Satellite telemetry data feature selection method and device based on multi-objective optimization algorithm and medium

Abstract

Embodiments of the present application disclose a satellite telemetry data feature selection method and device based on a multi-objective optimization algorithm and a medium; the method can include: splicing the original telemetry data of the satellite, and screening out constant values in numerical data and time alignment, and based on the Rayleigh criterion, the 53H algorithm and the median filtering algorithm, removing outliers and noise in the original telemetry data to obtain preprocessed telemetry data; improving the traditional fast non-dominated sorting genetic algorithm NSGA-II according to the non-uniform arithmetic crossover operator and the mutation operator based on the fitness level; determining the objective function for feature selection according to the feature quantity, the root mean square error and the correlation function; based on the objective function, selecting feature data for data completion, satellite state prediction and fault diagnosis from the preprocessed telemetry data by using the improved NSGA-II algorithm.

Description

Technical Field

[0001] This invention relates to the field of spacecraft telemetry data processing technology, and in particular to a method, apparatus and medium for satellite telemetry data feature selection based on a multi-objective optimization algorithm. Background Technology

[0002] Satellite telemetry data can have hundreds or even thousands of dimensions and exhibit diverse feature variations, yet some features still share similar patterns of change. During feature selection from satellite telemetry data, noise, outliers, and the size of the telemetry variable alignment window all affect the accuracy of feature selection. For example, noise and outliers can negatively impact correlation calculations, especially when satellites undergo attitude and orbit maneuvers, where abrupt changes in outliers can make it difficult to identify and recognize telemetry parameters that are otherwise correlated. Furthermore, an excessively small correlation alignment window can also lead to significant errors in the correlation calculation of telemetry features.

[0003] Currently, conventional methods typically select features based on a single objective. For example, they score each feature according to the divergence or correlation of satellite telemetry data and filter features by setting a fixed threshold or a preset number of features to be selected. In addition, there are some methods based on mutual information, maximum information coefficient, etc., which determine the correlation between features of variables by calculating the mutual information value and maximum information coefficient value between variables.

[0004] For the aforementioned conventional approaches, firstly, using a fixed threshold for feature selection can lead to the difficulty in discovering certain periodically or trend-related telemetry parameters; and secondly, using only the total number of features to be selected ignores the correlation between variables, which undoubtedly significantly reduces the accuracy of feature selection. Secondly, using mutual information or maximum information coefficients for feature selection results in excessive computational complexity due to the lack of search space constraints, affecting the efficiency of feature selection. Summary of the Invention

[0005] In view of this, embodiments of the present invention aim to provide a method, apparatus, and medium for feature selection of satellite telemetry data based on a multi-objective optimization algorithm; which can improve the efficiency and accuracy of feature selection of satellite telemetry data.

[0006] The technical solution of this invention is implemented as follows:

[0007] In a first aspect, embodiments of the present invention provide a satellite telemetry data feature selection method based on a multi-objective optimization algorithm, the method comprising:

[0008] The raw telemetry data from the satellite is stitched together, and constant values ​​in the numerical data are removed and time alignment is performed. Then, based on the Laida criterion, the 53H algorithm and the median filtering algorithm, outliers and noise in the raw telemetry data are removed to obtain preprocessed telemetry data.

[0009] The traditional fast non-dominated sorting genetic algorithm NSGA-II is improved by using a non-uniform arithmetic crossover operator and a mutation operator based on fitness level.

[0010] The objective function for feature selection is determined based on the number of features, root mean square error, and correlation function.

[0011] Based on the objective function, the improved NSGA-II algorithm is used to select feature data from the preprocessed telemetry data for data completion, satellite status prediction, and fault diagnosis.

[0012] Secondly, embodiments of the present invention provide a satellite telemetry data feature selection device based on a multi-objective optimization algorithm. The device includes: a preprocessing section, an improved NSGA-II algorithm section, an objective function determination section, and a selection section; wherein...

[0013] The preprocessing section is configured to: stitch together the original telemetry data of the satellite, and after filtering and removing constant values ​​and time alignment in the numerical data, remove outliers and noise in the original telemetry data based on the Laida criterion, the 53H algorithm and the median filtering algorithm to obtain the preprocessed telemetry data.

[0014] The improved part of the NSGA-II algorithm is configured to improve the traditional fast non-dominated sorting genetic algorithm NSGA-II based on the non-uniform arithmetic crossover operator and the mutation operator based on fitness level.

[0015] The objective function determination part is configured to determine the objective function for feature selection based on the number of features, root mean square error, and correlation function.

[0016] The selection portion is configured to select feature data from the preprocessed telemetry data for data completion, satellite status prediction, and fault diagnosis based on the objective function and using the improved NSGA-II algorithm.

[0017] Thirdly, embodiments of the present invention provide a computing device, the computing device comprising: a communication interface, a memory, and a processor; the various components are coupled together via a bus system; wherein...

[0018] The communication interface is used for receiving and sending signals during the process of sending and receiving information with other external network elements;

[0019] The memory is used to store computer programs that can run on the processor;

[0020] The processor is configured to execute the steps of the satellite telemetry data feature selection method based on a multi-objective optimization algorithm described in the first aspect when running the computer program.

[0021] Fourthly, embodiments of the present invention provide a computer storage medium storing a satellite telemetry data feature selection program based on a multi-objective optimization algorithm. When the satellite telemetry data feature selection program based on the multi-objective optimization algorithm is executed by at least one processor, it implements the steps of the satellite telemetry data feature selection method based on the multi-objective optimization algorithm described in the first aspect.

[0022] This invention provides a method, apparatus, and medium for feature selection of satellite telemetry data based on a multi-objective optimization algorithm. First, preprocessing the raw telemetry data removes numerical data, outliers, and noise, improving robustness and avoiding adverse effects on subsequent correlation calculations, thus enhancing feature identification and selection accuracy. Second, the traditional NSGA-II algorithm is improved using a non-uniform arithmetic crossover operator and a mutation operator based on fitness levels, enhancing search capability (population diversity) and preventing premature convergence. Finally, the objective function comprehensively considers the relationship between the number of selected features, the classification effect of the regression model, and the fitness function based on standardized mutual information, thereby improving the efficiency and accuracy of feature selection in satellite telemetry data. Attached Figure Description

[0023] Figure 1 This is a schematic flowchart of a satellite telemetry data feature selection method based on a multi-objective optimization algorithm provided in an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of a process for preprocessing satellite telemetry data provided in an embodiment of the present invention;

[0025] Figure 3 A schematic diagram comparing unstandardized raw data and data standardized with zero mean, provided for embodiments of the present invention;

[0026] Figure 4 This is a schematic diagram comparing the original signal, the outlier-removed signal, and the outlier-removed denoised signal provided in an embodiment of the present invention.

[0027] Figure 5 A schematic diagram of simulation experiment results provided for embodiments of the present invention;

[0028] Figure 6This is a schematic diagram of a satellite telemetry data feature selection device based on a multi-objective optimization algorithm provided in an embodiment of the present invention;

[0029] Figure 7 This is a schematic diagram of another satellite telemetry data feature selection device based on a multi-objective optimization algorithm provided in an embodiment of the present invention;

[0030] Figure 8 This is a schematic diagram of the specific hardware structure of a computing device provided in an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0032] See Figure 1 This invention illustrates a satellite telemetry data feature selection method based on a multi-objective optimization algorithm, provided by an embodiment of the present invention. The method includes:

[0033] S101: The raw telemetry data of the satellite is stitched together, and after filtering and removing constant values ​​and time alignment in the numerical data, outliers and noise in the raw telemetry data are removed based on the Laida criterion, the 53H algorithm and the median filtering algorithm to obtain the preprocessed telemetry data.

[0034] S102: Improve the traditional fast non-dominated sorting genetic algorithm (NSGA-II) based on the non-uniform arithmetic crossover operator and the mutation operator based on fitness level;

[0035] S103: Determine the objective function for feature selection based on the number of features, root mean square error, and correlation function;

[0036] S104: Based on the objective function, the improved NSGA-II algorithm is used to select feature data from the preprocessed telemetry data for data completion, satellite status prediction and fault diagnosis.

[0037] The above technical solution firstly preprocesses the raw telemetry data to remove constant values, outliers, and noise from numerical data, improving robustness and avoiding adverse effects on subsequent correlation calculations, thus enhancing feature identification and selection accuracy. Secondly, it improves the traditional NSGA-II algorithm by using a non-uniform arithmetic crossover operator and a mutation operator based on fitness levels, enhancing search capability (i.e., population diversity) and preventing premature convergence. Finally, for the objective function, it comprehensively considers the relationship between the number of selected features, the classification effect of the regression model, and the fitness function based on standardized mutual information, thereby improving the efficiency and accuracy of feature selection in satellite telemetry data.

[0038] for Figure 1 The technical solutions shown, in some examples, see [link to example]. Figure 2 Step S101 describes stitching together the raw telemetry data from the satellite, filtering out constant values ​​and time alignment in the numerical data, and then removing outliers and noise from the raw telemetry data based on the Laida criterion, the 53H algorithm, and the median filtering algorithm to obtain preprocessed telemetry data, including:

[0039] S1011: The original telemetry data sequence obtained by splicing the original telemetry data is used to remove numerical data from the original telemetry data by zero-mean standardization for each feature.

[0040] S1012: Time-align the telemetry data sequence after removing numerical data and remove duplicates;

[0041] S1013: According to the Raida criterion, outliers in the time-aligned and duplicate-removed telemetry data sequence are first removed;

[0042] S1014: The isolated outliers in the telemetry data sequence after the first removal are removed a second time according to the 53H algorithm;

[0043] S1015: The telemetry data after the second round of filtering is subjected to median filtering to remove noise, thereby obtaining the preprocessed telemetry data.

[0044] In the example above, specifically, after obtaining the raw telemetry data from the satellite, the first step should be to concatenate this data. It should be noted that since the large amount of engineering source code and constant values ​​contained in the numerical data are not very meaningful for feature selection, it is also necessary to filter out all the numerical data. Specifically, zero-mean normalization is applied to each feature of the raw telemetry data, and the transformation function is:

[0045]

[0046] Where, μn It is the mean of feature n, σ n It is the standard deviation of feature n, n i It is the i-th data sample in feature n. It is the value of sample i after zero-mean standardization; such as Figure 3 As shown, it compares the unstandardized raw data and the zero-mean standardized data of a certain telemetry parameter of the satellite power system within a set time period. It can be seen that the zero-mean standardized data has good robustness to outliers and is beneficial for feature selection and dimensionality reduction of subsequent data.

[0047] Next, due to the misalignment of initial and subsequent recording times in the telemetry data from different satellite subsystems, the effectiveness of correlation analysis between telemetry parameters is affected. Furthermore, during satellite transit, multiple observation stations may simultaneously conduct telemetry, leading to duplicate data. Therefore, it is preferable to remove these duplicates before feature selection. Specifically, in this embodiment, it is preferable to first define a timestamp of time 0 and convert all telemetry data to a timestamp starting from this time. Then, the telemetry data from the satellite's power and other subsystems are filtered, retaining all non-duplicate data in chronological order.

[0048] Secondly, since outliers and noise are prevalent in real telemetry data, they can adversely affect correlation analysis and fault detection. Therefore, satellite telemetry data should be processed to remove outliers and reduce noise. Specifically, as described in S1013, the outlier points in the time-aligned and duplicate-removed telemetry data sequence are initially removed according to the Raida criterion. The specific implementation process may include: for the telemetry data time series X = {x1, x2, ..., x...} n}, calculate its mean x and standard deviation σ, if the data at a certain time x i (i = 1, 2, ..., n) satisfies Then remove that point, thus completing the first removal.

[0049] For the telemetry data after the initial removal, as described in S1014, a second removal of isolated outliers in the telemetry data sequence after the initial removal is performed according to the 53H algorithm. The specific implementation process may include: for the telemetry data time series X = {x1, x2, ..., x...}, ... n}, construct a new sequence in the following way. First, take the median value of x1, x2, ..., x5 as... Next, we take the median of x2, x3, ..., x6 as... And so on until x is used n-4 ,x n-3 ,…,x n The median as Until then; similarly, by calculating sequence X 1 The sequence is obtained by taking the median of three adjacent time points. Finally, through calculation Obtain sequence Subsequently, given a threshold k, if If true, then x i Treat it as a wild value, and use it The replacement is used to remove isolated wild points.

[0050] For the telemetry data after secondary rejection, S1015 describes performing median filtering on the telemetry data after secondary rejection to remove noise and obtain the preprocessed telemetry data. The specific implementation process may include: replacing the value of a point in the digital sequence with the median value of all points in a neighborhood N of that point, smoothing data fluctuations, thereby reducing the interference of noise points on the real signal; defining the input original time sequence X = {x1, x2, ..., x...} n After median filtering, the output time series Y = {y1, y2, ..., y} is obtained. n If the neighborhood N = 2c + 1, where c is a natural number, then we have: y i+k =median{x i ,x i+1 ,…,x i+N-1 The median(·) function sorts the data and then takes the median value.

[0051] Through the above implementation process, this embodiment of the invention integrates the Laida criterion, the 53H algorithm, and the median filtering algorithm to remove outliers and noise from the original telemetry data while retaining fault data caused by the fault. See also Figure 4 It shows the original signal, outlier-removed signal, and outlier-removed noise signal of the satellite power system's set telemetry parameters. It can be seen that after the original signal is processed by the Laida criterion and the 53H algorithm, most outliers can be removed, and the noise in the original signal can be further removed by the median filtering algorithm.

[0052] for Figure 1 The technical solution shown, in some examples, involves improving the traditional NSGA-II algorithm based on the non-uniform arithmetic crossover operator and the mutation operator based on fitness levels, including:

[0053] The non-uniform arithmetic crossover operator replaces the analog binary crossover operator in the traditional NSGA-II algorithm, and the level of the evolved individual is incorporated into the coefficient of the non-uniform arithmetic crossover operator.

[0054] The mutation operator in the traditional NSGA-II algorithm is replaced with a mutation operator based on fitness level, according to the individual fitness classification.

[0055] In the above example, specifically, the goal of the NSGA-II algorithm is to select N suitable individuals from the current M individuals, which is consistent with the idea of ​​feature selection. Therefore, this embodiment of the invention preferably uses the NSGA-II algorithm as the basic framework for feature selection of satellite telemetry data. The traditional NSGA-II algorithm mainly includes three parts: elite strategy selection, fast non-dominated sorting, and crowding calculation. In detail, the NSGA-II algorithm first finds all Pareto solutions in the current population M, marks them as the first layer F1, and deletes them; then it continues to find Pareto solutions for the remaining individuals, marks them as the second layer F2, and deletes them; this operation is repeated until all individuals are assigned to a layer. Next, for individuals in each layer, they are sorted in descending order according to the calculated crowding of each individual and put back into the set N until N is full. As can be seen from the above detailed description, the NSGA-II algorithm is essentially based on the genetic algorithm and still uses the basic operators in the genetic algorithm, which to some extent affect its application effect.

[0056] In view of this, embodiments of the present invention preferably use a non-uniform arithmetic crossover operator to replace the simulated binary crossover operator in the traditional NSGA-II algorithm, and take the level of the evolved individuals into the coefficients of the operator. Specifically, after applying the non-uniform arithmetic operator crossover, the decision variable values ​​of the two offspring individuals can be expressed as:

[0057]

[0058]

[0059] Where ξ represents the crossover operator coefficient, and when ξ is not a constant, it indicates non-uniform arithmetic crossover. Considering that during the evolutionary process, the algorithm aims to retain more superior individuals and individuals with better distribution in the offspring, preferably, in this embodiment of the invention, ξ is designed as:

[0060]

[0061] in, Representing the father generation Non-dominated sorting order, Representing the father generation The crowding distance. Specifically, in the early stages of the algorithm, the parent generation... and father generation The non-dominated levels are generally unequal, and the ξ value at this time can preserve the genes of better individuals; as the algorithm iterates, the parent generation... and father generation When the non-dominated levels tend to be consistent, the ξ value at this point can preserve the genes of individuals with better distribution. Therefore, this operator improves the search capability of the algorithm and the diversity of the population.

[0062] Furthermore, considering that the mutation operation in the NSGA-II algorithm aims to enable the genetic algorithm to perform local random searches and maintain population diversity, selecting a mutation operator that better aligns with the decision-maker's preferences allows the algorithm to maintain efficient local search capabilities within a suitable space. This is similar to the attention mechanism in neural networks, allowing the algorithm to focus on the region the decision-maker wants to focus on, thus reducing the difficulty of the final decision. Therefore, addressing the problem of insufficient population diversity or premature convergence caused by mutation operators in the traditional NSGA-II algorithm, this invention proposes a mutation operator based on fitness levels, dividing individuals into low-fitness, medium-fitness, and high-fitness categories as follows:

[0063]

[0064] Among them, f max This represents the maximum fitness value in the middle. α and β are the set mutation probabilities; it can be seen from the above formula that the individual fitness value is based on... and The standard is divided into three categories: low fitness, medium fitness, and high fitness, and individuals with different fitness levels are assigned different mutation probabilities. This design can ensure population diversity while solving the problem of premature convergence of traditional mutation operators.

[0065] In summary, the traditional NSGA-II algorithm is improved based on the non-uniform arithmetic crossover operator and the adaptive preference mutation operator, so that the region of focus for decision-makers is given by reference points. The introduction of the new non-uniform arithmetic crossover operator and the adaptive mutation operator can effectively improve the local search capability and the quality of the non-dominated solution set of the algorithm.

[0066] for Figure 1 In some examples of the technical solutions shown, the step of determining the objective function for feature selection based on the number of features, root mean square error, and correlation function includes:

[0067] The first index function is determined based on the number of features selected:

[0068]

[0069] Where n is the dimension of the final solution set N, N k This represents the k-th solution in the solution set N;

[0070] The second index function is determined based on the root mean square error:

[0071]

[0072] Where L is the total number of data to be predicted, and y i and Let represent the true value and the model prediction value of the i-th sample, respectively;

[0073] Based on minimizing redundancy and maximizing relevance, the third indicator function is determined using the redundancy between features and the relevance between features and labels:

[0074]

[0075] Here, the label of the data to be predicted is denoted as y, and the selected feature set is denoted as N = {N1, N2, ..., N}. n}, f 31 (L1) represents the redundancy calculation function, f 32 (L2) represents the correlation calculation function. NMI(N i N j ) represents the selected feature N i and N j Standardized mutual information between them, NMI(N k ,y) represents feature N k Standardized mutual information between label y and label y;

[0076] Based on the principle of minimization, the objective function for feature selection is determined using the first, second, and third indicator functions as follows:

[0077]

[0078] Regarding the above example, it should be noted that, as the final decision result, the first indicator function, being the most critical indicator, can be given more preference by setting appropriate reference points, such as weights, so that the improved NSGA-II algorithm pays more attention to this target during execution. Secondly, since the purpose of feature selection from satellite telemetry data is for regression analysis, this embodiment of the invention uses the root mean square error (RMSE), the basic evaluation indicator of the regression model, as the second indicator function for feature selection. That is, the smaller the RMSE, the more representative the selected features are, reflecting more of the intrinsic properties of the data to be regressed. Next, another principle of feature selection is "maximum-relevance and minimum-redundancy" (MRMR). That is, if the label of the data to be predicted is y, and the set of selected features is denoted as N = {N1, N2, ..., N...} nThe purpose of this principle is to make the features in set N as independent as possible, while making the features in set N and the label y as correlated as possible. This criterion can be mathematically expressed using the cross-correlation evaluation index NMI. Therefore, we consider this correlation function as another factor affecting the result of the multi-objective optimization feature selection algorithm, thus obtaining a third index function. Since the goal of the multi-objective optimization problem is to minimize the objective function, and the feature selection process aims to minimize NMI(N... k Therefore, the second term in the third index function is designed to be 1-NMI(N) as large as possible. k Thus, by minimizing the third index function, the specific location information of the selected feature can be obtained.

[0079] Based on the above example, the step of selecting feature data for data completion, satellite status prediction, and fault diagnosis from the preprocessed telemetry data using the improved NSGA-II algorithm based on the objective function includes:

[0080] The improved NSGA-II algorithm, combined with the objective function minimization target, selects feature data from the preprocessed telemetry data for data completion, satellite status prediction, and fault diagnosis.

[0081] To demonstrate the effectiveness of the technical solution proposed in the embodiments of the present invention, a simulation experiment was conducted by applying the technical solution proposed in the embodiments of the present invention to satellite telemetry data. The parameter settings of the improved NSGA-II algorithm are shown in Table 1:

[0082] Population size pop Number of iterations gen Reference point R Critical threshold σ <![CDATA[Mutation probability k1]]> 500 300 <![CDATA[[5;3.5×10 -4 ;5]]]> 0.5 0.1

[0083] The goal of the simulation experiment is to complete the missing data of the satellite power subsystem. For the preprocessed telemetry data feature Z, the number and location of features are selected using the technical solution proposed in this invention, thereby obtaining the optimized leading edge surface, as shown below. Figure 5 As shown in the figure, it can be seen that when the number of features is selected as 5, the Pareto solution is closest to the reference point R. Therefore, the 5 features C, F, M, P and Q that are most relevant to feature Z can be obtained in the end.

[0084] Based on the same inventive concept as the aforementioned technical solution, see [link to inventive concept]. Figure 6 This invention illustrates a satellite telemetry data feature selection device 60 based on a multi-objective optimization algorithm, comprising: a preprocessing section 601, an NSGA-II algorithm improvement section 602, an objective function determination section 603, and a selection section 604; wherein,

[0085] The preprocessing section 601 is configured to: stitch together the original telemetry data of the satellite, and after filtering and removing constant values ​​and time alignment in the numerical data, remove outliers and noise in the original telemetry data based on the Laida criterion, the 53H algorithm and the median filtering algorithm to obtain the preprocessed telemetry data.

[0086] The NSGA-II algorithm improvement part 602 is configured to improve the traditional fast non-dominated sorting genetic algorithm NSGA-II based on the non-uniform arithmetic crossover operator and the mutation operator based on fitness level.

[0087] The objective function determination part 603 is configured to determine the objective function for feature selection based on the number of features, root mean square error, and correlation function.

[0088] The selection portion 604 is configured to select feature data from the preprocessed telemetry data for data completion, satellite status prediction, and fault diagnosis based on the objective function and using the improved NSGA-II algorithm.

[0089] See in some examples Figure 7 The preprocessing section 601 includes: a removal unit 6011, a first rejection unit 6012, a second rejection unit 6013, and a noise reduction unit 6014; wherein,

[0090] The removal unit 6011 is configured to remove numerical data from the original telemetry data sequence obtained by splicing the original telemetry data, using zero-mean normalization for each feature; and to perform time alignment and remove duplicates from the telemetry data sequence after removing numerical data.

[0091] The first elimination unit 6012 is configured to perform the first elimination of outliers in the time-aligned and duplicate-removed telemetry data sequence according to the Raida criterion;

[0092] The second elimination unit 6013 is configured to perform a second elimination of isolated outliers in the telemetry data sequence after the first elimination according to the 53H algorithm;

[0093] The denoising unit 6014 is configured to perform median filtering on the telemetry data after secondary rejection to remove noise, thereby obtaining the preprocessed telemetry data.

[0094] In some examples, the NSGA-II algorithm improvement section 602 is configured as follows:

[0095] The non-uniform arithmetic crossover operator replaces the analog binary crossover operator in the traditional NSGA-II algorithm, and the level of the evolved individual is incorporated into the coefficient of the non-uniform arithmetic crossover operator.

[0096] The mutation operator in the traditional NSGA-II algorithm is replaced with a mutation operator based on fitness level, according to the individual fitness classification.

[0097] In some examples, the objective function determination part 603 is configured as follows:

[0098] The first index function is determined based on the number of features selected:

[0099]

[0100] Where n is the dimension of the final solution set N, N k This represents the k-th solution in the solution set N;

[0101] The second index function is determined based on the root mean square error:

[0102]

[0103] Where L is the total number of data to be predicted, and y i and Let represent the true value and the model prediction value of the i-th sample, respectively;

[0104] Based on minimizing redundancy and maximizing relevance, the third indicator function is determined using the redundancy between features and the relevance between features and labels:

[0105]

[0106] Here, the label of the data to be predicted is denoted as y, and the selected feature set is denoted as N = {N1, N2, ..., N}. n}, f 31 (L1) represents the redundancy calculation function, f 32 (L2) represents the correlation calculation function. NMI(N i N j ) represents the selected feature N i and N j Standardized mutual information between them, NMI(N k ,y) represents feature N k Standardized mutual information between label y and label y;

[0107] Based on the principle of minimization, the objective function for feature selection is determined using the first, second, and third indicator functions as follows:

[0108]

[0109] In some examples, the selection portion 604 is configured to select feature data from the preprocessed telemetry data for data completion, satellite status prediction, and fault diagnosis using the improved NSGA-II algorithm and the objective function minimized.

[0110] Understandably, in this embodiment, "part" can be a part of a circuit, a part of a processor, a part of a program or software, etc., and can also be a unit, a module, or a non-modular one.

[0111] Furthermore, in this embodiment, the components can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional module.

[0112] If the integrated unit is implemented as a software functional module and not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the method described in this embodiment. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0113] Therefore, this embodiment provides a computer storage medium storing a satellite telemetry data feature selection program based on a multi-objective optimization algorithm. When the satellite telemetry data feature selection program based on the multi-objective optimization algorithm is executed by at least one processor, it implements the steps of the satellite telemetry data feature selection method based on the multi-objective optimization algorithm described in the above technical solution.

[0114] Based on the aforementioned satellite telemetry data feature selection device 60 based on a multi-objective optimization algorithm and the computer storage medium, see [link to relevant documentation]. Figure 8This illustration shows the specific hardware structure of a computing device 80, provided by an embodiment of the present invention, capable of implementing the aforementioned satellite telemetry data feature selection device 60 based on a multi-objective optimization algorithm. The computing device 80 can be a wireless device, mobile or cellular phone (including so-called smartphones), personal digital assistant (PDA), video game console (including video display, mobile video game device, mobile video conferencing unit), laptop computer, desktop computer, set-top box, tablet computing device, e-book reader, fixed or mobile media player, etc. The computing device 80 includes: a communication interface 801, a memory 802, and a processor 803; the various components are coupled together through a bus system 804. It is understood that the bus system 804 is used to realize the connection and communication between these components. In addition to a data bus, the bus system 804 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 8 The general designated all buses as Bus System 804.

[0115] The communication interface 801 is used for receiving and sending signals during the process of sending and receiving information with other external network elements;

[0116] The memory 802 is used to store computer programs that can run on the processor 803;

[0117] The processor 803 is used to execute the steps of the satellite telemetry data feature selection method based on multi-objective optimization algorithm described in the foregoing technical solution when running the computer program, which will not be repeated here.

[0118] It is understood that the memory 802 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 802 of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0119] The processor 803 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 803 or by software instructions. The processor 803 can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory 802, and the processor 803 reads the information in memory 802 and, in conjunction with its hardware, completes the steps of the above method.

[0120] It is understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described herein, or combinations thereof.

[0121] For software implementation, the techniques described herein can be achieved through modules (e.g., procedures, functions, etc.) that perform the functions described herein. The software code can be stored in memory and executed by a processor. The memory can be implemented within the processor or externally.

[0122] It is understood that the exemplary technical solutions of the satellite telemetry data feature selection device 60 and computing device 80 based on the multi-objective optimization algorithm described above belong to the same concept as the technical solutions of the satellite telemetry data feature selection method based on the multi-objective optimization algorithm described above. Therefore, all details not described in detail above regarding the technical solutions of the satellite telemetry data feature selection device 60 and computing device 80 based on the multi-objective optimization algorithm can be found in the description of the technical solutions of the satellite telemetry data feature selection method based on the multi-objective optimization algorithm described above. This embodiment of the invention will not elaborate further on these details.

[0123] It should be noted that the technical solutions described in the embodiments of the present invention can be combined arbitrarily without conflict.

[0124] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A satellite telemetry data feature selection method based on a multi-objective optimization algorithm, characterized in that, The method includes: The raw telemetry data from the satellite is stitched together, and constant values ​​in the numerical data are removed and time alignment is performed. Then, based on the Laida criterion, the 53H algorithm and the median filtering algorithm, outliers and noise in the raw telemetry data are removed to obtain preprocessed telemetry data. The traditional fast non-dominated sorting genetic algorithm NSGA-II is improved by using a non-uniform arithmetic crossover operator and a mutation operator based on fitness level. The objective function for feature selection is determined based on the number of features, root mean square error, and correlation function. Based on the objective function, the improved NSGA-II algorithm is used to select feature data from the preprocessed telemetry data for data completion, satellite status prediction and fault diagnosis. The step of determining the objective function for feature selection based on the number of features, root mean square error, and correlation function includes: The first index function is determined based on the number of features selected: wherein, is the final solution set of dimensionality, denotes the th solution in the solution set ; The second index function is determined based on the root mean square error: wherein, is the total number of data to be predicted, and respectively represent the true value and the model predicted value of the first i sample. Based on minimizing redundancy and maximizing relevance, the third indicator function is determined using the redundancy between features and the relevance between features and labels: The labels of the data to be predicted are denoted as follows: Let the selected feature set be denoted as , The function represents the redundancy calculation. This represents the correlation calculation function. Indicates the selected feature and Standardized mutual information between them Representation of features and tags Standardized mutual information between them; Based on the principle of minimization, the objective function for feature selection is determined using the first, second, and third indicator functions as follows: 。 2. The method of claim 1, wherein, The process involves stitching together the raw satellite telemetry data, filtering out constant values ​​and time alignment in the numerical data, and then removing outliers and noise from the raw telemetry data based on the Laida criterion, the 53H algorithm, and the median filtering algorithm to obtain preprocessed telemetry data, including: The original telemetry data sequence obtained by splicing the original telemetry data is then processed by zero-mean standardization to remove numerical data from the original telemetry data for each feature. The telemetry data sequence after removing numerical data is time-aligned and duplicates are removed. Outliers in the time-aligned and duplicate-removed telemetry data sequence were initially removed according to the Laida criterion. The isolated outliers in the telemetry data sequence after the first round of removal are removed a second time according to the 53H algorithm. The telemetry data after the second round of filtering is subjected to median filtering to remove noise, thus obtaining the preprocessed telemetry data.

3. The method of claim 1, wherein, The improvement of the traditional NSGA-II algorithm based on the non-uniform arithmetic crossover operator and the mutation operator based on fitness level includes: The non-uniform arithmetic crossover operator replaces the analog binary crossover operator in the traditional NSGA-II algorithm, and the level of the evolved individual is incorporated into the coefficient of the non-uniform arithmetic crossover operator. The mutation operator in the traditional NSGA-II algorithm is replaced with a mutation operator based on fitness level, according to the individual fitness classification.

4. The method of claim 1, wherein, The step of selecting feature data from the preprocessed telemetry data for data completion, satellite status prediction, and fault diagnosis based on the objective function and using the improved NSGA-II algorithm includes: The improved NSGA-II algorithm, combined with the objective function minimization target, selects feature data from the preprocessed telemetry data for data completion, satellite status prediction, and fault diagnosis.

5. A satellite telemetry data feature selection device based on a multi-objective optimization algorithm, characterized by, The device includes: a preprocessing section, an NSGA-II algorithm improvement section, an objective function determination section, and a selection section; wherein... The preprocessing section is configured to: stitch together the original telemetry data of the satellite, and after filtering and removing constant values ​​and time alignment in the numerical data, remove outliers and noise in the original telemetry data based on the Laida criterion, the 53H algorithm and the median filtering algorithm to obtain the preprocessed telemetry data. The improved part of the NSGA-II algorithm is configured to improve the traditional fast non-dominated sorting genetic algorithm NSGA-II based on the non-uniform arithmetic crossover operator and the mutation operator based on fitness level. The objective function determination part is configured to determine the objective function for feature selection based on the number of features, root mean square error, and correlation function. The selection portion is configured to select feature data for data completion, satellite status prediction, and fault diagnosis from the preprocessed telemetry data based on the objective function and using the improved NSGA-II algorithm. The objective function determination part is configured as follows: The first index function is determined based on the number of features selected: wherein, is the final solution set of dimensionality, denotes the th solution in the solution set ; The second index function is determined based on the root mean square error: in, It is the total number of data to be predicted. and They represent the first i The true value and the model prediction value of each sample; Based on minimizing redundancy and maximizing relevance, the third indicator function is determined using the redundancy between features and the relevance between features and labels: The labels of the data to be predicted are denoted as follows: Let the selected feature set be denoted as , The function represents the redundancy calculation. This represents the correlation calculation function. Indicates the selected feature and Standardized mutual information between them Representation of features and tags Standardized mutual information between them; Based on the principle of minimization, the objective function for feature selection is determined using the first, second, and third indicator functions as follows: 。 6. The apparatus of claim 5, wherein, The improved part of the NSGA-II algorithm is configured as follows: The non-uniform arithmetic crossover operator replaces the analog binary crossover operator in the traditional NSGA-II algorithm, and the level of the evolved individual is incorporated into the coefficient of the non-uniform arithmetic crossover operator. The mutation operator in the traditional NSGA-II algorithm is replaced with a mutation operator based on fitness level, according to the individual fitness classification.

7. A computing device, comprising: The computing device includes: a communication interface, a memory, and a processor; the various components are coupled together via a bus system; wherein... The communication interface is used for receiving and sending signals during the process of sending and receiving information with other external network elements; The memory is used to store computer programs that can run on the processor; The processor is configured to, when running the computer program, execute the steps of the satellite telemetry data feature selection method based on a multi-objective optimization algorithm as described in any one of claims 1 to 4.

8. A computer storage medium, characterized in that The computer storage medium stores a satellite telemetry data feature selection program based on a multi-objective optimization algorithm. When the satellite telemetry data feature selection program based on the multi-objective optimization algorithm is executed by at least one processor, it implements the steps of the satellite telemetry data feature selection method based on the multi-objective optimization algorithm as described in any one of claims 1 to 4.

Images