Airbag type rocket sublevel package recovery method and device and storage medium

Through variational inference and domain adaptation technology, the problem of insufficient adaptability of rocket sub-level recycling methods in different environments and tasks is solved, an efficient and accurate recycling process is achieved, and the robustness and efficiency of the system are improved.

CN119048071AInactive Publication Date: 2024-11-29ZHONGBEI UNIV
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411058476.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-29
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing rocket sub-level recycling methods are difficult to adapt to changes in different environments and tasks, resulting in low recycling accuracy, poor robustness, high cost, and difficult to effectively coordinate and optimize multiple tasks.

Method used

Variable inference and domain adaptation technology are adopted to adjust model parameters in real time, dynamically adapt to environmental changes, realize multi-task optimization and knowledge sharing, and improve the robustness and efficiency of the recycling system.

Benefits of technology

Improves the accuracy and success rate of rocket sub-level recycling, ensures the best time for airbag deployment and position correction, optimizes overall system efficiency and performance, and adapts to different environments and task requirements.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119048071A_ABST
    Figure CN119048071A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of rocket launching, and discloses an air bag type rocket sublevel package recycling method and device and a storage medium. The method comprises the following steps: carrying out probability modeling on a rocket sub-level trajectory by using variational inference to obtain a variational inference model; optimizing the variational inference model through domain self-adaption; in a new environment, a landing point of a rocket sub-level is predicted through a variational inference model, the uncertainty of prediction is evaluated, and then data features in the new environment are adjusted through domain adaptation, so that the data features are aligned with feature distribution of an original environment; optimizing the variational inference model in combination with domain adaptation by adopting a multi-task optimization mode and an environment transfer learning mode; and according to the optimized variational inference model, performing decision making on each task in the rocket sub-level package recovery process, and performing package recovery according to a decision making result. According to the method, the recovery efficiency and performance are improved, and a plurality of tasks can be effectively coordinated and optimized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of rocket launching technology, and specifically relates to a method, equipment and storage medium for recovering airbag-type rocket sub-stage packages. Background Art

[0002] With the continuous development of aerospace technology, rocket recovery has become one of the key technologies to increase the reuse rate of spacecraft and reduce costs. In the launch cost of modern rockets, the fuel cost itself accounts for no more than 0.5% of the total cost, so rocket sub-stage recovery technology can effectively change the cost structure of rocket launches and provide support for large-scale satellite networking missions.

[0003] Existing rocket sub-stage recovery methods mainly include hard landing, cushioned landing and parachute landing. During recovery, the trajectory of the rocket sub-stage needs to be calculated and planned, but the algorithms in the existing technology are not adaptable enough and cannot adapt to environmental changes. For example, in the recovery mission of rocket sub-stages in different environments (such as deserts, oceans, mountains, etc.), the environmental conditions will be significantly different, and the existing algorithms are difficult to adapt to all environmental changes. In addition, it is also difficult to adapt to changes in missions.

[0004] For example, different tasks may involve different recovery requirements and strategies, such as different airbag deployment timing and trajectory prediction requirements, and existing algorithms are difficult to respond flexibly. In addition, existing algorithms also have the problem of poor robustness. On the one hand, due to the limited ability of the algorithm to handle the noise and uncertainty in the sensor data, it is easy to cause inaccurate predictions and wrong decisions. On the other hand, the algorithm model is poorly stable and may not be able to maintain stable performance when facing extreme conditions or abnormal data, resulting in recovery failure. Therefore, existing recovery methods may have complex structures, high costs, or low landing accuracy and low recovery efficiency. Summary of the invention

[0005] In order to solve the technical problems of high rocket recovery cost and low recovery efficiency in the prior art, the present invention proposes a new type of airbag-type rocket sub-stage package recovery method, equipment and storage medium, so as to optimize the recovery process by combining lightweight artificial intelligence technology.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for recovering an airbag-type rocket sub-stage package, comprising the following steps:

[0007] Step 1: Use variational inference to perform probabilistic modeling on the rocket sub-stage trajectory and obtain a variational inference model;

[0008] Step 2: During the recovery process, based on the real-time collection of environmental data and rocket sub-stage status data, the parameters of the variational model are dynamically adjusted through domain adaptation technology, so that the variational inference model can adapt to the current environment in real time; after the recovery mission is completed, the variational inference model is retrained based on the collected data, and the parameters of the variational inference model are adjusted through domain adaptation;

[0009] Step 3: In the new environment, the landing point of the rocket substage is predicted by the variational inference model and the uncertainty of the prediction is evaluated. Then, the data features in the new environment are adjusted through domain adaptation to align them with the feature distribution of the original environment.

[0010] Step 4: Use multi-task optimization and environment transfer learning methods, combined with domain adaptation to optimize the variational inference model;

[0011] Step 5: Based on the optimized variational inference model, make decisions on each task in the rocket sub-stage package recovery process, and recover the package based on the decision results.

[0012] The specific method of step 4 is:

[0013] In different task stages of the variational inference model, feature sharing and parameter migration are performed through domain adaptation to achieve multi-task optimization;

[0014] In different environments, the parameters of the variational inference model are transferred and adjusted through domain adaptation.

[0015] In step 2, during the recycling process, environmental data is collected in real time, and the collected environmental data includes wind speed and ambient temperature.

[0016] The task stages of the variational inference model include a trajectory prediction stage and an airbag deployment timing decision stage.

[0017] The variational inference model includes the following computational steps:

[0018] Calculate the volume stress of the airbag;

[0019] Calculate the strength factor of the airbag;

[0020] Adjusting domain adaptation parameters based on collected environmental data;

[0021] Initialize probability distribution parameters;

[0022] The rotation speed θ is calculated based on the position P of the recovery sub-stage.

[0023] The calculation formula of the volume stress and intensity factor of the airbag is:

[0024] σ=S / V;

[0025] F = k*σ;

[0026] Among them, V represents the volume of the airbag, S represents the material strength, and k represents the aerodynamic constant.

[0027] The calculation formula of the domain adaptation parameter is:

[0028] γ=γ0+a*W+b*X;

[0029] Among them, γ0 represents the reference value, W represents the ambient temperature, X represents the wind speed, and a and b represent the adjustment coefficients.

[0030] The formula for calculating the rotation speed is:

[0031] θ=VP / r;

[0032] Wherein, VP represents the rotation speed and r represents the distance of the recycling sub-stage relative to the rotation axis.

[0033] In addition, the present invention also provides a terminal device, comprising: a processor and a memory communicatively connected to the processor;

[0034] The memory stores computer-executable instructions;

[0035] The processor executes the computer-executable instructions stored in the memory to implement the airbag-type rocket sub-stage packaging and recovery method.

[0036] In addition, the present invention also provides a computer-readable storage medium, in which computer execution instructions are stored. When the computer execution instructions are executed by a processor, they are used to implement the described method of recovering an airbag-type rocket sub-stage package.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. The present invention models the trajectory and environmental uncertainty of rocket substages through variational inference, and dynamically adjusts the model parameters in combination with domain adaptation technology, so that the recovery system can maintain high performance under different environmental conditions. This solves the problem that the recovery method in the prior art is difficult to adapt to all environments, resulting in reduced recovery accuracy and success rate;

[0039] 2. The present invention combines variational inference and domain adaptation technology to collect and process sensor data in real time. By dynamically updating the prediction model, the real-time and accuracy of the recovery decision are improved, the optimal time for airbag deployment and position correction is ensured, and the efficiency of recovery is improved.

[0040] 3. In addition, the rocket sub-stage recovery process involves multiple tasks (such as trajectory prediction, airbag deployment timing decision, etc.). The present invention uses variational inference technology to model each task, and realizes knowledge sharing and collaborative optimization between tasks through domain adaptation technology, thereby improving the efficiency and performance of the overall system, and solving the problem that the existing technology is difficult to effectively coordinate and optimize multiple tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic flow chart of an airbag-type rocket sub-stage packaging recovery method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0043] An embodiment of the present invention provides a method for recovering a balloon-type rocket sub-stage package, which is based on variational inference and domain adaptation to achieve trajectory prediction of balloon-type rocket sub-stages and prediction of airbag deployment timing. Among them, variational inference is an approximate inference method, which is solved by converting the calculation problem of the posterior distribution into an optimization problem, and is mainly used to process complex probability models. The deployment timing of the airbag is crucial to the success of the recovery. Through variational inference, the system can calculate the optimal airbag deployment timing based on the current trajectory prediction and landing point distribution, ensuring that the airbag can be effectively deployed when the sub-stage approaches the ground to provide buffer protection. And domain adaptation (Domain Adaptation, DA) is an algorithm that enables the model to perform well in different but related fields by learning shared representations. In different launch and recovery missions, rocket sub-stages may face different environmental conditions, such as wind speed, temperature, terrain, etc. The use of domain adaptation technology can help the recovery system adapt the model learned from one environment to a new environment during the training process, so that the system can still maintain efficient recovery performance under different environmental conditions. Specifically, such as Figure 1 As shown, an embodiment of the present invention provides an airbag rocket sub-stage package recovery method, comprising the following steps:

[0044] Step 1: Joint modeling and inference: Use variational inference to probabilistically model the rocket sub-stage trajectory and obtain a variational inference model.

[0045] Variational inference can be used to model the probability of rocket sub-stage trajectories, environmental factors, and uncertainties, while domain adaptation techniques can be used to adjust the performance of these models in different environments. The initial model is built through variational inference, and domain adaptation techniques are used to fine-tune the model in different tasks or environments to better adapt it to specific mission requirements.

[0046] During the initial modeling, variational inference is used to probabilistically model the trajectory of the rocket substage to capture the potential uncertainties under different environmental conditions. When performing tasks in different environments, such as desert environments and ocean environments, domain adaptation techniques are used to adjust the parameters of the variational inference model to adapt it to the data distribution and environmental conditions in specific scenarios.

[0047] Step 2: Domain-adaptive variational inference: During the recovery process, the parameters of the variational model are dynamically adjusted through domain adaptation technology based on the real-time collection of environmental data and rocket sub-stage status data, so that the variational inference model can adapt to the current environment in real time; after the recovery mission is completed, the variational inference model is retrained based on the collected data, and the parameters of the variational inference model are adjusted through domain adaptation.

[0048] Specifically, through domain adaptation technology, variational inference can dynamically adjust model parameters to adapt to different environments and tasks. In the variational inference process, domain adaptation technology is used to adaptively adjust each new data to improve the robustness and accuracy of the model. Therefore, this step 2 specifically includes two steps:

[0049] a. Online adjustment: During the recovery process, environmental data (such as wind speed and temperature) and rocket sub-stage status data are collected in real time, and the variational inference model is dynamically adjusted through domain adaptation technology to ensure that the model can adapt to the current environment in real time.

[0050] b. Batch training: After each recycling task, the collected data is used to retrain and domain adapt the variational inference model so that the model performs better in the next task.

[0051] Step 3: Uncertainty evaluation and domain alignment: In the new environment, the landing point of the rocket substage is predicted through the variational inference model and the uncertainty of the prediction is evaluated. Then, the data features in the new environment are adjusted through domain adaptation to align them with the feature distribution of the original environment.

[0052] The uncertainty assessment provided by variational inference can help identify the degree of adaptability of the model in the new environment. By aligning the data features in different environments through domain adaptation technology, the model can be better generalized to different environments. Therefore, in the new environment application, the variational inference model is used to predict the landing point of the rocket sub-stage and evaluate the uncertainty of the prediction. Then, through domain adaptation technology, the data features in the new environment are adjusted to align with the feature distribution in the original environment, which can improve the prediction accuracy of the variational inference model in the new environment.

[0053] Step 4: Collaborative optimization and transfer learning: Use multi-task optimization and environment transfer learning methods, combined with domain adaptation to optimize the variational inference model.

[0054] The specific method of step 4 is:

[0055] Multi-task optimization: In different task stages of the variational inference model, such as trajectory prediction and airbag deployment timing decision stage, feature sharing and parameter migration are performed through domain adaptation to achieve multi-task optimization and improve the performance of each task;

[0056] Environmental transfer learning: In different environments, the parameters of the variational inference model are transferred and adjusted through domain adaptation, and the modeling advantages of variational inference are combined to improve the adaptability and prediction accuracy of the model in the new environment.

[0057] Step 5: Predict and recover through the model: Based on the optimized variational inference model, make decisions on each task in the rocket sub-stage package recovery process, and recover the package based on the decision results.

[0058] Specifically, in this embodiment, the variational inference model includes the following input parameters:

[0059] 1. Airbag deployment time: T;

[0060] 2. Airbag volume: V;

[0061] 3. Airbag material strength: S;

[0062] 4. Recycling environment temperature: W;

[0063] 5. Recycling environment wind speed: X;

[0064] 6. Aerodynamic constant: k;

[0065] 7. Recycle the sub-level center of mass position: P;

[0066] 8. Domain adaptation parameter: γ;

[0067] 9. Probability distribution parameter: ɑ;

[0068] 10. Recovery sub-stage rotation speed: θ;

[0069] In addition, the variational inference model includes the following computational steps:

[0070] S1 calculates the volume stress of the airbag, and the calculation formula is:

[0071] σ=S / V; (1)

[0072] Among them, V represents the volume of the airbag, and S represents the material strength;

[0073] S3 calculates the strength factor of the airbag, and the calculation formula is:

[0074] F=k*σ; (2)

[0075] Where k represents the aerodynamic constant;

[0076] S3 adjusts the domain adaptation parameter according to the collected environmental data. The calculation formula of the domain adaptation parameter is:

[0077] γ=γ0+a*W+b*X; (3)

[0078] Among them, γ0 represents the reference value, W represents the ambient temperature, X represents the wind speed, and a and b represent the adjustment coefficients.

[0079] S4 initializes the probability distribution parameters; the probability distribution parameters ɑ are initialized according to a specific statistical method. The present invention uses maximum likelihood estimation (Maximum Likelihood Estimation, MLE). For maximum likelihood estimation, the present invention uses sample statistics in an existing data set to estimate parameters.

[0080] S5 calculates the rotation speed θ according to the position P of the recycling sub-stage. The position P of the recycling sub-stage changes with time, and the rotation speed θ is calculated according to the change of its position. Considering P as a vector (x, y) on a two-dimensional plane, the rotation speed θ of the sub-stage can be expressed as θ = dθ / dt, that is, the rate of change of the angle over time. The rotation speed θ can be estimated by calculating the rate of change of P. Assuming that the rate of change of position P is VP, the rotation speed θ can be expressed as:

[0081] θ=VP / r;(4)

[0082] where r is the distance of the recycling substage relative to the axis of rotation.

[0083] In an exemplary embodiment, the present invention further provides a terminal device. The terminal device includes: a processor and a memory in communication with the processor, the memory storing computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the above-mentioned method for recovering a balloon-type rocket sub-stage package.

[0084] In an exemplary embodiment, a computer-readable storage medium is also provided, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement a method for recovering an airbag-type rocket sub-stage package as described above.

[0085] The feasibility of the present invention has been demonstrated through a series of experiments, simulations and practical applications.

[0086] 1. Experimental verification: Field tests of the recovery of balloon-type rocket sub-stages were carried out through a large number of tests and experiments in laboratory environments, including experiments on the ground or in simulated environments, as well as experiments in real environments.

[0087] 2. Simulation verification: A simulation study was conducted using numerical simulation technology to verify the effectiveness and feasibility of the invention.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for recovering sub-stage packages of airbag rockets, It is characterized in that The following steps are involved: Step 1: Use variational inference to perform probabilistic modeling on the rocket sub-stage trajectory and obtain a variational inference model; Step 2: During the recovery process, based on the real-time collection of environmental data and rocket sub-stage status data, the parameters of the variational model are dynamically adjusted through domain adaptation technology, so that the variational inference model can adapt to the current environment in real time; after the recovery mission is completed, the variational inference model is retrained based on the collected data, and the parameters of the variational inference model are adjusted through domain adaptation; Step 3: In the new environment, the landing point of the rocket substage is predicted by the variational inference model and the uncertainty of the prediction is evaluated. Then, the data features in the new environment are adjusted through domain adaptation to align them with the feature distribution of the original environment. Step 4: Use multi-task optimization and environment transfer learning methods, combined with domain adaptation to optimize the variational inference model; Step 5: Based on the optimized variational inference model, make decisions on each task in the rocket sub-stage package recovery process, and recover the package based on the decision results.

2. A method for recovering an airbag rocket substage package according to claim 1, It is characterized in that The specific method of step 4 is: In different task stages of the variational inference model, feature sharing and parameter migration are performed through domain adaptation to achieve multi-task optimization; In different environments, the parameters of the variational inference model are transferred and adjusted through domain adaptation.

3. A method for recovering an airbag rocket substage package according to claim 1, It is characterized in that In step 2, during the recycling process, environmental data is collected in real time, and the collected environmental data includes wind speed and ambient temperature.

4. A method for recovering an airbag rocket substage package according to claim 1, It is characterized in that The task stages of the variational inference model include a trajectory prediction stage and an airbag deployment timing decision stage.

5. The method for recovering an airbag rocket substage package according to claim 1, It is characterized in that The variational inference model includes the following computational steps: Calculate the volume stress of the airbag; Calculate the strength factor of the airbag; Adjusting domain adaptation parameters based on collected environmental data; Initialize probability distribution parameters; The rotation speed θ is calculated based on the position P of the recovery sub-stage.

6. A method for recovering an airbag rocket substage package according to claim 5, It is characterized in that The calculation formula of the volume stress and intensity factor of the airbag is: σ=S / V; F = k*σ; Among them, V represents the volume of the airbag, S represents the material strength, and k represents the aerodynamic constant.

7. The method for recovering an airbag rocket substage package according to claim 5, It is characterized in that The calculation formula of the domain adaptation parameter is: γ=γ0+a*W+b*X; Among them, γ0 represents the reference value, W represents the ambient temperature, X represents the wind speed, and a and b represent the adjustment coefficients.

8. The method for recovering a sub-stage of an airbag rocket according to claim 5, It is characterized in that The formula for calculating the rotation speed is: θ=VP / r; Wherein, VP represents the rotation speed and r represents the distance of the recycling sub-stage relative to the rotation axis.

9. A terminal device, It is characterized in that include: A processor and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement an airbag-type rocket sub-stage wrapping and recovery method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement a method for wrapping and recovering an airbag-type rocket substage as described in any one of claims 1 to 8.

Citation Information

Cited By

  • Liquid rocket engine mass center dynamic measurement system based on ultrasonic waves

    CN119756688A