Construction method of solid engine charge storage performance prediction system

Abstract

The application discloses a construction method of a solid engine charge storage performance prediction system. The method firstly collects and processes the performance of the solid engine in the storage process to form a related performance change database, and then establishes a related virtual model. The related parameters obtained by processing the solid engine are brought into the model to obtain a virtual model which is the same as the actual solid engine state change, which is equivalent to a digital twin model of the solid engine. The twin model can accurately predict the performance change of the subsequent solid engine physical model in the storage process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of solid engine charge design, and particularly relates to a construction method of a solid engine charge storage performance prediction system. BACKGROUND

[0002] As an important engine, a solid engine charge is filled with solid propellant, but meanwhile, the engine has characteristic loss during storage. Generally, to study the change of storage performance, the engine needs to be tested by ignition, which is a great loss to the research cost. Therefore, a method is needed to predict the change of the performance of the solid engine charge through certain parameters, so as to reduce the development cost of the solid engine and improve the reliability. If the change of the storage performance of the solid engine can be predicted, the same type of solid engine under the same condition can be better stored, and the performance of the engine can be comprehensively and specifically understood without ignition, so that the related research can be better carried out. SUMMARY

[0003] The present application aims at solving the above-mentioned problems of the prior art, and provides a construction method of a solid engine charge storage performance prediction system. The method first collects and processes the performance of the solid engine during storage to form a related performance change database, and then establishes a related virtual model. The related parameters of the solid engine obtained by processing are brought into the model to obtain a virtual model with the same state change as the actual solid engine, which is equivalent to a digital twin model of the solid engine. The twin model can make a good prediction of the performance change of the subsequent solid engine physical model during storage.

[0004] To solve the above-mentioned technical problems, the technical scheme adopted by the present application is as follows:

[0005] The construction method of the solid engine charge storage performance prediction system comprises the following steps:

[0006] Step 1, a data acquisition and conversion system is built, and then the engine state data is acquired, which specifically comprises:

[0007] Step 1A, a plurality of sensors are arranged on the engine entity to realize the acquisition and transmission of the state data of the entity engine;

[0008] Step 1B, a data conversion device is arranged to convert the monitoring data obtained by different sensors into standard current or voltage signals according to a unified rule; wherein the input end of the data conversion device is connected to the output end of the sensor whose data needs to be converted;

[0009] Step 2, a real-time database is constructed, which specifically comprises:

[0010] Step 2A, constructing a state awareness module; at the same time, obtaining solid engine historical data;

[0011] Step 2B, processing engine state data based on the state awareness module, and obtaining monitoring data after processing;

[0012] Step 3, constructing a solid engine digital twin model, specifically including:

[0013] Step 3A, physical modeling: constructing a solid engine geometric model according to the physical size and assembly relationship of the solid engine;

[0014] Step 3B, modeling of grain rheological model, modeling of grain performance degradation model, modeling of boundary viscosity model, and modeling of crack physical model;

[0015] Step 3C, assigning the monitoring data to the solid engine geometric model, the grain rheological model, the grain performance degradation model, the boundary viscosity model, and the crack physical model, respectively, to obtain a solid engine digital twin model;

[0016] Step 3D, based on the real-time database assigned to the solid engine digital twin model, simulating the solid engine digital twin model, and outputting simulation results;

[0017] Step 4, training and prediction of the solid engine digital twin model, specifically including:

[0018] Step 4A, training the solid engine digital twin model based on the result data in the historical data, and then generating a data migration model through transfer learning;

[0019] Step 4B, training the solid engine digital twin model based on the simulation data in Step 3D, and inputting the data output by the solid engine digital twin model into the data migration model generated in Step 4A through transfer learning, to correct the data migration model;

[0020] Step 4C, inputting the monitoring data obtained in Step 2B into the data migration model generated in Step 4B to output a prediction scheme;

[0021] The prediction scheme includes multiple charge storage performance prediction levels and the weights of each influencing component in the monitoring data in different charge storage performance prediction levels.

[0022] Preferably, Step 1A includes temperature sensors, strain sensors, flexible sensors, humidity sensors, and vibration sensors.

[0023] Preferably, the processing of the data by the state perception module comprises data acquisition, storage, preprocessing, working condition acquisition, feature quantity extraction, feature quantity selection and feature quantity fusion in sequence.

[0024] Preferably, in step 4, the influencing components include temperature, humidity, vibration and aging degree.

[0025] Preferably, in step 3C, an updating method is further included to update the monitoring data for simulation input in real time.

[0026] Preferably, in step 3C, a consistency checking method is further included to check whether the monitoring data for simulation input and the actual simulation input data are consistent.

[0027] The present application has the following beneficial effects:

[0028] The solid engine digital twin model collects, transmits, integrates data and performs simulation calculation on full-stage load conditions, correlates and analyzes multi-source heterogeneous data such as online monitoring data (real-time database) and offline data (historical database) (training process of the solid engine digital twin model), mines information and knowledge hidden behind the data, and realizes the purpose of individualized performance evaluation of each engine. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 It is a construction method principle diagram of a solid engine charge storage performance prediction system.

[0030] Figure 2 It is a solid engine digital twin model construction diagram.

[0031] Figure 3 It is a real-time database generation diagram.

[0032] Figure 4 It is a prediction scheme generation diagram.

[0033] Figure 5 It is a prediction scheme architecture diagram. DETAILED DESCRIPTION

[0034] The present application will be further described in detail below in combination with the drawings and specific preferred embodiments.

[0035] In the description of the present application, it should be understood that the terms "left side", "right side", "upper part", "lower part" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and "first", "second" and the like do not represent the importance of the parts, and therefore cannot be understood as a limitation on the present application. The specific dimensions used in the embodiments are only for the purpose of illustrating the technical solutions and do not limit the protection scope of the present application.

[0036] Based on the solid rocket engine storage influencing factors, digital twin function analysis, as Figures 1-5 The construction method of a solid rocket engine charge storage performance prediction system is shown, and the solid rocket engine charge storage performance prediction system structure based on digital twin is shown. According to different levels of data information state, the system structure construction process is divided into: solid rocket engine entity, data acquisition and conversion system, solid rocket engine database, solid rocket engine digital model, data and model fusion driven performance prediction method (updating mechanism method).

[0037] Step 1, build a data acquisition and conversion system, and then acquire engine state data, which specifically includes:

[0038] Step 1A, multiple sensors are arranged on the engine entity to realize the acquisition and transmission of the state data of the entity engine.

[0039] Among them, the data acquisition and conversion system includes temperature sensor, strain sensor, flexible sensor, humidity sensor and vibration sensor. Among them, the strain sensor and the flexible sensor are used to acquire shape strain.

[0040] Step 1B, set up a data conversion device to convert the monitoring data obtained by different sensors into standard current or voltage signals according to unified rules, and pre-process the collected signals to complete the noise reduction, gross error elimination and other preprocessing work of the collected information, laying a foundation for the following data fusion. Among them, the output data of the temperature sensor and the humidity sensor need to be converted.

[0041] Among them, the input end of the data conversion device is connected to the output end of the sensor whose data needs to be converted.

[0042] Step 2, build a real-time database, which specifically includes:

[0043] Step 2A, build a state perception module; at the same time, acquire historical data of the solid rocket engine;

[0044] Step 2B, process the engine state data based on the state perception module, and obtain monitoring data after processing.

[0045] The historical database is an integration of the once real-time database, which is already processed data and does not need to be put into the state-aware module for processing again.

[0046] The processing of the data by the state-aware module includes data acquisition, storage, preprocessing, working condition acquisition, feature quantity extraction, feature quantity selection and feature quantity fusion in sequence. The state-aware module finally realizes the processing of the real-time database.

[0047] Step 3, constructing a solid rocket engine digital twin model, specifically including:

[0048] Step 3A, physical modeling: constructing a solid rocket engine geometric model according to the physical size and assembly relationship of the solid rocket engine;

[0049] Step 3B, modeling of grain rheological model, modeling of grain performance degradation model, modeling of boundary viscous model and modeling of crack physical model;

[0050] Step 3C, assigning the monitoring data to the solid rocket engine geometric model, the grain rheological model, the grain performance degradation model, the boundary viscous model and the crack physical model respectively, thereby obtaining the solid rocket engine digital twin model.

[0051] It also includes an updating method and a consistency checking method. The updating method updates the monitoring data used for simulation input in real time.

[0052] The consistency checking method is used to check whether the monitoring data used for simulation input and the actual simulation input data are consistent.

[0053] Step 3D, based on the monitoring data assigned to the solid rocket engine digital twin model, simulating the solid rocket engine digital twin model and outputting simulation result data;

[0054] Step 4, training and prediction of the solid rocket engine digital twin model.

[0055] For the material properties and change rules of the solid rocket engine charge storage stage, a storage performance prediction method based on transfer learning is adopted. Based on the historical data analysis of the performance degradation rules of the solid rocket engine charge, a simulation data driven model is trained, and then the trained model is migrated to the actual application environment to receive real-time state data monitored by the sensor. The trained model can quickly and accurately output the prediction results without going through the process of data algorithm reconstruction and model training, which is a time-consuming and complex process.

[0056] Specifically including:

[0057] Step 4A, training the solid rocket engine digital twin model based on the result data in the historical data, and then generating a data migration model through transfer learning.

[0058] The historical data includes state data and result data.

[0059] The state data includes temperature, humidity, vibration, and aging degree.

[0060] The result data includes defect feature data and performance degradation data.

[0061] Step 4B, based on the simulation data in step 3D, the solid engine digital twin model is trained, and the data output by the solid engine digital twin model is input into the data migration model generated in step 4A to correct the data migration model.

[0062] Step 4C, input the monitoring data obtained in step 2B into the data migration model generated in step 4B to output the prediction scheme.

[0063] The prediction scheme includes multiple charge storage performance prediction levels and the weights of each influencing component in the monitoring data in different charge storage performance prediction levels.

[0064] If the performance detection of the solid engine is completed, the monitoring is distinguished from the time point: the propellant casting process, the transportation process, the storage process, and the use process. At the same time, according to the different physical quantities, the weights of temperature, stress, vibration, and other load quantities are also distinguished in different processes.

[0065] These factors can cause interface debonding, propellant deformation, and engine state changes such as mechanical properties. These changes affect the availability of the engine's propellant performance, and in severe cases, they can cause the engine to fail. In order to increase the maintenance cost of the solid engine, improve the service life of the engine, and increase the reliability of use, monitoring and performance evaluation of the solid engine are particularly critical.

[0066] Developing solid engine charge performance analysis refers to analyzing the performance changes of the engine from production to ignition within this time period. The task profile and load experienced in different life cycle processes are different, and the response characteristics exhibited by the engine are different, and the focus of attention is also different.

[0067] The propellant deforms under its own gravity, and the deformation further causes the deformation of the shell. Therefore, when detecting the storage environment temperature and humidity of the engine, the deformation of the engine shell and the propellant, and the deformation of the liner should be monitored.

[0068] In the transportation process, in addition to the deformation of the shell and the propellant, the external vibration transmitted to the shell through the support structure and the external stress should also be focused on.

[0069] In use, the deformation of the engine housing and the propellant grain in various directions, the force state of the support point need to be focused on, the interlayer strain caused by the inclination will cause damage to the adiabatic lining, and the interlayer stress is the focus of monitoring.

[0070] As Figure 5 indicated, the prediction scheme includes various propellant storage performance prediction levels, and the weight of each influencing component in the real-time database in different propellant storage performance prediction levels.

[0071] According to the application actuality in different working conditions, the data-driven model is effectively fused with the state monitoring data by using the transfer learning algorithm, so as to improve the accuracy and reliability of the solid engine propellant storage performance prediction.

[0072] From the foregoing analysis, there are many factors causing the degradation of the storage performance of the solid engine propellant, and each factor is interlaced with each other, so the propellant storage performance prediction scheme needs to be comprehensively considered from the system level. Based on the analytic hierarchy process, an analysis method is established, and the basic idea is to decompose all elements contained in the decision problem, and respectively assign them to target, criterion, scheme and other levels. Based on the divided levels, corresponding qualitative analysis and quantitative analysis are performed on the decision problem to form the evaluation decision. The operation process of the solid engine propellant storage performance prediction scheme is as shown in Figure 1 .

[0073] Verification:

[0074] Based on the digital twinning, the solid engine propellant storage performance prediction scheme is used for state monitoring of some factors of a solid engine, and state analysis is performed.

[0075] Four three-axis acceleration sensors are installed for vibration test monitoring of the engine, which are respectively installed on the left side of the front support of the transportation device, the right side of the front support of the transportation device, the left side of the rear support of the transportation device and the right side of the rear support of the transportation device. The X axis is the forward direction of the vehicle, the Y axis is the vertical upward direction, and the Z axis is the lateral direction. The monitoring test lasts for 5 hours.

[0076] The change curve of the vehicle speed with time and the history curve of the three-axis acceleration of the four vibration measuring points during transportation are obtained. Through calculation and analysis, it can be known that the energy distribution of the x-axis direction and the z-axis direction signals is much smaller than that of the Y axis. Based on the above results, the influence of transportation vibration on the mechanical properties of the solid engine propellant is mainly in the Y axis direction (vertical direction). Based on this, the engine performance is evaluated.

[0077] The preferred embodiments of the application are described in detail above, but the application is not limited to the specific details in the above embodiments, and various equivalent transformations can be made to the technical solutions of the application within the technical concept range of the application, and these equivalent transformations all belong to the protection range of the application.

Claims

1. A method for constructing a solid propellant storage performance prediction system for a solid rocket motor, characterized in that, Includes the following steps: Step 1: Build a data acquisition and conversion system to obtain engine status data, specifically including: Step 1A: Install various sensors on the engine to collect and transmit the engine's status data. Step 1B: Set up a data conversion device to convert the monitoring data acquired by different sensors into standard current or voltage signals according to a unified rule; wherein, the input end of the data conversion device is connected to the output end of the sensor whose data needs to be converted; Step 2: Build a real-time database, which includes: Step 2A: Construct a state-aware module; Step 2B: Process the engine status data based on the status perception module, obtain the monitoring data after processing, and build a real-time database; Step 3: Construct a digital twin model of the solid rocket motor, specifically including: Step 3A, Physical Modeling: Construct a geometric model of the solid rocket motor based on its solid engine dimensions and assembly relationships; Step 3B: Modeling the rheological model of the propellant grain, modeling the performance degradation model of the propellant grain, modeling the boundary viscosity model, and modeling the crack physical model; Step 3C: Assign the monitoring data to the solid rocket motor geometric model, propellant rheological model, propellant performance degradation model, boundary viscosity model, and crack physical model respectively, thereby obtaining the solid rocket motor digital twin model; Step 3D: Based on the monitoring data already assigned to the solid rocket motor digital twin model, simulate the solid rocket motor digital twin model and output simulation data; Step 4, training and prediction of the solid rocket motor digital twin model, specifically includes: Step 4A: Train the digital twin model of the solid rocket motor based on the result data in the historical data, and then generate a data transfer model through transfer learning; Step 4B: Train the solid rocket motor digital twin model based on the simulation data in Step 3D. The data output by the solid rocket motor digital twin model is transferred and input into the data transfer model generated in Step 4A to correct the data transfer model. Step 4C: Input the monitoring data obtained in Step 2B into the data migration model generated in Step 4B to output the predictive scheme; The prediction scheme includes multiple levels of prediction for the storage performance of the drug charge and the weights of each influencing component in the monitoring data at different levels of prediction for the storage performance of the drug charge.

2. The method for constructing a solid propellant storage performance prediction system according to claim 1, characterized in that, Step 1A includes a temperature sensor, a strain sensor, a flexible sensor, a humidity sensor, and a vibration sensor.

3. The method for constructing a solid propellant storage performance prediction system according to claim 1, characterized in that, The state perception module processes data sequentially as follows: data acquisition, storage, preprocessing, operating condition acquisition, feature extraction, feature selection, and feature fusion.

4. The method for constructing a solid propellant storage performance prediction system according to claim 1, characterized in that, In step 4, the influencing factors include temperature, humidity, vibration, and degree of aging.

5. The method for constructing a solid propellant storage performance prediction system according to claim 1, characterized in that, Step 3C also includes an update method to update the monitoring data used for simulation input in real time.

6. The method for constructing a solid propellant storage performance prediction system according to claim 1, characterized in that, Step 3C also includes a consistency check to verify whether the monitoring data used for simulation input is consistent with the actual simulation input data.

Images