CN115906644A - Marine propeller numerical map construction method based on machine learning - Google Patents
Marine propeller numerical map construction method based on machine learning Download PDFInfo
- Publication number
- CN115906644A CN115906644A CN202211508963.3A CN202211508963A CN115906644A CN 115906644 A CN115906644 A CN 115906644A CN 202211508963 A CN202211508963 A CN 202211508963A CN 115906644 A CN115906644 A CN 115906644A
- Authority
- CN
- China
- Prior art keywords
- propeller
- parameters
- performance
- sample
- design parameters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Abstract
The application discloses a marine propeller numerical map construction method based on machine learning, which relates to the technical field of ships and adopts a viscous flow numerical calculation model to calculate performance parameters of a propeller with each group of sample design parameters at each sample advancing speed, then each group of sample design parameters and the sample advancing speed are used as input, corresponding performance parameters are used as output, a performance agent prediction model is obtained by utilizing a plurality of groups of sample design parameters based on machine learning model training, namely, performance parameters corresponding to different propeller schemes in a map parameter range can be obtained by utilizing the performance agent prediction model, and a marine propeller numerical map is constructed. The method combines the characteristics of high efficiency and high precision of the viscous flow numerical method and the super-strong nonlinear fitting capability of the machine learning model, does not need to depend on a large number of physical model tests, is simple in construction method, and is high in numerical precision of the constructed propeller map.
Description
Technical Field
The application relates to the technical field of ships, in particular to a marine propeller numerical map construction method based on machine learning.
Background
The propeller is an important part of the ship, and the design of the propeller greatly influences the rapidity, the safety, the vibration performance, the noise performance and the like of the ship. The design method of the marine propeller adopted at present mainly comprises a map design method and a circulation theory design method, wherein the map design method is simple to operate and relatively accurate in hydrodynamic force result, and designers can quickly and effectively obtain a mature scheme with relatively stable performance, so that the requirements of rapidity and intellectualization of propeller design are met.
However, at present, the development of the marine propeller map mainly depends on a physical model test means, and needs to be processed through the processing of a propeller physical model, a water tank test, test data processing and the like, and generally, a mature propeller map comprises dozens or even hundreds of series propeller schemes, so that the construction of the marine propeller map usually needs to consume more time and cost, the efficiency is low, and the popularization and the application of a map design method are not facilitated.
Disclosure of Invention
In order to solve the problems and the technical requirements, the applicant provides a method for constructing a numerical map of a marine propeller based on machine learning, and the technical scheme of the method is as follows:
a marine propeller numerical map construction method based on machine learning comprises the following steps:
constructing a sample data set, wherein the sample data set comprises sample design parameters of a plurality of groups of propellers;
for each group of sample design parameters in the sample data set, calculating performance parameters of the propeller with the sample design parameters at the advancing speed of each sample by adopting a viscous flow numerical calculation model;
taking each group of sample design parameters and sample advance speed as input, and corresponding performance parameters as output, and training by using a sample data set based on a machine learning model to obtain a performance agent forecasting model;
and obtaining performance parameters corresponding to different propeller schemes in the spectrum parameter range by using the performance agent forecasting model, and constructing to obtain the numerical spectrum of the marine propeller.
The further technical scheme is that the performance parameters of the propeller calculated by adopting the viscous flow numerical calculation model comprise open water hydrodynamic performance parameters and cavitation performance parameters, and the performance proxy prediction model obtained by training comprises the following steps:
taking each group of sample design parameters and sample advance speed as input, and corresponding open water hydrodynamic performance parameters as output, and training by using a sample data set based on a machine learning model to obtain a hydrodynamic agent forecasting model;
taking the design parameters and the sample advancing speed of each group of samples as input and the corresponding cavitation performance parameters as output, and training by utilizing a sample data set based on a machine learning model to obtain a cavitation agent forecasting model;
and the constructed hydrodynamic agent forecasting model is used for obtaining open water hydrodynamic performance parameters corresponding to different propeller schemes in the map parameter range, and the constructed cavitation agent forecasting model is used for obtaining cavitation performance parameters corresponding to different propeller schemes in the map parameter range, so that the constructed ship propeller numerical map reflects the open water hydrodynamic performance parameters and the cavitation performance parameters corresponding to the different propeller schemes in the map parameter range.
The method for obtaining the performance parameters corresponding to different propeller schemes in the map parameter range by using the performance agent forecasting model comprises the following further technical scheme:
determining a disc surface ratio and a lateral inclination parameter, carrying out value dispersion on a pitch ratio in a map parameter range, and constructing and obtaining a group of candidate design parameters based on the set disc surface ratio, the lateral inclination parameter and the value of each pitch ratio;
forecasting by using a hydrodynamic agent forecasting model to obtain open water hydrodynamic performance parameters of each group of candidate design parameters, and forecasting the rapidity of a real ship by using a parameter matching method based on the open water hydrodynamic performance parameters to obtain the rapid performance and the corresponding propeller diameter of each group of candidate design parameters;
selecting a group of candidate design parameters meeting the design requirements as a propeller scheme according to the rapid performance of each group of candidate design parameters and the corresponding propeller diameter, and calculating cavitation performance parameters of the propeller scheme by using a cavitation agent prediction model;
and when the performance parameters of the propeller scheme meet the performance design requirements, outputting the propeller scheme, otherwise, adjusting at least one of the disc surface ratio and the lateral inclination parameters, and re-executing the step of determining the disc surface ratio and the lateral inclination parameters.
The further technical scheme is that the method for selecting a group of candidate design parameters meeting the design requirements as a propeller scheme comprises the following steps:
a set of candidate design parameters for optimal hydrodynamic efficiency in fast performance is selected as a propeller solution, or a set of candidate design parameters for a target diameter corresponding to a propeller diameter is selected as a propeller solution.
The method for calculating the performance parameters of the propeller with the sample design parameters under different sample advancing speeds for each group of sample design parameters comprises the following steps:
selecting a plurality of fitting speed points in a speed range corresponding to the sample design parameters according to sampling intervals, and respectively calculating performance parameters of the propeller with the sample design parameters at each fitting speed point;
performing curve fitting on the corresponding relation between the plurality of groups of fitting acceleration points and the performance parameters to obtain a performance parameter curve corresponding to the design parameters of the sample;
carrying out interpolation selection in the acceleration range according to working condition intervals to obtain a plurality of sample acceleration, wherein the working condition intervals are smaller than sampling intervals, and the number of the obtained sample acceleration is larger than that of the fitting acceleration points;
and determining and obtaining the performance parameters of the propeller with the sample design parameters at the advancing speed of each sample through the performance parameter curve.
The method for obtaining the performance agent forecasting model by utilizing the sample data set based on the machine learning model training comprises the following steps:
and training the sample data set based on various different machine learning models respectively to obtain a plurality of candidate models, and selecting the candidate model with the optimal forecasting performance as a performance agent forecasting model.
The further technical scheme is that the adopted machine learning model is a nonlinear machine learning regression model and comprises a random forest and a feedforward neural network.
The further technical scheme is that the method for constructing the sample data set comprises the following steps:
based on the propeller design parameters of the parent type propeller, at least one of the disc surface ratio, the pitch ratio and the sideslip parameters in the propeller design parameters is changed, other propeller design parameters are kept unchanged, a plurality of groups of sample design parameters are obtained, and a sample data set is constructed and obtained.
The further technical scheme is that the method for obtaining the multiple groups of sample design parameters comprises the following steps:
performing parameter expansion towards two sides by taking 0.1Z as a center, and selecting n1 uniformly distributed disk surface ratio parameter discrete points; selecting n2 uniformly distributed disk surface ratio parameter discrete points within the range of 0.5-1.5, and selecting n3 uniformly distributed lateral inclination parameter discrete points within the range of 0-360/Z degrees; wherein Z represents the number of propeller blades of the parent propeller, and n1, n2 and n3 all exceed a predetermined threshold;
and (3) carrying out parameter combination on the n1 disc surface ratio parameter discrete points, the n2 disc surface ratio parameter discrete points and the n3 skew parameter discrete points, and combining other propeller design parameters of the parent propeller to obtain n1 x n2 x n3 groups of sample design parameters.
The further technical scheme is that the method also comprises the following steps:
and adjusting to obtain model parameters of the viscous flow numerical calculation model meeting the calculation precision by utilizing the propeller design parameters of the female propeller and the performance parameters of the female propeller.
The beneficial technical effect of this application is:
the application discloses a marine propeller numerical map construction method based on machine learning, the method is realized by combining a viscous flow numerical method with a machine learning model, the CFD technology is higher in prediction efficiency and prediction precision of performance parameters, and the superstrong nonlinear fitting capacity of the machine learning model is far beyond the traditional regression means, so that the method can construct a marine propeller hydrodynamic force and cavitation numerical map with high precision without depending on a large number of physical model tests, the rapid design and selection of the marine propeller can be realized, the numerical precision of the propeller map is high, the evaluation result is reliable, the engineering practical value is high, the propeller map development cost is favorably reduced greatly, and the map design efficiency and the design precision of the marine propeller are improved.
Drawings
Fig. 1 is a flowchart of a method for constructing a ship propeller numerical map according to an embodiment of the present application.
FIG. 2 is a flow diagram of a method to a performance agent forecasting model trained in one embodiment of the present application.
Fig. 3 is a schematic diagram of an example in which a thrust coefficient curve and a torque coefficient curve are obtained by respectively fitting the thrust coefficient and the torque coefficient at the fitted speed point, and then interpolation is performed according to working condition intervals to obtain the thrust coefficient and the torque coefficient corresponding to the advance speed of each sample.
FIG. 4 is a flow chart of a method of obtaining a propeller solution and its corresponding performance parameters within a map parameter range.
Detailed Description
The following description of the embodiments of the present application will be made with reference to the accompanying drawings.
The application discloses a marine propeller numerical map construction method based on machine learning, please refer to a flow chart shown in fig. 1, and the method comprises the following steps:
In one embodiment, a sample data set is constructed based on propeller design parameters of a parent propeller, the parent propeller refers to a propeller with excellent comprehensive performance, and the parent propeller is selected according to actual needs. And then, on the basis of the propeller design parameters of the female propeller, changing at least one of the disc surface ratio, the pitch ratio and the sideslip parameters in the propeller design parameters of the female propeller, keeping other propeller design parameters unchanged, obtaining a plurality of groups of sample design parameters, and constructing to obtain a sample data set.
Compared with a traditional propeller numerical map, the propeller numerical map has the advantages that the consideration on the parameter of the sideslip parameter is also increased, the disc surface ratio, the pitch ratio and the sideslip parameter are parameters which have larger influence on the performance parameter of the propeller, and the influence of other propeller design parameters on the performance parameter is smaller, so that other propeller design parameters are kept unchanged, and the calculation amount is reduced as far as possible on the basis of not sacrificing the accuracy.
On the basis of propeller design parameters of a female propeller, multidimensional series scheme expansion design is carried out aiming at the disc surface ratio, the pitch ratio and the sideslip, and the propeller parameter expansion range is subject to the requirement of covering common design working conditions of the ship propeller. In one embodiment, parameter expansion is carried out towards two sides by taking 0.1Z as a center, and n1 uniformly distributed disk surface ratio parameter discrete points are selected; n2 uniformly distributed disk surface ratio parameter discrete points are selected within the range of 0.5-1.5, and n3 uniformly distributed skew parameter discrete points are selected within the range of 0-360/Z degrees. Wherein Z represents the number of blades of the parent propeller. n1, n2 and n3 each exceed a predetermined threshold, such as each exceed 5. And (3) carrying out parameter combination on the n1 disc surface ratio parameter discrete points, the n2 disc surface ratio parameter discrete points and the n3 skew parameter discrete points, and combining other propeller design parameters of the parent propeller, thereby expanding and obtaining n1 x n2 x n3 groups of sample design parameters.
And 2, for each group of sample design parameters in the sample data set, calculating the performance parameters of the propeller with the sample design parameters at the advancing speed of each sample by adopting a viscous flow numerical calculation model.
In one embodiment, in order to improve the accuracy of a calculation result, a model parameter of a viscous flow numerical calculation model meeting the calculation precision is obtained by adjusting the propeller design parameter of the female propeller and the performance parameter of the female propeller.
And then calculating the performance parameters of the propeller of each sample design parameter at the advancing speed of each sample by using a viscous flow numerical calculation model. The performance parameters comprise open water hydrodynamic performance parameters and cavitation performance parameters, and the open water hydrodynamic performance parameters comprise a thrust coefficient K T And a torque coefficient K Q The cavitation performance parameter is the lowest pressure coefficient of the vane surface.
For each propeller with sample design parameters, if the number of the sample advance speed selections is too small, the sample amount is too small, and the accuracy of subsequent machine learning is affected, but if the number of the sample advance speed selections is too large, the time consumption for respectively calculating by adopting a viscous flow numerical calculation model is longer.
Therefore, in order to balance the calculation amount and the sample amount, in one embodiment, for each set of sample design parameters, the method for calculating the performance parameters of the propeller with the sample design parameters at different sample advance speeds comprises the following steps:
(1) And selecting a plurality of fitting speed points in the speed range corresponding to the sample design parameters according to sampling intervals, and respectively calculating the performance parameters of the propeller with the sample design parameters at each fitting speed point.
The corresponding speed ranges of different sample design parameters may be different, and the corresponding thrust coefficient is not less than 0, that is, the thrust coefficient of the propeller having the sample design parameter in the corresponding speed range is always greater than or equal to 0. The sampling interval is set according to actual conditions.
(2) And performing curve fitting on the corresponding relation between the plurality of groups of fitted speed points and the performance parameters to obtain a performance parameter curve corresponding to the sample design parameters, wherein the performance parameter curve corresponding to the sample design parameters is a continuous curve of the performance parameters of the propeller with the sample design parameters at different speeds. The curve fitting of this step may be in the form of polynomial fitting.
As mentioned above, in the present application, the performance parameters include open water hydrodynamic performance parameters and cavitation performance parameters, and therefore the performance parameter curve obtained in this step includes a hydrodynamic curve and a cavitation curve, the hydrodynamic curve is a continuous curve of open water hydrodynamic performance parameters of the propeller having the sample design parameters at different forward speeds, and the cavitation curve is a continuous curve of cavitation performance parameters of the propeller having the sample design parameters at different forward speeds.
(3) And carrying out interpolation selection according to working condition intervals in the acceleration range to obtain a plurality of sample acceleration rates, wherein the working condition intervals are smaller than the sampling intervals, and the number of the obtained sample acceleration rates is larger than the number of the fitting acceleration points.
(4) And directly determining and obtaining the performance parameters of the propeller with the sample design parameters at the advancing speed of each sample through the performance parameter curve. Therefore, performance parameters do not need to be calculated for the advancing speed of each sample, but the effect of sample size expansion is achieved, and subsequent training of a machine learning model is facilitated.
For example, referring to the example shown in fig. 3, first, 12 fitting speed advance points are selected according to the sampling interval Δ J =0.1, and the thrust coefficient K at each fitting speed advance point is calculated respectively T And coefficient of torque K Q . Then, the thrust coefficient K is obtained by fitting by taking the advancing speed J as a variable T A thrust coefficient curve changing along with the advancing speed J, and a torque coefficient K obtained by fitting Q Torque coefficient curve as a function of speed J. Then, interpolation selection is carried out according to the working condition interval delta J =0.025 to obtain 46 sample advancing speeds, and a thrust coefficient K corresponding to the 46 sample advancing speeds is obtained by utilizing a thrust coefficient curve T Obtaining torque coefficient K corresponding to the advancing speeds of 46 samples by using a torque coefficient curve Q 。
And 3, taking the design parameters and the sample advancing speed of each group of samples as input and the corresponding performance parameters as output, and training by utilizing the sample data set based on a machine learning model to obtain a performance agent forecasting model.
The method comprises the following steps: and taking the design parameters and the sample advancing speed of each group of samples as input and the corresponding open water hydrodynamic performance parameters as output, and training by utilizing the sample data set based on a machine learning model to obtain a hydrodynamic agent forecasting model. And taking the design parameters and the sample advancing speed of each group of samples as input and the corresponding cavitation performance parameters as output, and training by utilizing the sample data set based on a machine learning model to obtain a cavitation agent forecasting model.
Whether a hydrodynamic agent forecasting model or a cavitation agent forecasting model is obtained through training, a plurality of candidate models are obtained through training respectively based on a plurality of different machine learning models by utilizing a sample data set. Fig. 2 illustrates an example of separate training using P different machine learning models. In one embodiment, the machine learning model employed is a non-linear machine learning regression model, and the machine learning model employed typically includes random forests and feed-forward neural networks. And then selecting the candidate model with the optimal prediction performance as a performance agent prediction model, wherein the predicted Mean Square Error (MSE) can be adopted to measure the prediction performance of the candidate model, namely the candidate model with the minimum MSE is selected as the performance agent prediction model.
And 4, obtaining performance parameters corresponding to different propeller schemes in the map parameter range by using the performance agent forecasting model, and constructing and obtaining the numerical map of the marine propeller. Each propeller scheme is a set of propeller design parameters.
As described above, the trained performance proxy prediction models include a hydrodynamic proxy prediction model and a cavitation proxy prediction model, the hydrodynamic proxy prediction model is used to obtain open water hydrodynamic performance parameters corresponding to different propeller schemes within a map parameter range, and the cavitation proxy prediction model is used to obtain cavitation performance parameters corresponding to different propeller schemes within the map parameter range. The numerical map of the marine propeller obtained by construction reflects open water hydrodynamic performance parameters and cavitation performance parameters corresponding to different propeller schemes within the map parameter range. Considering that the open water hydrodynamic performance parameters of the propeller scheme are relatively more important performance parameters in the propeller design process and the problem of matching of operating points needs to be considered, in one embodiment, the method for obtaining the performance parameters corresponding to different propeller schemes within the spectrum parameter range by using the performance surrogate prediction model comprises the following steps:
firstly, determining the disk surface ratio and the lateral inclination parameters, and inputting the values of the disk surface ratio and the lateral inclination parameters as specified values. And then carrying out value dispersion on the pitch ratio within the spectrum parameter range, wherein the discrete density of the step reaches a density threshold value, namely carrying out intensive value dispersion on the pitch ratio.
And constructing to obtain a group of candidate design parameters based on the set disc surface ratio, the set sideslip parameters and the values of each pitch ratio obtained through dispersion, wherein the same disc surface ratio and sideslip parameters are combined with different pitch ratios to obtain a plurality of groups of different candidate design parameters.
Forecasting by using a hydrodynamic agent forecasting model to obtain open water hydrodynamic performance parameters of each group of candidate design parameters, and utilizing K based on the open water hydrodynamic performance parameters T /J 2 And forecasting the rapidity of the real ship by using a parameter matching method to obtain the rapid performance of each group of candidate design parameters and the corresponding propeller diameter, wherein the rapid performance comprises hydrodynamic efficiency, navigational speed, rotating speed, working point and the like.
And selecting a group of candidate design parameters meeting the design requirements as a propeller scheme according to the rapid performance of each group of candidate design parameters and the corresponding propeller diameter. In one embodiment, a set of candidate design parameters for optimal hydrodynamic efficiency in fast performance is selected as a propeller solution. Or in one embodiment, the target diameter required by the propeller is also input as a specified value, and a set of candidate design parameters with the corresponding propeller diameter as the target diameter is selected as a propeller scheme. After a propeller scheme is obtained by selecting a pitch ratio, a cavitation performance parameter of the propeller scheme can be calculated by using a cavitation surrogate prediction model.
When the performance parameters of the propeller scheme obtained at the moment meet the performance design requirements, the propeller scheme is output, and meanwhile, the open water hydrodynamic performance parameters, the cavitation performance parameters, the rapid performance and the corresponding propeller diameter of the propeller scheme can also be output. And when the performance parameters of the propeller scheme obtained at the moment do not meet the performance design requirements, adjusting at least one of the disc surface ratio and the sideslip parameters, and re-executing the step of determining the disc surface ratio and the sideslip parameters until the propeller scheme is output. The performance design requirements indicate requirements to be met by open water hydrodynamic performance parameters and cavitation performance parameters.
In practical application, the method of the present application can be implemented by programming in a programming language such as C/python.
What has been described above is only a preferred embodiment of the present application, and the present application is not limited to the above examples. It is to be understood that other modifications and variations directly derived or suggested to those skilled in the art without departing from the spirit and concepts of the present application are to be considered as included within the scope of the present application.
Claims (10)
1. A marine propeller numerical map construction method based on machine learning is characterized by comprising the following steps:
constructing a sample data set, wherein the sample data set comprises sample design parameters of a plurality of groups of propellers;
for each group of sample design parameters in the sample data set, adopting a viscous flow numerical calculation model to calculate performance parameters of the propeller with the sample design parameters at the advancing speed of each sample;
taking each group of sample design parameters and sample advancing speed as input, and corresponding performance parameters as output, and training by using the sample data set based on a machine learning model to obtain a performance agent forecasting model;
and obtaining performance parameters corresponding to different propeller schemes in the spectrum parameter range by using the performance agent forecasting model, and constructing to obtain the numerical spectrum of the marine propeller.
2. The method of claim 1, wherein the performance parameters of the propeller calculated by the viscous flow numerical calculation model comprise open water hydrodynamic performance parameters and cavitation performance parameters, and the training of the performance proxy prediction model comprises:
taking each group of sample design parameters and sample advancing speed as input, and corresponding open water hydrodynamic performance parameters as output, and training by using the sample data set based on a machine learning model to obtain a hydrodynamic agent forecasting model;
taking each group of sample design parameters and sample advancing speed as input, and corresponding cavitation performance parameters as output, and training by using the sample data set based on a machine learning model to obtain a cavitation agent forecasting model;
the hydrodynamic agent forecasting model is used for obtaining open water hydrodynamic performance parameters corresponding to different propeller schemes within a map parameter range, the cavitation agent forecasting model is used for obtaining cavitation performance parameters corresponding to different propeller schemes within the map parameter range, and the ship propeller numerical map is constructed to reflect the open water hydrodynamic performance parameters and the cavitation performance parameters corresponding to the different propeller schemes within the map parameter range.
3. The method of claim 2, wherein the method of using the performance proxy prediction model to derive performance parameters corresponding to different propeller schemes within a map parameter range comprises:
determining a disc surface ratio and a lateral inclination parameter, carrying out value dispersion on the pitch ratio in a map parameter range, and constructing and obtaining a group of candidate design parameters based on the set disc surface ratio, the lateral inclination parameter and the value of each pitch ratio;
forecasting by using the hydrodynamic agent forecasting model to obtain open water hydrodynamic performance parameters of each group of candidate design parameters, and forecasting the rapidity of a real ship by using a parameter matching method based on the open water hydrodynamic performance parameters to obtain the rapid performance of each group of candidate design parameters and the corresponding propeller diameter;
selecting a group of candidate design parameters meeting design requirements as a propeller scheme according to the rapid performance of each group of candidate design parameters and the corresponding propeller diameter, and calculating to obtain cavitation performance parameters of the propeller scheme by using the cavitation proxy forecasting model;
and when the performance parameters of the propeller scheme meet the performance design requirements, outputting the propeller scheme, otherwise, adjusting at least one of the disc surface ratio and the sideslip parameters, and re-executing the step of determining the disc surface ratio and the sideslip parameters.
4. The method of claim 3, wherein selecting a set of candidate design parameters that meet the design requirements as a propeller solution comprises:
a set of candidate design parameters for optimal hydrodynamic efficiency in fast performance is selected as a propeller solution, or a set of candidate design parameters for a target diameter corresponding to a propeller diameter is selected as a propeller solution.
5. The method of claim 1, wherein for each set of sample design parameters, the method of calculating performance parameters for a propeller having the sample design parameters at different sample approach speeds comprises:
selecting a plurality of fitting speed points in a speed range corresponding to the sample design parameters according to sampling intervals, and respectively calculating performance parameters of the propeller with the sample design parameters at each fitting speed point;
performing curve fitting on the corresponding relation between the plurality of groups of fitting acceleration points and the performance parameters to obtain a performance parameter curve corresponding to the sample design parameters;
performing interpolation selection in the acceleration range according to working condition intervals to obtain a plurality of sample acceleration, wherein the working condition intervals are smaller than the sampling intervals, and the number of the obtained sample acceleration is larger than the number of the fitting acceleration points;
and determining and obtaining the performance parameters of the propeller with the sample design parameters at the advancing speed of each sample through the performance parameter curve.
6. The method of claim 1, wherein the method of deriving a performance agent prediction model based on machine learning model training using the sample data set comprises:
and training the sample data set based on multiple different machine learning models respectively to obtain multiple candidate models, and selecting the candidate model with the optimal forecasting performance as the performance agent forecasting model.
7. The method of claim 6, wherein the machine learning model used is a non-linear machine learning regression model, and wherein the machine learning model used comprises a random forest and a feed-forward neural network.
8. The method of claim 1, wherein constructing the sample data set comprises:
and on the basis of the propeller design parameters of the parent propeller, changing at least one of the disc surface ratio, the pitch ratio and the sideslip parameters in the propeller design parameters, keeping other propeller design parameters unchanged, obtaining a plurality of groups of sample design parameters, and constructing to obtain the sample data set.
9. The method of claim 8, wherein obtaining a plurality of sets of sample design parameters comprises:
carrying out parameter expansion towards two sides by taking 0.1Z as a center, and selecting n1 uniformly distributed disk surface ratio parameter discrete points; selecting n2 uniformly distributed disk surface ratio parameter discrete points within the range of 0.5-1.5, and selecting n3 uniformly distributed lateral inclination parameter discrete points within the range of 0-360/Z degrees; wherein Z represents the number of propeller blades of the female propeller, and n1, n2 and n3 all exceed a predetermined threshold;
and performing parameter combination on the n1 disc ratio parameter discrete points, the n2 disc ratio parameter discrete points and the n3 skew parameter discrete points, and combining other propeller design parameters of the parent propeller to obtain n1 x n2 x n3 groups of sample design parameters.
10. The method of claim 8, further comprising:
and adjusting to obtain the model parameters of the viscous flow numerical calculation model meeting the calculation precision by utilizing the propeller design parameters of the female propeller and the performance parameters of the female propeller.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211508963.3A CN115906644A (en) | 2022-11-29 | 2022-11-29 | Marine propeller numerical map construction method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211508963.3A CN115906644A (en) | 2022-11-29 | 2022-11-29 | Marine propeller numerical map construction method based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115906644A true CN115906644A (en) | 2023-04-04 |
Family
ID=86474467
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211508963.3A Pending CN115906644A (en) | 2022-11-29 | 2022-11-29 | Marine propeller numerical map construction method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115906644A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116776464A (en) * | 2023-06-09 | 2023-09-19 | 武汉理工大学 | Method and system for generating profile pedigree of specific-route river-sea direct container ship |
-
2022
- 2022-11-29 CN CN202211508963.3A patent/CN115906644A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116776464A (en) * | 2023-06-09 | 2023-09-19 | 武汉理工大学 | Method and system for generating profile pedigree of specific-route river-sea direct container ship |
| CN116776464B (en) * | 2023-06-09 | 2024-01-30 | 武汉理工大学 | Method and system for generating profile pedigree of specific-route river-sea direct container ship |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107944648B (en) | Large ship speed and oil consumption rate prediction method | |
| KR102206324B1 (en) | Methods for ascertaining a model of a starting variable of a technical system | |
| CN115906644A (en) | Marine propeller numerical map construction method based on machine learning | |
| CN111898212B (en) | Impeller mechanical profile design optimization method based on BezierGAN and Bayesian optimization | |
| CN109992889B (en) | Wind power plant model building method and system and wake value calculating method and system | |
| CN115048874B (en) | Aircraft design parameter estimation method based on machine learning | |
| Saad et al. | A new kriging–bat algorithm for solving computationally expensive black-box global optimization problems | |
| CN112650053A (en) | Genetic algorithm optimization-based motor PID self-tuning method for BP neural network | |
| CN113777931A (en) | Icing wing type pneumatic model construction method, device, equipment and medium | |
| CN114429090A (en) | Gas compressor blade robustness design method based on data driving | |
| CN113919221A (en) | Fan load prediction and analysis method and device based on BP neural network and storage medium | |
| CN111310341B (en) | Fan operation parameter determining method, device, equipment and readable storage medium | |
| CN116292057A (en) | Control method, device, equipment and medium for variable-speed pumped storage unit | |
| CN116305574A (en) | Test design method based on neural network model | |
| Chu et al. | Improved fast model migration method for centrifugal compressor based on bayesian algorithm and Gaussian process model | |
| CN116123029A (en) | Pitch angle control method and device, electronic equipment and wind turbine generator | |
| CN113294297B (en) | Variable weight adjusting method for wind turbine generator nonlinear model prediction torque control | |
| CN115438312A (en) | Probability distribution model establishment and application method and medium for new energy output | |
| CN114357602A (en) | Design method, device and equipment of propeller | |
| CN116776600B (en) | Wind turbine blade optimal design method and system based on self-adaptive proxy model | |
| CN115809506B (en) | Method and system for forecasting linear wave load in ship structural design | |
| CN112632726A (en) | Flow field reconstruction method for aeroelasticity simulation of impeller mechanical blade | |
| He et al. | Aerodynamic data fusion with a multi-fidelity surrogate modeling method | |
| Vedam et al. | Evaluation of Gradient and Curvature-Based Adaptive Mesh Refinement for Viscous Transonic Flows | |
| CN111219257A (en) | Turbofan engine direct data drive control method based on adaptive enhancement algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |