CN-122021332-A - Method for predicting spraying characteristics of internal combustion engine

CN122021332ACN 122021332 ACN122021332 ACN 122021332ACN-122021332-A

Abstract

The invention discloses a spray characteristic prediction method of an internal combustion engine, which comprises the steps of establishing an original data set comprising data of working condition parameters and data of spray characteristic parameters, calculating deviation between an actual measured value and a calculated value of an existing spray characteristic formula, correcting the existing spray characteristic formula, constructing a physical constraint system of spray characteristics, adopting a Gaussian process regression method to carry out extrapolation expansion, adopting a generated countermeasure network to carry out interpolation enhancement, and carrying out deep learning to construct a spray prediction model, wherein the method corrects the existing spray characteristic formula through deviation, improves the accuracy of data expansion, effectively solves the limitation of small sample experimental data through data enhancement, utilizes the Gaussian process regression extrapolation with priori knowledge to expand coverage limit working conditions, adds the correction formula and the physical constraint into the generated countermeasure network, further complements details, eliminates data break points, and improves the generalization capability, prediction precision and physical rationality of the deep learning model.

Inventors

QIN JING
LIANG YONGSEN
Ren Yige
PEI YIQIANG
Su liwang
Bo Yaqing

Assignees

天津大学

Dates

Publication Date: 20260512
Application Date: 20260206

Claims (8)

1. A method for predicting spray characteristics of an internal combustion engine, comprising the steps of: S1, acquiring spray test data, and establishing an original data set comprising data of working condition parameters and data of spray characteristic parameters; s2, introducing an existing spray characteristic formula of the spray characteristic parameter, calculating the deviation between the actual measurement value and the calculated value of the spray characteristic parameter, and correcting the existing spray characteristic formula: ; Wherein, the For the deviation of the spray characteristic parameter of item i, The existing spray characteristic formula for the spray characteristic parameter of item i, A spray characteristic correction formula for the spray characteristic parameter of item i; s3, constructing a physical constraint system of spray characteristics based on a spray mechanism and the original data set, wherein the physical constraint system comprises trend constraint, association constraint and limit constraint, the trend constraint is used for constraining the variation trend of the spray characteristic parameters along with the core parameters of working conditions, and the association constraint is used for correcting a formula by using the spray characteristics Establishing a core control and physical constraint equation for a core carrier, wherein the limit constraint is a physical boundary value of the spray characteristic parameter; S4, setting a working condition parameter boundary of a limit working condition, taking the spray characteristic correction formula F i as a priori formula, and adopting a Gaussian process regression method to carry out extrapolation expansion on the original data set based on the physical constraint system to obtain an extrapolation expansion data set covering the full working condition; S5, interpolating and enhancing the extrapolated expansion data set by adopting a generation countermeasure network to complement details and eliminate break points, obtaining a training data set, and embedding the prior formula and the physical constraint system into a total loss function of a generator; S6, inputting the data of the training data set into a deep learning network model, learning a change rule of spray characteristics and a relation between the working condition parameters and the spray characteristic parameters, embedding a physical constraint system, and training and verifying a spray prediction model.
2. The method for predicting spray characteristics of an internal combustion engine according to claim 1, wherein said operating condition parameter comprises ambient pressure Ambient temperature Injection pressure And fuel injector aperture The spray characteristic parameter comprises a spray penetration distance Cone angle of spray And spray particle size 。
3. The method for predicting spray characteristics of an internal combustion engine according to claim 2, wherein said deviation amount is fitted to a linear relation relating to said operating condition parameter: ; Wherein, the 、、、 For the influence coefficients of the individual operating parameters, The deviation amount of the spray characteristic parameter of the i-th item.
4. The method for predicting spray characteristics of an internal combustion engine according to any one of claims 1 to 3, wherein the spray is divided into a near spray hole region, a crushing region and a far spray hole region, different main core control and physical constraint equations are respectively configured according to the dominant physical processes of the three regions in step S3, the main core control and physical constraint equations of the near spray hole region comprise a multiphase flow volume fraction transportation equation and a momentum conservation equation, the main core control and physical constraint equations of the crushing region comprise a liquid drop crushing model and a gas-liquid two-phase momentum exchange equation, and the main core control and physical constraint equations of the far spray hole region comprise a gas phase component transportation equation, a drag force equation and a liquid drop evaporation model.
5. The method for predicting spray characteristics of internal combustion engine according to claim 4, wherein in step S4, the partition physical constraint losses are converted into residual terms respectively, and are integrated into a negative log likelihood function of Gaussian process regression, and a final loss function of a unified optimization target is established : ; Wherein, the In order for the likelihood loss to be a function of, For the physical constraint of the global equilibrium coefficient, 、 And Residual errors of the near spray hole area, the crushing area and the far spray hole area are respectively, 、、 And the contribution weight coefficients are respectively the near spray hole area, the crushing area and the far spray hole area.
6. The method for predicting spray characteristics of an internal combustion engine according to claim 5, wherein said contribution weight coefficients derive contribution degrees of three regions according to said inputted operating condition parameters, and adaptively adjust partition constraint intensities.
7. The method for predicting spray characteristics of an internal combustion engine according to claim 5, wherein said total loss function of said generator in step S5 The method comprises the following steps: ; Wherein, the In order to combat the loss of this, The data is lost and, in response to the loss of data, To generate a loss of deviation of the data from the a priori formula map, As a total loss of physical constraints of the partition, 、 And Is the weight coefficient of each loss.
8. The method for predicting spray characteristics of an internal combustion engine according to claim 7, wherein the total loss of the zoned physical constraint is obtained by weighted summing residuals for each zone: ; Wherein, the 、 And And the regional weight coefficients of the near spray hole region, the crushing region and the far spray hole region are respectively adjusted according to the influence degree of different regions on the spray performance parameters.

Description

Method for predicting spraying characteristics of internal combustion engine Technical Field The invention relates to the technical field of simulation calculation of internal combustion engines, in particular to a spray characteristic prediction method of an internal combustion engine. Background The development process and atomization quality of the spray are used as key links for influencing combustion, and can directly influence the mixing effect of fuel oil and air so as to influence combustion efficiency and pollutant emission, so that an accurate and efficient spray characteristic prediction model is established for meeting the requirements of combustion prediction on rapidity and accuracy, and becomes a necessary foundation for further study of the combustion process and reduction of emission; In the prior art, a method for predicting the diesel oil spraying penetration distance with variable oil injection rate is provided, which is characterized in that a method for predicting the diesel oil spraying penetration distance with variable oil injection rate is provided, according to the application number 202010556843.5, a jet mechanics theory is adopted, a variation analytic expression of the effective injection speed along with the injection speed is established, the effective injection speed is replaced by the injection speed and is introduced into a spray model suitable for calculating the constant oil injection rate, the spray model suitable for predicting the penetration distance is obtained, a method for predicting the in-cylinder diesel oil spraying penetration distance with the application number 202311080763.7 is provided, the method for predicting the in-cylinder diesel oil spraying penetration distance is provided, the acceleration of the fuel drops is calculated according to the sauter mean diameter and the mass of the fuel drops, finally, the expression of the liquid spraying penetration distance relative to time is calculated according to the liquid spraying penetration distance prediction model, the prediction accuracy is greatly improved, the result is more close to the real situation, a three-dimensional model of the liquid-liquid coaxial injector with the application number 202827380. X is firstly established, the three-dimensional model of the liquid coaxial injector is divided, the continuous phase is established, the conversion mechanism is based on the preset, the impulse characteristic is improved, the dynamic fluid injector is calculated, the impulse fluid flow is calculated, the simulation result is improved, the simulation result is achieved, and the simulation result is achieved through the simulation is realized, and the method is based on the method. The spray prediction method in the prior art still has the following defects and limitations that 1, under the condition of limited conditions, the collected data volume is limited, a great amount of time and energy are consumed for collecting and expanding a data set through CFD simulation or experiment, the accuracy of data generated by the data set obtained through data expansion is influenced by a priori formula and a posterior formula, but the calculation result of the existing spray characteristic formula has deviation from the actual state, and the expansion accuracy of the data set is influenced; 2, the variable parameters are single, the predicted scene of the spray characteristic is limited, and the spray characteristic is influenced by the coupling of the multidimensional variable parameters such as the structure of the oil sprayer, the ambient pressure, the ambient temperature and the like; 3, the generalization capability of the model is insufficient, the model is built based on specific working condition parameters, and when the variation amplitude exceeds a verification interval, the prediction accuracy of the model is reduced; the calculation cost is large, the iterative calculation amount under the rapid change of the variable parameters is large, the time consumption of the data regression fitting calculation is long, the front simulation calculation occupies calculation resources, the calculation efficiency is improved, and the real-time rapid accurate prediction requirement of dynamic spraying is difficult to adapt; therefore, there is a need for a spray characteristic prediction method that effectively expands a data set based on a modified spray characteristic formula and a physical constraint system, and that is coupled with multi-dimensional parameter effects, and that establishes an accurate and rapid prediction model through machine learning. Disclosure of Invention The invention aims to overcome the defects of the prior art and provide a method for predicting the spraying characteristics of an internal combustion engine, so as to solve the problems in the prior art. In order to achieve the above object, the technical scheme of the present invention is as follows: a method of predicting spray characte