Temperature soft measurement system and method of star sensor

Abstract

The invention discloses a temperature soft measurement system and method of a star sensor. The measurement system comprises an interface module; the limited temperature collection point data set module is used for acquiring the temperature of a key node during the in-orbit operation period of the star sensor; the prediction node input module is used for inputting node information of which the temperature needs to be predicted; the thermal analysis simulation database is used for storing thermal analysis simulation data of the star sensor; the ground and on-orbit test database is used for storing existing temperature data of the star sensor; the thermal analysis model correction module is used for adjusting the thermal analysis simulation model based on the temperature data; the machine learning training module is used for training and constructing a neural network model for star sensor temperature soft measurement by utilizing a machine learning algorithm based on the temperature data; and the predicted point temperature output module is used for analyzing and outputting the predicted point temperature based on the star sensor temperature soft measurement neural network model constructed by the machine learning training module.

Description

Technical Field

[0001] The invention relates to a temperature soft measurement system and method for a star sensor, belonging to the technical field of space thermal control. Background Art

[0002] The star sensor is small in size, and the traditional solution of collecting temperature data through temperature sensors is complex in wiring and has low space utilization.

[0003] With the vigorous development of deep space exploration technology, people have higher and higher requirements for the control accuracy of space optoelectronic equipment. As a typical space optoelectronic equipment, the observation accuracy of star sensors has reached the milli-arcsecond level. To further improve its effective observation accuracy, it is necessary to comprehensively analyze the coupling effect of different node temperatures on performance, and to optimize and improve performance by adaptively compensating for the effects according to the degree of influence. Therefore, in order to find out the coupling effect of different node temperatures on the performance of star sensors, it is usually necessary to first obtain the global temperature field, and then obtain the key nodes that mainly affect the performance of star sensors through sensitivity analysis of temperature sampling points to arrange and optimize the temperature sampling points. The existing space thermal control technology cannot directly obtain the global temperature field during the on-orbit operation phase, and can only collect a limited number of temperature field data through temperature sensors arranged at specific locations. The node temperature without temperature sensors usually needs to be interpolated from the thermal balance test data of the ground prototype phase based on the experience of engineers, and the accuracy is difficult to guarantee.

[0004] However, in the above analysis process, the key parameters required for supporting thermal analysis model correction and flight mission prediction obtained during the thermal balance test phase of the star sensor prototype are usually set based on the experience of engineers, which makes it difficult to cover all the key temperature sampling points and parameters during the actual on-orbit operation. In addition, the existing space temperature measurement technology usually requires a large number of temperature sensors to be arranged at the temperature sampling points on the surface of the star sensor to obtain the most comprehensive node temperature. However, the space utilization rate of the existing sensors and wires is low, the star sensor is small in size, and the area where the temperature measurement area can be arranged is limited. Even some sensitive areas cannot be arranged with sensors to obtain their temperature data due to the small space, which will seriously affect the effect of temperature measurement and even temperature control.

[0005] In view of this, how to solve the above-mentioned technical defects has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the invention

[0006] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and solve the accuracy and effectiveness problems of star sensitive temperature measurement.

[0007] The purpose of the present invention is achieved through the following technical solutions:

[0008] A temperature soft measurement system for a star sensor, comprising:

[0009] Interface module;

[0010] The limited temperature sampling point data set module is used to obtain the temperature of key nodes during the on-orbit operation of the star sensor;

[0011] Prediction node input module, used to input node information of which temperature needs to be predicted;

[0012] Thermal analysis simulation database, used to store thermal analysis simulation data of star sensors;

[0013] Ground and on-orbit test database, used to store the temperature data of key nodes during the thermal balance test of the star sensor prototype and on-orbit operation;

[0014] A thermal analysis model correction module, used to receive temperature data from ground and on-orbit test databases via the interface module to adjust the thermal analysis simulation model;

[0015] A machine learning training module is used to receive temperature data from a thermal analysis simulation database and a ground and on-orbit test database via the interface module, and to train and construct a neural network model for soft measurement of temperature of a star sensor using a machine learning algorithm;

[0016] The predicted point temperature output module is used to analyze and output the predicted point temperature based on the star sensor temperature soft measurement neural network model constructed by the machine learning training module.

[0017] In one embodiment of the present invention, the limited temperature sampling point data set module is specifically used for real-time temperature collection of a limited number of key nodes through temperature sensors when the star sensor is in orbit.

[0018] In one embodiment of the present invention, the prediction node input module is specifically used to input node information for which temperature prediction is required, including node location, transient time and other information, and transmit it to the prediction point temperature output module through the interface module for temperature prediction.

[0019] In one embodiment of the present invention, the thermal analysis simulation database is specifically used to store simulation test data of the thermal analysis simulation model of the star sensor under different working conditions and before and after the model is corrected.

[0020] In one embodiment of the present invention, the ground and on-orbit test database includes:

[0021] The ground thermal balance test data storage unit is used to collect and store the key node temperature data of the thermal balance test of the star sensor during the ground prototype design stage.

[0022] The on-orbit test data storage unit is used to collect and store the temperature data of key nodes during the on-orbit operation of the star sensor.

[0023] In one embodiment of the present invention, the thermal analysis model correction module is specifically used to receive temperature data from the ground and on-orbit test databases via the interface module to adjust the thermal analysis simulation model, reduce its error, and improve its accuracy.

[0024] In one embodiment of the present invention, the interface module, the limited temperature sampling point data set module, and the ground and on-orbit test database are constructed using C++ language.

[0025] In one embodiment of the present invention, the prediction node input module, the thermal analysis simulation database, the thermal analysis model correction module, the machine learning training module, and the prediction point temperature output module are obtained through MATLAB language.

[0026] A temperature soft measurement method for a star sensor, comprising:

[0027] The thermal analysis simulation database module is used to collect and store simulation test data of the thermal analysis simulation model of the star sensor under different working conditions and before and after model correction;

[0028] The ground and on-orbit test databases are used to collect and store the temperature data of key nodes of the star sensor during the thermal balance test in the ground prototype design phase and the on-orbit operation phase;

[0029] The thermal analysis model correction module receives the temperature data of the ground and on-orbit test databases through the interface module to adjust the thermal analysis simulation model, reduce its error, and improve its accuracy, including three steps: parameter analysis, determination of correction targets, and optimization analysis under given targets;

[0030] The machine learning training module uses a machine learning algorithm to train the temperature data of the thermal analysis simulation database and the ground and on-orbit test databases received via the interface module and construct a neural network model for soft measurement of the temperature of the star sensor; the limited temperature sampling point data set module is used to obtain the temperature data of a limited number of key node temperature sampling points during the on-orbit operation of the star sensor in real time;

[0031] The node information that needs to predict the temperature, such as the node position, transient time, etc., is transmitted to the neural network model for soft measurement of the star sensor temperature through the interface module through the prediction node input module;

[0032] The predicted point temperature output module receives the temperature data of a limited number of key node temperature sampling points and the node information requiring temperature prediction during the on-orbit operation of the star sensor via the interface module, and the predicted point temperature is obtained by analyzing and calculating the neural network model of the star sensor temperature soft measurement constructed by the machine learning training module.

[0033] A temperature soft measurement device for a star sensor, comprising:

[0034] Temperature sensor, used to collect the temperature of key nodes during the star sensor thermal balance test or on-orbit operation;

[0035] Memory for storing computer programs;

[0036] A processor is used to implement the steps of the temperature soft measurement method of the star sensor when executing the computer program.

[0037] A computer-readable storage device, characterized in that the computer-readable storage medium is specifically used to store a computer program, and the computer program is specifically used to be executed by a processor to implement the steps of the temperature soft measurement method of the star sensor.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] (1) Effectively improve the efficiency of on-orbit temperature data collection of on-board sensors;

[0040] (2) Effectively reduce the cost of on-orbit temperature data collection for star sensors;

[0041] (3) The space occupied by the temperature sensor in the star sensor is reduced;

[0042] (4) The intelligence level of the thermal control system of the star sensor has been improved;

[0043] (5) The temperature field of the entire star sensor can be obtained with limited temperature sampling points, which greatly reduces the difficulty and cost of obtaining the temperature field of the entire star sensor;

[0044] (6) The reliability of obtaining the transient temperature field of the entire star sensor is effectively improved. The fewer the temperature sampling points, the better the synchronization of the temperature acquisition time and the less space is occupied. The more the temperature sampling points, the worse the synchronization and the more space is occupied. Therefore, the full temperature field data can be obtained with limited temperature sampling points, which can better obtain the transient temperature of the entire star sensor.

[0045] (7) The present invention realizes the modeling of the nonlinear thermodynamic coupling relationship between the limited temperature sampling points and the prediction nodes through a machine learning algorithm, which can effectively predict the global temperature field based on the limited temperature sampling point data, improve the efficiency of on-orbit temperature data collection of the star sensor, and reduce costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of a temperature soft measurement system for a star sensor provided in an embodiment of the present application;

[0047] Figure 2 A schematic flow chart of a temperature soft-sensing method for a star sensor provided in an embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0049] Figure 1 A schematic diagram of a temperature soft measurement system for a star sensor provided in an embodiment of the present application; the temperature soft measurement system comprises:

[0050] Interface module 10;

[0051] The limited temperature sampling point data set module 20 is used to obtain the temperature of key nodes during the on-orbit operation of the star sensor;

[0052] The prediction node input module 30 is used to input the node information of which the temperature needs to be predicted;

[0053] A thermal analysis simulation database 40, used for storing thermal analysis simulation data of star sensors;

[0054] The ground and on-orbit test database 50 is used to store the temperature data of key nodes during the thermal balance test of the star sensor prototype and the on-orbit operation;

[0055] The thermal analysis model correction module 60 is used to adjust the thermal analysis simulation model by receiving the temperature data from the ground and on-orbit test databases via the interface module;

[0056] A machine learning training module 70 is used to receive temperature data from a thermal analysis simulation database and a ground and on-orbit test database via the interface module, and to train and construct a neural network model for soft measurement of temperature of a star sensor using a machine learning algorithm;

[0057] The predicted point temperature output module 80 is used to analyze and output the predicted point temperature based on the star sensor temperature soft measurement neural network model constructed by the machine learning training module.

[0058] Specifically, in the working mode of the temperature soft measurement system, the limited temperature point data set module 20, the prediction node input module 30, the thermal analysis simulation database 40, the ground and on-orbit test database 50, the thermal analysis model correction module 60, the machine learning training module 70 and the prediction point temperature output module 80 are operated in sequence. Among them, the interface module 10 provides an interface for the other seven modules, is mainly responsible for communication and data transmission, and provides an interface for the automatic operation of the finite element simulation software, which is to operate the finite element simulation software.

[0059] The limited temperature sampling point data set module 20 is responsible for real-time acquisition of a limited number of key node temperature data of the star sensor during its on-orbit operation through temperature sensors arranged on the surface of the star sensor. For different star sensors and operating states, the corresponding types and quantities of temperature sensors may be different. In addition, this application does not make a sole limitation on the specific types and quantities of temperature sensors, and can be set differently according to actual needs. For example, platinum resistance is used as a temperature measurement sensor for components with high temperature control accuracy requirements, thermocouples are used for temperature measurement for components with low temperature control accuracy requirements, and as many temperature sensors as possible are arranged for components with high temperature gradient requirements, and the number of temperature sensors is reduced otherwise.

[0060] The prediction node input module 30 is responsible for inputting node information of which temperature needs to be predicted, such as node position, transient time and other information, and transmits it to the neural network model for soft measurement of star sensor temperature through the interface module 10 .

[0061] Among them, the neural network model for soft measurement of star sensor temperature is a machine learning training module that uses a machine learning algorithm to train the temperature data of the thermal analysis simulation database 40 and the ground and on-orbit test database 50 received via the interface module 10, and models the nonlinear thermodynamic coupling relationship between the finite temperature sampling points and the prediction nodes in the database through a neural network, thereby obtaining a nonlinear relationship between the finite temperature sampling points and the prediction node temperatures, and then transmits the node information that needs to be predicted temperature to the neural network model for soft measurement of star sensor temperature through the interface module for analysis and calculation to obtain the corresponding prediction point temperature.

[0062] Specifically, in a specific implementation, the thermal analysis simulation database 40 performs transient thermal analysis on the thermal analysis simulation model of the star sensor. The duration and temperature sampling period of the transient thermal analysis are not limited in this application and can be set differently according to actual needs. For example, if the analysis duration is set to 36000s and the temperature sampling period is 10s, 3600 sets of transient thermal analysis simulation result data will be generated after the complete simulation period, and each set of data contains complete temperature field information, that is, all node temperature information.

[0063] The specific implementation process of using thermal analysis simulation software to simulate and obtain simulation results will not be described in detail in this application, and reference can be made to the prior art.

[0064] In addition, the ground and on-orbit test database 50 collects and stores the temperature data of key nodes of the star sensor during the thermal balance test of the ground prototype design stage and the on-orbit operation stage through the temperature sensor, and transmits the temperature data collected by the ground and on-orbit test database 50 to the thermal analysis model correction module through the interface module 10, thereby correcting the thermal analysis simulation model of the star sensor. Then, the thermal analysis simulation is performed based on the corrected thermal analysis model to further update the thermal analysis simulation database.

[0065] Similarly, the specific implementation process of obtaining temperature data through temperature sensors and telemetry during the ground and on-orbit test phases will not be described in detail in this application, and reference will be made to the prior art.

[0066] Specifically, in this embodiment, firstly, the simulation result data of the corresponding node is screened out from the modified thermal analysis simulation model according to the actual limited temperature sampling point position and number of the star sensor, and then the node temperature without the temperature sensor is screened out from the simulation result as needed, so that the actual limited temperature sampling point position and temperature and other information can be retrieved from the result of the thermal analysis simulation database 40, and the thermal analysis simulation result of the predicted node can also be checked. For example, if the transient thermal analysis simulation time is set to 36000s, the temperature sampling cycle is 10s, then 3600 groups of thermal analysis simulation result data will be output, the limited temperature sampling point is set to 36, and the number of nodes to be predicted is 10, then in the 3600 groups of thermal analysis simulation result data, each group of data contains 36 limited temperature sampling point data and 10 node temperature values ​​to be predicted obtained by thermal analysis simulation. Subsequently, the temperature data in the thermal analysis s-true database 40 and the ground and on-orbit test database 50 are transmitted to the machine learning training module 70 through the interface module 10, and the nonlinear thermodynamic relationship between the 36 finite temperature sampling points and the 10 prediction nodes is modeled by a neural network through a machine learning algorithm, thereby obtaining a neural network model for soft measurement of the temperature of the star sensor, and continuously optimizing the neural network model through the on-orbit temperature data of the finite temperature sampling point data set module 20. Finally, the finite temperature sampling point data collected in real time by the finite temperature sampling point data set module 20 at any time and the node information of the node that needs to be predicted temperature input by the prediction node input module 30 are transmitted to the neural network model through the interface module 10, and the predicted node temperature is obtained by analysis and calculation, and then the result is output to the prediction point temperature output module 80 through the interface module 10.

[0067] Further, in a specific implementation, the interface module 10, the limited temperature sampling point data set module 20, and the ground and on-orbit test database 50 are constructed by C++ language. The prediction node input module 30, the thermal analysis simulation database 40, the thermal analysis model correction module 60, the machine learning training module 70, and the prediction point temperature output module 80 are obtained by MATLAB language. Of course, in addition to constructing each module based on the C++ and MATLAB languages, the above modules can also be constructed based on other computer languages, such as Python, C#, etc.

[0068] In summary, the soft measurement system of the star sensor provided by the present application includes an interface module 10; a limited temperature sampling point data set module 20, which is used to obtain the temperature of key nodes during the on-orbit operation of the star sensor; a prediction node input module 30, which is used to input the node information of the node whose temperature needs to be predicted; a thermal analysis simulation database 40, which is used to store the thermal analysis simulation data of the star sensor; a ground and on-orbit test database 50, which is used to store the temperature data of the key nodes during the thermal balance test of the initial sample of the star sensor and the on-orbit operation; a thermal analysis model correction module 60, which is used to receive the temperature data of the ground and on-orbit test databases via the interface module to adjust the thermal analysis simulation model, reduce its error, and improve its accuracy; a machine learning training module 70, which is used to receive the temperature data of the thermal analysis simulation database and the ground and on-orbit test databases via the interface module, and use the machine learning algorithm to train and construct a neural network model for soft measurement of the star sensor temperature; a prediction point temperature output module 80, which is used to analyze and output the prediction point temperature based on the star sensor temperature soft measurement neural network model constructed by the machine learning training module. The temperature soft measurement system uses machine learning algorithms to model the nonlinear thermodynamic coupling relationship between limited temperature sampling points and prediction nodes. It can effectively predict the global temperature field based on the data from limited temperature sampling points, improve the efficiency of on-orbit temperature data collection of star sensors, and reduce costs.

[0069] The present application also provides a temperature soft measurement method for a star sensor, and the method described below can be referenced in correspondence with the system described above. Figure 2 A schematic flow chart of a temperature soft measurement method for a star sensor provided in an embodiment of the present application, the method comprising:

[0070] S1. Collect and store simulation test data of the thermal analysis simulation model of the star sensor under different working conditions and before and after model correction through the thermal analysis simulation database module;

[0071] S2. Collect and store the preset node temperature data of the star sensor in the existing thermal balance test on the ground and in the orbit operation stage through the ground and on-orbit test database;

[0072] S3, receiving the temperature data of the ground and on-orbit test databases through the interface module through the thermal analysis model correction module to adjust the thermal analysis simulation model, reduce its error, and improve its accuracy, including three steps: parameter analysis, determination of correction targets, and optimization analysis under given targets;

[0073] S4, using a machine learning algorithm to train the temperature data of the thermal analysis simulation database and the ground and on-orbit test databases received via the interface module through the machine learning training module, and construct a neural network model for soft measurement of star sensor temperature;

[0074] S5, transmitting the node information of which the temperature needs to be predicted, such as the node position, transient time and other information, to the neural network model for soft measurement of the star sensor temperature through the interface module through the prediction node input module;

[0075] S6. Real-time acquisition of temperature data of a limited number of key node temperature sampling points during the on-orbit operation of the star sensor is carried out through the limited temperature sampling point data set module; receiving the temperature data of a limited number of key node temperature sampling points during the on-orbit operation of the star sensor and the node information of the temperature that needs to be predicted through the interface module through the predicted point temperature output module, and analyzing and calculating the predicted point temperature through the neural network model of the star sensor temperature soft measurement constructed by the machine learning training module.

[0076] On the basis of the above embodiment, optionally, a machine learning training module utilizes a machine learning algorithm to train the temperature data of the thermal analysis simulation database and the ground and on-orbit test databases received via the interface module and construct a neural network model for soft measurement of star sensor temperature, wherein the machine learning algorithm may include different types such as BP neural network, RBF neural network, graph neural network, etc. This application does not make a sole limitation and can be differently set according to actual needs.

[0077] The present application also provides a temperature soft measurement device for a star sensor, the device comprising: a temperature sensor, a memory and a processor; wherein the temperature sensor is used to collect the temperature of key nodes during a thermal balance test of the star sensor or during on-orbit operation; the memory is used to store a computer program; and the processor is used to implement the following steps when executing the computer program:

[0078] The temperature data of a limited number of key node temperature sampling points during the on-orbit operation of the star sensor are acquired in real time through the limited temperature sampling point data set module 20; the node information that needs to be predicted in temperature, such as node position, transient time and other information, is transmitted to the neural network model for soft measurement of the temperature of the star sensor through the interface module 10 through the prediction node input module 30; the simulation test data of the thermal analysis simulation model of the star sensor under different working conditions and before and after model correction are collected and stored through the thermal analysis simulation database 40; the key node temperature data of the star sensor in the thermal balance test of the ground prototype design stage and the on-orbit operation stage are collected and stored through the ground and on-orbit test database 50; the thermal analysis model correction module 60 receives the ground and on-orbit test database 50 through the interface module 10 The temperature data is used to adjust the thermal analysis simulation model to reduce its error and improve its accuracy, which includes three steps: parameter analysis, determination of correction targets, and optimization analysis under given targets; the temperature data of the thermal analysis simulation database 40 and the ground and on-orbit test database 50 received via the interface module 10 are trained using a machine learning algorithm through a machine learning training module, and a neural network model for soft measurement of star sensor temperature is constructed; the temperature data of a limited number of key node temperature sampling points and the node information requiring temperature prediction during the on-orbit operation of the star sensor are received via the interface module 10 through a prediction point temperature output module 80, and the prediction point temperature is calculated and analyzed through the neural network model of soft measurement of star sensor temperature constructed by the machine learning training module 70.

[0079] For an introduction to the equipment provided in this application, please refer to the embodiments of the above-mentioned system, and this application will not go into details here.

[0080] The present application also provides a computer-readable storage medium, on which a computer program required for a star sensor temperature soft measurement system is stored. When the computer program is executed by a processor, the following steps are implemented:

[0081] The temperature data of a limited number of key node temperature sampling points during the on-orbit operation of the star sensor are acquired in real time through the limited temperature sampling point data set module 20; the node information that needs to be predicted in temperature, such as node position, transient time and other information, is transmitted to the neural network model for soft measurement of the temperature of the star sensor through the interface module 10 through the prediction node input module 30; the simulation test data of the thermal analysis simulation model of the star sensor under different working conditions and before and after model correction are collected and stored through the thermal analysis simulation database 40; the key node temperature data of the star sensor in the thermal balance test of the ground prototype design stage and the on-orbit operation stage are collected and stored through the ground and on-orbit test database 50; the thermal analysis model correction module 60 receives the ground and on-orbit test database 50 through the interface module 10 The temperature data is used to adjust the thermal analysis simulation model to reduce its error and improve its accuracy, which includes three steps: parameter analysis, determination of correction targets, and optimization analysis under given targets; the temperature data of the thermal analysis simulation database 40 and the ground and on-orbit test database 50 received via the interface module 10 are trained using a machine learning algorithm through a machine learning training module, and a neural network model for soft measurement of star sensor temperature is constructed; the temperature data of a limited number of key node temperature sampling points and the node information requiring temperature prediction during the on-orbit operation of the star sensor are received via the interface module 10 through a prediction point temperature output module 80, and the prediction point temperature is calculated and analyzed through the neural network model of soft measurement of star sensor temperature constructed by the machine learning training module 70.

[0082] The computer calibration storage medium may include: a USB flash drive, an optical disk, a mobile hard disk, a random access memory (RAM), a read-only memory (ROM), and other media that can store program codes. This application does not make a sole limitation on the specific implementation scenario, and different settings can be made according to actual needs.

[0083] For an introduction to the computer calibration storage medium provided by the present invention, please refer to the above-mentioned system embodiment, and the present invention will not be elaborated here.

[0084] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

[0085] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. A temperature soft measurement system for a star sensor, characterized in that: include Interface module; The limited temperature sampling point data set module is used to obtain the preset node temperature during the on-orbit operation of the star sensor; Prediction node input module, used to input node information of which temperature needs to be predicted; Thermal analysis simulation database, used to store thermal analysis simulation data of star sensors; Ground and on-orbit test database, used to store the existing thermal balance test data of star sensors and preset node temperature data during on-orbit operation; A thermal analysis model correction module, used to receive temperature data from ground and on-orbit test databases via the interface module to adjust the thermal analysis simulation model; A machine learning training module is used to receive temperature data from a thermal analysis simulation database and a ground and on-orbit test database via the interface module, and to train and construct a neural network model for soft measurement of temperature of a star sensor using a machine learning algorithm; The predicted point temperature output module is used to analyze and output the predicted point temperature based on the star sensor temperature soft measurement neural network model constructed by the machine learning training module.

2. The star sensor temperature soft measurement system according to claim 1, characterized in that: The limited temperature sampling point data set module is specifically used for real-time temperature collection of a limited number of key nodes through temperature sensors during on-orbit operation of star sensors.

3. The star sensor temperature soft measurement system according to claim 1, characterized in that: The prediction node input module is specifically used to input the node information of the temperature to be predicted, including the node position and transient time, and transmit the node information to the prediction point temperature output module through the interface module for temperature prediction.

4. The star sensor temperature soft measurement system according to claim 1, characterized in that: The thermal analysis simulation database is specifically used to store simulation test data of the thermal analysis simulation model of the star sensor under different working conditions and before and after the model is corrected.

5. The star sensor temperature soft measurement system according to claim 1, characterized in that: The ground and on-orbit test database includes: A ground thermal balance test data storage unit is used to collect and store preset node temperature data of the thermal balance test of the star sensor in the ground prototype design stage; The on-orbit test data storage unit is used to collect and store preset node temperature data during the on-orbit operation of the star sensor.

6. The star sensor temperature soft measurement system according to claim 1, characterized in that: The machine learning training module models the nonlinear thermodynamic coupling relationship between the finite temperature sampling points and the prediction nodes in the database through a neural network, thereby obtaining the nonlinear relationship between the finite temperature sampling points and the prediction node temperatures. Then, the node information whose temperature needs to be predicted is transmitted to the neural network model used for star sensor temperature soft measurement through the interface module to analyze and calculate the corresponding prediction point temperature.

7. The star sensor temperature soft measurement system according to claim 1, characterized in that: According to the actual limited temperature sampling point position and number of the star sensor, the simulation result data of the corresponding node is filtered out from the modified thermal analysis simulation model, and then the node temperature without temperature sensor is filtered out from the simulation result as needed, the actual limited temperature sampling point position and temperature information are retrieved from the results of the thermal analysis simulation database, and the thermal analysis simulation results of the predicted node are checked.

8. The star sensor temperature soft measurement system according to claim 1, characterized in that: The machine learning training module uses a machine learning algorithm to perform neural network modeling on the nonlinear thermodynamic relationship between the limited temperature sampling points and the prediction nodes, and then obtains a neural network model for soft measurement of star sensor temperature, and continuously optimizes the neural network model through the on-orbit temperature data of the limited temperature sampling point data set module.

9. A measurement method based on the star sensor temperature soft measurement system according to any one of claims 1 to 8, characterized in that: include: The thermal analysis simulation database module is used to collect and store simulation test data of the thermal analysis simulation model of the star sensor under different working conditions and before and after model correction; The ground and on-orbit test databases are used to collect and store the preset node temperature data of the star sensor during the existing thermal balance test on the ground and the on-orbit operation stage; The thermal analysis model correction module receives the temperature data of the ground and on-orbit test databases via the interface module to adjust the thermal analysis simulation model; Using a machine learning algorithm through a machine learning training module to train the temperature data of the thermal analysis simulation database and the ground and on-orbit test databases received through the interface module and construct a neural network model for soft measurement of star sensor temperature; The node information that needs to predict the temperature is transmitted to the neural network model used for the soft measurement of the star sensor temperature through the prediction node input module; The predicted point temperature output module receives the temperature data of a limited number of key node temperature sampling points and the node information requiring temperature prediction during the on-orbit operation of the star sensor via the interface module, and the predicted point temperature is obtained by analyzing and calculating the neural network model of the star sensor temperature soft measurement constructed by the machine learning training module.

10. A computer-readable storage medium having computer program instructions stored thereon, wherein when the computer program instructions are loaded and executed by a processor, the processor is caused to execute the method according to claim 9.

Images