CN117688849A - Marine emission virtual test method based on hierarchical sampling agent model - Google Patents
Marine emission virtual test method based on hierarchical sampling agent model Download PDFInfo
- Publication number
- CN117688849A CN117688849A CN202410149365.4A CN202410149365A CN117688849A CN 117688849 A CN117688849 A CN 117688849A CN 202410149365 A CN202410149365 A CN 202410149365A CN 117688849 A CN117688849 A CN 117688849A
- Authority
- CN
- China
- Prior art keywords
- marine
- model
- emission
- launching
- rigid
- 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.)
- Granted
Links
- 238000005070 sampling Methods 0.000 title claims abstract description 48
- 238000013461 design Methods 0.000 claims abstract description 21
- 230000008878 coupling Effects 0.000 claims abstract description 19
- 238000010168 coupling process Methods 0.000 claims abstract description 19
- 238000005859 coupling reaction Methods 0.000 claims abstract description 19
- 238000004364 calculation method Methods 0.000 claims description 43
- 238000012549 training Methods 0.000 claims description 35
- 239000011159 matrix material Substances 0.000 claims description 31
- 230000000694 effects Effects 0.000 claims description 29
- 238000013528 artificial neural network Methods 0.000 claims description 26
- 239000011248 coating agent Substances 0.000 claims description 19
- 238000000576 coating method Methods 0.000 claims description 19
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 238000000034 method Methods 0.000 claims description 18
- 239000003795 chemical substances by application Substances 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 13
- 230000033001 locomotion Effects 0.000 claims description 12
- 238000011160 research Methods 0.000 claims description 12
- 238000009413 insulation Methods 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 9
- 239000007787 solid Substances 0.000 claims description 9
- 230000005540 biological transmission Effects 0.000 claims description 6
- 230000003993 interaction Effects 0.000 claims description 6
- 238000004088 simulation Methods 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 6
- 238000006073 displacement reaction Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000005381 potential energy Methods 0.000 claims description 3
- 238000005096 rolling process Methods 0.000 claims description 3
- 238000010206 sensitivity analysis Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 2
- 238000012827 research and development Methods 0.000 abstract description 12
- 238000010586 diagram Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 2
- 239000000945 filler Substances 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000005309 stochastic process Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a marine launching virtual test method based on a layered sampling agent model, which belongs to the technical field of launching, wherein a layered sampling agent model is formed by coupling launching phases of a marine launching movable platform and key structural parameters of a launching system, a large sample virtual test is rapidly completed, dynamic characteristic response parameters of a mass launching system are obtained, and finally launching safety of the marine movable platform is analyzed and evaluated. The marine emission virtual test method based on the layered sampling agent model can obtain the envelope characteristic of the emission dynamic characteristic more accurately, effectively guide the design, improve the research and development efficiency, shorten the research and development period and reduce the research and development cost.
Description
Technical Field
The invention relates to the technical field of emission, in particular to an offshore emission virtual test method based on a hierarchical sampling agent model.
Background
The motion of the mobile platform of rocket-borne mobile platform launching systems is induced by wind and wave loading, and the waves are usually described by a complex stochastic process. The complex wind and wave load excitation acts on the offshore launching movable platform, and the wind and wave load is transmitted to the rocket through the movable platform and the offshore launching system structure. The wind wave load, the offshore launching movable platform and the launching system structure are mutually coupled, so that the rocket offshore launching process shows extremely strong nonlinear coupling effect. The problem that the safety boundary and the launching dynamics characteristic envelope boundary are difficult to be found exists in the launching of the high-sea-state marine moving platform, and the high-sea-state launching risk is high.
In general, the limit sea condition of the high sea condition is difficult to reproduce reliably, and in the design stage, the sea launching dynamics characteristic of the carrier rocket of the high sea condition is analyzed, so that the high sea condition can be reproduced by a virtual simulation test method. However, as the number of the six-degree-of-freedom sea wave random phase combinations is large, the possible emission boundary working conditions of the movable platform are extremely large, and great difficulty is brought to comprehensive, safe and reliable quantitative analysis of the dynamic characteristics of the marine launching system.
The traditional research of the dynamic characteristics of the marine launching system is generally based on rigid-flexible coupling system dynamic theory, and the dynamic response of the launching system for determining the structural parameters under the extreme value combination of the swinging motion boundary of the movable platform is analyzed, so that the dynamic response is used as the design theoretical basis of the launching system, the envelope boundary of the dynamic characteristics of the marine launching system is determined, the determination of a marine launching safety window is guided, and the design risk is released.
Disclosure of Invention
The invention aims to provide an offshore emission virtual test method based on a layered sampling agent model, which is used for coupling the emission phase of an offshore emission movable platform and key structural parameters of an emission system, rapidly completing a large sample virtual test through the layered sampling agent model, acquiring dynamic characteristic response parameters of a massive emission system, more accurately obtaining envelope characteristics of the emission dynamic characteristics, effectively guiding design, improving research and development efficiency, shortening research and development period and reducing research and development cost.
In order to achieve the above purpose, the invention provides an offshore emission virtual test method based on a hierarchical sampling agent model, which comprises the following steps:
s1, determining structural parameters which need to be coupled with a transmitting phase of a moving platform in the research of the transmitting dynamic characteristics of the marine moving platform, and establishing a dynamic parameterized model of a multi-rigid-body marine moving platform transmitting system;
s2, parameterizing the emission phase moment in the multi-rigid-body marine mobile platform emission system dynamics parameterized model established in the step S1 to complete multi-rigid-body marine mobile platform emission system dynamics model modeling, and obtaining a multi-rigid-body marine mobile platform emission system dynamics model with parameterized structure and phase;
s3, sampling the transmitting phase time parameter in the step S2 and the parameter of the parameterized structure in the step S1 by adopting a sampling method, wherein each sampling result corresponds to a calculation file;
s4, carrying out emission phase large sample calculation of the multi-rigid-body marine moving platform emission system dynamics model under the standard structural parameters based on the structure and phase parameterized multi-rigid-body marine moving platform emission system dynamics model in the step S2 and the sampling result in the step S3 to form an emission phase large sample database;
s5, building three hidden layers by using an artificial neural network algorithm, training the emission phase large sample database in the step S4 to form an emission phase large sample proxy model, and acquiring phase boundaries of emission dynamic characteristic bad boundary working conditions according to the emission phase large sample proxy model, wherein each working condition phase boundary is obtained to obtain a calculation file;
s6, under the severe boundary working condition determined in the step S5, carrying out structural parameter large sample calculation of a multi-rigid-body marine dynamic platform launching system dynamics model on the parameterized structural model in the step S1, sequentially calling a calculation file obtained in the step S5 of calculating the multi-body system dynamics calculation general solver by using an operating system instruction script, obtaining a simulation analysis result, storing a model of each input solver and a corresponding output result in a folder named according to parameters, and forming a structural parameter large sample database by using all result folders and result data;
s7, building three hidden layers by using an artificial neural network algorithm, and obtaining a structural parameter large sample proxy model by taking the emission structural parameter large sample database obtained in the step S6 as input, and obtaining structural parameters corresponding to the bad boundary working conditions of emission dynamics according to the structural parameter large sample proxy model;
s8, forming a dynamic proxy model of the multi-rigid-body marine dynamic platform launching system based on the launching phase large sample proxy model in the step S5 and the structural parameter large sample proxy model in the step S7, and obtaining a global multi-parameter combined severe working condition of the multi-rigid-body marine dynamic platform launching system dynamics;
s9, invoking a sobol algorithm to perform multi-parameter sensitivity analysis based on the dynamic proxy model of the multi-rigid-body marine mobile platform launching system in the step S8 by using python, so as to obtain the influence of each parameter on the dynamic characteristics of the multi-rigid-body marine mobile platform launching system;
s10, establishing a dynamic model of the rigid-flexible coupling marine mobile platform launching system, and taking boundary conditions of a global multi-parameter combination severe working condition and a traditional typical extremum combination working condition of the multi-rigid-body marine mobile platform launching system dynamics in the step S8 as boundary input parameters for researching dynamic characteristics of the rigid-flexible coupling marine mobile platform launching system together to develop dynamic characteristics of the rigid-flexible coupling marine mobile platform launching system to obtain dynamic characteristics research results of the rigid-flexible coupling marine mobile platform launching system;
s11, analyzing the dynamic characteristic research result of the multi-rigid-body marine mobile platform launching system in the step S9 and the dynamic characteristic research result of the rigid-flexible coupling marine mobile platform launching system in the step S10, and completing the marine mobile platform launching safety analysis and evaluation.
Preferably, in step S1, the structural parameters include: rocket geometry, size and mass characteristics, launch barrel inner radius, adapter size and stiffness characteristics.
Preferably, in step S1, a dynamic parameterized model of the multi-rigid-body marine mobile platform launching system is established, including the following steps:
s101, geometric modeling: (1) According to the actual size, modeling is carried out on a movable platform, a supporting cylinder, a transmitting cylinder matrix, a tail cover, an adapter matrix and a low-pressure chamber module in the offshore movable platform transmitting system by using three-dimensional modeling software, and model simplification is carried out according to the actual structure and the relative position of a transmitting device;
wherein the transmitting cylinder matrix is a solid cylinder; the adapter matrix comprises 4 circular rings with 1/4 circular arc structures, 3 circular rings in total, and 12 adapter matrices, wherein the inner diameter of the adapter matrix is smaller than the minimum value of the outer diameter parameterization of the rocket thermal insulation coating, and the outer diameter of the adapter matrix is larger than the maximum value of the inner diameter parameterization of the launching tube;
(2) Geometrically modeling the rocket by using multi-body dynamics simulation software; the rocket body surface is a heat-insulating coating, 3 adapters are arranged on the heat-insulating coating surface, and two groups of parameters of the radius of the rocket heat-insulating coating and the inner radius of the launcher are defined as design variables: defining a design variable alpha, wherein the radius of the rocket thermal insulation coating is represented by alpha, and completing a rocket geometric model by combining the rocket length and the segmentation size of a given length; defining a design variable beta, establishing a solid cylinder with a radius beta at the coaxial position with the transmitting cylinder matrix, performing Boolean subtraction operation on the transmitting cylinder matrix and the solid cylinder with the radius beta through geometry create shape csg command to obtain a hollow cylinder with the inner radius beta, and completing modeling of the transmitting cylinder; performing Boolean subtraction operation on the adapter matrix, the rocket thermal insulation coating and the adapter matrix and the launch canister respectively through geometry create shape csg commands to complete geometric modeling of the adapter;
s102, defining quality characteristics: setting mass, barycenter coordinates and rotational inertia parameters for the rocket and the movable platform respectively; setting density, young modulus and Poisson ratio parameters for the supporting cylinder, the transmitting cylinder, the tail cover and the adapter respectively;
s103, defining a connection relation: the movable platform is fixedly connected with the supporting cylinder, the supporting cylinder is fixedly connected with the launching cylinder, the adapter is fixedly connected with the rocket, the tail cover is in contact relation with the launching cylinder and the rocket, and the adapters are in contact relation with the launching cylinder.
Preferably, in step S2, the phase of the transmitting time is parameterized, which specifically includes: the six-degree-of-freedom motion of the movable platform is simulated by a sine function, and the generalized degree-of-freedom motion at the moment of transmitting is expressed as follows:
(1)
wherein,,/>the values of (1) respectively correspond to rolling, pitching, bowing, swaying, pitching and swaying of the movable platform; />、/>、/>Respectively representing the corresponding generalized degree-of-freedom motion amplitude, angular frequency and phase; />Is a time term.
Preferably, in step S4, the emission phase subsamples of the dynamic model of the multi-rigid-body marine mobile platform emission system are calculated, and the specific operations are as follows: and (3) circularly and sequentially calling the calculation files obtained in the step (S3) of calculating the general solver of the multi-body system dynamics calculation by using an operating system instruction script, obtaining simulation analysis results, storing a model and a corresponding output result of each input general solver of the multi-body system dynamics calculation in a folder named according to parameters, and forming a transmitting phase large sample database by using all the result folders and the result data;
the general solver formula for the dynamics calculation of the multi-body system is as follows:
(2)
wherein,;/>,/>for kinetic energy item->Is a potential energy item; />A displacement term for a directional component.
Preferably, in step S5, the emission dynamics are bad boundary conditions, including: maximum pitch angle working condition, maximum pitch angle speed working condition, maximum yaw angle speed working condition and maximum collision load working condition of rocket in the process of launching the rocket at the moment of leaving the rocket frame.
Preferably, in step S5, three hidden layers are built by using an artificial neural network algorithm, the emission phase large sample database in step S4 is trained to form an emission phase large sample proxy model, and the phase boundary of the emission dynamic characteristic bad boundary working condition is obtained according to the emission phase large sample proxy model, which specifically comprises the following steps: using the emission phase large sub-sample database obtained in the step S4 as input, wherein 70% is selected as a training set, 20% is selected as a verification set, and 10% is selected as a test set;
the training set database is directly presented to the neural network during training, and the neural network carries out parameter adjustment according to the error of the training set database; the verification set data is used for measuring the generalization capability of the network and stopping training the network when the generalization stops improving; the test set has no influence on the training of the network and is used for evaluating the error of the network after the training; establishing a neural network prediction agent model based on the sample point calculation data, testing the neural network through a test set, and determining an error of the neural network;
by R 2 Measuring the accuracy of the neural network, and calculating the obtained R 2 Comparing the training error with a preset proxy model error, stopping training when the training error is smaller than the preset error, and otherwise, continuing training until the preset error requirement is met; and obtaining a transmitting phase large sub-sample proxy model, and obtaining phase boundaries of the working conditions with bad transmitting dynamic characteristics according to the transmitting phase large sub-sample proxy model, wherein each working condition phase boundary obtains a calculation file.
Preferably, in step S9, the sobol algorithm will first solve the total variance representing the total variation of the model output, and output the main effect, interaction effect and error term decomposed into each parameter;
the total variance is equal to the variance matrix of the output quantity:
(3)
the main effect index represents the effect of each individual parameter on the output variable:
(4)
the total effect index represents the total influence of the parameter on the output variable after the individual parameter and the interaction effect of the parameter and other parameters are considered, and the calculation method is as follows:
(5)
wherein,S k represent the firstkMain effect index of each parameter;S Tk is the firstkTotal effect index of individual parameters;VarX k is shown inX k Performing variance calculation under the condition of (1);E ~Xk representation pairX k Marginalizing, i.e. removingX k Is a desired value of (2);Var(Y|X k ) Is shown inX k Under the condition of (2)YIs a conditional variance of (2);Var(Y) Representing model outputYIs the total variance of (2);X k is the firstkA parameter;Yis output;
error term: the error between the analog output and the true output is used in the sobol algorithm by the total errorTo estimate the main effect of the input variables:
(6)
for analog output, +.>Is the desire for true output.
Therefore, the invention adopts the above-mentioned marine emission virtual test method based on the hierarchical sampling agent model, and has the following technical effects:
(1) According to the invention, the transmission phase of the marine transmission movable platform and key structural parameters of the transmission system are coupled, a large sample virtual test is rapidly completed through a layered sampling agent model, and the dynamic characteristic response parameters of the massive transmission system are obtained, so that the envelope characteristic of the transmission dynamic characteristic can be more accurately obtained, the design is effectively guided, the research and development efficiency is improved, the research and development period is shortened, and the research and development cost is reduced;
(2) Acquiring dynamic response characteristic parameters of an offshore launching system under the conditions of mass launching phase combination and launching system structure parameter combination, acquiring a global solution of severe working conditions under the condition of multi-parameter combination, powerfully supporting the work of a design stage, and fully releasing design risks;
(3) The safe boundary of rocket marine movable platform launching is accurately obtained, engineering practice is effectively guided, and launching risks are fully released.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
FIG. 1 is a flow chart of a method for marine launching virtual test based on a hierarchical sampling proxy model of the present invention;
FIG. 2 is a diagram of a component topology;
FIG. 3 is a schematic illustration of Latin hypercube sampling and optimized Latin hypercube sampling; wherein (a) in fig. 3 is a latin hypercube schematic; fig. 3 (b) is a schematic diagram of optimized latin hypercube sampling.
Detailed Description
The technical scheme of the invention is further described below through the attached drawings and the embodiments.
Unless defined otherwise, technical or scientific terms used herein should be given the ordinary meaning as understood by one of ordinary skill in the art to which this invention belongs.
Example 1
Referring to fig. 1, a flowchart of a marine emission virtual test method based on a hierarchical sampling agent model according to the present invention is shown, and specific steps are as follows:
s1, determining rocket geometric shape, size and mass characteristics, inner radius of a launch barrel and adapter size and rigidity characteristics. Establishing a parameterized model of the multi-rigid-body marine mobile platform launching system, which comprises the following steps:
the parameterized model building step of the multi-rigid-body marine mobile platform launching system is as follows:
s101, geometric modeling
(1) The three-dimensional modeling software Ug models a moving platform, a supporting cylinder, a transmitting cylinder matrix (solid cylinder), a tail cover and an adapter matrix (a circular ring consisting of 4 blocks of 1/4 circular arc structures, 3 circular rings in total, 12 adapter matrices, the inner diameter of the adapter matrix is smaller than the minimum value of the outer diameter parameterization of the rocket thermal insulation coating, the outer diameter of the adapter matrix is larger than the maximum value of the inner diameter parameterization of the transmitting cylinder) and a low-pressure chamber module in an offshore moving platform transmitting system through a sketch, a stretching and rotating command group. The rocket body surface is a heat-insulating coating, and 3 adapters are arranged on the heat-insulating coating surface.
Two groups of parameters of the radius of the rocket thermal insulation coating and the inner radius of the launching tube are defined as design variables: defining a design variable alpha, wherein the radius of the rocket thermal insulation coating is represented by alpha, and completing a rocket geometric model by combining the rocket length with a given length and the sectional size; defining a design variable beta, establishing a solid cylinder with a radius beta at the coaxial position with the transmitting cylinder matrix, performing Boolean subtraction operation on the transmitting cylinder matrix and the solid cylinder with the radius beta through geometry create shape csg command to obtain a hollow cylinder with the inner radius beta, and completing modeling of the transmitting cylinder; and performing Boolean subtraction operation on the adapter matrix and the rocket thermal insulation coating and the adapter matrix and the launch canister respectively through geometry create shape csg commands to complete geometric modeling of the adapter.
(2) And (3) importing the model built in the step (1) into multi-rigid-body dynamics solving software adams in an x_t format.
S102, defining quality characteristics for part in adams
Setting mass, barycenter coordinates and rotational inertia parameters for the rocket and the movable platform respectively; and setting density, young modulus and Poisson ratio parameters for the supporting cylinder, the transmitting cylinder, the tail cover and the adapter respectively.
S103, defining connection relation in adams
As shown in fig. 2, the movable platform is fixedly connected with the supporting cylinder, the supporting cylinder is fixedly connected with the launching cylinder, the adapter is fixedly connected with the rocket, the tail cover is in contact relation with the launching cylinder and the rocket, and the adapters are in contact relation with the launching cylinder;
s2, in the dynamic parameterization model of the multi-rigid-body marine moving platform launching system established in the step S1, the moving platform movement speed is set, namely, the launching moment phase is parameterized, the six-degree-of-freedom movement of the moving platform is simulated by a sine function, and the rolling movement of the moving platform at the launching moment is expressed as 1.5 degrees sin (1.05 degrees sin)t+φ 1 ) Pitching 0.4 °. Sin (1.05 ×)t+φ 2 ) Heave 0.2 sin (1.05)t+φ 3 ),φ i ,i=1, 2,3; respectively, the corresponding motion phases, the specific values of which are obtained by sampling in step S3. The geometric model, the contact relation model and the boundary condition of the multi-rigid-body marine moving platform launching system are constructed, the multi-rigid-body marine moving platform launching system dynamics model is formed based on the multi-rigid-body system dynamics equation, and the structure and the phase parameters are obtainedA dynamic model of the system is launched by the multi-rigid-body marine dynamic platform;
and S3, sampling the transmitting phase time parameter in the step S2 and the parameter of the parameterized structure in the step S1 by adopting a sampling method. Five design variables, roll phaseφ 1 Pitching movementφ 2 Pitching movementφ 3, The radius of the rocket thermal insulation coating is alpha, the radius beta of the launching cylinder is within the range of [0,2 pi ]],[0,2*pi],[0,2*pi],[1.2,2],[1.5,2.4]. Each sampling result corresponds to a calculation file;
in addition, the optimized Latin hypercube sampling method can be adopted for sampling. As shown in fig. 3, a schematic diagram of the latin hypercube sampling method and the optimized latin hypercube sampling method is shown.
The optimized Latin hypercube sampling method is a random sampling technology and aims to reduce the correlation between input variables so as to improve the accuracy of Monte Carlo simulation. The randomness and uniformity of the sample are ensured by equally dividing the value range of each component into the same intervals and randomly extracting a value in each interval. The optimal Latin hypercube optimizes the location of the sampling points by:
1. and analyzing the probability distribution condition of the input variables and the value range of each input variable.
2. According to the probability distribution condition of the input variable, the initial sampling point is randomly generated in the value range.
3. The positions of the sampling points are continuously adjusted by an iterative optimization algorithm (objective function=weight 1+weight 2+correlation index; the spatial filler index includes Kolmogorov-Smirnov statistic (K-S statistic) and cramer statistic, etc., and the correlation index includes Spearman correlation coefficient and Pearson correlation coefficient, etc.) to maximize the spatial filler of the sampling points and minimize the correlation between the input variables.
4. It results in an optimal set of sampling points that are uniformly distributed in space and that have less correlation with each other, thus making the sampling more efficient and accurate.
S4, based on the parameterized multi-rigid-body marine dynamic platform launching system dynamics model in the step S2 and the sampling result in the step S3, launching phase large sample calculation of the multi-rigid-body marine dynamic platform launching system dynamics model under standard structural parameters is carried out, an operating system instruction script (bat file) is utilized to circularly and sequentially call the calculation file obtained in the multi-body system dynamics calculation general solver calculation S3, a simulation analysis result is obtained, a multi-body system dynamics calculation general solver core formula is shown in the formula (1), the operating system instruction script is firstly written in a txt postscript text format, and the operating system instruction script is written in a form of 'disk drive/multi-stage file path/solver disk drive/multi-stage file path/solving calculation file auxiliary instruction' (auxiliary instruction is used for confirming whether background solving), a solver is called, after writing is completed, the txt postscript text file is saved and closed, then the txt postfix is changed into the bat format, and then the operating system instruction script (bat file) is double-clicked, so that the system instruction can be operated. Storing the model of each input multi-body system dynamics calculation general solver and the corresponding output result in a folder named according to parameters, wherein all the result folders and the result data form a transmitting phase large sample database;
(1)
wherein,;/>,/>for kinetic energy item->Is a potential energy item; />A displacement term that is a component of a certain direction; />Is a time term.
S5, constructing three hidden layers in the isight by using an artificial neural network algorithm (Artificial Neural Networks, ANN), and taking the emission phase large sample database obtained in the step S4 as input, wherein 70% is selected as a training set, 20% is selected as a verification set, and 10% is selected as a test set. The training set database is directly presented to the neural network during training, and the neural network carries out parameter adjustment according to the error of the training set database; the verification set data is used for measuring the generalization capability of the network and stopping training the network when the generalization stops improving; the test set has no impact on the training of the network for evaluating the error of the network after training.
Based on the sample point calculation data, a neural network prediction agent model is established, the neural network is tested through a test set, the error is determined, and R is adopted 2 For measuring accuracy of neural network, R 2 The larger the fitting effect, the better and the more accurate the prediction result. R is calculated 2 And comparing the training with a preset proxy model error, stopping training when the training is smaller than the preset error, and otherwise, continuing training until the preset error requirement is met. Obtaining a transmitting phase large sub-sample proxy model, and obtaining phase boundaries of the working conditions with bad transmitting dynamic characteristics according to the transmitting phase large sub-sample proxy model, wherein each working condition phase boundary obtains a calculation file;
s6, under the severe boundary working condition determined in the S5, carrying out structural parameter large sample calculation of a multi-rigid-body marine dynamic platform launching system dynamics model on the parameterized structural model in the S1, sequentially calling a calculation file obtained in the multi-body system dynamics calculation general solver calculation S5 by utilizing an operating system instruction script (bat) and obtaining a simulation analysis result, storing a model of each input solver and a corresponding output result in a folder named according to parameters, and forming a structural parameter large sample database by using all result folders and result data;
s7, building three hidden layers by using an artificial neural network algorithm, and obtaining a structural parameter large sample proxy model by taking the emission structural parameter large sample database obtained in the step S6 as input, and obtaining structural parameters corresponding to the bad boundary working conditions of emission dynamics according to the structural parameter large sample proxy model;
s8, forming a multi-rigid-body marine movable platform launching system dynamics proxy model based on the launching phase large sample proxy model in the step S5 and the structural parameter large sample proxy model in the step S7, and obtaining a global multi-parameter combined severe working condition of the multi-rigid-body marine movable platform launching system dynamics;
s9, invoking a sobol algorithm to perform multi-parameter sensitivity analysis based on the dynamic proxy model of the multi-rigid-body marine dynamic platform transmitting system in the step S8 by using python; the sobol algorithm will first solve for the total variance representing the total variation of the model output, which can be decomposed into the main effect, interaction effect and error term for each parameter.
The total variance is equal to the variance matrix of the output quantity:
(3)
the main effect index represents the effect of each individual parameter on the output variable:
(4)
the total effect index represents the total influence of the parameter on the output variable after the individual parameter and the interaction effect of the parameter and other parameters are considered, and the calculation method is as follows:
(5)
wherein,S k represent the firstkMain effect index of each parameter;S Tk is the firstkTotal effect index of individual parameters;VarX k is shown inX k Performing variance calculation under the condition of (1);E ~Xk representation pairX k Marginalizing, i.e. removingX k Is a desired value of (2);Var(Y|X k ) Is shown inX k Under the condition of (2)YIs a conditional variance of (2);Var(Y) Representing model outputYIs the total variance of (2);X k is the firstkA parameter;Yis output.
Error term: the error between the analog output and the true output is used in the sobol algorithm by the total errorTo estimate the main effect of the input variables:
(6)
for analog output, +.>Is the desire for true output.
The influence of each parameter on the dynamic characteristics of the multi-rigid-body marine dynamic platform launching system is obtained;
s10, establishing a dynamic model of the rigid-flexible coupling marine mobile platform launching system, and taking boundary conditions of a global multi-parameter combination severe working condition and a traditional typical extremum combination working condition of the multi-rigid-body marine mobile platform launching system dynamics in the step S8 as boundary input parameters for researching dynamic characteristics of the rigid-flexible coupling marine mobile platform launching system together to develop dynamic characteristics of the rigid-flexible coupling marine mobile platform launching system to obtain dynamic characteristics research results of the rigid-flexible coupling marine mobile platform launching system;
s11, analyzing the dynamic characteristic research result of the multi-rigid-body marine mobile platform launching system in the step S9 and the dynamic characteristic research result of the rigid-flexible coupling marine mobile platform launching system in the step S10, and completing the marine mobile platform launching safety analysis and evaluation.
Therefore, the marine launching virtual test method based on the layered sampling agent model is used for coupling the launching phase of the marine launching movable platform and the key structural parameters of the launching system, and the layered sampling agent model is used for rapidly completing the virtual test of the large sample to obtain the dynamic characteristic response parameters of the mass launching system, so that the envelope characteristic of the launching dynamic characteristic can be obtained more accurately, the design is guided effectively, the research and development efficiency is improved, the research and development period is shortened, and the research and development cost is reduced; acquiring dynamic response characteristic parameters of an offshore launching system under the conditions of mass launching phase combination and launching system structure parameter combination, acquiring a global solution of severe working conditions under the condition of multi-parameter combination, powerfully supporting the work of a design stage, and fully releasing design risks; the safe boundary of rocket marine movable platform launching is accurately obtained, engineering practice is effectively guided, and launching risks are fully released.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention and not for limiting it, and although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that: the technical scheme of the invention can be modified or replaced by the same, and the modified technical scheme cannot deviate from the spirit and scope of the technical scheme of the invention.
Claims (8)
1. The marine emission virtual test method based on the hierarchical sampling agent model is characterized by comprising the following steps of:
s1, determining structural parameters which need to be coupled with a transmitting phase of a moving platform in the research of the transmitting dynamic characteristics of the marine moving platform, and establishing a dynamic parameterized model of a multi-rigid-body marine moving platform transmitting system;
s2, parameterizing the emission phase moment in the multi-rigid-body marine mobile platform emission system dynamics parameterized model established in the step S1 to complete multi-rigid-body marine mobile platform emission system dynamics model modeling, and obtaining a multi-rigid-body marine mobile platform emission system dynamics model with parameterized structure and phase;
s3, sampling the transmitting phase time parameter in the step S2 and the parameter of the parameterized structure in the step S1 by adopting a sampling method, wherein each sampling result corresponds to a calculation file;
s4, carrying out emission phase large sample calculation of the multi-rigid-body marine moving platform emission system dynamics model under the standard structural parameters based on the structure and phase parameterized multi-rigid-body marine moving platform emission system dynamics model in the step S2 and the sampling result in the step S3 to form an emission phase large sample database;
s5, building three hidden layers by using an artificial neural network algorithm, training the emission phase large sample database in the step S4 to form an emission phase large sample proxy model, and acquiring phase boundaries of emission dynamic characteristic bad boundary working conditions according to the emission phase large sample proxy model, wherein each working condition phase boundary is obtained to obtain a calculation file;
s6, under the severe boundary working condition determined in the step S5, carrying out structural parameter large sample calculation of a multi-rigid-body marine dynamic platform launching system dynamics model on the parameterized structural model in the step S1, sequentially calling a calculation file obtained in the step S5 of calculating the multi-body system dynamics calculation general solver by using an operating system instruction script, obtaining a simulation analysis result, storing a model of each input solver and a corresponding output result in a folder named according to parameters, and forming a structural parameter large sample database by using all result folders and result data;
s7, building three hidden layers by using an artificial neural network algorithm, and obtaining a structural parameter large sample proxy model by taking the emission structural parameter large sample database obtained in the step S6 as input, and obtaining structural parameters corresponding to the bad boundary working conditions of emission dynamics according to the structural parameter large sample proxy model;
s8, forming a dynamic proxy model of the multi-rigid-body marine dynamic platform launching system based on the launching phase large sample proxy model in the step S5 and the structural parameter large sample proxy model in the step S7, and obtaining a global multi-parameter combined severe working condition of the multi-rigid-body marine dynamic platform launching system dynamics;
s9, invoking a sobol algorithm to perform multi-parameter sensitivity analysis based on the dynamic proxy model of the multi-rigid-body marine mobile platform launching system in the step S8 by using python, so as to obtain the influence of each parameter on the dynamic characteristics of the multi-rigid-body marine mobile platform launching system;
s10, establishing a dynamic model of the rigid-flexible coupling marine mobile platform launching system, and taking boundary conditions of a global multi-parameter combination severe working condition and a traditional typical extremum combination working condition of the multi-rigid-body marine mobile platform launching system dynamics in the step S8 as boundary input parameters for researching dynamic characteristics of the rigid-flexible coupling marine mobile platform launching system together to develop dynamic characteristics of the rigid-flexible coupling marine mobile platform launching system to obtain dynamic characteristics research results of the rigid-flexible coupling marine mobile platform launching system;
s11, analyzing the dynamic characteristic research result of the multi-rigid-body marine mobile platform launching system in the step S9 and the dynamic characteristic research result of the rigid-flexible coupling marine mobile platform launching system in the step S10, and completing the marine mobile platform launching safety analysis and evaluation.
2. The method for marine transmission virtual test according to claim 1, wherein in step S1, the structural parameters include: rocket geometry, size and mass characteristics, launch barrel inner radius, adapter size and stiffness characteristics.
3. The method for virtually testing marine launching according to claim 2, wherein in step S1, a dynamic parameterized model of the multi-rigid-body marine moving platform launching system is established, comprising the steps of:
s101, geometric modeling: (1) According to the actual size, modeling is carried out on a movable platform, a supporting cylinder, a transmitting cylinder matrix, a tail cover, an adapter matrix and a low-pressure chamber module in the offshore movable platform transmitting system by using three-dimensional modeling software, and model simplification is carried out according to the actual structure and the relative position of a transmitting device;
wherein the transmitting cylinder matrix is a solid cylinder; the adapter matrix comprises 4 circular rings with 1/4 circular arc structures, 3 circular rings in total, and 12 adapter matrices, wherein the inner diameter of the adapter matrix is smaller than the minimum value of the outer diameter parameterization of the rocket thermal insulation coating, and the outer diameter of the adapter matrix is larger than the maximum value of the inner diameter parameterization of the launching tube;
(2) Geometrically modeling the rocket by using multi-body dynamics simulation software; the rocket body surface is a heat-insulating coating, 3 adapters are arranged on the heat-insulating coating surface, and two groups of parameters of the radius of the rocket heat-insulating coating and the inner radius of the launcher are defined as design variables: defining a design variable alpha, wherein the radius of the rocket thermal insulation coating is represented by alpha, and completing a rocket geometric model by combining the rocket length and the segmentation size of a given length; defining a design variable beta, establishing a solid cylinder with a radius beta at the coaxial position with the transmitting cylinder matrix, performing Boolean subtraction operation on the transmitting cylinder matrix and the solid cylinder with the radius beta through geometry create shape csg command to obtain a hollow cylinder with the inner radius beta, and completing modeling of the transmitting cylinder; performing Boolean subtraction operation on the adapter matrix, the rocket thermal insulation coating and the adapter matrix and the launch canister respectively through geometry create shape csg commands to complete geometric modeling of the adapter;
s102, defining quality characteristics: setting mass, barycenter coordinates and rotational inertia parameters for the rocket and the movable platform respectively; setting density, young modulus and Poisson ratio parameters for the supporting cylinder, the transmitting cylinder, the tail cover and the adapter respectively;
s103, defining a connection relation: the movable platform is fixedly connected with the supporting cylinder, the supporting cylinder is fixedly connected with the launching cylinder, the adapter is fixedly connected with the rocket, the tail cover is in contact relation with the launching cylinder and the rocket, and the adapters are in contact relation with the launching cylinder.
4. The method for marine emission virtual test based on hierarchical sampling proxy model according to claim 3, wherein in step S2, the phase of the emission time is parameterized, specifically: the six-degree-of-freedom motion of the movable platform is simulated by a sine function, and the generalized degree-of-freedom motion at the moment of transmitting is expressed as follows:
(1)
wherein,,/>the values of (1) respectively correspond to rolling, pitching, bowing, swaying, pitching and swaying of the movable platform;、/>、/>respectively representing the corresponding generalized degree-of-freedom motion amplitude, angular frequency and phase; />Is a time term.
5. The method for virtually testing marine emissions based on a hierarchical sampling proxy model according to claim 4, wherein in step S4, the emission phase subsamples of the multi-rigid body marine mobile platform emission system dynamics model are calculated, specifically as: and (3) circularly and sequentially calling the calculation files obtained in the step (S3) of calculating the general solver of the multi-body system dynamics calculation by using an operating system instruction script, obtaining simulation analysis results, storing a model and a corresponding output result of each input general solver of the multi-body system dynamics calculation in a folder named according to parameters, and forming a transmitting phase large sample database by using all the result folders and the result data;
the general solver formula for the dynamics calculation of the multi-body system is as follows:
(2)
wherein,;/>,/>for kinetic energy item->Is a potential energy item; />A displacement term for a directional component.
6. The method for marine emission virtual test as defined in claim 5, wherein in step S5, the emission dynamics are bad boundary conditions, comprising: maximum pitch angle working condition, maximum pitch angle speed working condition, maximum yaw angle speed working condition and maximum collision load working condition of rocket in the process of launching the rocket at the moment of leaving the rocket frame.
7. The marine emission virtual test method based on the hierarchical sampling proxy model according to claim 6, wherein in step S5, three hidden layers are built by using an artificial neural network algorithm, the emission phase large sample database in step S4 is trained to form an emission phase large sample proxy model, and phase boundaries of poor boundary conditions of emission dynamics are obtained according to the emission phase large sample proxy model, and the method specifically comprises the following steps: using the emission phase large sub-sample database obtained in the step S4 as input, wherein 70% is selected as a training set, 20% is selected as a verification set, and 10% is selected as a test set;
the training set database is directly presented to the neural network during training, and the neural network carries out parameter adjustment according to the error of the training set database; the verification set data is used for measuring the generalization capability of the network and stopping training the network when the generalization stops improving; the test set has no influence on the training of the network and is used for evaluating the error of the network after the training; establishing a neural network prediction agent model based on the sample point calculation data, testing the neural network through a test set, and determining an error of the neural network;
by R 2 Measuring the accuracy of the neural network, and calculating the obtained R 2 Comparing the training error with a preset proxy model error, stopping training when the training error is smaller than the preset error, and otherwise, continuing training until the preset error requirement is met; and obtaining a transmitting phase large sub-sample proxy model, and obtaining phase boundaries of the working conditions with bad transmitting dynamic characteristics according to the transmitting phase large sub-sample proxy model, wherein each working condition phase boundary obtains a calculation file.
8. The method for marine launching virtual test based on the hierarchical sampling proxy model according to claim 7, wherein in step S9, the sobol algorithm solves the total variance representing the total variation of the model output first, and the output is decomposed into main effect, interaction effect and error term of each parameter;
the total variance is equal to the variance matrix of the output quantity:
(3)
the main effect index represents the effect of each individual parameter on the output variable:
(4)
the total effect index represents the total influence of the parameter on the output variable after the individual parameter and the interaction effect of the parameter and other parameters are considered, and the calculation method is as follows:
(5)
wherein,S k represent the firstkMain effect index of each parameter;S Tk is the firstkTotal effect index of individual parameters;VarX k is shown inX k Is carried out under the condition of (2)Calculating a difference;E ~Xk representation pairX k Marginalizing, i.e. removingX k Is a desired value of (2);Var(Y|X k ) Is shown inX k Under the condition of (2)YIs a conditional variance of (2);Var(Y) Representing model outputYIs the total variance of (2);X k is the firstkA parameter;Yis output;
error term: the error between the analog output and the true output is used in the sobol algorithm by the total errorTo estimate the main effect of the input variables:
(6)
for analog output, +.>Is the desire for true output.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410149365.4A CN117688849B (en) | 2024-02-02 | 2024-02-02 | Marine emission virtual test method based on hierarchical sampling agent model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410149365.4A CN117688849B (en) | 2024-02-02 | 2024-02-02 | Marine emission virtual test method based on hierarchical sampling agent model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117688849A true CN117688849A (en) | 2024-03-12 |
CN117688849B CN117688849B (en) | 2024-05-07 |
Family
ID=90137497
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410149365.4A Active CN117688849B (en) | 2024-02-02 | 2024-02-02 | Marine emission virtual test method based on hierarchical sampling agent model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117688849B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10562599B1 (en) * | 2018-09-06 | 2020-02-18 | Active Inertia Offshore | Rocket launch platform stabilization system |
CN115640648A (en) * | 2022-10-28 | 2023-01-24 | 北京理工大学 | Carrier rocket marine launching system dynamics modeling method |
CN115833912A (en) * | 2022-11-17 | 2023-03-21 | 中国船舶集团有限公司第七〇九研究所 | Marine satellite emission command control system and method |
-
2024
- 2024-02-02 CN CN202410149365.4A patent/CN117688849B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10562599B1 (en) * | 2018-09-06 | 2020-02-18 | Active Inertia Offshore | Rocket launch platform stabilization system |
CN115640648A (en) * | 2022-10-28 | 2023-01-24 | 北京理工大学 | Carrier rocket marine launching system dynamics modeling method |
CN115833912A (en) * | 2022-11-17 | 2023-03-21 | 中国船舶集团有限公司第七〇九研究所 | Marine satellite emission command control system and method |
Non-Patent Citations (1)
Title |
---|
张玉涛;薛飞;李鹏;: "火箭助飞鱼雷海上落点测控系统测量误差分析", 鱼雷技术, no. 05, 15 October 2012 (2012-10-15) * |
Also Published As
Publication number | Publication date |
---|---|
CN117688849B (en) | 2024-05-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109543243B (en) | ADAMS-based carrier-borne aircraft arresting ship dynamics simulation analysis method | |
CN102508692B (en) | Simulation and verification method of control method of near space aircraft | |
CN116029219B (en) | Aerodynamic heat prediction method, device, equipment and storage medium for aircraft | |
CN113449474B (en) | Improved gray wolf algorithm optimized BP neural network pipe forming quality prediction method | |
CN113654756B (en) | Active real-time mixed model test method for offshore floating type fan | |
CN112597702A (en) | Pneumatic modeling generation type confrontation network model training method based on radial basis function | |
CN117688849B (en) | Marine emission virtual test method based on hierarchical sampling agent model | |
CN111027250A (en) | Special-shaped curved surface reinforced shell modeling method based on grid deformation technology | |
CN110083946B (en) | Multi-state model correction method based on unconstrained optimization model | |
CN116933386A (en) | Aircraft pneumatic data fusion method based on MCOK proxy model | |
CN110210072B (en) | Method for solving high-dimensional optimization problem based on approximate model and differential evolution algorithm | |
CN114474004B (en) | Error compensation planning control strategy for multi-factor coupling vehicle-mounted building robot | |
CN114638048A (en) | Three-dimensional spray pipe flow field rapid prediction and sensitivity parameter analysis method and device | |
CN110096798A (en) | A kind of method of multimode FEM updating | |
CN115470675A (en) | Prediction method and system for crack propagation path of heavy-load pressure vessel | |
CN108509684B (en) | Steering engine and dynamic load simulator matching design method | |
CN114970305A (en) | Prediction method for magnetic particle distribution of magnetic control software robot | |
CN107292015B (en) | Neural network algorithm-based underwater vehicle equilibrium submerged model evaluation method | |
CN114254416A (en) | Soil stress-strain relation determination method based on long-term and short-term memory deep learning | |
CN111199779A (en) | Virtual drug screening method and device based on molecular docking | |
Kim et al. | Material property-estimate technique based on natural frequency for updating finite element model of orthotropic beams | |
CN111967198B (en) | Novel high-bearing-capacity lightweight design method for composite material structure | |
CN116052814B (en) | J integral parameter acquisition method for viscoelastic material crack growth simulation | |
CN115563844B (en) | Model correction method based on shape correlation coefficient of frequency response function | |
CN117313441B (en) | Amphibious vehicle emission proxy model prediction method based on random sea wave spectrum |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |