A rocket attitude prediction method and system

Abstract

The present application relates to the field of aerospace technology, and in particular to a method and system for predicting rocket attitude, including: collecting rocket characteristic data at the current moment, and preprocessing the rocket characteristic data at the current moment; inputting the preprocessed rocket characteristic data into a rocket attitude prediction model to predict the rocket attitude at the next moment; correcting the prediction result of the rocket attitude at the next moment according to the accuracy of the rocket attitude comprehensive evaluation value at the previously predetermined moment; and adjusting the rocket attitude at the next moment according to the corrected prediction result of the rocket attitude at the next moment. The present application can improve the accuracy of rocket attitude prediction under the premise of meeting the real-time requirements.

Description

Technical Field

[0001] The present application relates to the field of aerospace technology, and in particular to a method and system for predicting a rocket attitude. Background Art

[0002] Rocket attitude refers to the position and direction of a rocket in space. Accurately predicting rocket attitude is crucial to ensure that the rocket flies on the intended orbit and successfully completes the launch mission (such as sending a satellite into the correct orbit, achieving precise pointing of a space probe, etc.). Rocket attitude is usually described by several angles, such as pitch angle (the up and down angle of the rocket head relative to the horizontal plane), yaw angle (the angle of the rocket head deviating left and right from the horizontal plane), and roll angle (the angle of the rocket rotating around its own axis).

[0003] The rocket attitude prediction needs to be performed in real time during the rocket flight in order to adjust the rocket's attitude control strategy in a timely manner. However, real-time prediction requires completing complex computing tasks within a limited time, which requires the rocket to have a large amount of computing resources. For example, some high-precision fluid dynamics calculation methods are used to calculate the impact of aerodynamics on attitude. The amount of calculation is very large, but due to the limitations of weight, power consumption and other factors, the computing power of the computing equipment on the rocket is relatively limited, so it is difficult to accurately predict the rocket attitude in the real-time environment of the rocket flight.

[0004] Therefore, how to improve the accuracy of rocket attitude prediction while meeting real-time requirements is a technical problem that technical personnel in this field urgently need to solve. Summary of the invention

[0005] The present application provides a method and system for predicting a rocket attitude, so as to improve the accuracy of rocket attitude prediction while meeting real-time requirements.

[0006] In order to solve the above technical problems, this application provides the following technical solutions:

[0007] A method for predicting a rocket attitude comprises the following steps: step S110, collecting rocket characteristic data at the current moment, and preprocessing the rocket characteristic data at the current moment; step S120, inputting the preprocessed rocket characteristic data into a rocket attitude prediction model, and predicting the rocket attitude at the next moment; step S130, correcting the prediction result of the rocket attitude at the next moment according to the accuracy of the comprehensive evaluation value of the rocket attitude at the previously predetermined moment; step S140, adjusting the rocket attitude at the next moment according to the corrected prediction result of the rocket attitude at the next moment.

[0008] In the method for predicting the rocket attitude as described above, preferably, the preprocessing includes: eliminating abnormal rocket characteristic data and unifying the magnitude of normal rocket characteristic data to obtain normal rocket characteristic data with unified magnitude.

[0009] In the method for predicting the rocket attitude as described above, preferably, eliminating abnormal rocket characteristic data includes the following steps: obtaining the launch phase of the rocket at the current moment according to the spatial position of the rocket at the current moment and the predetermined trajectory of the rocket; retrieving the normal screening range of each type of rocket characteristic data in the launch phase, and judging whether each type of rocket characteristic data collected at the current moment is within the normal screening range of the corresponding rocket characteristic data; if it is within the normal screening range of the corresponding rocket characteristic data, judging the rocket characteristic data at the current moment to be normal rocket characteristic data, otherwise judging the rocket characteristic data at the current moment to be abnormal rocket characteristic data, and eliminating the abnormal rocket characteristic data.

[0010] The method for predicting the rocket attitude as described above, wherein preferably, the training of the rocket attitude prediction model includes the following steps: step S121, collecting the rocket feature data of each historical moment, and constructing a historical rocket feature data training set; step S122, collecting the rocket attitude comprehensive evaluation value of each historical moment, and constructing a historical rocket attitude comprehensive evaluation value training set; step S123, constructing a rocket attitude prediction model, and training the rocket attitude prediction model through the historical rocket feature data training set and the historical rocket attitude comprehensive evaluation value training set.

[0011] In the rocket attitude prediction method as described above, it is preferred that before constructing the historical rocket characteristic data training set and the historical rocket attitude comprehensive evaluation value training set, the magnitudes of the historical rocket characteristic data and the historical rocket attitude comprehensive evaluation values ​​should be unified.

[0012] A rocket attitude prediction system comprises: a data acquisition unit, a preprocessing unit, a prediction input unit, a correction unit and an attitude adjustment unit; the data acquisition unit acquires rocket characteristic data at the current moment, and the preprocessing unit preprocesses the rocket characteristic data at the current moment; the prediction input unit inputs the preprocessed rocket characteristic data into a rocket attitude prediction model to predict the rocket attitude at the next moment; the correction unit corrects the prediction result of the rocket attitude at the next moment according to the accuracy of the rocket attitude comprehensive evaluation value at the previously predetermined moment; the attitude adjustment unit adjusts the rocket attitude at the next moment according to the corrected prediction result of the rocket attitude at the next moment.

[0013] In the rocket attitude prediction system as described above, preferably, the pre-processing unit eliminates abnormal rocket characteristic data and unifies the magnitude of normal rocket characteristic data to obtain normal rocket characteristic data with unified magnitude.

[0014] In the rocket attitude prediction system as described above, preferably, the pre-processing unit obtains the launch phase of the rocket at the current moment according to the spatial position of the rocket at the current moment and the predetermined trajectory of the rocket; the pre-processing unit retrieves the normal screening range of each rocket characteristic data of the launch phase, and determines whether each rocket characteristic data collected at the current moment is within the normal screening range of the corresponding rocket characteristic data; if it is within the normal screening range of the corresponding rocket characteristic data, the rocket characteristic data at the current moment is determined to be normal rocket characteristic data; otherwise, the rocket characteristic data at the current moment is determined to be abnormal rocket characteristic data, and the abnormal rocket characteristic data is discarded.

[0015] The rocket attitude prediction system as described above, wherein preferably, the rocket attitude prediction model is obtained by training the rocket attitude prediction model training system, and the rocket attitude prediction model training system comprises: a training data collection unit, a model construction unit and a model training unit; the training data collection unit collects rocket feature data at each historical moment and constructs a historical rocket feature data training set; the training data collection unit collects the rocket attitude comprehensive evaluation value at each historical moment and constructs a historical rocket attitude comprehensive evaluation value training set; the model construction unit constructs the rocket attitude prediction model, and the model training unit trains the rocket attitude prediction model through the historical rocket feature data training set and the historical rocket attitude comprehensive evaluation value training set.

[0016] The rocket attitude prediction system as described above, wherein preferably, the rocket attitude prediction model training system also includes: a training data preprocessing unit; before constructing the historical rocket feature data training set and the historical rocket attitude comprehensive evaluation value training set, the training data preprocessing unit must unify the magnitude of the historical rocket feature data and the historical rocket attitude comprehensive evaluation value.

[0017] Compared with the above-mentioned background technology, the rocket attitude prediction in the present application applies a rocket attitude prediction model that has been trained with a large amount of data on the ground. Therefore, the present application can meet the real-time requirements of rocket attitude prediction, and the rocket attitude prediction model is trained with a large amount of data, so it can ensure the accuracy of rocket attitude prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0019] Figure 1is a flow chart of a method for predicting a rocket attitude provided in an embodiment of the present application;

[0020] Figure 2 is a flow chart of the rocket attitude prediction model training provided in the embodiment of the present application;

[0021] Figure 3 is a schematic diagram of a rocket attitude prediction system provided in an embodiment of the present application;

[0022] Figure 4 It is a schematic diagram of a rocket attitude prediction model training system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. In addition, spatial relationship terms such as "upper", "lower", "front", "back", "left" and "right" are used to facilitate the description to explain the positional relationship between two components. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention. Embodiment 1

[0024] like Figure 1 As shown, the present application provides a method for predicting a rocket attitude, comprising the following steps:

[0025] Step S110, collecting the rocket characteristic data at the current moment, and preprocessing the rocket characteristic data at the current moment.

[0026] After the rocket is launched, various sensors on the rocket will collect rocket characteristic data in real time, including: rocket attitude data, such as pitch angle, yaw angle, roll angle, etc., and rocket status data, such as: engine thrust data (thrust size, direction, change law, etc.), aerodynamic data (aerodynamic coefficients determined by flight speed, altitude, angle of attack, sideslip angle, etc.), environmental data (atmospheric density, temperature, solar radiation intensity, etc.), etc. Among them, the rocket status data is the data that will affect the rocket's attitude.

[0027] The rocket's flight control system will pre-process the rocket's characteristic data (rocket attitude data and rocket status data) at the current moment to eliminate abnormal rocket characteristic data and unify the magnitude of normal rocket characteristic data, thereby obtaining normal rocket characteristic data after magnitude unification, avoiding the influence of abnormal rocket characteristic data on the subsequent rocket attitude prediction, and avoiding the influence of rocket characteristic data of different magnitudes on the accuracy and speed of rocket attitude prediction.

[0028] Specifically, the rocket's flight control system will determine the launch phase of the rocket at the current moment based on the rocket's current spatial position and the rocket's planned trajectory, such as the first sub-stage of the climb phase, the second sub-stage of the climb phase, the first sub-stage of the orbital entry phase, the second sub-stage of the orbital entry phase, etc. Specifically, during the rocket's design phase, the rocket's planned trajectory will be pre-divided into Launch phases, each launch phase is pre-divided into In the sub-stage, while collecting the rocket's characteristic data at the current moment, various sensors on the rocket also collect the rocket's position data at the current moment. The rocket's flight control system will compare the rocket's position data at the current moment with the predetermined trajectory. The launch phase By comparing the sub-stages, the launch stage that falls into the rocket position data at the current moment is determined as the launch stage that the rocket is in at the current moment.

[0029] The rocket's flight control system will call up the normal screening range of each type of rocket characteristic data in the launch phase, and determine whether each type of rocket characteristic data collected at the current moment is within the normal screening range of the corresponding rocket characteristic data. If it is within the normal screening range of the corresponding rocket characteristic data, then the rocket characteristic data at the current moment is determined to be normal rocket characteristic data; otherwise, the rocket characteristic data at the current moment is determined to be abnormal rocket characteristic data, indicating that the data obviously does not conform to the laws of physics and must be eliminated.

[0030] in, For the The first launch phase The sub-stage Normal data screening range for the characteristic data of this type of rocket; For the The first launch phase The sub-stage The normal average value of the characteristic data of the rocket; For the The first launch phase The sub-stage Normal standard deviation of the characteristic data of the rocket; is the adjustment coefficient, ; , ;

[0031] For the In the first launch The first launch phase The sub-stage rocket characteristic data, is the total number of launches, For the The adjustment value of the next emission relative to the current emission;

[0032] ;

[0033] The current rocket launch design parameters, For the The first rocket launch design parameters, which may be weight parameters, trajectory parameters, etc. of the rocket; For the The weight coefficients of the design parameters; is the number of design parameters; is the type parameter value of the current rocket launch, For the The type parameter value of the launch rocket.

[0034] After eliminating abnormal rocket characteristic data, the magnitude of normal rocket characteristic data is unified to avoid excessive differences in data ranges between rocket characteristic data of different magnitudes, thereby avoiding deviations in measurement when predicting rocket attitude.

[0035] Specifically, the rocket's flight control system is based on the formula , unify the magnitude of rocket characteristic data;

[0036] in, To obtain the rocket characteristic data after the order of magnitude unification; For rocket characteristic data (rocket attitude data) before the order of magnitude unification and rocket status data); It is the minimum value in the data set composed of the same rocket characteristic data; It is the maximum value in the data set composed of the same rocket characteristic data.

[0037] Step S120: input the preprocessed rocket characteristic data into the rocket attitude prediction model to predict the rocket attitude at the next moment.

[0038] After preprocessing the rocket characteristic data at the current moment, eliminating abnormal rocket characteristic data and unifying the magnitude, the rocket's flight control system will send the preprocessed rocket characteristic data (rocket attitude data and rocket status data ) is input into the rocket attitude prediction model, thereby obtaining the corresponding output from the rocket attitude prediction model , and use this output as the prediction result of the rocket attitude at the next moment. For the current moment The first rocket attitude data, For the current moment The second rocket attitude data, For the current moment No. Rocket attitude data, For the current moment The first rocket status data, For the current moment The second rocket status data, For the current moment No. Rocket status data, For the next moment The comprehensive evaluation value of the rocket attitude.

[0039] Among them, the rocket attitude prediction model is obtained by training a large amount of historical rocket feature data and historical rocket attitude comprehensive evaluation values. Figure 2 As shown, the following steps are included:

[0040] Step S121: Collect rocket feature data at each historical moment and construct a historical rocket feature data training set.

[0041] Each time a rocket is launched, the rocket characteristic data (rocket attitude data and rocket status data) during the launch process will be recorded in the computer device. Therefore, the computer device collects the rocket characteristic data (rocket attitude data and rocket status data) recorded from multiple successful rocket launches to construct a historical rocket characteristic data training set. ,

[0042] in, The first rocket attitude data at the first historical moment, The second rocket attitude data for the first historical moment, The first rocket status data at the first historical moment, The status data of the second rocket at the first historical moment; The first rocket attitude data at the second historical moment, The second rocket attitude data at the second historical moment, The first rocket status data at the second historical moment, This is the 2nd rocket status data for the 2nd historical moment.

[0043] Before constructing the historical rocket feature data training set, the magnitude of the historical rocket feature data must be unified (as above). The data with unified magnitude can make the model training process more stable and efficient, speed up the model convergence speed, and reduce training time.

[0044] Step S122: Collect the comprehensive evaluation value of the rocket attitude at each historical moment and construct a training set of the comprehensive evaluation values ​​of the historical rocket attitude.

[0045] Each time a rocket is launched, the computer device will record the comprehensive evaluation value of the rocket attitude during the launch process. The comprehensive evaluation value of the rocket attitude can generally indicate the attitude of the rocket. Therefore, the computer device collects the comprehensive evaluation values ​​of the rocket attitude recorded by multiple successful rocket launches, thereby constructing a historical rocket attitude comprehensive evaluation value training set. ,in, is the comprehensive evaluation value of the rocket's posture at the first historical moment, This is the comprehensive evaluation value of the rocket's posture at the second historical moment.

[0046] Before constructing the training set of historical rocket attitude comprehensive evaluation values, the magnitude of the historical rocket attitude comprehensive evaluation values ​​must be unified (as above). The data with unified magnitude can also make the model training process more stable and efficient, speed up the convergence of the model, and reduce the training time.

[0047] Step S123: construct a rocket attitude prediction model, and train the rocket attitude prediction model using a historical rocket feature data training set and a historical rocket attitude comprehensive evaluation value training set.

[0048] The rocket attitude prediction model constructed by the computer equipment is as follows:

[0049] ,

[0050] in, For the moment Comprehensive evaluation value of rocket attitude; For the moment No. rocket attitude data; is the number of rocket attitude data; is the weight of rocket attitude data on rocket attitude prediction; For the moment No. Rocket status data; is the number of rocket status data; is the weight of rocket status data on rocket attitude prediction; is the weight from the input layer to the hidden layer; is the bias corresponding to the input layer to the hidden layer; is the weight from the hidden layer to the output layer; is the offset from the hidden layer to the output layer; is the activation function, .

[0051] The computer device concentrates historical rocket characteristic data training, and uses the rocket attitude data and rocket status data of each historical moment as the input of the rocket attitude prediction model. The computer device concentrates historical rocket attitude comprehensive evaluation value training, and uses the rocket attitude comprehensive evaluation value of the next historical moment as the output of the rocket attitude prediction model, thereby training the rocket attitude prediction model. The trained rocket attitude prediction model can then be configured in the rocket's flight control system, so that the rocket's flight control system can use the trained rocket attitude prediction model to predict the rocket's attitude.

[0052] Step S130: Correct the prediction result of the rocket attitude at the next moment according to the accuracy of the comprehensive evaluation value of the rocket attitude at the previously predetermined moment.

[0053] After the rocket is launched, the rocket's flight control system will continuously predict the rocket's attitude comprehensive evaluation value at different times, and will verify the accuracy of the predicted rocket attitude comprehensive evaluation value at each moment in the future. Specifically, the accuracy of the predicted rocket attitude comprehensive evaluation value at each moment is verified based on the difference between the actual rocket attitude comprehensive evaluation value that occurs next and the previously predicted rocket attitude comprehensive evaluation value. Therefore, when the rocket's flight control system obtains the next moment Comprehensive evaluation value of rocket attitude After that, the rocket's flight control system will make a decision on the next moment based on the accuracy of the comprehensive evaluation value of the rocket's attitude at the previously scheduled moment. Comprehensive evaluation value of rocket attitude Make corrections.

[0054] Specifically, the rocket's flight control system is based on the formula ,

[0055] For the next moment Comprehensive evaluation value of rocket attitude Make corrections; among them, The next moment after correction Comprehensive evaluation value of rocket attitude; For the moment The accuracy of the comprehensive evaluation value of rocket attitude; For the moment The weight of the rocket attitude comprehensive evaluation value accuracy is , Is a positive integer.

[0056] Step S140: adjusting the rocket attitude at the next moment according to the corrected prediction result of the rocket attitude at the next moment.

[0057] The rocket's flight control system has a preset standard comprehensive evaluation value for the rocket's attitude at each moment. After obtaining the corrected comprehensive evaluation value of the rocket's attitude at the next moment, the rocket's flight control system will compare the corrected comprehensive evaluation value of the rocket's attitude at the next moment with the standard comprehensive evaluation value of the rocket's attitude at the next moment. If the difference exceeds the threshold, the rocket's attitude at the next moment must be adjusted in advance. If the difference does not exceed the threshold, the rocket's attitude at the next moment does not need to be adjusted.

[0058] Embodiment 2

[0059] like Figure 3 As shown, the present application provides a rocket attitude prediction system 300, including: a data acquisition unit 310, a preprocessing unit 320, a prediction input unit 330, a correction unit 340 and an attitude adjustment unit 350.

[0060] The data acquisition unit 310 acquires the rocket characteristic data at the current moment, and the preprocessing unit 320 preprocesses the rocket characteristic data at the current moment.

[0061] After the rocket is launched, the data acquisition unit 310 (various sensors on the rocket) will collect rocket characteristic data in real time, including: rocket attitude data, such as pitch angle, yaw angle, roll angle, etc., and rocket status data, such as: engine thrust data (thrust size, direction, change law, etc.), aerodynamic data (aerodynamic coefficients determined by flight speed, altitude, angle of attack, sideslip angle, etc.), environmental data (atmospheric density, temperature, solar radiation intensity, etc.), etc., among which the rocket status data is the data that will affect the rocket attitude.

[0062] The preprocessing unit 320 will preprocess the rocket characteristic data (rocket attitude data and rocket status data) at the current moment to eliminate abnormal rocket characteristic data and unify the magnitude of normal rocket characteristic data, thereby obtaining normal rocket characteristic data after magnitude unification, avoiding the influence of abnormal rocket characteristic data on the subsequent rocket attitude prediction, and avoiding the influence of rocket characteristic data of different magnitudes on the accuracy and speed of rocket attitude prediction.

[0063] Specifically, the preprocessing unit 320 will obtain the launch phase of the rocket at the current moment according to the space position of the rocket at the current moment and the rocket's predetermined trajectory, for example: the first sub-stage of the climbing phase, the second sub-stage of the climbing phase, the first sub-stage of the orbiting phase, the second sub-stage of the orbiting phase, etc. Specifically, in the design stage of the rocket, the rocket's predetermined trajectory will be pre-divided into Launch phases, each launch phase is pre-divided into In the sub-stage, when the data acquisition unit 310 collects the rocket characteristic data at the current moment, the data acquisition unit 310 also collects the rocket position data at the current moment, and the pre-processing unit 320 compares the rocket position data at the current moment with the predetermined trajectory. The launch phase By comparing the sub-stages, the launch stage that falls into the rocket position data at the current moment is determined as the launch stage that the rocket is in at the current moment.

[0064] The pre-processing unit 320 will retrieve the normal screening range of each type of rocket characteristic data in the launch phase, and determine whether each type of rocket characteristic data collected at the current moment is within the normal screening range of the corresponding rocket characteristic data. If it is within the normal screening range of the corresponding rocket characteristic data, then the rocket characteristic data at the current moment is determined to be normal rocket characteristic data; otherwise, the rocket characteristic data at the current moment is determined to be abnormal rocket characteristic data, indicating that the data obviously does not conform to the laws of physics and these data must be eliminated.

[0065] in, For the The first launch phase The sub-stage Normal data screening range for the characteristic data of this type of rocket; For the The first launch phase The sub-stage The normal average value of the characteristic data of the rocket; For the The first launch phase The sub-stage Normal standard deviation of the characteristic data of the rocket; is the adjustment coefficient, ; , ;

[0066] For the In the first launch The first launch phase The sub-stage rocket characteristic data, is the total number of launches, For the The adjustment value of the next emission relative to the current emission;

[0067] ;

[0068] The current rocket launch design parameters, For the The first rocket launch design parameters, which may be weight parameters, trajectory parameters, etc. of the rocket; For the The weight coefficients of the design parameters; is the number of design parameters; is the type parameter value of the current rocket launch, For the The type parameter value of the launch rocket.

[0069] After eliminating abnormal rocket characteristic data, the preprocessing unit 320 will also unify the magnitude of normal rocket characteristic data to avoid excessive differences in data ranges between rocket characteristic data of different magnitudes, thereby avoiding deviations in measurement when predicting rocket attitude.

[0070] Specifically, the preprocessing unit 320 uses the formula , unify the magnitude of rocket characteristic data;

[0071] in, To obtain the rocket characteristic data after the order of magnitude unification; For rocket characteristic data (rocket attitude data) before the order of magnitude unification and rocket status data); It is the minimum value in the data set composed of the same rocket characteristic data; It is the maximum value in the data set composed of the same rocket characteristic data.

[0072] The prediction input unit 330 inputs the preprocessed rocket feature data into the rocket attitude prediction model to predict the rocket attitude at the next moment.

[0073] After preprocessing the rocket feature data at the current moment, removing abnormal rocket feature data and unifying the magnitude, the prediction input unit 330 will convert the preprocessed rocket feature data (rocket attitude data) into and rocket status data ) is input into the rocket attitude prediction model, thereby obtaining the corresponding output from the rocket attitude prediction model , and use this output as the prediction result of the rocket attitude at the next moment. For the current moment The first rocket attitude data, For the current moment The second rocket attitude data, For the current moment No. Rocket attitude data, For the current moment The first rocket status data, For the current moment The second rocket status data, For the current moment No. Rocket status data, For the next moment The comprehensive evaluation value of the rocket attitude.

[0074] The rocket attitude prediction model is obtained by training the rocket attitude prediction model training system 400 through a large amount of historical rocket feature data and historical rocket attitude comprehensive evaluation values. Figure 4 As shown, the rocket attitude prediction model training system 400 includes: a training data collection unit 410, a model construction unit 420 and a model training unit 430.

[0075] The training data collection unit 410 collects rocket feature data at each historical moment and constructs a historical rocket feature data training set.

[0076] Each time a rocket is launched, the rocket characteristic data (rocket attitude data and rocket status data) during the launch process will be recorded in the computer device. Therefore, the training data collection unit 410 collects the rocket characteristic data (rocket attitude data and rocket status data) recorded from multiple successful rocket launches, thereby constructing a historical rocket characteristic data training set. ,

[0077] in, The first rocket attitude data at the first historical moment, The second rocket attitude data for the first historical moment, The first rocket status data at the first historical moment, The status data of the second rocket at the first historical moment; The first rocket attitude data at the second historical moment, The second rocket attitude data at the second historical moment, The first rocket status data at the second historical moment, This is the 2nd rocket status data for the 2nd historical moment.

[0078] The training data collection unit 410 collects the comprehensive evaluation value of the rocket attitude at each historical moment and constructs a training set of historical rocket attitude comprehensive evaluation values.

[0079] Each time a rocket is launched, the computer device will record the comprehensive evaluation value of the rocket attitude during the launch process. The comprehensive evaluation value of the rocket attitude can generally indicate the attitude of the rocket. Therefore, the computer device collects the comprehensive evaluation values ​​of the rocket attitude recorded by multiple successful rocket launches, thereby constructing a historical rocket attitude comprehensive evaluation value training set. ,in, is the comprehensive evaluation value of the rocket's posture at the first historical moment, This is the comprehensive evaluation value of the rocket's posture at the second historical moment.

[0080] On the basis of the above, the rocket attitude prediction model training system 400 also includes: a training data preprocessing unit 440; before constructing the historical rocket feature data training set and the historical rocket attitude comprehensive evaluation value training set, the training data preprocessing unit 440 must unify the magnitudes of the historical rocket feature data and the historical rocket attitude comprehensive evaluation values ​​(as above). The data with unified magnitudes can make the model training process more stable and efficient, accelerate the convergence speed of the model, and reduce the training time.

[0081] The model building unit 420 builds a rocket attitude prediction model, and the model training unit 430 trains the rocket attitude prediction model through a historical rocket feature data training set and a historical rocket attitude comprehensive evaluation value training set.

[0082] The rocket attitude prediction model constructed by the model construction unit 420 is as follows:

[0083] ,

[0084] in, For the moment Comprehensive evaluation value of rocket attitude; For the moment No. rocket attitude data; is the number of rocket attitude data; is the weight of rocket attitude data on rocket attitude prediction; For the moment No. Rocket status data; is the number of rocket status data; is the weight of rocket status data on rocket attitude prediction; is the weight from the input layer to the hidden layer; is the bias corresponding to the input layer to the hidden layer; is the weight from the hidden layer to the output layer; is the offset from the hidden layer to the output layer; is the activation function, .

[0085] The model training unit 430 collects historical rocket feature data into a training set, and uses the rocket attitude data and rocket status data at each historical moment as the input of the rocket attitude prediction model. The model training unit 430 collects historical rocket attitude comprehensive evaluation value training sets, and uses the rocket attitude comprehensive evaluation value at the next historical moment as the output of the rocket attitude prediction model, thereby training the rocket attitude prediction model. The trained rocket attitude prediction model can then be configured in the rocket attitude prediction system 300, so that the rocket attitude prediction system 300 can use the trained rocket attitude prediction model to predict the rocket attitude.

[0086] The correction unit 340 corrects the prediction result of the rocket attitude at the next moment according to the accuracy of the comprehensive evaluation value of the rocket attitude at the previously predetermined moment.

[0087] After the rocket is launched, the prediction input unit 330 will continuously input the pre-processed rocket feature data into the rocket attitude prediction model, so as to predict the comprehensive evaluation value of the rocket attitude at different times, and the rocket attitude prediction system 300 will then verify the accuracy of the predicted comprehensive evaluation value of the rocket attitude at each moment, specifically according to the difference between the actual comprehensive evaluation value of the rocket attitude that occurs next and the comprehensive evaluation value of the rocket attitude that is predicted before, to verify the accuracy of the predicted comprehensive evaluation value of the rocket attitude at each moment, so as to obtain the next moment Comprehensive evaluation value of rocket attitude After that, the correction unit 340 will adjust the next moment according to the accuracy of the rocket attitude comprehensive evaluation value at the previous predetermined moment. Comprehensive evaluation value of rocket attitude Make corrections.

[0088] Specifically, the correction unit 340 is based on the formula ,

[0089] For the next moment Comprehensive evaluation value of rocket attitude Make corrections; among them, The next moment after correction Comprehensive evaluation value of rocket attitude; For the moment The accuracy of the comprehensive evaluation value of rocket attitude; For the moment The weight of the rocket attitude comprehensive evaluation value accuracy is , Is a positive integer.

[0090] The attitude adjustment unit 350 adjusts the rocket attitude at the next moment according to the corrected prediction result of the rocket attitude at the next moment.

[0091] In the rocket attitude prediction system 300, a standard comprehensive evaluation value of the rocket attitude at each moment is preset. After obtaining the corrected comprehensive evaluation value of the rocket attitude at the next moment, the attitude adjustment unit 350 compares the corrected comprehensive evaluation value of the rocket attitude at the next moment with the standard comprehensive evaluation value of the rocket attitude at the next moment. If the difference exceeds the threshold, the attitude adjustment unit 350 will pre-adjust the rocket attitude at the next moment. If the difference does not exceed the threshold, the attitude adjustment unit 350 does not need to adjust the rocket attitude at the next moment.

[0092] The rocket attitude prediction in this application applies a rocket attitude prediction model that has been trained with a large amount of data on the ground. Therefore, this application can meet the real-time requirements of rocket attitude prediction, and the rocket attitude prediction model is trained with a large amount of data, so it can ensure the accuracy of rocket attitude prediction.

[0093] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0094] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for predicting rocket attitude, characterized in that: The steps include: Step S110, collecting rocket characteristic data at the current moment, and preprocessing the rocket characteristic data at the current moment; Step S120, inputting the preprocessed rocket characteristic data into the rocket attitude prediction model to predict the rocket attitude at the next moment; Step S130, correcting the prediction result of the rocket attitude at the next moment according to the accuracy of the comprehensive evaluation value of the rocket attitude at the previously predetermined moment; According to the formula , for the next moment Comprehensive evaluation value of rocket attitude Make corrections; among them, The next moment after correction Comprehensive evaluation value of rocket attitude; For the moment The accuracy of the comprehensive evaluation value of rocket attitude; For the moment The weight of the rocket attitude comprehensive evaluation value accuracy, value , is a positive integer; Step S140, adjusting the rocket attitude at the next moment according to the corrected prediction result of the rocket attitude at the next moment; The training of the rocket attitude prediction model includes the following steps: Step S121, collecting rocket feature data at each historical moment, and constructing a historical rocket feature data training set; Step S122, collecting the rocket attitude comprehensive evaluation value at each historical moment, and constructing a historical rocket attitude comprehensive evaluation value training set; Step S123, constructing a rocket attitude prediction model, and training the rocket attitude prediction model through a historical rocket feature data training set and a historical rocket attitude comprehensive evaluation value training set; The constructed rocket attitude prediction model is as follows: ； in, For the moment Comprehensive evaluation value of rocket attitude; For the moment No. Rocket attitude data; is the number of rocket attitude data; is the weight of rocket attitude data on rocket attitude prediction; For the moment No. Rocket status data; is the number of rocket status data; is the weight of rocket status data on rocket attitude prediction; is the weight from the input layer to the hidden layer; is the bias corresponding to the input layer to the hidden layer; is the weight from the hidden layer to the output layer; is the offset from the hidden layer to the output layer; is the activation function, .

2. The method for predicting rocket attitude according to claim 1, characterized in that: The preprocessing includes: eliminating abnormal rocket characteristic data and unifying the magnitude of normal rocket characteristic data to obtain normal rocket characteristic data after the magnitude is unified.

3. The method for predicting rocket attitude according to claim 2, characterized in that: Eliminating abnormal rocket characteristic data includes the following steps: According to the current spatial position of the rocket and the rocket's predetermined trajectory, the launch phase of the rocket at the current moment is obtained; Retrieve the normal screening range of each rocket characteristic data of the launch phase, and determine whether each rocket characteristic data collected at the current moment is within the normal screening range of the corresponding rocket characteristic data; If it is within the normal screening range of the corresponding rocket characteristic data, then the rocket characteristic data at the current moment is determined to be normal rocket characteristic data; otherwise, the rocket characteristic data at the current moment is determined to be abnormal rocket characteristic data, and the abnormal rocket characteristic data is eliminated.

4. The method for predicting rocket attitude according to claim 1, characterized in that: Before constructing the training set of historical rocket characteristic data and the training set of historical rocket attitude comprehensive evaluation values, the magnitudes of historical rocket characteristic data and historical rocket attitude comprehensive evaluation values ​​must be unified.

5. A rocket attitude prediction system, characterized in that: include: Data acquisition unit, preprocessing unit, prediction input unit, correction unit and posture adjustment unit; The data acquisition unit acquires rocket characteristic data at the current moment, and the preprocessing unit preprocesses the rocket characteristic data at the current moment; The prediction input unit inputs the preprocessed rocket characteristic data into the rocket attitude prediction model to predict the rocket attitude at the next moment; The correction unit corrects the prediction result of the rocket attitude at the next moment according to the accuracy of the comprehensive evaluation value of the rocket attitude at the previously predetermined moment; Correction unit according to the formula , for the next moment Comprehensive evaluation value of rocket attitude Make corrections; among them, The next moment after correction Comprehensive evaluation value of rocket attitude; For the moment The accuracy of the comprehensive evaluation value of rocket attitude; For the moment The weight of the rocket attitude comprehensive evaluation value accuracy, value , is a positive integer; The attitude adjustment unit adjusts the rocket attitude at the next moment according to the corrected prediction result of the rocket attitude at the next moment; The rocket attitude prediction model is obtained by training a rocket attitude prediction model training system, and the rocket attitude prediction model training system includes: a training data collection unit, a model building unit and a model training unit; The training data collection unit collects rocket feature data at each historical moment and constructs a historical rocket feature data training set; The training data collection unit collects the comprehensive evaluation value of the rocket attitude at each historical moment and constructs a training set of the comprehensive evaluation value of the historical rocket attitude; The model building unit builds a rocket attitude prediction model, and the model training unit trains the rocket attitude prediction model through a historical rocket feature data training set and a historical rocket attitude comprehensive evaluation value training set; The rocket attitude prediction model constructed by the model building unit is as follows: ； in, For the moment Comprehensive evaluation value of rocket attitude; For the moment No. Rocket attitude data; is the number of rocket attitude data; is the weight of rocket attitude data on rocket attitude prediction; For the moment No. Rocket status data; is the number of rocket status data; is the weight of rocket status data on rocket attitude prediction; is the weight from the input layer to the hidden layer; is the bias corresponding to the input layer to the hidden layer; is the weight from the hidden layer to the output layer; is the offset from the hidden layer to the output layer; is the activation function, .

6. The rocket attitude prediction system according to claim 5, characterized in that: The preprocessing unit eliminates abnormal rocket characteristic data and unifies the magnitude of normal rocket characteristic data to obtain normal rocket characteristic data after the magnitude is unified.

7. The rocket attitude prediction system according to claim 6, characterized in that: The preprocessing unit obtains the launch phase of the rocket at the current moment according to the spatial position of the rocket at the current moment and the predetermined trajectory of the rocket; The preprocessing unit retrieves the normal screening range of each type of rocket characteristic data in the launch phase, and determines whether each type of rocket characteristic data collected at the current moment is within the normal screening range of the corresponding rocket characteristic data. If it is within the normal screening range of the corresponding rocket characteristic data, the rocket characteristic data at the current moment is determined to be normal rocket characteristic data; otherwise, the rocket characteristic data at the current moment is determined to be abnormal rocket characteristic data, and the abnormal rocket characteristic data is discarded.

8. The rocket attitude prediction system according to claim 5, characterized in that: The rocket attitude prediction model training system also includes: a training data preprocessing unit; before constructing a historical rocket characteristic data training set and a historical rocket attitude comprehensive evaluation value training set, the training data preprocessing unit must unify the magnitude of the historical rocket characteristic data and the historical rocket attitude comprehensive evaluation value.

Images