A method for efficient optimization design of rotor blade aerodynamic configuration
By using a variable reliability proxy model and rotor aerodynamic characteristic analysis with different levels of precision, the problems of high computational load and low efficiency in rotor blade aerodynamic layout optimization design were solved, achieving high-efficiency optimization and improving hovering efficiency and forward lift-to-drag ratio.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA HELICOPTER RES & DEV INST
- Filing Date
- 2023-11-13
- Publication Date
- 2026-07-24
AI Technical Summary
Existing methods for optimizing rotor blade aerodynamic layouts suffer from high costs and low efficiency in terms of computational complexity and time complexity, especially when using high-precision CFD analysis, where the large number of sample points leads to a huge computational burden.
A variable confidence surrogate model and rotor aerodynamic characteristic analysis models with different analysis precisions are adopted. By generating a small number of high confidence and multiple low confidence sample points, and combining low-precision and high-precision analysis methods, a surrogate model is constructed and iteratively optimized to reduce the number of calls to high-precision analysis.
It improves the computational efficiency of rotor blade aerodynamic layout optimization design, enhances hovering efficiency and forward lift-to-drag ratio, and achieves efficient optimization design.
Smart Images
Figure CN117421828B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of helicopter rotor aerodynamic design technology, and specifically relates to an efficient optimization design method for rotor blade aerodynamic layout. Background Technology
[0002] Rotor blade aerodynamic layout design refers to the design of aerodynamics / noise as the design goal, while taking into account structural characteristics. Based on the two-dimensional airfoil, the three-dimensional shape of the rotor blade is obtained through positive and negative torsion, chord length variation, and blade tip shape. The blade aerodynamic layout optimization design method directly determines whether a layout scheme that meets the performance index requirements can be designed.
[0003] Currently, the industry generally adopts rotor aerodynamic layout optimization design technology based on optimization algorithms to replace manual computation. This involves using optimization algorithms to find the optimal solution based on a sufficient number of sample points. To improve accuracy, this type of rotor aerodynamic layout optimization design technology often uses high-precision CFD-based methods for aerodynamic analysis. However, since CFD methods have long computation cycles, calculating a large number of sample points would result in a huge computational load, making the cost and time unacceptable.
[0004] An efficient optimization design method for rotor blade aerodynamic layout employs a variable confidence model-based optimization approach. In each iteration, a small number of high-confidence sample points and multiple low-confidence sample points are simultaneously generated. These are then combined with both CFD-based high-precision rotor aerodynamic analysis and free-wake-based low-precision rotor aerodynamic analysis to generate response values, establishing surrogate models with different confidence levels. The high-confidence surrogate models are then corrected using the low-confidence surrogate models. During the optimization iteration process, the number of calls to the time-consuming high-precision rotor aerodynamic characteristic analysis program is significantly reduced, improving optimization efficiency while maintaining accuracy, thus demonstrating high engineering practical value. Summary of the Invention
[0005] Unlike traditional blade aerodynamic layout optimization design methods, this invention employs a surrogate model with variable confidence and rotor aerodynamic characteristic analysis models with different analysis accuracies. Combined with an optimization process suitable for blade aerodynamic layout design, it constructs an efficient optimization design method for rotor blade aerodynamic layout, effectively improving the computational efficiency of blade aerodynamic layout optimization design.
[0006] A highly efficient optimization design method for rotor blade aerodynamic layout is described in the following specific implementation:
[0007] (1) Define the optimization status and objectives;
[0008] (2) Determine the optimal design parameters and range for the blades;
[0009] (3) Within the range of blade optimization design parameters, the blade optimization design parameter sample point set is obtained by sampling multiple proxy model layers with different confidence levels through experimental design. The blade aerodynamic shape is constructed based on the blade parameterization model. Then, the low confidence sample point and high confidence sample point in the sample point set are calculated and evaluated by rotor aerodynamic characteristic analysis models with different analysis accuracy.
[0010] (4) Build a low-confidence proxy model on the low-confidence dataset and introduce it as a trend model into the next layer of high-confidence proxy model modeling process. Repeat this process to complete the modeling of the variable-confidence proxy model.
[0011] (5) Solve the sub-optimization problem on the surrogate model with varying confidence level, find the optimal solution, and then use the found optimal solution as a new sample point. Continue to use rotor aerodynamic characteristic analysis models with different analysis precision to calculate its response value and add it to the high confidence dataset and low confidence dataset.
[0012] (6) Update the agent model with variable credibility;
[0013] (7) Repeat (4)-(6) until the optimization converges.
[0014] The specific steps (1) are as follows: According to the requirements and indicators of rotor blade optimization design, the flight states that need to be calculated and analyzed are the hovering state and the level flight state, and the corresponding optimization objectives for analysis and evaluation are the hovering efficiency and the rotor forward lift-to-drag ratio.
[0015] Step (2) specifically involves: based on the baseline rotor shape, selecting four blade shape parameter variables for optimization design: chord length increase starting position, with an optimization range of 0.6R to 0.8R; chord length at the sweep start position, with an optimization range of 1.1C to 1.3C; sweep start position, with an optimization range of 0.8R to 0.95R; and the distance from the trailing edge point at 1.0R to the horizontal trailing edge line, with an optimization range of 0C to 0.5C. Here, R is the rotor radius, and C is the chord length at the starting position of the blade's main airfoil section.
[0016] The multiple proxy model layers in step (3) are specifically divided into a high-confidence proxy model layer and a low-confidence proxy model layer. The high-confidence proxy model layer is jointly established by the parameters of high-confidence sample points and the response values calculated by the rotor aerodynamic characteristic analysis model with high analysis accuracy. The low-confidence proxy model layer is jointly established by the parameters of low-confidence sample points and the response values calculated by the rotor aerodynamic characteristic analysis model with low analysis accuracy.
[0017] The specific analysis models of rotor aerodynamic characteristics with different analysis precision in step (3) are as follows: different analysis methods with different calculation precision are used to calculate and evaluate the sample points, including high-precision analysis methods based on CFD and low-precision analysis methods based on free wakes. The high-precision analysis methods take longer, while the low-precision analysis methods take shorter.
[0018] The specific sample point set of blade optimization design parameters in step (3) is as follows: the sample point set of blade optimization design parameters includes four types of parameters: the starting position of chord length increase, the chord length at the starting position of sweep, the starting position of sweep, and the distance from the trailing edge point at 1.0R to the horizontal trailing edge line. Each set of parameters can be constructed by the blade parameterization model to form different blade aerodynamic shapes.
[0019] The low confidence dataset in step (4) is specifically composed of a set of low confidence sample points and corresponding low confidence response values.
[0020] The new sample point in step (5) is specifically obtained by optimizing the new sample point on the surrogate model with varying confidence. If the sample point is obtained by optimizing the low confidence surrogate model, its response value is calculated by the rotor aerodynamic characteristic analysis model with low analysis accuracy. If the sample point is obtained by optimizing the high confidence surrogate model, its response value is calculated by the rotor aerodynamic characteristic analysis model with high analysis accuracy.
[0021] The calculation flow of the traditional method and the method of this invention is as follows: Figure 1 As shown.
[0022] Technical effects of the present invention:
[0023] This invention optimizes the aerodynamic layout of the rotor blades with the goal of improving hovering efficiency and forward lift-to-drag ratio. A comparison is made between traditional methods and the method described in this invention. Figure 2 It can be seen that both methods have good optimization trend effects and produce a "Pareto" front. However, the method of this invention can reduce the need to call the time-consuming high-precision rotor aerodynamic characteristic analysis model by generating fewer high-confidence sample point sets, thereby effectively improving the computational efficiency of blade aerodynamic layout optimization design. Figure 3 and Figure 4 As can be seen, the optimized solution obtained by the method of the present invention improves both rotor hovering efficiency and forward lift-to-drag ratio compared with the benchmark solution, proving the effectiveness of the present invention. Attached Figure Description
[0024] Figure 1 A comparison of the optimization process for rotor blade aerodynamic layout; wherein: (a) traditional calculation method; (b) the calculation method of the present invention;
[0025] Figure 2To optimize the comparison of response value results for sample points; where: (a) traditional calculation method; (b) the calculation method of this invention;
[0026] Figure 3 The blade optimization scheme obtained using the present invention includes: (a) a baseline scheme; and (b) an optimized scheme.
[0027] Figure 4 The aerodynamic performance of the blade optimization scheme obtained in this invention is compared with that of the benchmark scheme; wherein: (a) hovering efficiency; (b) forward lift drag ratio. Detailed Implementation
[0028] The present invention will be further described below with reference to the accompanying drawings:
[0029] like Figure 1 (b) such as Figure 2 (b) Figure 3 , Figure 4 As shown, a specific implementation method for an efficient optimization design of rotor blade aerodynamic layout is as follows:
[0030] (1) Formulate optimization states and objectives; Based on the requirements and indicators of rotor blade optimization design, it is determined that the flight states that need to be calculated and analyzed are hovering state and level flight state, and the corresponding optimization objectives for analysis and evaluation are hovering efficiency and rotor forward lift-to-drag ratio.
[0031] (2) Establish the blade optimization design parameters and ranges; based on the baseline rotor shape, select four blade shape parameter variables for optimization design: chord length increase starting position, optimization range 0.6R~0.8R; chord length at the sweep start position, optimization range 1.1C~1.3C; sweep start position, optimization range 0.8R~0.95R; distance from the trailing edge point at 1.0R to the horizontal trailing edge line, optimization range 0C~0.5C. R is the rotor radius, and C is the chord length at the starting position of the blade's main airfoil section.
[0032] (3) Within the blade optimization design parameters and range, the blade optimization design parameter sample point set is obtained by sampling multiple proxy model layers with different confidence levels through experimental design. The blade aerodynamic shape is constructed based on the blade parameterization model. Then, the low confidence sample point and high confidence sample point in the sample point set are calculated and evaluated by rotor aerodynamic characteristic analysis models with different analysis accuracy.
[0033] The multiple proxy model layers are divided into a high-confidence proxy model layer and a low-confidence proxy model layer. The high-confidence proxy model layer is jointly established by the parameters of high-confidence sample points and the response values calculated by the rotor aerodynamic characteristic analysis model with high analysis accuracy. The low-confidence proxy model layer is jointly established by the parameters of low-confidence sample points and the response values calculated by the rotor aerodynamic characteristic analysis model with low analysis accuracy.
[0034] The sample point set of blade optimization design parameters includes four types of parameters: the starting position of chord length increase, the chord length at the starting position of sweep, the starting position of sweep, and the distance from the trailing edge point at 1.0R to the horizontal trailing edge line. Each set of parameters can be used to construct different blade aerodynamic shapes through blade parameterization models.
[0035] The rotor aerodynamic characteristic analysis models with different levels of analysis precision are: a high-precision analysis method based on CFD and a low-precision analysis method based on free wake. The high-precision analysis method takes longer, while the low-precision analysis method takes less time. The high-confidence sample points are calculated using the high-precision analysis method based on CFD, while the low-confidence sample points are calculated using the low-precision analysis method based on free wake.
[0036] (4) On the low confidence dataset, the sample points and their corresponding response values are fitted by fitting method to construct an approximate model, thereby establishing a low confidence proxy model, and introducing it as a trend model into the next layer of high confidence proxy model modeling process. This process is repeated continuously to complete the modeling of the variable confidence proxy model; the low confidence dataset is composed of a set of low confidence sample points and corresponding low confidence response values.
[0037] (5) Solve the sub-optimization problem on the surrogate model with varying confidence level, find the optimal solution, and then use the found optimal solution as a new sample point. Continue to use rotor aerodynamic characteristic analysis models with different analysis precision to calculate its response value and add it to the high confidence dataset and the low confidence dataset. The new sample point is: a new sample point obtained by optimization on the surrogate model with varying confidence level. If the sample point is obtained by optimization from the low confidence surrogate model, its response value is calculated by the rotor aerodynamic characteristic analysis model with low analysis precision. If it is obtained by optimization from the high confidence surrogate model, its response value is calculated by the rotor aerodynamic characteristic analysis model with high analysis precision.
[0038] (6) Update the agent model with variable credibility;
[0039] (7) Repeat (4)-(6) until the optimization converges.
Claims
1. A method for efficient optimization design of rotor blade aerodynamic layout, characterized in that, The specific implementation method is as follows: (1) Define the optimization status and objectives; (2) Determine the optimal design parameters and range for the blades; (3) Within the range of blade optimization design parameters, the blade optimization design parameter sample point set is obtained by sampling multiple proxy model layers with different confidence levels through experimental design. The blade aerodynamic shape is constructed based on the blade parameterization model. Then, the low confidence sample point and high confidence sample point in the sample point set are calculated and evaluated by rotor aerodynamic characteristic analysis models with different analysis accuracy. (4) Build a low-confidence proxy model on the low-confidence dataset and introduce it as a trend model into the next layer of high-confidence proxy model modeling process. Repeat this process to complete the modeling of the variable-confidence proxy model. (5) Solve the sub-optimization problem on the surrogate model with varying confidence level, find the optimal solution, and then use the found optimal solution as a new sample point. Continue to use rotor aerodynamic characteristic analysis models with different analysis precision to calculate its response value and add it to the high confidence dataset and low confidence dataset. (6) Update the agent model with variable credibility; (7) Repeat (4)-(6) until the optimization converges.
2. The efficient optimization design method for rotor blade aerodynamic layout according to claim 1, characterized in that, The specific steps (1) are as follows: formulate the optimization state and target; according to the rotor blade optimization design requirements and indicators, clarify that the flight state that needs to be calculated and analyzed is the hovering state and the level flight state, and the corresponding optimization targets for analysis and evaluation are the hovering efficiency and the rotor forward lift-to-drag ratio.
3. The efficient optimization design method for rotor blade aerodynamic layout according to claim 1, characterized in that, The specific steps (2) are as follows: formulate the blade optimization design parameters and ranges; based on the reference rotor shape, select four blade shape parameter variables for optimization design: chord length increase starting position, optimization range is 0.6R~0.8R; chord length at the sweep start position, optimization range is 1.1C~1.3C; sweep start position, optimization range is 0.8R~0.95R; distance from the trailing edge point at 1.0R to the horizontal trailing edge line, optimization range is 0C~0.5C; R is the rotor radius and C is the chord length at the starting position of the blade main airfoil section.
4. The efficient optimization design method for rotor blade aerodynamic layout according to claim 1, characterized in that, The multiple proxy model layers in step (3) are specifically divided into a high-confidence proxy model layer and a low-confidence proxy model layer. The high-confidence proxy model layer is jointly established by the parameters of high-confidence sample points and the response values calculated by the rotor aerodynamic characteristic analysis model with high analysis accuracy. The low-confidence proxy model layer is jointly established by the parameters of low-confidence sample points and the response values calculated by the rotor aerodynamic characteristic analysis model with low analysis accuracy.
5. The efficient optimization design method for rotor blade aerodynamic layout according to claim 1, characterized in that, The specific analysis models of rotor aerodynamic characteristics with different analysis precision in step (3) are as follows: different analysis methods with different calculation precision are used to calculate and evaluate the sample points, including high-precision analysis methods based on CFD and low-precision analysis methods based on free wakes. The high-precision analysis methods take longer, while the low-precision analysis methods take shorter.
6. The efficient optimization design method for rotor blade aerodynamic layout according to claim 1, characterized in that, The specific sample point set of blade optimization design parameters in step (3) is as follows: the sample point set of blade optimization design parameters includes four types of parameters: the starting position of chord length increase, the chord length at the starting position of sweep, the starting position of sweep, and the distance from the trailing edge point at 1.0R to the horizontal trailing edge line. Each set of parameters can be constructed by the blade parameterization model to form different blade aerodynamic shapes.
7. The efficient optimization design method for rotor blade aerodynamic layout according to claim 1, characterized in that, The low confidence dataset in step (4) is specifically composed of a set of low confidence sample points and corresponding low confidence response values.
8. The efficient optimization design method for rotor blade aerodynamic layout according to claim 1, characterized in that, The new sample point in step (5) is specifically obtained by optimizing the new sample point on the surrogate model with varying confidence. If the sample point is obtained by optimizing the low confidence surrogate model, its response value is calculated by the rotor aerodynamic characteristic analysis model with low analysis accuracy. If the sample point is obtained by optimizing the high confidence surrogate model, its response value is calculated by the rotor aerodynamic characteristic analysis model with high analysis accuracy.