CN-122021294-A - Multi-target design method for automobile air conditioner HVAC system based on sensitivity analysis and collaborative optimization

Abstract

The invention relates to the technical field of design and optimization of an automobile air conditioning system, in particular to a multi-objective design method of an automobile air conditioning HVAC system based on sensitivity analysis and collaborative optimization, which comprises the steps of carrying out sensitivity analysis on a plurality of design parameters of a centrifugal fan through orthogonal test design, and screening out design variables respectively related to air output and pneumatic noise intensity; and constructing an improved collaborative optimization strategy framework, setting a system level optimization target as a weighted normalization function of air output maximization and noise minimization, relaxing a consistency constraint from an equation to an inequality by introducing a relaxation factor, carrying out iterative optimization by taking the difference of design variables between a system level and a discipline level as the discipline level target, and finally obtaining the optimal balance design of the centrifugal fan of the HVAC system.

Inventors

• LAI CHENGUANG
• DUAN MENGHUA
• HUANG SHUNQIAO
• FENG JINYANG
• ZHANG SHAOSONG
• QIN LING
• CHEN DAI
• ZHANG QIANWEN
• XIA CHUNBO
• LIU SHICHUN
• FENG SHUAI
• WANG QINGYU
• FU HANG
• HE ZIQIANG
• SONG JIE
• XU LEI
• WANG QINGYANG

Assignees

• 中国汽车工程研究院股份有限公司
• 重庆理工大学

Dates

Publication Date

20260512

Application Date

20260127

Claims (10)

1. 1. A multi-target design method for an automobile air conditioner HVAC system based on sensitivity analysis and collaborative optimization is characterized by comprising the following steps: The method comprises the steps of S1, parameterized modeling and sensitivity analysis, namely determining initial design parameters of a centrifugal fan, sampling all the design parameters at different levels by adopting an orthogonal test design method, and obtaining the air output and the aerodynamic noise value of each sample point through calculation of fluid dynamics and calculation of aerodynamic acoustic numerical simulation; step S2, constructing and optimizing a single subject approximation model, namely, aiming at an air output subject and a noise subject, respectively carrying out DOE sampling in a design variable subset range corresponding to the subject approximation model, generating a sample geometric model through a grid deformation technology, carrying out numerical simulation, respectively constructing a Kriging approximation model of the air output and the pneumatic noise relative to the corresponding design variables based on sample data, and carrying out single-target optimization on each objective approximation model by adopting a multi-island genetic algorithm to obtain a maximum value Q_max and a minimum value Q_min of the air output and a maximum value N_max and a minimum value of the pneumatic noise so as to determine the optimization range of each target; Step 3, collaborative optimization solution, namely constructing a system-level and discipline-level double-layer optimization framework, wherein the system-level optimizer takes a weighted normalization function of air output maximization and pneumatic noise minimization as an optimization target, and the discipline-level optimizer comprises an air output discipline optimizer and a noise discipline optimizer, wherein the objective functions of the air output discipline optimizer and the noise discipline optimizer are respectively difference norms for minimizing discipline variables and system-level shared variables, and meet respective design variable range constraints; and step S4, result verification, namely combining final design variables obtained by the collaborative optimization strategy, generating a final geometric model through a grid deformation technology, carrying out CFD+CAE numerical simulation verification, and comparing the error of the optimization result and the predicted value of the approximate model.
2. 2. The method for multi-objective design of an HVAC system of an automobile air conditioner based on sensitivity analysis and co-optimization of claim 1, wherein the design parameters of the centrifugal fan in step S1 comprise at least four of an impeller inner diameter, a blade inlet angle, a blade outlet angle, an outlet width, a volute tongue gap, and a blade number.
3. 3. The multi-objective design method for the HVAC system of the automobile air conditioner based on the sensitivity analysis and the collaborative optimization of the invention as set forth in claim 1, wherein the DOE sampling in the step S2 is an optimal Latin hypercube method.
4. 4. The multi-objective design method for the HVAC system of the automobile air conditioner based on the sensitivity analysis and the collaborative optimization of the claim 1 is characterized in that the step S2 is characterized in that after the Kriging approximate model is constructed, the model precision is evaluated by adopting a cross validation method, and the determination coefficient is required Greater than 0.9.
5. 5. The multi-objective design method for the HVAC system of the automobile air conditioner based on the sensitivity analysis and the collaborative optimization of the invention as set forth in claim 1, wherein in the step S3, the system level and the two subject levels both adopt a multi-island genetic algorithm as an optimizer.
6. 6. The method for designing the multiple targets of the HVAC system of the automobile air conditioner based on the sensitivity analysis and the collaborative optimization of claim 5, wherein the parameter setting of the multiple island genetic algorithm comprises the sub-population scale, the island number, the genetic algebra, the crossover rate, the mutation rate and the inter-island mobility.
7. 7. The multi-objective design method for the HVAC system of the automobile air conditioner based on the sensitivity analysis and the collaborative optimization according to claim 1, wherein in the step S3, the mathematical model of the system-level optimizer is as follows: In the above, the above-mentioned method, The output air quantity value transmitted from the output air discipline level to the system level is shown, A sound pressure level value representing the noise discipline level transferred to the system level, For constructing the objective function, the principle of the formula is normalization processing of two targets, thereby measuring the size of the targets, and finally adding weight factors in the front of each term of the formula ; The mathematical model of the single disciplinary level optimizer of the air output is as follows: In the above, the above-mentioned method, The output air quantity value transmitted from the output air discipline level to the system level is shown, The air output value of the system level is represented, Is composed of variables input by the system level and coupling variables of the discipline level, For each variable at the system level, Is each variable of the scientific grade of the air output; the single discipline level optimizer mathematical model of noise level is as follows: In the above, the above-mentioned method, A sound pressure level value representing the noise discipline level transferred to the system level, A sound pressure level value representing the system level, Is composed of variables input by the system level and coupling variables of the discipline level, For each variable at the system level, Are individual variables of the noise class.
8. 8. The method for multi-objective design of an HVAC system of an automobile air conditioner based on sensitivity analysis and collaborative optimization according to claim 7, wherein the weight factors A and B are adjustable between 0.4 and 0.6 according to actual design requirements to balance the emphasis degree of air output and noise.
9. 9. The multi-objective design method for the HVAC system of the automotive air conditioner based on sensitivity analysis and collaborative optimization according to claim 7, wherein the method comprises the following steps: Introducing a relaxation factor for constraint of a system level and constraint functions of the system level are consistent A kind of electronic device , 。
10. 10. The method for multi-objective design of an HVAC system of an automobile air conditioner based on sensitivity analysis and collaborative optimization according to claim 9, wherein the method is characterized by relaxation factors The range of the values is as follows: 。

Description

Multi-target design method for automobile air conditioner HVAC system based on sensitivity analysis and collaborative optimization Technical Field The invention relates to the technical field of design and optimization of automobile air conditioning systems, in particular to a multi-target design method of an automobile air conditioning HVAC system based on sensitivity analysis and collaborative optimization. Background The automobile air conditioning system is a key component for guaranteeing the thermal comfort and the air quality of the passenger cabin. The heat supply ventilation and air conditioning system is at the heart, and the centrifugal fan is used as a key part of the HVAC system and is responsible for driving air circulation, and the performance of the heat supply ventilation and air conditioning system directly determines the air output, energy consumption and noise level of the system. With the rapid development of electric automobiles, noise sources in the automobiles are fundamentally changed, and after traditional engine noise disappears, aerodynamic noise generated by the operation of an HVAC system becomes a primary factor affecting the NVH performance of a passenger cabin. The performance of centrifugal fans, particularly the air output and the pneumatic noise, have a relative relationship. Increasing the fan speed or increasing the flow channel size can increase the air output, but tends to exacerbate airflow separation, vortex shedding and pressure pulsation, resulting in significant increases in broadband vortex noise and discrete rotational noise. Therefore, how to effectively inhibit aerodynamic noise on the premise of ensuring enough air output is a core challenge of the design of the HVAC system of the modern automobile. Traditional centrifugal fan designs rely on designer experience and a large number of "design-sample-test" cycles, are costly, long-lived, and have difficulty finding the optimal solution on the performance pareto front. In recent years, numerical simulation technology based on computational fluid dynamics and computational aero-acoustic provides a powerful tool for fan performance prediction. However, the simulation is directly embedded into the optimization loop, and the problems of huge calculation cost and low optimization efficiency are faced. In addition, fan design involves multiple geometric parameters that affect different performance objectives to varying degrees. Traditional parameterization research or multi-objective optimization algorithm usually treat all variables equally, ignores the relevance strength between the variables and the targets, leads to high optimization dimension and low efficiency, and is easy to fall into local optimum. Disclosure of Invention Aiming at the problems in the prior art, the invention provides a multi-target design method for an automobile air conditioner HVAC system based on sensitivity analysis and collaborative optimization. The method can accurately identify key design variables, remarkably reduce the dimensionality of an optimization problem, and intelligently find the optimal balance point between the air output and the pneumatic noise, so that the development period is shortened, and the comprehensive performance of the product is greatly improved. The technical scheme adopted for solving the technical problems is that the multi-target design method of the automobile air conditioner HVAC system based on sensitivity analysis and collaborative optimization comprises the following steps: the method comprises the steps of S1, parameterizing modeling and sensitivity analysis, namely determining initial design parameters of a centrifugal fan, sampling all design parameters at different levels by adopting an orthogonal test design method, obtaining air output and aerodynamic noise values of all sample points by calculating fluid dynamics and calculating aerodynamic acoustic numerical simulation, systematically exploring independent and interactive influences of a plurality of design variables on two targets of the air output and the noise by utilizing the high efficiency of orthogonal test design and minimum simulation times, calculating the sensitivity of all design parameters on the targets of the air output and the aerodynamic noise based on a range analysis method, and dividing a design variable subset which is strongly related to the air output and a design variable subset which is strongly related to the aerodynamic noise according to the sensitivity; S2, constructing and optimizing a single subject approximation model, namely, respectively sampling DOE (degree of freedom) in the range of a corresponding design variable subset aiming at an air output subject and a noise subject, generating a sample geometric model through a grid deformation technology, and carrying out numerical simulation; based on sample data, respectively constructing Kriging approximate models of the air output and the pneumatic noise about