Spacecraft environment test sensor cutting method and device, computer equipment and computer readable storage medium

Abstract

The invention provides a spacecraft environment test sensor clipping method, which comprises the following steps of: pasting sensors inside and outside a cabin of a preorder test spacecraft in a spacecraft to be tested, and executing an environment test to obtain response data; according to the sensor pasting positions, spacecraft measuring point classification is carried out; cutting off corresponding in-cabin test points and out-cabin necessary test points in a subsequent test spacecraft; a KMeans clustering model with the preset number of reserved measuring points as the classification number is established, and the corresponding outboard optional measuring points in the spacecraft are cut in the mode that one measuring point is reserved for each category through the KMeans clustering model; establishing a spacecraft response prediction model based on machine learning, and taking the response data in the preorder test spacecraft as a test set and a training set; and the average error and the maximum error of response prediction of the measurement points output in the test set are adopted as model evaluation parameters, and the number of the preset reserved measurement points is adjusted to determine the reserved measurement points of the subsequent test spacecraft. Therefore, the response of each part of the spacecraft can be accurately obtained, the spacecraft environment test can be safely and quickly carried out, and the thermal control measures on the outer surface of the spacecraft are protected.

Description

Technical Field

[0001] The present invention relates to the field of spacecraft test technology, and in particular to a spacecraft environmental test sensor tailoring method, device, computer equipment and computer-readable storage medium. Background Art

[0002] The Internet spacecraft constellation has an extremely large scale, a tight development cycle and huge economic potential. Relevant manufacturers at home and abroad are transforming their development model towards low-cost, mass production.

[0003] Environmental testing is a common means of detecting spacecraft faults, including mechanical testing (such as vibration testing, noise testing, etc.) and thermal testing (such as thermal vacuum testing, thermal cycle testing, etc.). It is a must-do project for every star in the current Internet spacecraft constellation. During environmental testing, sensors need to be arranged at typical locations of various equipment and panels of the spacecraft, such as vibration sensors during mechanical testing and thermocouples during thermal testing, to measure the spacecraft response and formulate test conditions. Usually, the number of spacecraft sensors is no less than 100, of which about one-third is inside the spacecraft. Before and after the test, the spacecraft panel needs to be opened to paste and remove the sensors, which inevitably prolongs the spacecraft test time, which does not meet the requirements of the spacecraft pulsating production mode. In addition, the outer surface of the spacecraft needs to be treated with white paint, OSR sheets or thermal control multi-layer coating to solve the problem of thermal control of the spacecraft on orbit. Pasting sensors on the outer surface of the spacecraft may affect the thermal control measures of the outer surface, thereby affecting the on-orbit life of the spacecraft.

[0004] At present, the common method for trimming spacecraft sensors is to trim the in-cabin measurement points or the measurement points that are not involved in the formulation of test conditions. For trimming in-cabin measurement points, the in-cabin measurement points at the corresponding positions can be replaced by out-cabin measurement points, or the responses of out-cabin measurement points can be used to predict the responses of in-cabin measurement points. The former method cannot cover all in-cabin measurement points, such as the measurement points on the spacecraft partition. Both methods cannot reduce or may even increase the number of out-cabin measurement points, and cannot reduce the risk of affecting the life of thermal control measures on the outer surface of the spacecraft. For trimming measurement points that are not involved in the formulation of test conditions, the trimming principle is based on historical data. Not only is it impossible to know the corresponding response of the trimmed measurement points, but also the response differences between spacecraft caused by differences in process operations are not considered. There is a risk that the corresponding response of the trimmed measurement points is too large. Therefore, there is an urgent need for a method that can solve the problems existing in the existing technology. Summary of the invention

[0005] The purpose of the present invention is to provide a method, device, computer equipment and computer-readable storage medium for cutting spacecraft environmental test sensors. The combination of unsupervised and supervised learning algorithms can achieve accurate acquisition of responses from various parts of the spacecraft when more than 60% of the sensors are cut, and spacecraft tests can be carried out safely and quickly, and thermal control measures for the outer surface of the spacecraft can be protected.

[0006] In order to achieve the above-mentioned object, on the one hand, the present invention provides a method for tailoring a spacecraft environmental test sensor, comprising the steps of:

[0007] affixing sensors inside and outside a cabin of a preceding test spacecraft among the spacecraft to be tested, and performing an environmental test on the preceding test spacecraft to obtain response data of the preceding test spacecraft;

[0008] Classifying the spacecraft measurement points according to the sensor pasting positions; wherein the types of spacecraft measurement points include in-cabin measurement points, mandatory out-cabin measurement points, and optional out-cabin measurement points;

[0009] Cutting out the in-cabin measuring points and the out-cabin necessary measuring points corresponding to the subsequent test spacecraft in the spacecraft to be tested;

[0010] Establish a KMeans clustering model with the preset number of retained measurement points as the number of classifications;

[0011] The optional measuring points outside the cabin are trimmed by retaining one measuring point in each category through the KMeans clustering model;

[0012] Establishing a spacecraft response prediction model based on machine learning, using the response data of some spacecraft in the preceding test spacecraft as a test set, and using the response data of the remaining spacecraft in the preceding test spacecraft as a training set;

[0013] The spacecraft response prediction model is trained based on the training set, and the average error and maximum error of the output measurement point response prediction in the test set are used as model evaluation parameters. The reserved measurement points of the subsequent test spacecraft that meet the preset errors are determined by adjusting the preset number of reserved measurement points.

[0014] Optionally, the step of attaching sensors inside and outside the cabin of a preceding test spacecraft in the spacecraft to be tested includes:

[0015] If the sensor is a mechanical sensor, the sensor is pasted at the root of all equipment in the cabin, the root of rigid equipment outside the cabin, the root of flexible equipment outside the cabin and the maximum response point of the preceding test spacecraft of the spacecraft to be tested; wherein the rigid equipment is a device whose fundamental frequency is not less than the upper limit of the mechanical test frequency, and the flexible equipment is a device whose fundamental frequency is less than the upper limit of the mechanical test frequency;

[0016] If the sensor is a thermal sensor, the sensor is pasted on the cabin surface corresponding to the ends of the heat pipe, the root of the high heat consumption equipment, and the highest temperature point in the high heat consumption equipment of the preceding test spacecraft of the spacecraft to be tested.

[0017] Optionally, the data structure of the KMeans clustering model input takes each optional measuring point outside the cabin as a sample, and each sample is formed by splicing data within boundary conditions under all test conditions of all the previous test spacecraft.

[0018] Optionally, the data structure of the KMeans clustering model input is a matrix of m rows and (a×d×s×n) columns; wherein m is the number of optional measuring points outside the cabin, a is the number of tests within the boundary conditions in the test, s is the number of test conditions for each of the preceding test spacecraft, n is the number of preceding test spacecraft, and d is the number of measurement groups of the sensor at the measuring points.

[0019] Optionally, the spacecraft response prediction model is trained based on a machine learning algorithm of a random forest or a neural network.

[0020] Optionally, the data structure of the test set uses each response data within the boundary conditions under all test conditions of some spacecraft in the previous test spacecraft as a sample, and the responses corresponding to all retained measurement points are used as model inputs, and the responses corresponding to the cropped measurement points are used as model outputs;

[0021] The data structure of the training set takes each response data within the boundary conditions under all test conditions relative to the remaining spacecraft in the previous test spacecraft as a sample, and the responses corresponding to all retained measurement points are used as model inputs, and the responses corresponding to the cropped measurement points are used as model outputs.

[0022] Optionally, the average error and the maximum error are calculated based on the following formula:

[0023]

[0024] Among them, Error1 and Error2 are the average error and the maximum error respectively, a T and a P are the spacecraft response test value and the predicted value, respectively, and are both (a×s)-row d×(b+c)-column matrices; a is the number of tests within the boundary conditions in the test, s is the number of test conditions for each of the preceding test spacecraft, n is the number of preceding test spacecraft, d is the number of measurement groups of the sensor at the measuring point; b is the number of retained side points, and c is the number of cropped measuring points.

[0025] On the other hand, based on the same inventive structure, the present invention also provides a spacecraft environmental test sensor cutting device, comprising:

[0026] The test unit is used to attach sensors inside and outside the cabin of a previous test spacecraft in the spacecraft to be tested, and perform environmental tests on the previous test spacecraft to obtain response data of the previous test spacecraft.

[0027] A classification unit, used for classifying the spacecraft measurement points according to the sensor pasting position; wherein the spacecraft measurement point types include in-cabin measurement points, out-cabin mandatory measurement points and out-cabin optional measurement points;

[0028] A first execution unit is used to cut out the in-cabin measurement points and the out-cabin must-cut measurement points corresponding to the subsequent test spacecraft in the spacecraft to be tested;

[0029] The first computing unit is used to establish a KMeans clustering model with a preset number of retained measurement points as the number of classifications;

[0030] A second execution unit is used to trim the optional measuring points outside the cabin by retaining one measuring point in each category through the KMeans clustering model;

[0031] a second computing unit, configured to establish a spacecraft response prediction model based on machine learning, using the response data of some spacecraft in the preceding test spacecraft as a test set, and the response data of the remaining spacecraft in the preceding test spacecraft as a training set;

[0032] A determination unit is used to train the spacecraft response prediction model based on the training set, and use the average error and maximum error of the output measurement point response prediction in the test set as model evaluation parameters, and adjust the preset number of retained measurement points to determine the retained measurement points of the subsequent test spacecraft that meet the preset errors.

[0033] In order to achieve another object of the present invention, the present invention further provides a computer device, comprising:

[0034] a memory for storing instructions executed by at least one processor;

[0035] The processor is used to execute the instructions stored in the memory to execute the above-mentioned spacecraft environmental test sensor tailoring method.

[0036] In order to achieve another inventive object of the present invention, the present invention also provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by a processor, they implement the spacecraft environmental test sensor trimming method described in any one of the above items.

[0037] The spacecraft environmental test sensor trimming method of the present invention trims all the measuring points inside the cabin and the necessary measuring points outside the cabin, thereby avoiding the need to open the cabin before and after the environmental test and protecting the thermal control measures on the outer surface of the spacecraft, and significantly improving the test effect and safety; and the present invention provides an unsupervised KMeans clustering machine learning model, as well as model evaluation criteria for average error and maximum error, which clarifies the basis for trimming optional measuring points outside the spacecraft cabin, thereby achieving the technical effect of trimming more than 60% of the sensors; at the same time, it also provides a supervised machine learning model, clarifies the data structure of the model training, and can achieve accurate prediction of the response at the trimmed measuring points of the spacecraft in subsequent tests under decoupling from the spacecraft response direction, and significantly improves the model prediction efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A flowchart showing the steps of the spacecraft environmental test sensor tailoring method provided by an embodiment of the present invention is shown;

[0039] Figure 2 The predicted error case of the test set of the spacecraft environmental test sensor tailoring method provided by an embodiment of the present invention is shown;

[0040] Figure 3 A system structure diagram of the spacecraft environmental test sensor cutting device provided by an embodiment of the present invention is shown;

[0041] Figure 4 A schematic diagram of the structure of the computer device provided in one embodiment of the present invention is shown. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0043] It should be noted that references to "one embodiment", "embodiment", "example embodiment", etc. in this specification refer to the embodiment described, which may include specific features, structures or characteristics, but not every embodiment must include these specific features, structures or characteristics. In addition, such expressions do not refer to the same embodiment. Furthermore, when describing specific features, structures or characteristics in conjunction with an embodiment, whether or not there is an explicit description, it has been shown that it is within the knowledge of those skilled in the art to combine such features, structures or characteristics into other embodiments.

[0044] In addition, certain words are used in the specification and subsequent claims to refer to specific components or parts. Those with ordinary knowledge in the relevant field should understand that manufacturers can use different nouns or terms to refer to the same component or part. This specification and subsequent claims do not use differences in names as a way to distinguish components or parts, but use differences in the functions of components or parts as the criteria for distinction. "Including" and "including" mentioned throughout the specification and subsequent claims are open-ended terms and should be interpreted as "including but not limited to". In addition, the word "connected" here includes any direct and indirect electrical connection means. Indirect electrical connection means include connection through other devices.

[0045] Embodiment 1

[0046] Figure 1 A method for tailoring a spacecraft environmental test sensor provided by an embodiment of the present invention is shown. The specific implementation of the method may include the following steps:

[0047] S101: Paste sensors inside and outside the cabin of a preceding test spacecraft among the spacecraft to be tested, and perform an environmental test on the preceding test spacecraft to obtain response data of the preceding test spacecraft. In this embodiment, a number of spacecraft to be tested are specifically divided into preceding test spacecraft and subsequent test spacecraft, and then sensors are pasted at corresponding positions inside and outside the cabin of the preceding test spacecraft, and an environmental test is performed on the preceding test spacecraft to obtain corresponding response data. The environmental tests to be performed in this embodiment specifically refer to mechanical tests and thermal tests; mechanical tests include vibration tests, noise tests, impact tests, etc., and the sensors are mechanical sensors, such as vibration sensors; thermal tests include thermal balance tests, thermal vacuum tests, thermal cycle tests, etc., and the sensors are thermal sensors, such as thermocouples, etc.

[0048] Specifically, in the mechanical test of the spacecraft, if the sensor is a mechanical sensor, the sensor can be pasted at the root of all equipment in the cabin of the previous test spacecraft, the root of the rigid equipment outside the cabin, the root of the flexible equipment outside the cabin, and the maximum response point; wherein the rigid equipment is a device with a fundamental frequency not less than the upper limit of the mechanical test frequency, and the flexible equipment is a device with a fundamental frequency less than the upper limit of the mechanical test frequency. In the thermal test of the spacecraft, if the sensor is a thermal sensor, the sensor is pasted on the cabin plate surface corresponding to the ends of the heat pipe of the previous test spacecraft to be tested, the root of the high heat consumption equipment, and the highest temperature of the high heat consumption equipment.

[0049] The spacecraft to be tested is a mass-produced spacecraft, a spacecraft on the same platform, or several spacecraft of similar types, etc.; the sensor can be a mechanical sensor or a thermal sensor.

[0050] In an optional implementation, mass-produced spacecraft and mechanical sensors are taken as an example for explanation. It is assumed that there are a total of 100 mass-produced spacecraft, the first 3 spacecraft are used as pre-sequence test spacecraft, and the remaining 97 spacecraft are used as post-sequence test spacecraft. A total of 135 mechanical sensors are pasted on each of the first 3 spacecraft, among which 35 mechanical sensors are pasted on the root of the equipment inside the spacecraft cabin (such as the payload integrated processor), 40 mechanical sensors are pasted on the root of the rigid equipment outside the spacecraft cabin (such as the laser terminal, star sensor), and 60 mechanical sensors are pasted on the root of the spacecraft flexible equipment and the maximum response point (such as the solar wing, the feeding antenna); there are three measurement groups of the mechanical sensors at the corresponding measuring points, that is, one mechanical sensor corresponds to three groups of data, corresponding to the three directions of X, Y, and Z respectively; if the sensor is a thermal sensor, then one thermal sensor only needs to measure one group of data at the corresponding measuring point.

[0051] In this example, sinusoidal vibration tests in the X, Y, and Z directions are carried out sequentially on each of the three preceding test spacecraft. The tests in each direction are carried out at small magnitude, large magnitude, and small magnitude respectively, for a total of 9 working conditions. The test frequency range is 5 to 100 Hz, and the response data of the 135 mechanical sensors are measured under each working condition. The measured response data specifically have the frequency (in Hz) as the horizontal coordinate and the acceleration amplitude (in g) as the vertical coordinate.

[0052] S102: Classifying the spacecraft measurement points according to the sensor pasting positions; wherein the types of spacecraft measurement points include in-cabin measurement points, mandatory measurement points outside the cabin, and optional measurement points outside the cabin; specifically, in-cabin measurement points are measurement points that can be operated after opening the spacecraft cabin panel; mandatory measurement points outside the cabin are measurement points that can be operated without opening the spacecraft cabin panel, but the operation has an adverse effect on the life of the spacecraft; and optional measurement points outside the cabin are measurement points other than in-cabin measurement points and mandatory measurement points outside the cabin.

[0053] In the specific implementation provided in this embodiment, there are 35 measuring points inside the cabin, 20 mandatory measuring points outside the cabin, which are the measuring points pasted on the white paint on the outer surface of the laser terminal, and 80 optional measuring points outside the cabin.

[0054] S103: Cut off the in-cabin measurement points and the must-cut measurement points outside the cabin that correspond to the subsequent test spacecraft in the spacecraft to be tested. That is, this embodiment will cut off the in-cabin measurement points and the must-cut measurement points outside the cabin classified by step S102 in the subsequent test spacecraft. Combined with the specific scheme provided by the above-mentioned embodiment, a total of 55 in-cabin measurement points and must-cut measurement points outside the cabin are cut off on the fourth subsequent test spacecraft and other subsequent spacecraft. The scheme provided by this embodiment cuts off all the in-cabin measurement points and must-cut measurement points outside the cabin by proposing a measurement point classification method based on the sensor pasting position, thereby avoiding the need to open the cabin before and after the test and protecting the thermal control measures on the outer surface of the spacecraft, and significantly improving the test effect and safety.

[0055] S104: Establishing a KMeans clustering model with a preset number of retained measurement points as the number of classifications; that is, presetting the number of retained measurement points, and establishing the KMeans clustering model with the preset number of retained measurement points as the number of classifications; the KMeans clustering model is a widely used unsupervised learning algorithm that can be used to divide data points into different clusters.

[0056] Preferably, the data structure input to the KMeans clustering model takes each optional measurement point outside the cabin as a sample, and each sample is composed of data within the boundary conditions of environmental tests under all test conditions of all previous test spacecraft. The boundary conditions refer to the range defined between the upper frequency limit and the lower frequency limit in mechanical tests, and the boundary conditions refer to the interval between the start and end time in thermal tests.

[0057] Specifically, the data structure of the KMeans clustering model input is a matrix of m rows and (a×d×s×n) columns; wherein m is the number of optional measuring points outside the cabin, a is the number of tests within the boundary conditions in the test, specifically, in the mechanical test, a is the number of tests within the upper and lower frequency limits in the mechanical test; s is the number of test conditions for each preceding test spacecraft, n is the number of preceding test spacecraft, and d is the number of measurement groups of the sensor at the measuring point; combined with the above example, it can be seen that the coefficient d corresponding to the mechanical sensor is 3, which represents the response of each optional measuring point outside the cabin in three axes.

[0058] The data structure of the KMeans clustering model input preferred in this embodiment (taking the measurement data of the mechanical sensor as an example) is shown in the following Table 1.

[0059] Table 1:

[0060]

[0061] In combination with the specific implementation methods provided in the above embodiment, in the table, m is the number of optional measurement points outside the cabin, which is equal to 80 here; a is the upper frequency limit f in the mechanical test. u=The number of measurements between the upper limit f1=100Hz and the lower limit f1=5Hz, which is equal to 500 here; s is the number of test conditions for each spacecraft, which is equal to 9 here; n is the number of preceding test spacecraft, which is equal to 3 here; that is, the data structure input to the KMeans clustering model of this embodiment is 80 rows and 40500 columns; the number of categories is the preset number of retained measuring points, which is 40 here.

[0062] S105: trimming the optional extravehicular measuring points by retaining one measuring point in each category through the KMeans clustering model; that is, retaining only one measuring point in each category and trimming the remaining optional extravehicular measuring points.

[0063] In combination with the specific implementation method provided above, the KMeans clustering model divides 80 optional extravehicular measuring points into 40 categories, randomly selects one from each category to form 40 reserved measuring points, and the remaining 95 measuring points are trimmed measuring points.

[0064] S106: Establish a spacecraft response prediction model based on machine learning, use the response data of some spacecraft in the previous test spacecraft as a test set, and use the response data of the remaining spacecraft in the previous test spacecraft as a training set.

[0065] Preferably, the data structure of the test set takes each response data within the boundary conditions under all test conditions of some spacecraft in the previous test spacecraft as a sample, and the corresponding responses of all retained measuring points are used as model input, and the responses corresponding to the cropped measuring points are used as model output; the data structure of the training set takes each response data within the boundary conditions under all test conditions of the remaining spacecraft in the previous test spacecraft as a sample, and the corresponding responses of all retained measuring points are used as model input, and the responses corresponding to the cropped measuring points are used as model output.

[0066] In specific implementation, the response data of the last spacecraft in the previous test spacecraft is used as the test set, and the corresponding response data of the remaining spacecraft are used as the test set; that is, in the mechanical test, the data structure of the test set uses each response data between the upper frequency limit and the lower frequency limit of the mechanical test under all test conditions of the last spacecraft in the previous test spacecraft as a sample, and the corresponding responses of all retained measuring points are used as model input, and the responses corresponding to the cropped measuring points are used as model output; the data structure of the training set is the same as that of the test set, but the difference between the two is that the data of the training set comes from the remaining spacecraft in the previous test spacecraft except the last spacecraft.

[0067] In thermal tests, the data structure of the test set takes each response data between the start and end of the thermal test under all test conditions of the last spacecraft in the previous test spacecraft as a sample, and the corresponding responses of all retained measuring points are used as model input, and the responses corresponding to the cropped measuring points are used as model output; correspondingly, the data structure of the training set is the same as that of the test set, but the difference between the two is that the data of the training set comes from the remaining spacecraft in the previous test spacecraft except the last spacecraft.

[0068] The data structure of the preferred training set in this embodiment (taking the measurement data of the mechanical sensor as an example) is shown in the following Table 2.

[0069] Table 2:

[0070]

[0071]

[0072] In combination with the specific implementation methods provided in the above embodiment, in the table, b is the number of retained measuring points, which is equal to 40 here; c is the number of cut measuring points, which is equal to 95 here; a is the frequency upper limit f in the mechanical test u =100Hz and the lower limit f1 =5Hz, which is equal to 500 here; s is the number of test conditions for each spacecraft, which is equal to 9 here; n is the number of previous test spacecraft, which is equal to 3 here; that is, the data structure of the training set input is 9000 rows and 120 columns, and the data structure of the training set output is 9000 rows and 258 columns.

[0073] The data structure of the preferred test set of this embodiment (taking the measurement data of the mechanical sensor as an example) is shown in the following Table 3.

[0074] Table 3:

[0075]

[0076] Using the same data structure as the training set above, it can be calculated that the data structure of the test set input is 4500 rows and 120 columns, and the data structure of the test set output is 4500 rows and 258 columns.

[0077] This embodiment preferably uses a machine learning algorithm that is good at processing nonlinear problems, such as a random forest and a neural network algorithm; that is, the spacecraft response prediction model is trained based on a random forest or a neural network machine learning algorithm. Specifically, in the solution provided in this embodiment, an MLP multilayer perceptron neural network algorithm is used, with 1 layer / 100 neurons per layer, and the activation function is a relu function.

[0078] S107: Training the spacecraft response prediction model based on the training set, and using the average error and maximum error of the output measurement point response prediction in the test set as model evaluation parameters, and adjusting the preset number of retained measurement points to determine the retained measurement points of the subsequent test spacecraft that meet the preset errors.

[0079] Furthermore, the average error and the maximum error are calculated based on the following formula:

[0080]

[0081] Among them, Error1 and Error2 are the average error and the maximum error respectively, a T and a P are the spacecraft response test value and the predicted value, respectively, and are both (a×s)-row d×(b+c)-column matrices; a is the number of tests within the boundary conditions in the test, s is the number of test conditions for each of the preceding test spacecraft, n is the number of preceding test spacecraft, d is the number of measurement groups of the sensor at the measuring point; b is the number of retained side points, and c is the number of cropped measuring points.

[0082] In the experiment taking the mechanical sensor as an example, one mechanical sensor corresponds to three sets of data, namely X, Y, and Z direction data, so d is equal to 3; in the experiment taking the thermal sensor as an example, since one thermal sensor only corresponds to one set of data, d is equal to 1.

[0083] In the solution provided in this embodiment, the responses of the star-sensing root measuring point AG06 in the three directions of X, Y, and Z are as follows: Figure 2 As shown in the figure, it can be seen that the predicted consistency is good. The calculated average error is 0.9g and the maximum error is 2g, which meets the preset average error of 1.5g and maximum error of 3g. If it is not satisfied, the preset number of retained measuring points can be adjusted to the original value + 2 until the preset error requirements are met.

[0084] In other examples, if a thermal sensor is used as an example for the experiment, the difference between the obtained measurement response data and the mechanical sensor is that the response corresponding to the thermal sensor will be presented in a way that the horizontal axis is time, that is, the boundary conditions are the start and end of time.

[0085] This embodiment clarifies the basis for cutting optional measuring points outside the spacecraft cabin by proposing an unsupervised KMeans clustering machine learning model and model evaluation criteria of average error and maximum error. Combined with the measuring point classification method of the sensor pasting position, it can achieve cutting of more than 60% of the sensors. In addition, by proposing a supervised machine learning model, the data structure of the model training is clarified, which can achieve accurate prediction of the response at the cut measuring points of the spacecraft in subsequent tests under decoupling from the mechanical direction and response direction of the spacecraft, and the model prediction efficiency is significantly improved.

[0086] Embodiment 2

[0087] Based on the same inventive structure, the present invention also provides a spacecraft environmental test sensor cutting device; see Figure 3 The device includes a testing unit 601, a dividing unit 602, a first executing unit 603, a first calculating unit 604, a second executing unit 605, a second calculating unit 606 and a determining unit 607, wherein:

[0088] The testing unit 601 is used to paste sensors inside and outside the cabin of the preceding test spacecraft in the spacecraft to be tested, and perform environmental tests on the preceding test spacecraft to obtain response data of the preceding test spacecraft; the classification unit 602 is used to classify the spacecraft measurement points according to the sensor pasting positions; wherein the spacecraft mechanical measurement point types include cabin measurement points, mandatory measurement points outside the cabin, and optional measurement points outside the cabin; the first execution unit 603 is used to cut off the corresponding cabin measurement points and mandatory measurement points outside the cabin in the subsequent test spacecraft in the spacecraft to be tested; the first calculation unit 604 is used to establish a KMeans clustering model with a preset number of retained measurement points as the number of classifications; the second execution unit 605 is used to The optional extravehicular measuring points are trimmed by retaining one measuring point in each category through the KMeans clustering model; the second computing unit 606 is used to establish a spacecraft response prediction model based on machine learning, and the response data of some spacecraft in the preceding test spacecraft are used as a test set, and the response data of the remaining spacecraft in the preceding test spacecraft are used as a training set; the determining unit 607 is used to train the spacecraft response prediction model based on the training set, and use the average error and maximum error of the output measuring point response prediction in the test set as model evaluation parameters, and determine the retained measuring points of the subsequent test spacecraft that meet the preset errors by adjusting the preset number of retained measuring points.

[0089] The specific processing methods of the functions to be implemented by each unit of this embodiment refer to the spacecraft environmental test sensor tailoring method provided in the above-mentioned first embodiment, which will not be repeated here.

[0090] Embodiment 3

[0091] See also Figure 4 The present invention further provides a computer device, the computer device comprising:

[0092] A memory 701, configured to store instructions executed by at least one processor;

[0093] The processor 702 is configured to execute instructions stored in the memory 701. Figure 1 The method shown.

[0094] Embodiment 4

[0095] A computer-readable storage medium is also provided, which stores one or more programs. When the one or more programs are executed by a processor, they implement any of the above-mentioned spacecraft environmental test sensor tailoring methods.

[0096] In summary, the embodiments provided by the present invention can effectively solve the technical problems in the prior art that most of the measurement points in the cabin are trimmed, there are a large number of external measurement points that affect the life of the spacecraft thermal control measures, and the response at the trimmed measurement points is unknown and there is a risk of excessive response.

[0097] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0098] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0099] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0100] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0101] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims

1. A method for tailoring a spacecraft environmental test sensor, characterized in that: Includes steps: affixing sensors inside and outside a cabin of a preceding test spacecraft among the spacecraft to be tested, and performing an environmental test on the preceding test spacecraft to obtain response data of the preceding test spacecraft; Classifying the spacecraft measurement points according to the sensor pasting positions; wherein the types of spacecraft measurement points include in-cabin measurement points, mandatory out-cabin measurement points, and optional out-cabin measurement points; Cutting out the in-cabin measuring points and the out-cabin necessary measuring points corresponding to the subsequent test spacecraft in the spacecraft to be tested; Establish a KMeans clustering model with the preset number of retained measurement points as the number of classifications; The optional measuring points outside the cabin are trimmed by retaining one measuring point in each category through the KMeans clustering model; Establishing a spacecraft response prediction model based on machine learning, using the response data of some spacecraft in the preceding test spacecraft as a test set, and using the response data of the remaining spacecraft in the preceding test spacecraft as a training set; The spacecraft response prediction model is trained based on the training set, and the average error and maximum error of the output measurement point response prediction in the test set are used as model evaluation parameters. The reserved measurement points of the subsequent test spacecraft that meet the preset errors are determined by adjusting the preset number of reserved measurement points.

2. The method for cutting a spacecraft environmental test sensor according to claim 1, characterized in that: The method of attaching sensors inside and outside the cabin of the preceding test spacecraft in the spacecraft to be tested comprises: If the sensor is a mechanical sensor, the sensor is pasted at the root of all equipment in the cabin, the root of rigid equipment outside the cabin, the root of flexible equipment outside the cabin and the maximum response point of the preceding test spacecraft of the spacecraft to be tested; wherein the rigid equipment is a device whose fundamental frequency is not less than the upper limit of the mechanical test frequency, and the flexible equipment is a device whose fundamental frequency is less than the upper limit of the mechanical test frequency; If the sensor is a thermal sensor, the sensor is pasted on the cabin surface corresponding to the ends of the heat pipe, the root of the high heat consumption equipment, and the highest temperature point in the high heat consumption equipment of the preceding test spacecraft of the spacecraft to be tested.

3. The method for cutting a spacecraft environmental test sensor according to claim 1, characterized in that: The data structure input to the KMeans clustering model takes each optional measuring point outside the cabin as a sample, and each sample is formed by splicing data within boundary conditions under all test conditions of all the previous test spacecraft.

4. The method for cutting a spacecraft environmental test sensor according to claim 3, characterized in that: The data structure of the KMeans clustering model input is a matrix of m rows and (a×d×s×n) columns; wherein m is the number of optional measuring points outside the cabin, a is the number of tests within the boundary conditions in the test, s is the number of test conditions for each of the preceding test spacecraft, n is the number of preceding test spacecraft, and d is the number of measurement groups of the sensor at the measuring points.

5. The method for cutting a spacecraft environmental test sensor according to claim 1, characterized in that: The spacecraft response prediction model is trained based on a machine learning algorithm of a random forest or a neural network.

6. The method for cutting a spacecraft environmental test sensor according to claim 1, characterized in that: The data structure of the test set uses each response data within the boundary conditions under all test conditions of some spacecraft in the previous test spacecraft as a sample, and the responses corresponding to all retained measurement points are used as model inputs, and the responses corresponding to the clipped measurement points are used as model outputs; The data structure of the training set takes each response data within the boundary conditions under all test conditions relative to the remaining spacecraft in the previous test spacecraft as a sample, and the responses corresponding to all retained measurement points are used as model inputs, and the responses corresponding to the cropped measurement points are used as model outputs.

7. The method for cutting a spacecraft environmental test sensor according to claim 1, characterized in that: The average error and the maximum error are calculated based on the following formula: Among them, Error1 and Error2 are the average error and the maximum error respectively, a T and a P are the spacecraft response test value and the predicted value, respectively, and are both (a×s)-row d×(b+c)-column matrices; a is the number of tests within the boundary conditions in the test, s is the number of test conditions for each of the preceding test spacecraft, n is the number of preceding test spacecraft, d is the number of measurement groups of the sensor at the measuring point; b is the number of retained side points, and c is the number of cropped measuring points.

8. A spacecraft environmental test sensor cutting device, characterized in that: Included are: The test unit is used to attach sensors inside and outside the cabin of a previous test spacecraft in the spacecraft to be tested, and perform environmental tests on the previous test spacecraft to obtain response data of the previous test spacecraft. A classification unit, used for classifying the spacecraft measurement points according to the sensor pasting position; wherein the spacecraft measurement point types include in-cabin measurement points, out-cabin mandatory measurement points and out-cabin optional measurement points; A first execution unit is used to cut out the in-cabin measurement points and the out-cabin must-cut measurement points corresponding to the subsequent test spacecraft in the spacecraft to be tested; The first computing unit is used to establish a KMeans clustering model with a preset number of retained measurement points as the number of classifications; A second execution unit is used to trim the optional measuring points outside the cabin by retaining one measuring point in each category through the KMeans clustering model; a second computing unit, configured to establish a spacecraft response prediction model based on machine learning, using the response data of some spacecraft in the preceding test spacecraft as a test set, and the response data of the remaining spacecraft in the preceding test spacecraft as a training set; A determination unit is used to train the spacecraft response prediction model based on the training set, and use the average error and maximum error of the output measurement point response prediction in the test set as model evaluation parameters, and adjust the preset number of retained measurement points to determine the retained measurement points of the subsequent test spacecraft that meet the preset errors.

9. A computer device, characterized in that: include: a memory for storing instructions executed by at least one processor; A processor, configured to execute instructions stored in a memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores one or more programs, and when the one or more programs are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images