A method and device for parallel optimization of geometry and control parameters of an aviation electromechanical system

Abstract

The present application provides a method and device for parallel optimization of geometry and control parameters of an aviation electromechanical system, which belongs to the field of aerospace. The method provided in the present application includes: constructing a first initial model of an aviation electromechanical system; constructing a geometry parameter-aerodynamic simulation model of a turbomachinery component; fusing the first initial model and the geometry parameter-aerodynamic simulation model to obtain a first fusion model; generating training samples according to the first fusion model, wherein the input samples in the training samples are geometry parameters and the output samples are three-dimensional characteristic matrices; training a neural network model based on the training samples; fusing the trained neural network model and the first initial model to obtain a parallel optimization model; and simultaneously generating the geometry and control parameters of the aviation electromechanical system based on the parallel optimization model. The method and device for parallel optimization of geometry and control parameters of an aviation electromechanical system provided in the present application can realize the parallel optimization of geometry and control parameters of an aviation electromechanical system.

Description

Technical Field

[0001] The present application relates to the field of aerospace technology, and in particular to a method and device for parallel optimization of geometry and control parameters of an aviation electromechanical system. Background Art

[0002] In the field of aviation electromechanical technology, integrated electromechanical systems have been widely used due to their high degree of integration, small size and weight. Integrated electromechanical systems involve multiple professional disciplines, have complex structures and changeable operating conditions. There is a lot of room for improvement in improving their performance from a design perspective. Multidisciplinary optimization of integrated electromechanical systems has become the focus of current research.

[0003] At present, the commonly used integrated electromechanical system optimization method is the serial optimization design method of geometric design parameters and control parameters. First, the geometric design parameters of key components are adjusted through expert experience to complete the design optimization of key components of the integrated electromechanical system. After the design of each component is completed, the intelligent algorithm is combined to complete the independent optimization of the control parameters of the system in different modes. However, there is a coupling relationship between the component geometric design parameters and the system control. Changes in the component geometric design parameters will lead to differences in the control optimization results, and changes in the control parameters will also cause changes in the component geometric optimization parameters. Therefore, the independent optimization of the geometric design parameters and the control parameters will result in the final optimization result not being the global optimal solution. The commonly used serial optimization design method at this stage has many limitations. There is an urgent need for a method for parallel optimization of geometric design parameters and control parameters to achieve global optimization of integrated electromechanical systems. Summary of the invention

[0004] In view of this, the present application provides a method and device for parallel optimization of geometry and control parameters of an aviation mechatronic system, which can realize parallel optimization of geometry and control parameters of an aviation mechatronic system.

[0005] Specifically, the present application is implemented through the following technical solutions:

[0006] The first aspect of the present application provides a method for parallel optimization of geometry and control parameters of an aviation electromechanical system, the method comprising:

[0007] Construct the first initial model of the aviation electromechanical system;

[0008] Construct geometric parameter-aerodynamic simulation model of turbomachinery components;

[0009] The first initial model and the geometric parameter-aerodynamic simulation model are fused to obtain a first fusion model, wherein a three-dimensional characteristic matrix is ​​calculated according to the geometric parameter-aerodynamic simulation model, the three-dimensional characteristic matrix includes geometric parameter changes, and an original three-dimensional characteristic matrix in the first initial model is replaced based on the three-dimensional characteristic matrix, the original three-dimensional characteristic matrix does not contain geometric parameter changes;

[0010] Generate training samples according to the first fusion model, wherein the input samples in the training samples are geometric parameters and the output samples are three-dimensional characteristic matrices;

[0011] Training a neural network model based on the training samples;

[0012] Fusion the trained neural network model and the first initial model to obtain a parallel optimization model;

[0013] The geometric and control parameters of the aero-mechanical system are simultaneously generated based on the parallel optimization model.

[0014] A second aspect of the present application provides a device for parallel optimization of geometry and control parameters of an aviation electromechanical system, the device comprising a construction module, a fusion module, a generation module and a training module;

[0015] The building module is used to build a first initial model of the aviation electromechanical system;

[0016] The construction module is also used to construct a geometric parameter-aerodynamic simulation model of the impeller machinery components;

[0017] The fusion module is used to fuse the first initial model and the geometric parameter-aerodynamic simulation model to obtain a first fusion model, wherein a three-dimensional characteristic matrix is ​​calculated according to the geometric parameter-aerodynamic simulation model, the three-dimensional characteristic matrix includes geometric parameter changes, and an original three-dimensional characteristic matrix in the first initial model is replaced based on the three-dimensional characteristic matrix, and the original three-dimensional characteristic matrix does not contain geometric parameter changes;

[0018] The generating module is used to generate training samples according to the first fusion model, wherein the input samples of the training samples are geometric parameters and the output samples are three-dimensional characteristic matrices;

[0019] The training module is used to train the neural network model based on the training samples;

[0020] The fusion module is also used to fuse the trained neural network model and the first initial model to obtain a parallel optimization model;

[0021] The generation module is also used to simultaneously generate the geometry and control parameters of the aviation electromechanical system based on the parallel optimization model.

[0022] The method and device for parallel optimization of geometry and control parameters of aviation electromechanical systems provided in this application, by constructing a first initial model, a geometry parameter-aerodynamic simulation model and performing a series of fusion operations, simultaneously considers the mutual influence of geometry parameters and control parameters during the optimization process, and can more comprehensively explore the parameter space, thereby obtaining the global optimal solution of the system and improving the design quality of electromechanical products. By establishing a geometry parameter-aerodynamic simulation model of turbomachinery components, parametric automatic simulation of turbomachinery such as turbines can be achieved, providing effective tool support for component geometry design optimization. In addition, the established parallel optimization model integrates the information of the geometric parameter-aerodynamic simulation model and the first initial model. By substituting the three-dimensional characteristic matrix calculated by the geometric parameter-aerodynamic simulation model into the impeller machinery part of the first initial model, replacing the original, relatively simple three-dimensional characteristic matrix, the key data transmission from the micro level of the component to the macro level of the system is realized, so that the first initial model can obtain more accurate component performance information that includes the influence of geometric parameter changes. Then, the neural network model trained based on the training samples containing geometric parameters and the three-dimensional characteristic matrix can effectively integrate the mapping relationship between geometric parameter changes and the three-dimensional physical properties of the components. In the optimization process, one-dimensional and three-dimensional physical information can be comprehensively considered, and the system efficiency can be globally optimized, thereby improving the optimization effect.

[0023] Specifically, the first initial model of the aviation electromechanical system and the geometric parameter-aerodynamic simulation model of the impeller machinery components were constructed respectively. Among them, the first initial model is based on the determination of the devices, system functions and control parameters contained in the aviation electromechanical system, and is established according to the thermodynamic principles and the parameters related to the operation state of the comprehensive electromechanical system. The model comprehensively considers the energy conversion, material flow and operation state relationship between the components from the system level, covering the macroscopic operation logic of the system. The geometric parameter-aerodynamic simulation model focuses on the impeller machinery components. Through the links of aerodynamic design, three-dimensional aerodynamic simulation, three-dimensional strength simulation and post-processing, it deeply explores the geometric shape, aerodynamic performance and strength characteristics of the components, and accurately obtains its performance under different working conditions, such as the flow-expansion ratio-speed characteristic matrix, the efficiency-expansion ratio-speed characteristic matrix, and provides detailed data for the microscopic performance analysis at the component level. The two models comprehensively consider the information of the system's macroscopic operation logic and the microscopic performance of the components, laying a solid foundation for subsequent optimization.

[0024] Then, the three-dimensional characteristic matrix (including the geometric parameter changes) calculated based on the geometric parameter-aerodynamic simulation model is substituted into the impeller machinery part of the first initial model to replace the original three-dimensional characteristic matrix (without geometric parameter changes), thereby realizing the fusion of the two models and obtaining the first fusion model. This process enables the optimization calculation of the first initial model to be directly affected by the geometric design information. When the geometric parameters change, the first fusion model can reflect the changes in the performance of the impeller machinery components in real time based on the new three-dimensional characteristic matrix, thereby driving the corresponding changes in the entire system model. There is no need to rebuild and optimize the system model after the geometric parameters are adjusted, which greatly improves the optimization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flowchart of Embodiment 1 of the method for parallel optimization of geometry and control parameters of an aviation electromechanical system provided in this application;

[0026] Figure 2 This is a schematic diagram of the structure of the second embodiment of the device for parallel optimization of geometry and control parameters of an aviation electromechanical system provided in this application. DETAILED DESCRIPTION

[0027] Here, exemplary embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application.

[0028] The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in this article refers to and includes any or all possible combinations of one or more associated listed items.

[0029] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0030] Specific embodiments are given below to introduce the technical solution of the present application in detail.

[0031] Figure 1This is a flow chart of the first embodiment of the method for parallel optimization of geometry and control parameters of an aviation electromechanical system provided by this application. Figure 1 , the method provided in this embodiment may include:

[0032] S101. Construct a first initial model of an aviation electromechanical system.

[0033] It should be noted that the first initial model is a mathematical model used to describe the overall operating characteristics of an aviation mechatronic system. It is constructed based on a comprehensive consideration of the architecture, functions, control parameters, thermodynamic principles, and parameters related to the operating status of the aviation mechatronic system. It can accurately describe the complex relationship between the architecture, functions, control parameters, and operating status of the aviation mechatronic system.

[0034] Specifically, constructing the first initial model of the aviation electromechanical system includes:

[0035] (1) Determine the devices included in the aviation electromechanical system, the functions of the system, and the control parameters of the system; the control parameters include the power plant speed and compressor inlet pressure.

[0036] It should be noted that according to the architecture of the aviation electromechanical system, it is determined that it includes a power unit, a semi-closed air cycle unit, a starter / generator integrated motor, and an accessory unit, among which the power unit further clearly includes a compressor, a combustion chamber, and a power turbine; the semi-closed air cycle unit includes a heat exchanger, a compressor, and a cooling turbine; and the accessory unit includes pipelines and valves. Since different devices have different functions and energy conversion relationships in the system, clarifying their composition helps to accurately describe the system's workflow and energy transfer path.

[0037] The main functions of the aviation electromechanical system are determined to be environmentally controlled cooling and power supply. These two functions are important bases for system design and optimization. For example, in the environmentally controlled cooling function, the work of cooling turbines and other components is involved, and their performance will affect the cooling effect; while the power supply function is related to components such as the starter / generator motor. The optimization of the system needs to ensure that the performance is improved while meeting these functional requirements.

[0038] In addition, the control parameter of the system is clearly defined as the power unit speed and compressor pressure These two control parameters have a key impact on the operating state of the system. For example, the speed of the power unit will affect the output power of the power turbine, and thus affect the energy distribution of the entire system; the pressure before the compressor will affect the working efficiency and air intake of the compressor, which is closely related to the performance of the system. In the subsequent model construction and optimization process, these two parameters will be adjusted as variables to find the optimal system operating state.

[0039] (2) establishing the first initial model according to thermodynamic principles and integrated electromechanical system operating state related parameters; the aviation electromechanical system operating state related parameters include flight state input information, engine bleed air volume, fuel consumption, power turbine exhaust temperature and compressor stability margin.

[0040] It should be noted that according to the principles of thermodynamics, a detailed energy and material balance analysis is performed on each component of the aviation electromechanical system. For example, in the power unit, the compressor compresses the air, which involves the compression process in thermodynamics. The work consumed and the change of the air state follow the laws of thermodynamics; the combustion process of the fuel in the combustion chamber involves the release and conversion of energy, which also needs to be modeled according to the principles of thermodynamics; the expansion work of the gas in the power turbine also follows the thermodynamic principle of expansion work. By applying these thermodynamic principles to the modeling of each component, the energy conversion and material flow process inside the component can be accurately described.

[0041] Fully consider the parameters related to the operating status of the aviation electromechanical system, such as flight status input information (which may include flight altitude, speed, atmospheric temperature, etc., which will affect the system's intake conditions and working environment), engine bleed air volume (its size is related to the system's energy input and distribution, affecting the working status of other components), fuel consumption (directly related to the system's energy consumption and efficiency), power turbine exhaust temperature (excessive exhaust temperature may affect component life and system performance, and is an important parameter that needs to be controlled during the optimization process) and compressor stability margin (a key indicator to ensure the stable operation of the compressor, which is crucial to the reliability of the system). Incorporating these parameters into the model can establish an accurate relationship between the system input and output, so that the performance of the system under different operating conditions can be simulated and analyzed in the model.

[0042] Specifically, the first initial model established is as follows:

[0043] ;

[0044] in, Indicates flight status input information. Indicates the engine bleed air volume, Indicates fuel consumption, represents the power turbine exhaust temperature, represents the compressor stability margin, Indicates the power unit speed, Indicates the pressure before the compressor.

[0045] S102, constructing a geometric parameter-aerodynamic simulation model of the impeller machinery components.

[0046] It should be noted that in the aviation electromechanical system, the impeller machinery components that are focused on in the construction of the geometric parameter-aerodynamic simulation model are the cooling turbine and the power turbine. This is because the cooling turbine plays a key role in the environmental control cooling function, and its performance directly affects the cooling effect; the power turbine is an important component that converts gas energy into mechanical energy, and has a key impact on the power output and energy distribution of the entire system. For example, the efficiency of the cooling turbine affects the degree of air cooling, and thus affects the temperature control of the system; the output power of the power turbine determines the amount of energy that the system can use to generate electricity and drive other components.

[0047] It should be noted that the construction of the geometric parameter-aerodynamic simulation model of the impeller machinery components includes four parts: aerodynamic design, three-dimensional aerodynamic simulation, three-dimensional strength simulation, and post-processing. Among them, the aerodynamic design part includes one-dimensional aerodynamic design, two-dimensional aerodynamic design, and three-dimensional aerodynamic design modules; the three-dimensional aerodynamic simulation part includes three-dimensional meshing, three-dimensional aerodynamic simulation, and calculation result processing modules; the three-dimensional strength simulation part includes three-dimensional finite element meshing, finite element simulation, and calculation result processing modules; the post-processing part includes automatic post-processing of simulation results and three-dimensional characteristic matrix output modules. Specifically, please refer to the description of the relevant technology for the construction process of each part, which will not be repeated here. In this step, the geometric parameter-aerodynamic simulation model of the impeller machinery components constructed can comprehensively and accurately simulate the geometric shape, aerodynamic performance, and strength characteristics of the impeller machinery components, and provide important basic data and technical support for the optimization design of aviation electromechanical systems.

[0048] S103 , fusing the first initial model and the geometric parameter-aerodynamic simulation model to obtain a first fusion model.

[0049] Among them, a three-dimensional characteristic matrix is ​​calculated according to the geometric parameter-aerodynamic simulation model, and the three-dimensional characteristic matrix includes geometric parameter changes. The original three-dimensional characteristic matrix in the first initial model is replaced based on the three-dimensional characteristic matrix, and the original three-dimensional characteristic matrix does not contain geometric parameter changes.

[0050] It should be noted that the first initial model and the geometric parameter-aerodynamic simulation model are fused to obtain a first fusion model, including:

[0051] (1) Calculating a three-dimensional characteristic matrix based on the geometric parameter-aerodynamic simulation model.

[0052] It should be noted that the three-dimensional characteristic matrix is ​​calculated according to the geometric parameter-aerodynamic simulation model, including:

[0053] (i) Identify the components of the aircraft electromechanical system for which the three-dimensional characteristic matrix needs to be calculated; the components include the cooling turbine and the power turbine.

[0054] Specifically, the reasons for selecting the cooling turbine and the power turbine as the components can be found in the previous description and will not be repeated here.

[0055] (ii) Perform aerodynamic design, 3D aerodynamic simulation, 3D strength simulation and post-processing based on the determined components.

[0056] Based on a series of simulation and processing processes, the performance of cooling turbine and power turbine components can be comprehensively improved to ensure their reliable operation in aviation electromechanical systems, provide strong support for the optimization of the entire system, and achieve an overall improvement in system performance.

[0057] (iii) obtaining a three-dimensional characteristic matrix of the cooling turbine and the power turbine components based on the simulation results; the three-dimensional characteristic matrix includes a flow-expansion ratio-speed characteristic matrix and an efficiency-expansion ratio-speed characteristic matrix.

[0058] It should be noted that by constructing the geometric parameters-aerodynamic simulation model of the turbomachinery components, the three-dimensional characteristic matrix of these components is calculated, including the flow-expansion ratio-speed characteristic matrix, the efficiency-expansion ratio-speed characteristic matrix, etc. These characteristic matrices accurately describe the performance of the components under different working conditions. For example, the flow-expansion ratio-speed characteristic matrix can reflect the flow change of the components under different speeds and expansion ratios, and the efficiency-expansion ratio-speed characteristic matrix shows the change of efficiency with these parameters. Among them, the three-dimensional characteristic matrix is ​​the key to achieve the parallel optimization of the geometry and control parameters of the aviation electromechanical system. It can more truly reflect the actual performance of the turbomachinery components, provide important data support for the optimization of the system, and help improve the accuracy and effectiveness of the optimization of the entire aviation electromechanical system.

[0059] (2) Substituting the calculated three-dimensional characteristic matrix into the impeller machinery of the first initial model to obtain the first initial model with modified geometric parameters.

[0060] It should be noted that, first of all, when constructing the first initial model of the aviation electromechanical system, the performance characteristics of the turbomachinery (cooling turbine and power turbine) were originally described based on certain theoretical assumptions or empirical data, and these descriptions constituted the original three-dimensional characteristic matrix (without geometric parameter changes). The three-dimensional characteristic matrix (including the impact of geometric parameter changes on component performance) calculated by constructing the geometric parameter-aerodynamic simulation model of the turbomachinery components is accurately substituted into the turbomachinery-related parts of the first initial model in this step.

[0061] The first initial model with changed geometric parameters can more realistically reflect the performance of the impeller machinery components in actual operation. Since the new three-dimensional characteristic matrix takes into account the influence of geometric parameter changes, such as the influence of changes in geometric parameters such as blade shape and angle on performance such as flow and efficiency, the model will be more accurate in calculating and predicting the working state of the impeller machinery. This provides a more reliable foundation for subsequent system analysis and optimization based on the model. In addition, in aviation electromechanical systems, there is a coupling relationship between component geometric design parameters and system control parameters. By substituting the three-dimensional characteristic matrix containing the influence of geometric parameter changes into the first initial model, conditions are created for the subsequent parallel optimization of geometric parameters and control parameters (such as power unit speed, compressor front pressure, etc.) that are considered simultaneously. During the optimization process, the model can more accurately evaluate how the system control parameters should be adjusted when the geometric parameters change, and the influence of control parameter changes on geometric parameter optimization, so that it is more likely to find the global optimal solution of the system.

[0062] S104. Generate training samples according to the first fusion model, wherein the input samples of the training samples are geometric parameters and the output samples are three-dimensional characteristic matrices.

[0063] It should be noted that the process of determining the geometric parameters in the training sample includes:

[0064] (1) Based on the sensitivity analysis method, the sensitivity of the power unit speed, compressor inlet pressure and flight status input information in the first fusion model to the engine bleed air volume is analyzed.

[0065] It should be noted that in the first fusion model, the power unit speed, the pressure before the compressor and the flight status input information are parameters that have an important impact on the system performance, and the engine bleed air volume is a key indicator of the system performance. The sensitivity of these three parameters to the engine bleed air volume is selected for analysis because there is a complex relationship between them, and this relationship directly affects the operating efficiency and performance of the system. For example, changes in the power unit speed will affect the working state of the compressor, and then affect the intake volume (engine bleed air volume); changes in the pressure before the compressor will affect the compression efficiency of the compressor, and also affect the engine bleed air volume; different flight status input information (such as flight altitude, speed, etc.) will lead to changes in atmospheric conditions, thereby affecting the engine's intake conditions and bleed air volume.

[0066] Through a specific sensitivity analysis algorithm, while keeping other parameters unchanged, a small change is applied to the power unit speed, compressor front pressure and flight status input information (for example, each time the value of a parameter is changed, an increase or decrease of a small percentage is made), and then the corresponding change in the engine bleed air volume is observed. The ratio of the change in the engine bleed air volume to the change in the parameter is calculated, and this ratio is the sensitivity of the parameter to the engine bleed air volume. For example, if the power unit speed increases by 1% and the engine bleed air volume increases by 0.5%, then the sensitivity of the power unit speed to the engine bleed air volume is 0.5. In this way, each parameter is analyzed one by one to obtain their respective sensitivity values ​​to the engine bleed air volume.

[0067] (2) Based on the analysis results, a preset number of geometric parameters with the highest sensitivity ranking are selected and determined as the geometric parameter change.

[0068] According to the results of the sensitivity analysis, the sensitivity of each parameter is ranked. A preset number of geometric parameters with the highest sensitivity ranking are selected and determined as geometric parameter changes. These selected geometric parameters are usually parameters that have a greater impact on the engine air intake volume. For example, the determined geometric parameter changes can be the outlet angle at the root of the power turbine guide vane, the installation angle at the root of the power turbine guide vane, the outlet angle in the middle of the power turbine guide vane, the installation angle in the middle of the power turbine guide vane, the outlet angle at the tip of the power turbine guide vane, and the installation angle at the tip of the power turbine guide vane.

[0069] The preset number is selected based on the characteristics of the system and the optimization requirements. The determination of this number needs to comprehensively consider factors such as computing resources, model complexity, and the importance of the impact on system performance. Selecting too many geometric parameters may result in excessive calculation and overly complex models, while too few geometric parameters may not fully reflect the characteristics of the system and affect the accuracy of the optimization results.

[0070] Specifically, generating training samples according to the first fusion model includes:

[0071] (1) Determine the number of sample points in the experimental design and set the optimal upper and lower limits of the geometric parameter changes.

[0072] It should be noted that the determination of the number of sample points needs to comprehensively consider computing resources, model complexity, and requirements for result accuracy. If the number of sample points is too small, it may not be able to fully cover various situations of geometric parameter changes, resulting in a lack of representativeness of the training samples; and if the number of sample points is too large, the amount of calculation and time cost will increase. Usually, a suitable number of sample points is determined based on experience or through some preliminary tests. For example, the determined number of sample points can be 100. After determining the number of sample points, based on the geometric parameter changes determined above (the outlet angle at the root of the power turbine guide vane, the installation angle at the root of the power turbine guide vane, the outlet angle in the middle of the power turbine guide vane, the installation angle in the middle of the power turbine guide vane, the outlet angle at the tip of the power turbine guide vane, and the installation angle at the tip of the power turbine guide vane), the optimized upper and lower limits of the changes in these six geometric parameters are set. The optimized upper and lower limits of the above six geometric parameters shown in this application are shown in Table 1:

[0073] Table 1 Optimization upper and lower limits of six geometric parameters

[0074] Unit: Degree Air outlet angle at the root of the guide Guide root installation angle Air outlet angle of guide center Guide middle installation angle Air outlet angle of guide tip Guide tip installation angle Upper limit 65 39 70 45 78 54 Lower limit 61 34 67 41 73 49

[0075] The optimization upper and lower limits of each geometric parameter in Table 1 can provide important constraints for subsequent experimental design, simulation calculations, etc. based on these geometric parameters.

[0076] (2) Apply the Latin hypercube method to obtain a sample group with the same number of sample points.

[0077] It should be noted that the Latin hypercube method is an efficient experimental design method, which is characterized by being able to evenly cover the entire parameter space with a small number of sample points. After determining the number of sample points and the upper and lower limits of the optimization of the geometric parameter changes, the Latin hypercube method is used to generate a sample group. This method generates a series of sample points according to certain rules based on the set parameter range, so that each geometric parameter can be selected with a relatively uniform probability within its optimization range, and the combination of different geometric parameters has good uniformity. In this way, the generated sample group can represent the various possibilities of geometric parameter changes to the greatest extent with a limited number of samples, providing diversified input data for subsequent simulations.

[0078] (3) Based on the first fusion model, the sample points in the obtained sample group are simulated to obtain the corresponding three-dimensional characteristic matrix to form the training sample.

[0079] It should be noted that each sample point (i.e., a set of geometric parameter values) in the sample group obtained by the Latin hypercube method is substituted into the first fusion model for simulation. During the simulation process, the first fusion model will calculate the performance of the corresponding turbomachinery components (cooling turbine and power turbine) based on the input geometric parameters and other fixed parameters in the model (such as power unit speed, compressor front pressure, flight state input information, etc.), and then obtain a three-dimensional characteristic matrix (including flow-expansion ratio-speed characteristic matrix, efficiency-expansion ratio-speed characteristic matrix, etc.). For 100 sample points, 100 sets of three-dimensional characteristic matrices will be obtained, and these matrices and the corresponding input geometric parameters together constitute the training samples.

[0080] S105: Training a neural network model based on the training samples.

[0081] It should be noted that the neural network model is trained based on the obtained training samples, so that the neural network can learn the complex mapping relationship between geometric parameters and three-dimensional characteristic matrices, which provides a basis for the subsequent parallel optimization of the geometry and control parameters of the aviation mechatronic system based on the trained neural network model.

[0082] Specifically, a multi-layer perceptron neural network model is selected to train the training samples, wherein the number of input layer nodes of the model is the same as the number of geometric parameter changes, and the number of output layer nodes is the same as the number of control parameters (power unit speed and compressor front pressure). Combined with the previous description, the number of input layer nodes is set to 6, the number of output layer nodes is set to 2, and in addition, the number of hidden layers is set to 3, and the number of nodes in each hidden layer is set to 64.

[0083] Specifically, the multi-layer perception neural network model is as follows:

[0084] ;

[0085] in, is the flow-expansion ratio-speed characteristic matrix, is the efficiency-expansion ratio-speed characteristic matrix, They are respectively the air outlet angle at the root of the power turbine guide vane, the installation angle at the root of the power turbine guide vane, the air outlet angle in the middle of the power turbine guide vane, the installation angle in the middle of the power turbine guide vane, the air outlet angle at the tip of the power turbine guide vane, and the installation angle at the tip of the power turbine guide vane.

[0086] S106, integrating the trained neural network model and the first initial model to obtain a parallel optimization model.

[0087] It should be noted that the parallel optimization model is obtained by integrating the trained neural network model and the first initial model, including: outputting a three-dimensional characteristic matrix based on the trained neural network model; replacing the original three-dimensional characteristic matrix in the first initial model with the three-dimensional characteristic matrix to obtain a parallel optimization model; the input of the parallel optimization model is the change in geometric parameters, the speed of the power unit, the pressure before the compressor and the flight state input information, and the output is the engine bleed air volume, fuel consumption, power turbine exhaust temperature and compressor stability margin. Specifically, the parallel optimization model is as follows:

[0088] ;

[0089] in, Indicates flight status input information. Indicates the engine bleed air volume, Indicates fuel consumption, represents the power turbine exhaust temperature, represents the compressor stability margin, Indicates the power unit speed, Indicates the pressure before the compressor. They are respectively the air outlet angle at the root of the power turbine guide vane, the installation angle at the root of the power turbine guide vane, the air outlet angle in the middle of the power turbine guide vane, the installation angle in the middle of the power turbine guide vane, the air outlet angle at the tip of the power turbine guide vane, and the installation angle at the tip of the power turbine guide vane.

[0090] It should also be noted that the geometric and control parameters of the aviation electromechanical system are simultaneously generated based on the parallel optimization model, and then include:

[0091] (1) Determine the operating constraints, operation optimization evaluation objectives, and optimization scope of the aviation electromechanical system.

[0092] It should be noted that the operating constraints are restrictions on some physical quantities of the aviation electromechanical system during operation. For example, the power turbine exhaust temperature cannot be too high, because too high exhaust temperature may cause the performance of the turbine blade material to deteriorate or even damage it, affecting the reliability and life of the system; the compressor stability margin must be within a certain range, in order to ensure that the compressor can work stably and avoid unstable phenomena such as surge that affect the normal operation of the system. These constraints are set based on the physical characteristics of the system and the requirements for safe operation. For example, the power turbine exhaust temperature is less than the constraint value of 750°C, and the compressor stability margin is greater than the constraint value of 8%.

[0093] The operation optimization evaluation target is a standard for measuring the performance of the aviation electromechanical system. Usually, the parameters that have a greater impact on the overall performance of the system are selected as the optimization target. For example, the operation optimization evaluation target of the aviation electromechanical system is determined to be the lowest system fuel compensation value (fuel consumption is directly related to aviation costs and endurance). The system fuel compensation value is equal to the sum of the engine bleed air compensation value and fuel consumption. By optimizing to minimize this value, the energy utilization efficiency of the system can be improved and the operating cost can be reduced. Specifically, the calculation formula is as follows:

[0094] ;

[0095] in, is the value of fuel consumption caused by unit bleed air volume for the selected engine; It is the system fuel compensation value; is the engine bleed air compensation value; is the engine bleed air volume; For fuel consumption.

[0096] Combining the above formula with the parallel optimization model, the calculation formula for the system fuel compensation value is obtained as follows:

[0097] ;

[0098] in, is the system fuel compensation value, Indicates flight status input information. Indicates the power unit speed, Indicates the pressure before the compressor. They are respectively the air outlet angle at the root of the power turbine guide vane, the installation angle at the root of the power turbine guide vane, the air outlet angle in the middle of the power turbine guide vane, the installation angle in the middle of the power turbine guide vane, the air outlet angle at the tip of the power turbine guide vane, and the installation angle at the tip of the power turbine guide vane.

[0099] The optimization range is a regulation of the value range of geometric parameters and control parameters. For geometric parameters, the reasonable value range should be determined according to the design requirements and physical characteristics of the components. For example, geometric parameters such as the blade angle of the impeller machinery components have their upper and lower design limits; for control parameters, such as the speed of the power unit and the pressure before the compressor, the value range is also set based on the working capacity and safety of the system. Reasonable determination of the optimization range can find the optimal parameter combination while ensuring the normal operation of the system.

[0100] (2) Determine an optimization weight coefficient based on the proportion of the working time of each operating point in the optimization range to the total flight time.

[0101] It should be noted that the requirements for system performance of aviation electromechanical systems under different flight conditions (such as take-off, cruising, landing, etc.) are different. The proportion of each operating point in the total flight time is different. For example, the cruising stage may account for most of the total flight time, while the take-off and landing stages account for a smaller proportion. Therefore, in the optimization process, all operating points cannot be treated equally, but the optimization weight coefficient should be determined according to their proportion in the total flight time. For operating points with a larger proportion (such as cruising conditions), a larger weight coefficient is given, so that more attention will be paid to the performance of the system under this condition during optimization; for operating points with a smaller proportion (such as take-off and landing conditions), the weight coefficient is relatively small, but it still needs to be considered to ensure the performance of the system throughout the flight process. The optimization weight coefficient determined in this way can make the optimization results of the system more in line with actual flight needs.

[0102] For example, to determine the optimization of the integrated electromechanical system within the full flight envelope, nine typical operating points are selected as flight status input information: Then, according to the proportion of the working time of each operating point in the flight envelope to the total flight time, the optimization weight coefficient is determined, which are Specifically, the flight status input information and the determined optimization weight coefficients are shown in Table 2:

[0103] Table 2 Flight status input information and determined optimization weight coefficients

[0104]

[0105] (3) Determine the optimization objective function based on the geometric parameters and control parameters.

[0106] Combined with the previous description, the geometric parameter optimization includes 6 parameters , the control parameter optimization includes 2 parameters , the optimization objective function is determined as:

[0107] ;

[0108] Among them, x and y represent the upper and lower limits of the optimization amount.

[0109] S107. Simultaneously generate geometry and control parameters of the aviation electromechanical system based on the parallel optimization model.

[0110] It should be noted that the geometric and control parameters of the aviation electromechanical system are simultaneously generated based on the parallel optimization model, including:

[0111] (1) Through the parallel optimization model, geometric parameters and control parameters are used as optimization particles.

[0112] It should be noted that in the parallel optimization model, the geometric parameters and control parameters are regarded as optimization particles, which represent possible solutions in the parameter space. By operating and optimizing these particles, the optimal combination of geometric and control parameters of the system is found. Among them, the chaos initialization strategy can be used to obtain the initialized particle group.

[0113] (2) Substitute the obtained particles into the optimization objective function to obtain the fitness value of the particles.

[0114] After obtaining these particles, they are substituted into the optimization objective function, which is a mathematical expression constructed by comprehensively considering factors such as system performance and operating constraints. The fitness value corresponding to each particle is obtained by calculation. This value reflects the degree of superiority or inferiority of the parameter combination represented by the particle to the system performance. For example, if the optimization goal is to minimize the fuel compensation value of the system, then the fitness value is related to the fuel compensation value under the parameter combination corresponding to the particle. The lower the value, the higher the fitness.

[0115] (3) According to the optimization objective function, the particles that do not meet the constraints are screened out, and the global optimal particles and local optimal particles are obtained according to the obtained fitness values.

[0116] According to the constraints set by the optimization objective function, the particles are screened and the particles that do not meet the constraints are removed. Among the remaining particles, the current global optimal particle and local optimal particle are determined according to the fitness value. Among them, the global optimal particle represents the optimal solution currently found in the entire parameter space, and the local optimal particle is the optimal solution in a certain local area.

[0117] (4) Substituting the global optimal particle and the local optimal particle into the particle swarm motion equation, adjusting the inertia weight according to the adaptive strategy, and then updating the speed and position of the swarm, and adopting the mutation strategy to generate the next generation of the swarm.

[0118] Substitute the global optimal particle and the local optimal particle into the particle swarm motion equation. The particle swarm motion equation describes the motion law of particles in the parameter space. By adjusting the inertia weight according to the adaptive strategy, the global search ability and local search ability of the particles can be balanced. Then, the speed and position of the population are updated, which is equivalent to moving particles in the parameter space to find a better solution. In addition, the mutation strategy is adopted. The mutation strategy can increase the diversity of the population and prevent the algorithm from falling into the local optimal solution too early, thereby generating the next generation of population.

[0119] (5) Determine whether the current population has reached the maximum number of iterations. If not, repeat the steps of obtaining the fitness value to generate the next generation population. If the maximum number of iterations is reached, the algorithm ends, the optimization process ends, the global optimal particle is output, and the optimization result is obtained.

[0120] Check whether the current population has reached the maximum number of iterations. If it has not reached the maximum number of iterations, repeat the above steps from obtaining the fitness value to generating the next generation population. Repeat this process continuously so that the particles continue to search and evolve in the parameter space and gradually approach the optimal solution. When the maximum number of iterations is reached, the algorithm ends and outputs the global optimal particle. The geometric parameters and control parameters corresponding to this global optimal particle are the optimal parameter combination of the aviation electromechanical system obtained after optimization, realizing the parallel optimization of the system geometry and control parameters. Specifically, the optimization results of the geometric parameters and control parameters are shown in Table 3:

[0121] Table 3 Optimization results of geometric parameters and control parameters

[0122]

[0123] The parallel optimization method for geometry and control parameters of an aviation electromechanical system provided in this embodiment comprehensively considers the mutual influence of geometry parameters and control parameters through parallel optimization. In the process of building a model, generating training samples, training neural networks and optimizing, the correlation between various parts of the system is fully considered, and the optimal solution can be found in the entire parameter space, thereby improving the overall performance of the aviation electromechanical system, solving the problem that the traditional serial sequential optimization method of first geometry design and then control design cannot obtain the global optimal solution of the system, improving the design quality of electromechanical products, and providing new ideas and methods for the component and control design of complex electromechanical systems. By determining that the aviation electromechanical system is optimized within the full flight envelope, selecting multiple typical operating points, and determining the optimization weight coefficient according to the proportion of working time of each operating point, compared with the optimization method that only considers a single operating point, this full flight envelope optimization can enable the aviation electromechanical system to maintain good performance throughout the flight process, improving the adaptability and practicality of the system. In addition, by establishing a geometric parameter-aerodynamic simulation model of turbomachinery components, parametric automatic simulation of turbomachinery such as turbines can be realized, providing effective tool support for component geometry design optimization. It should also be noted that the established parallel optimization model integrates the information of the geometric parameters-aerodynamic simulation model and the first initial model. During the optimization process, it can comprehensively consider one-dimensional and three-dimensional physical information, carry out global optimization of the system efficiency, and improve the optimization effect.

[0124] Corresponding to the aforementioned embodiment of a method for parallel optimization of geometry and control parameters of an aviation mechatronic system, the present application also provides an embodiment of a device for parallel optimization of geometry and control parameters of an aviation mechatronic system.

[0125] Figure 2 This is a schematic diagram of the structure of the second embodiment of the device for parallel optimization of geometry and control parameters of an aviation electromechanical system provided by this application. Figure 2, the device provided in this embodiment includes a construction module 210, a fusion module 220, a generation module 230 and a training module 240;

[0126] The construction module 210 is used to construct a first initial model of the aviation electromechanical system;

[0127] The construction module 210 is also used to construct a geometric parameter-aerodynamic simulation model of the impeller machinery component;

[0128] The fusion module 220 is used to fuse the first initial model and the geometric parameter-aerodynamic simulation model to obtain a first fusion model, wherein a three-dimensional characteristic matrix is ​​calculated according to the geometric parameter-aerodynamic simulation model, the three-dimensional characteristic matrix includes geometric parameter changes, and the original three-dimensional characteristic matrix in the first initial model is replaced based on the three-dimensional characteristic matrix, and the original three-dimensional characteristic matrix does not have geometric parameter changes;

[0129] The generating module 230 is used to generate training samples according to the first fusion model, wherein the input samples of the training samples are geometric parameters and the output samples are three-dimensional characteristic matrices;

[0130] The training module 240 is used to train a neural network model based on the training samples;

[0131] The fusion module 220 is also used to fuse the trained neural network model and the first initial model to obtain a parallel optimization model;

[0132] The generation module 230 is further configured to simultaneously generate the geometry and control parameters of the aviation electromechanical system based on the parallel optimization model.

[0133] The device of this embodiment can be used to perform Figure 1 The steps, specific implementation principles and implementation processes of the method embodiment shown are similar and will not be repeated here.

[0134] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0135] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0136] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for parallel optimization of geometry and control parameters of an aviation electromechanical system, characterized in that: The method comprises: Construct the first initial model of the aviation electromechanical system; Construct geometric parameter-aerodynamic simulation model of turbomachinery components; The first initial model and the geometric parameter-aerodynamic simulation model are fused to obtain a first fusion model, wherein a three-dimensional characteristic matrix is ​​calculated according to the geometric parameter-aerodynamic simulation model, the three-dimensional characteristic matrix includes geometric parameter changes, and an original three-dimensional characteristic matrix in the first initial model is replaced based on the three-dimensional characteristic matrix, the original three-dimensional characteristic matrix does not contain geometric parameter changes; Generate training samples according to the first fusion model, wherein the input samples in the training samples are geometric parameters and the output samples are three-dimensional characteristic matrices; Training a neural network model based on the training samples; Fusion the trained neural network model and the first initial model to obtain a parallel optimization model; Simultaneously generating geometric and control parameters of the aero-mechanical system based on the parallel optimization model; The fusion trained neural network model and the first initial model are used to obtain a parallel optimization model, including: Outputting a three-dimensional characteristic matrix based on the trained neural network model; The three-dimensional characteristic matrix replaces the original three-dimensional characteristic matrix in the first initial model to obtain a parallel optimization model; the input of the parallel optimization model is the geometric parameter change, the power unit speed, the compressor front pressure and the flight state input information, and the output is the engine bleed air volume, the fuel consumption, the power turbine exhaust temperature and the compressor stability margin; The first initial model of the aviation electromechanical system is constructed, comprising: Determine the devices included in the aviation electromechanical system, the functions of the system and the control parameters of the system; the control parameters include the power unit speed and the compressor inlet pressure; The first initial model is established according to thermodynamic principles and integrated electromechanical system operating state related parameters; the aviation electromechanical system operating state related parameters include flight state input information, engine bleed air volume, fuel consumption, power turbine exhaust temperature and compressor stability margin; The calculating of the three-dimensional characteristic matrix according to the geometric parameter-aerodynamic simulation model comprises: Determine the components of the aviation electromechanical system for which the three-dimensional characteristic matrix needs to be calculated; the components include a cooling turbine and a power turbine; Perform aerodynamic design, 3D aerodynamic simulation, 3D strength simulation and post-processing based on the determined components; A three-dimensional characteristic matrix of the cooling turbine and the power turbine components is obtained based on the simulation results; the three-dimensional characteristic matrix includes a flow-expansion ratio-speed characteristic matrix and an efficiency-expansion ratio-speed characteristic matrix.

2. The method according to claim 1, characterized in that The process of determining the geometric parameters in the training sample includes: Based on the sensitivity analysis method, the sensitivity of the power unit speed, compressor front pressure and flight state input information in the first fusion model to the engine bleed air volume is analyzed; Based on the analysis results, a preset number of geometric parameters with the highest sensitivity rankings are selected and determined as the geometric parameter changes.

3. The method according to claim 1, characterized in that The fusing the first initial model and the geometric parameter-aerodynamic simulation model to obtain a first fusion model includes: Calculate a three-dimensional characteristic matrix based on the geometric parameter-aerodynamic simulation model; The calculated three-dimensional characteristic matrix is ​​substituted into the impeller machinery of the first initial model to obtain the first initial model with modified geometric parameters.

4. The method according to claim 1, characterized in that: The method of simultaneously generating the geometry and control parameters of the aviation electromechanical system based on the parallel optimization model includes: By means of the parallel optimization model, the geometric parameters and the control parameters are used as optimization particles; Substitute the obtained particles into the optimization objective function to obtain the fitness value of the particles; According to the optimization objective function, the particles that do not meet the constraints are screened out, and the global optimal particles and local optimal particles at this time are obtained according to the obtained fitness values; Substituting the global optimal particle and the local optimal particle into the particle swarm motion equation, adjusting the inertia weight according to the adaptive strategy, then updating the speed and position of the swarm, and adopting the mutation strategy to generate the next generation of swarm; Determine whether the current population has reached the maximum number of iterations. If not, repeat the steps of obtaining the fitness value to generate the next generation population. If the maximum number of iterations is reached, the algorithm ends, the optimization process ends, the global optimal particle is output, and the optimization result is obtained.

5. The method according to claim 1, characterized in that The generating of training samples according to the first fusion model includes: Determine the number of sample points for the test design and set the upper and lower limits of the optimization of the geometric parameter changes; The Latin hypercube method was applied to obtain the same number of sample groups as the sample points; The sample points in the obtained sample group are simulated based on the first fusion model to obtain a corresponding three-dimensional characteristic matrix to form the training sample.

6. The method according to claim 1, characterized in that The method of simultaneously generating the geometry and control parameters of the aero-mechanical system based on the parallel optimization model, then comprises: Determine the operational constraints, operational optimization evaluation objectives, and optimization scope of the aviation electromechanical system; Determining an optimization weight coefficient based on the proportion of the working time of each operating point in the optimization range to the total flight time; The optimization objective function is determined based on the geometric parameters and control parameters.

7. A device for parallel optimization of geometry and control parameters of an aviation electromechanical system, characterized in that: The device comprises a construction module, a fusion module, a generation module and a training module; The building module is used to build a first initial model of the aviation electromechanical system; The construction module is also used to construct a geometric parameter-aerodynamic simulation model of the impeller machinery components; The fusion module is used to fuse the first initial model and the geometric parameter-aerodynamic simulation model to obtain a first fusion model, wherein a three-dimensional characteristic matrix is ​​calculated according to the geometric parameter-aerodynamic simulation model, the three-dimensional characteristic matrix includes geometric parameter changes, and an original three-dimensional characteristic matrix in the first initial model is replaced based on the three-dimensional characteristic matrix, and the original three-dimensional characteristic matrix does not contain geometric parameter changes; The generating module is used to generate training samples according to the first fusion model, wherein the input samples in the training samples are geometric parameters and the output samples are three-dimensional characteristic matrices; The training module is used to train the neural network model based on the training samples; The fusion module is also used to fuse the trained neural network model and the first initial model to obtain a parallel optimization model; The generation module is further used to simultaneously generate the geometry and control parameters of the aviation electromechanical system based on the parallel optimization model; The fusion trained neural network model and the first initial model are used to obtain a parallel optimization model, including: Outputting a three-dimensional characteristic matrix based on the trained neural network model; The three-dimensional characteristic matrix replaces the original three-dimensional characteristic matrix in the first initial model to obtain a parallel optimization model; the input of the parallel optimization model is the geometric parameter change, the power unit speed, the compressor front pressure and the flight state input information, and the output is the engine bleed air volume, the fuel consumption, the power turbine exhaust temperature and the compressor stability margin; The first initial model of the aviation electromechanical system is constructed, comprising: Determine the devices included in the aviation electromechanical system, the functions of the system and the control parameters of the system; the control parameters include the power unit speed and the compressor inlet pressure; The first initial model is established according to thermodynamic principles and integrated electromechanical system operating state related parameters; the aviation electromechanical system operating state related parameters include flight state input information, engine bleed air volume, fuel consumption, power turbine exhaust temperature and compressor stability margin; The calculating of the three-dimensional characteristic matrix according to the geometric parameter-aerodynamic simulation model comprises: Determine the components of the aviation electromechanical system for which the three-dimensional characteristic matrix needs to be calculated; the components include a cooling turbine and a power turbine; Perform aerodynamic design, 3D aerodynamic simulation, 3D strength simulation and post-processing based on the determined components; A three-dimensional characteristic matrix of the cooling turbine and the power turbine components is obtained based on the simulation results; the three-dimensional characteristic matrix includes a flow-expansion ratio-speed characteristic matrix and an efficiency-expansion ratio-speed characteristic matrix.

Images