology, archive update strategy, genetic algorithm and K-means clustering method to construct the MMPS algorithm. Based on this algorithm, high-dimensional multi-objective optimization is performed on the structural parameters of the impeller, volute and diffuser to improve the performance of the fuel cell centrifugal air compressor under multiple working conditions.

Images