CN111611294A - Star sensor data anomaly detection method - Google Patents
Star sensor data anomaly detection method Download PDFInfo
- Publication number
- CN111611294A CN111611294A CN202010361728.2A CN202010361728A CN111611294A CN 111611294 A CN111611294 A CN 111611294A CN 202010361728 A CN202010361728 A CN 202010361728A CN 111611294 A CN111611294 A CN 111611294A
- Authority
- CN
- China
- Prior art keywords
- data
- detection
- target
- quaternion
- telemetering
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2462—Approximate or statistical queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Fuzzy Systems (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention provides a star sensor data anomaly detection method, relates to an anomaly data detection method, and can solve the problems that satellite attitude and operation orbit are affected by star sensor working anomaly, the satellite attitude deviates from a control range and even turns over, heavy loss is caused, and more false alarms and false alarm leakage problems often occur. The specific technical scheme is as follows: selecting telemetry sample data to carry out pretreatment to obtain target telemetry data; selecting a confidence interval of the detection quantity of the target telemetering data obtained by the detection quantity through confidence analysis; and detecting the target telemetering data exceeding the confidence interval as abnormal data on line. The method is used for detecting and managing the data abnormity of the star sensor of the high orbit satellite.
Description
Technical Field
The disclosure relates to the field of spacecraft fault diagnosis, in particular to a star sensor data anomaly detection method.
Background
The spacecraft fault diagnosis is an important means for timely discovering and handling the spacecraft abnormity, and plays an important role in the long-term management of the spacecraft. The star sensor can provide accurate space orientation and reference for the satellite, has autonomous navigation capability, and is an important measurement component of a satellite attitude control subsystem. The satellite attitude and the operation orbit are influenced by the abnormal working of the star sensor, so that the satellite attitude deviates from the control range and even turns over, and the heavy loss is brought. In daily measurement and control, monitoring of relevant parameters of the star sensor is very important, and quaternion is a core attitude parameter of the star sensor. Machine learning is a multi-field interdiscipline, and a research computer simulates the learning behavior of human beings, acquires new knowledge or skills, and reorganizes the existing knowledge structure to continuously improve the performance of the knowledge structure. The machine learning is characterized in that the machine learning method is based on data, feature extraction and automatic updating, can be used for designing a time sequence model of a machine learning algorithm for mainly determining the change of the telemetering parameters according to time, analyzing the deviation of the model and the actually measured data, can be approximately used as the real change rule of the parameters in a certain deviation range, and is used for subsequent detection, prediction and evaluation.
The fault diagnosis logic and reasoning method mainly comprises a rule-based reasoning method, a model-based reasoning method and a case-based reasoning method, is applied to the actual technology such as threshold judgment, an expert system, logic judgment and the like, plays an important role, but still has the following limitations: the threshold for anomaly detection depends on the experience and design principle of a spacecraft development unit and is often rough;
the expert system has the problems of unscientific weight distribution, inaccurate logic, insufficient basis and the like, and the logic judgment is often ineffective in the face of wild value or sudden state conversion, so that more false alarms and false-alarms are often generated in the application of the spacecraft fault diagnosis system.
Disclosure of Invention
The invention establishes a star sensor data anomaly detection method, which is used for solving the problem that telemetry data detection cannot be dealt with based on a threshold judgment method in the existing diagnosis system. The problem that more false alarms and missed alarms often occur in the application of the spacecraft fault diagnosis system is solved.
The invention is based on the basic principle of machine learning, combines big data and basic ideas of data mining, takes mass telemetering data as objects, and randomly divides the telemetering data into a training set, a testing set and a detection set according to an independence principle through a telemetering data preprocessing algorithm. Two types of detection quantities are extracted according to the quaternion characteristics of the star sensor and used as identification characteristics of telemetering parameter abnormality, a machine learning fitting star sensor quaternion time sequence model is used for verifying and performing confidence coefficient analysis on the fitting model, and a method for moving a time window is designed for abnormality detection.
According to a first aspect of the embodiments of the present disclosure, a method for detecting data anomaly of a star sensor is provided, the method including:
selecting telemetry sample data to carry out pretreatment to obtain target telemetry data; selecting a confidence interval of the detection quantity of the target telemetering data obtained by the detection quantity through confidence analysis; detecting the target telemetering data exceeding the confidence interval as abnormal data on line;
the confidence interval is the confidence interval of the error detection quantity of the single-parameter model; or, a confidence interval of the correlation coefficient detection quantity;
wherein the detection quantity is selected as a single-parameter detection quantity; or, multi-dimensional telemetry correlation detection quantity selection.
In one embodiment, the telemetry data preprocessing comprises removing abnormal values of telemetry sample data, obtaining a target telemetry data segment, and performing cycle segmentation and diversity on the target telemetry data segment.
The abnormal value elimination is to eliminate an abnormal value near a zero value or eliminate an abnormal value exceeding the upper and lower limits of the telemetering data;
in one embodiment, the elimination of the abnormal value near the zero value refers to scanning from the time starting point of the target telemetering data at least in one moving time window period, sorting the data in the moving time window according to size, obtaining a median of a sorted sequence, adding a triple standard deviation of the data in the moving time window by taking the median as a reference to obtain an upper limit interval and a lower limit interval, and eliminating the data outside the upper limit interval and the lower limit interval;
the method for eliminating the abnormal values exceeding the upper limit and the lower limit of the telemetering data refers to the steps of carrying out statistical analysis on the quaternion telemetering data of the satellite star sensor through a computer program, determining the upper limit and the lower limit of a quaternion telemetering data sequence, and eliminating the abnormal values of the quaternion telemetering data except the upper limit and the lower limit.
In one embodiment, the period division of the target telemetering data segment means that a time sequence change period of the target telemetering data can be obtained through a machine learning mode, the target telemetering data segment is scanned from a starting point to an end point of the target telemetering data segment in a moving time window, the period of the selected telemetering data segment and the starting point and the end point of the period are identified by judging a maximum value or a minimum value in at least one period corresponding to monotonicity shown by data in the time window, and the telemetering data is divided according to the period;
the target telemetering data section is subjected to diversity, namely, the target telemetering data is divided into 3n groups according to a period, n is more than or equal to 10, and the 3n groups of data are randomly divided into 3 groups which are respectively a training set, a test set and a detection set; the training set is used for machine learning of a time sequence model of the star sensitive quaternion, the test set is used for a rechecking quaternion model, and the model is obtained by learning of the training set; the detection set telemetry data refers to a whole-cycle array with a fault time period which can be picked out in advance in at least one cycle or is real-time telemetry data of a satellite; the target telemetry data is detection set telemetry data.
In one embodiment, the detection set telemetry data is subjected to a fitting model to obtain a target telemetry data time sequence model;
the fitting model is a time sequence model of the target telemetering data obtained by extracting the periodic characteristics and the time sequence function characteristics of the target telemetering data by a machine learning method and adopting a computer trigonometric function fitting program.
In one embodiment, the fitting model value of the target telemetry data time sequence model is used as the detection quantity, and the difference value between the fitting model value and the target telemetry data is selected in at least one complete period in different time periods. In one embodiment, the detected amount is subjected to confidence analysis to obtain a confidence interval of the detected amount of the target telemetry data; and the confidence interval is obtained by carrying out statistical analysis on the detection quantity sequence of the telemetry data of the N samples, wherein N is more than or equal to 1, the mean value and the standard deviation of the detection quantity sequence are obtained, the significance test level is preset, and the confidence interval of the error detection quantity is obtained through calculation and is the confidence interval of the detection quantity of the target telemetry data.
Preferably, the online detection is carried out in at least one moving time window, the difference value between the fitting model value and the target telemetering data is monitored, and when the difference value exceeds the confidence interval range of the detection amount of the target telemetering data, the target telemetering data is judged to be abnormal; and judging the abnormal value as abnormal value more than 10 times, and judging the target telemetering data as abnormal data by monotonically increasing more than 30 times.
According to the second aspect of the embodiment of the disclosure, in the confidence interval of the detection quantity of the target telemetering data, the correlation coefficient of pairwise correlation of the quaternion telemetering data in at least one period in the same time period is selected as the detection quantity, the correlation coefficients of M samples are calculated, wherein M is larger than or equal to 1, the significance test level is preset according to a statistical method, and the confidence interval of the quaternion correlation coefficient under the level is obtained.
Preferably, the confidence interval of the quaternion correlation coefficient under the horizontal is used for online detection
Judging abnormal data when the target telemetering data exceeds the confidence interval range of the quaternion correlation coefficient under the level;
the abnormal data is the quaternion sample telemetering data of at least one period in the same time period, and the correlation coefficient of the quaternion is obtained and is the target telemetering data.
According to the method and the device, the problem faced by the field of fault diagnosis can be effectively solved through the anomaly detection technology based on machine learning, the telemetering data is an important basis for truly reflecting the state of the satellite, the machine learning starts from massive historical telemetering data, the potential information of the historical telemetering data is fully mined, and anomalies can be found sensitively and accurately.
The invention extracts the quaternion telemetering data characteristics of the satellite star sensor by a machine learning method, models a quaternion time sequence curve under the normal working state of the satellite, compares a model value with the telemetering data in the fault period of the satellite star sensor, can fully utilize the self information of the data to carry out abnormal detection, obtains obvious effect, and has the detection accuracy higher than 95 percent when the given detection level is 5 percent.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and together with the description, serve to explain the principles of the disclosure.
FIG. 1 is a diagram of a method for detecting data anomaly of a star sensor according to an embodiment of the present disclosure;
FIG. 2 is a structural diagram of a single-parameter detection data anomaly detection method for a star sensor according to an embodiment of the present disclosure;
FIG. 3 is a diagram of a pre-processing structure of telemetry sample data of a star sensor according to an embodiment of the disclosure;
FIG. 4 is a logical structure diagram of confidence intervals of error detection quantities of a single-parameter model of a star sensor according to an embodiment of the present disclosure;
FIG. 5 is a logical structure diagram of confidence interval determination abnormal data of error detection quantity of a single-parameter model of a star sensor according to an embodiment of the present disclosure;
FIG. 6 is a logical structure diagram of a confidence interval of multi-dimensional telemetering correlation detection data of a star sensor according to an embodiment of the present disclosure;
FIG. 7 is a logic structure diagram for determining an anomaly in a confidence interval of single multi-dimensional telemetry associated detection data according to an embodiment of the present disclosure;
FIG. 8 is a logical structure diagram of a confidence interval determination anomaly data selected by a multi-dimensional telemetry associated detection quantity according to an embodiment of the present disclosure;
FIG. 9 is a graphical representation of telemetry outliers near a zero value after format conversion of telemetry data in an embodiment of the disclosure.
Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
The embodiment of the disclosure provides a method for detecting abnormal data of a star sensor by using quaternion abnormality detection of a high-orbit satellite star sensor, which is shown in figure 1,
the first embodiment,
101, selecting telemetry sample data to carry out pretreatment to obtain target telemetry data;
in one embodiment, the telemetry data preprocessing comprises the steps of removing abnormal values of telemetry sample data to obtain a target telemetry data section, and carrying out periodic segmentation and diversity on the target telemetry data section;
the abnormal value elimination has two modes, one mode is elimination of the abnormal value near the zero value, and the other mode is elimination of the abnormal value exceeding the upper limit and the lower limit of the telemetering data;
in one embodiment, the elimination of the abnormal value near the zero value refers to scanning from the time starting point of the target telemetering data at least in one moving time window period, sorting the data in the moving time window according to size, obtaining a median of a sorted sequence, adding a triple standard deviation of the data in the moving time window by taking the median as a reference to obtain an upper limit interval and a lower limit interval, and eliminating the data outside the upper limit interval and the lower limit interval; such as: eliminating points near zero appearing suddenly on the sine function curve and having telemetry data less than 10-5The abnormal value is shown in fig. 9, which illustrates the abnormal value situation near the zero value of the telemetry data in the embodiment of the disclosure, and the diagram illustrates the case that the telemetry value of the quaternion of the star sensor is near zero, which is a target that we need to reject, and the normal telemetry value following the time sequence change should be shown as a curve trend in the diagram, and a single value or a plurality of values near zero in the diagram are/is a rejection target.
In one embodiment, the other method is to remove the abnormal values exceeding the upper limit and the lower limit of the telemetering data, namely performing statistical analysis on the quaternion telemetering data of the satellite star sensor through a computer program to determine the upper limit and the lower limit of a quaternion telemetering data sequence and removing the abnormal values of the quaternion telemetering data outside the upper limit and the lower limit.
In one embodiment, the period division of the target telemetry data segment means that a time sequence change period of the target telemetry data is obtained through a machine learning mode, scanning is carried out from a starting point to an end point of the target telemetry data segment in a moving time window, a maximum value or a minimum value in at least one period corresponding to monotonicity of data in the moving time window is judged, the monotonicity changing place is the maximum value or the minimum value in one period, and the length between two adjacent maximum values or minimum values is one period. Identifying the period of the selected telemetry data segment and the starting point and the ending point of the period, and dividing the telemetry data according to the period;
in one embodiment, the target telemetering data section is subjected to diversity, namely, the target telemetering data is divided into 3n groups according to a period, n is more than or equal to 10, and the 3n groups of data are randomly divided into 3 groups which are respectively a training set, a test set and a detection set; the training set is used for machine learning of a time sequence model of the star sensitive quaternion, the test set is used for a rechecking quaternion model, and the model is obtained by learning of the training set; the detection set telemetering data is a whole-cycle array with a fault time period which can be picked out in advance in a period at least greater than or equal to one cycle, or is real-time telemetering data of a satellite; the detection set telemetry data is target telemetry data.
102, selecting a confidence interval of the detected quantity of the target telemetering data obtained by confidence analysis of the detected quantity; two modes are available for selecting the detection quantity;
in one embodiment, the first detection amount is selected as a single-parameter detection amount, and through confidence degree analysis, the detection set telemetering data passes through a fitting model to obtain a target telemetering data time sequence model; the fitting model is a time sequence model obtained by extracting the periodic characteristics and the time sequence function characteristics of the target telemetering data by a machine learning method and adopting a computer trigonometric function fitting program.
And selecting the difference value between the fitting model value and the target telemetering data as a detection quantity in at least one complete period in different time periods according to the fitting model value of the target telemetering data time sequence model. Obtaining a confidence interval with the confidence interval being the error detection quantity of the single-parameter model;
the confidence interval is obtained by performing statistical analysis on the detection quantity sequence of the telemetry data of the N samples to obtain the mean value and standard deviation of the detection quantity sequence, and presetting the significance test level, wherein the significance test level is usually given artificially and is not calculated. According to a statistical theory, the detection level and the detection interval have a corresponding relation, and the detection interval is usually wider given a larger detection level through a formula and a table look-up; and calculating to obtain a confidence interval of the error detection amount, wherein the confidence interval of the error detection amount is the confidence interval of the detection amount of the target telemetering data, and N is more than or equal to 1.
The second detection quantity selection is multi-dimensional telemetering associated detection quantity selection, and confidence intervals of the correlation coefficient detection quantities are obtained through confidence degree analysis;
selecting a confidence interval of the detection quantity of the target telemetering data obtained by the detection quantity through confidence analysis; selecting correlation coefficients related to each other of the quaternion telemetering data in at least one period in the same time period as detection quantity, calculating the correlation coefficients of M samples, presetting significance test level according to a statistical method, and obtaining a confidence interval of the quaternion correlation coefficients in the level, wherein M is more than or equal to 1.
103, detecting abnormal data of the target telemetering data exceeding the confidence interval on line
In one embodiment, there are two ways to detect that the target telemetry data exceeds the confidence interval as abnormal data on line;
in the first mode, the confidence interval of the target telemetering data detection amount is used as the confidence interval of the error detection amount for online detection, the difference value between the fitting model value and the target telemetering data is monitored at least in one moving time window, and abnormal data of the target telemetering data is judged when the difference value exceeds the confidence interval range of the target telemetering data detection amount; and judging the abnormal value to be more than 10 times, and judging the target telemetering data to be abnormal data more than 30 times in monotone increment of the abnormal value.
In the second mode, the confidence interval based on the detection quantity of the target telemetering data is the confidence interval of the quaternion correlation coefficient under the horizontal condition, and the confidence interval of the quaternion correlation coefficient under the horizontal condition is used for judging abnormal data when the target telemetering data exceeds the confidence interval range of the quaternion correlation coefficient under the horizontal condition through online detection; the abnormal data is quaternion sample telemetry data of at least one period in the same time period, and the correlation coefficient of the quaternion is calculated and is target telemetry data.
According to the star sensor data anomaly detection method provided by the embodiment, a quaternion time sequence curve in a normal working state of a satellite is modeled, a fitting model value is compared with remote target telemetry data in a satellite star sensor fault period, the self information of the data can be fully utilized to carry out anomaly detection, an obvious effect is obtained, the detection accuracy is higher than 95%, and the problems of more false alarms and false alarm omission which often occur in the application of a spacecraft fault diagnosis system are well solved.
Example two
A method for detecting data anomalies of a star sensor, as shown in fig. 2, in one embodiment.
201. The method selects the telemetering sample data to carry out preprocessing and comprises data format conversion, data cleaning, data period division and data diversity. As shown in figure 3 of the drawings,
2011. telemetry sample data format conversion
From the data perspective, satellite telemetering data is uneven and discontinuous, certain interference is caused to machine learning, and the telemetering data needs to be uniformly depicted according to time granularity. From a modeling perspective, the satellite telemetry data time series is datatime type, which is not favorable for mathematical expression. In order to adapt to the requirement of model fitting, the starting moment of the telemetry sample is selected as a starting point, datetime type data is converted into numerical type data, and the telemetry time data is uniformly sampled.
The quaternion q of the satellite A star sensor is completed by adopting a computer program0、q1、q2、q3Telemetry data format conversion: selecting a period starting point TsiAs a time reference, this time is subtracted from the subsequent time, and the time-series numerical conversion is completed in units of seconds.
2012. Telemetry sample data cleaning
The satellite works in a severe space environment, a telemetry data transmission path is far, signals are weak, and certain measurement errors exist in the satellite and ground equipment, so that errors, miscalues and outliers exist in the satellite telemetry data measurement value. If the satellite telemetering data is not cleaned, the machine learning algorithm can bring the error information into the satellite telemetering data, wrong knowledge can be learned, and the fitted model can generate errors, so that the satellite telemetering data needs to be cleaned, and data which can truly reflect the change rule of the telemetering parameters can be obtained. Outlier culling includes two categories: such as rejecting outliers that exceed ± 1 and rejecting outliers that differ from neighbors by more than 0.01 and by less than 0.01.
2013. Telemetry target data cycle partitioning and diversity
The telemetering target data of the quaternion of the star sensor has time sequence periodicity, and the time sequence change period of the telemetering data of the star sensor can be obtained in a machine learning mode. Scanning from the starting point to the end point of the target telemetering data segment by adopting a moving time window, identifying the period of the selected telemetering data segment and the starting point and the end point of the period by judging monotonicity and the maximum value or the minimum value shown by the data in the time window, and dividing the telemetering data according to the period. Dividing the sample telemetering data into 3n groups (n is more than or equal to 10) according to the period, and randomly dividing the 3n groups of data into 3 groups in order to ensure the independence of the diversity data: the method comprises the steps of obtaining a training set, a test set and a detection set, wherein the training set is used for machine learning of a time sequence model of a star sensitive quaternion, the test set is used for rechecking a quaternion model, and the quaternion model is obtained by learning of the training set; the detection set is a whole-cycle array which can pick out fault time periods in advance in at least one cycle or more, and can also be real-time telemetering data of a satellite.
In one embodiment, the start of a cycle of the sequence is determined based on the monotonicity and extrema of the telemetry data, and for a cycle, the maximum or minimum is determined, the fitted model is a sinusoidal function of half cycles, where the monotonicity changes corresponding to the maximum or minimum, and a cycle length is between two adjacent maxima or minima. But for a plurality of cycles, is referred to as a maximum or minimum, i.e., collectively as an extremum; the period is divided, for example, 100 groups of whole period data (at least one period and must satisfy the whole period) of the satellite a in the normal working state are selected to form a data set R, and the data set R is randomly selected and divided into three groups: training set T, testing set C and detecting set D, and marking as follows:
The three data sets are independent, i.e.
T∪C∪D＝R
202. The detection set telemetering data is subjected to model fitting to obtain a target telemetering data time sequence model, the model fitting is based on machine learning, the data characteristics are extracted by adopting a machine learning method, and a time sequence model (time function) of the star sensor quaternion is obtained through fitting. From the characteristics of satellite orbits, quaternion telemetering data of the high-orbit satellite star sensor is changed into a half-period trigonometric function, and the general form of a time sequence function is set as follows:
q(t)＝a(sinωt+b)+c+N0(t)
wherein q (t) represents a quaternion, ω is a quaternion period, N0And (t) the modeling is negligible high-order noise quantity, and a, b and c are undetermined coefficients and can be obtained by learning and carrying out statistical averaging on historical telemetering data of the satellite star sensor quaternion. Because datetime type time data (in the shape of '2018-09-0519: 51: 12.953') is inconvenient for mathematical expression, a certain time starting point T0 of the telemetric data is selected as a zero point, and the datetime type data is changed into a form easy for mathematical expression according to the concept similar to 'product seconds' in aerospace. The period of the quaternion time series model is calculated using a program that identifies the start and end of the period of the telemetry data and fits a function to the period.
In one embodiment, the training set T is sent to a learning machine, and a python trigonometric function fitting program is used for fitting the quaternion time sequence function model of the star sensor to obtain the period of the function and the values of the coefficients a, b and c. Quaternion q of star sensor0For example, the partial elements of the sample are as follows:
...
42108,0.7060563325
42109,0.7060572448
42110,0.7060581562
42111,0.7060590667
42112,0.7060599762
the sample periodic time sequence function model is obtained by fitting as follows:
the fitting coefficients are:
| coefficient to be estimated | Fitting value of coefficient to be estimated |
| a | 0.706512158432 |
| b | 0.0000487812804619 |
| c | 0.0000288515321792 |
And after calculating to obtain the fitting function period and the coefficient of each sample, respectively carrying out statistical averaging on coefficient results obtained by fitting each sample, and taking the statistical average of each coefficient as a final fitting result.
203. Selecting single parameter detection quantity and analyzing confidence degree to obtain a confidence interval;
according to the characteristics of the telemetering data of the star sensor, two methods different from the conventional method for artificially setting the detection threshold are designed: the two methods of single parameter detection respectively correspond to two detection quantities, namely error detection quantity, and the part consists of two parts of detection quantity selection and confidence degree analysis; as shown in fig. 4.
2031 single parameter telemetry data detection volume selection,
selecting single parameter detection quantity, measuring the accuracy of the fitting model by using errors of the real model and the fitting model, and defining a time function with errors according to the error quantity:
(t)＝q0(t)-q0 *(t)
in the above formula, q0(t) a real time function representing the quaternion of the star sensor, q0 *And (t) represents a fitting time function of the quaternion of the star sensor, and the error quantity is a detection quantity of a single-parameter detection method.
In one embodiment, a sequence of the detection quantity is obtained, time sequences corresponding to the telemetry data sequences of the test set C are respectively substituted into the fitting model to obtain a quaternary digital model value sequence of the star sensor, and the value sequence is respectively combined with the respective time sequences to obtain a fitting model set G. And subtracting the quaternion telemetering data sequence of the star sensor in the test set C from the analog value sequence in the fitting matrix G according to corresponding time to obtain an error sequence E, wherein the error sequence is a reflection of the difference between the fitting model and the real telemetering data, and the smaller the value of the error sequence is, the higher the coincidence degree of the fitting model and the telemetering data is, the more accurate the fitting model is. The fitting model set G is:
the error sequence is represented as follows:
wherein
2032 Single parameter telemetry data confidence analysis
Due to the error detection amount of the single parameter model and the fact that the data of the test set and the data of the training set are different in time period, the single parameter model and the training set can be considered to be independent. And taking the test set data as real data of the quaternion of the star sensor in the test period, taking data obtained by calculating the fitting model according to the test period as model data, and subtracting the two rows of data according to corresponding moments to obtain an error sequence of the model in the test period, wherein the error sequence reflects the conformity degree of the fitting function and the real function. According to statistical knowledge, a significance test level alpha is preset, an error sequence obtained by N samples is taken, N is larger than or equal to 1, a confidence interval of error detection quantity is obtained through calculation, and the confidence coefficient of a fitting model is 1-alpha in the confidence interval.
2033. Obtaining single parameter telemetry data confidence interval
In one embodiment, the confidence analysis is performed by taking the distribution function of the overall Δ as
Obey a normal distribution, i.e. Δ to N (μ)1，σ1) Wherein
According to the statistics t distribution related conclusion, the calculation is carried out to obtain
In the above formula, the first and second carbon atoms are,
is composed of
Sample mean of (1), S1Is composed of
The standard deviation of the sample of (2) is obtained from statistical knowledge1One confidence level of 1- α:
taking alpha as 0.05 and n as 10, calculating the mean value and standard deviation of each detection quantity sequence, obtaining the following table,
and calculating to obtain a confidence interval (-0.00007, 0.00019) with a confidence level of 0.95 about the detection quantity e, wherein the interval can be used as a detection quantity judgment threshold to carry out abnormal detection on the quaternion of the star sensor.
204. The on-line detection is shown in figure 5,
in one embodiment of the present invention,
and determining an abnormality when the difference between the fitting model value of at least one moving time window and the target telemetry data exceeds the confidence interval range of the detection amount of the target telemetry data.
2041. The single parameter detection based on the fitting model error adopts a method of moving a time window, the size of the time window can be set according to the detection requirement, if abnormality is found within 5 minutes, the window is selected within 300 seconds, after the detection window is set, the period starting point of the current real-time telemetering data is selected as the origin, the time window moves along with the natural time, the difference value between the model value in the time window and the telemetering data is compared, if the difference value exceeds the single parameter model error detection amount confidence interval, the difference value is marked as an abnormal value, and when the difference value exceeds the threshold and reaches the specified times, the telemetering data is judged to be abnormal. Because the satellite works in a severe environment and the quaternion telemetering data of the star sensor possibly has data loss or individual outlier, the moving time window detection method designed by the invention has certain fault-tolerant capability for the condition that the telemetering data has a few outliers, and the detection is real-time.
In one embodiment, the detection set D is taken as online data, and a part of fault period data is supplemented to form a detection data set D*And detecting by adopting a moving time window, assuming that the length of the moving time window is 5min, the end point of the window is the current time, and the starting point of the window is 300 seconds ago.
2042. Taking the starting point of the current period of the detection set as the origin, moving a time window along with natural time, monitoring the difference value between the fitting model value and the real-time telemetering data, and judging abnormality when the difference value exceeds the detection quantity confidence interval range; the difference value exceeds the upper value and the lower value of the confidence interval for more than 10 times;
2043. if the difference value is monotonously increased and the abnormal value exceeds 30, the telemetering data is considered to be abnormal, and an alarm popup window is opened.
The invention provides a method for detecting data abnormality of a star sensor, which is a high-orbit satellite star sensor quaternion abnormality detection method based on machine learning, can fully utilize historical telemetering data resources, solves the problems of imprecise abnormality detection technology based on experience threshold, more false alarms and missed alarms, improves the satellite fault detection capability, and provides a new solution for detecting the abnormality of other parameters and components of a satellite.
EXAMPLE III
On-line detection of the confidence interval of the multi-dimensional telemetering associated detection quantity, judging abnormal data when the target telemetering data exceeds the confidence interval range of the quaternion related data under the indicated level; as shown in figure 6 of the drawings,
301. selecting telemetering sample data for preprocessing
The processing method and steps are the same as those of the method and steps for preprocessing the telemetry sample data 201 in fig. 2 to obtain the target telemetry data, and are not repeated here.
302. Selecting the multi-dimensional telemetering associated detection quantity and analyzing the confidence degree to obtain a confidence interval,
after the telemetering sample data is preprocessed, selecting a correlation coefficient detection quantity, and selecting a multidimensional telemetering correlation detection quantity, wherein if a satellite is in a normal working state, the working parameters of components with correlation have determined correlation, and the correlation coefficient is represented as the size of the correlation coefficient in mathematics, when the correlation coefficient approaches to 1 or-1, the correlation between the two parameters is very strong, and when the correlation coefficient approaches to 0, the correlation between the two parameters is very weak or irrelevant. The correlation coefficient between every two quaternions of the star sensor can be used for detecting whether the correlation between the quaternions of the star sensor and the quaternion is normal or not, and accordingly a normalized 4-order covariance matrix can be established for detecting satellite parameter abnormality.
The detection quantity of the correlation coefficient is based on statistical knowledge, and the correlation coefficient calculation formula of two columns of data i, j is as follows:
in the above formula, SijIs the sample covariance, Si，SjAre each qi，qjSelecting the quaternion of the whole-cycle star sensor at different time intervals without faults as a training set, respectively calculating the sample correlation coefficient of the fusion quaternion of the star sensor to obtain a corresponding quaternion correlation coefficient sequence, presetting a significance test level β, and calculating a confidence interval of the correlation coefficient detection quantity, wherein the confidence degree of the quaternion correlation coefficient of the star sensor on the confidence interval is 1- β.
3021. Multidimensional telemetry dependent detection quantity selection, as shown in FIG. 7
And detecting the condition of the correlation coefficient of the target telemetering data in the whole period according to the calculated confidence interval of the significance test level beta of the quaternion correlation coefficient, and judging that the satellite is abnormal when the correlation coefficient calculated by the detection centralized telemetering data, namely the target telemetering data exceeds the confidence interval.
It should be noted that, if the quaternion sample data of the selected star sensor is less than 1 cycle, the calculated correlation coefficient cannot be used for detection due to incomplete expression of the related information, so that in real-time detection, at least one cycle of data needs to be taken from the detection moment backwards for correlation calculation, and the detection is non-real-time.
In one embodiment, the training set D comprises the quaternion q of the star sensor of the satellite A0、q1、q2、q3In a suitable combination, are
T*：{(time,q0、q1、q2、q3)i},i∈(1,2，...，m)
The formula of the respective correlation coefficient of quaternions in the same time period is set as
In the above formula, i, j is 0,1,2,3, which is a sequence of quaternion telemetry values at the same time, SijIs the sample covariance, Si，SjIs the sample standard deviation. Respectively calculating sample correlation coefficients r of quaternions of the star sensor01，r02，r03，r12，r13，r23The value interval of the correlation coefficient is between-1 and 1, wherein 1 represents that two samples are completely correlated, -1 represents that two samples are completely negatively correlated, and 0 represents that two samples are not correlated.
In one embodiment, the quaternion respective correlation coefficients in the same period are the correlation coefficient of quaternion q0 and quaternion q1, the correlation coefficient of quaternion q0 and quaternion q2, the correlation coefficient of quaternion q0 and quaternion q3, the correlation coefficient of quaternion q1 and quaternion q2, the correlation coefficient of quaternion q1 and quaternion q3, and the correlation coefficient of quaternion q2 and quaternion q 3.
The closer to zero, the weaker the correlation between the two samples, and the closer to 1 or-1, the stronger the correlation between the two variables. And taking the quaternion correlation coefficient of the satellite A star sensor as the detection quantity of the multi-dimensional telemetering correlation detection, and when the detection quantity is in the conditions of positive correlation and negative correlation (a boundary +/-1), the confidence interval is unilateral.
3022. Confidence analysis
In one embodiment, assuming that the correlation coefficient between the sensor quaternions of the satellite A star in the fault-free period is normally distributed at (-1, 1), the integration region is compressed from (— ∞, + ∞) to (-1, 1). The correlation coefficient between the fusion quaternions has the problem of double-sided confidence interval (near 0) and the problem of single-sided confidence interval (r)ij→ 1). Following is a negative correlation quaternion q0And q is1For example, the detected quantity r is analyzed12The confidence of the detection interval of (1).
3023. Multi-dimensional telemetry association confidence interval
Let the distribution function of the population P be F (r)12(ii) a ) Obey a normal distribution, i.e. P to N (mu)2，σ2)，P(r12Containing unknown parameters, statistics determined from the training set T for a preset confidence level β
Satisfy the requirement of
According to statistically relevant conclusions, μ2The confidence interval of (a) is:
taking m as 10, calculating the correlation coefficient of each sample as follows:
by calculation, obtain
S2When m is 10, t is t, t is 0.0032, β is 0.05, and table look-upβ(9) 1.8331, a confidence interval (-1, -0.994847) with a confidence level of 0.95 is obtained, i.e. the quaternion q0And q is1The correlation coefficient of (a) has a 95% probability of being within the interval (-1, -0.994847), which can be used to threshold for fused quaternion anomaly detection.
303. On-line detection
Judging abnormal data when the target telemetering data exceeds the confidence interval range of the quaternion correlation coefficient under the inspection level; as shown in fig. 8
3031. Acquiring correlation coefficients of quaternions by using the quaternion sample telemetry data of at least one period in the same time period;
3032. and judging abnormal data when the range of the confidence interval of the quaternion correlation coefficient under the inspection level is exceeded, acquiring a confidence interval corresponding to the detection quantity through learning according to the significance inspection level beta of the quaternion correlation coefficient, comparing the confidence interval with the correlation coefficient of the target telemetering data in the whole period, and judging that the satellite is abnormal when the corresponding quaternion correlation coefficient exceeds the confidence interval.
It should be noted that, if the quaternion sample data of the selected star sensor is less than 1 cycle, the calculated correlation coefficient cannot be used for detection due to incomplete expression of the related information, so that in real-time detection, at least one cycle of data needs to be taken from the detection moment backwards for correlation calculation, and the detection is non-real-time.
In one embodiment, the on-line data of the detection set D is used for calculating the correlation coefficient of each quaternion of the detection set D, the correlation coefficients are respectively compared with the confidence intervals corresponding to the quaternion, and the detection set D falling outside the confidence intervals is judged to be abnormal. Loading the data of abnormal period, and calculating to obtain r01When the correlation coefficient falls outside the confidence interval, the detection routine determines an anomaly.
The invention provides a method for detecting data abnormality of a star sensor, which is a high-orbit satellite star sensor quaternion abnormality detection method based on machine learning, can fully utilize historical telemetering data resources, solves the problems of imprecise abnormality detection technology based on experience threshold, more false alarms and missed alarms, improves the satellite fault detection capability, and provides a new solution for detecting the abnormality of other parameters and components of a satellite.
Example four,
In one embodiment, model fusion is performed separately on the basis of the second embodiment or the third embodiment.
The quaternion time sequence model (target telemetering data time sequence model) of the star sensor obtained by data is not invariable, quaternion telemetering data are generated in real time, the new data also express the change rule of quaternion, the characteristics expressed by the new data are extracted by a learning machine and added into the model through data fusion to express the real-time performance of machine learning, and the step is composed of innovation judgment and model fusion.
1. Judging innovation, namely, judging that normal telemetering data is fused into all normal data sequences;
and detecting newly generated telemetering data, and adding the telemetering data into the learning machine as innovation after confirming that the telemetering data is not abnormal data. For example, after the newly downloaded satellite sensor telemetering data of the satellite A is confirmed to be abnormal in two detection modes, the satellite A is added into a white list in a whole period mode, the telemetering data with abnormality is added into a suspected abnormal yellow list in the whole period mode, and the telemetering data with abnormality is confirmed to be fault data which is added into a black list for future reference.
2. Model fusion
And after innovation judgment, fusing normal telemetering data into all normal data sequences, taking the part of data as a training set, training a learning machine to extract data characteristics and knowledge, and fusing the data characteristics and knowledge into a model to obtain a new model for subsequent detection. Such as: and inputting the newly added data of the white list into the learning machine, and fusing the newly added data of the white list with the previous model to obtain a new model.
The invention provides a method for detecting data abnormality of a star sensor, which comprises the steps of fitting a time sequence model of a quaternion of a satellite star sensor by machine learning of massive historical telemetering data, extracting the characteristics of the quaternion telemetering data, and establishing a quaternion correlation model, wherein compared with the prior modeling, the historical telemetering data are fully utilized; two detection quantities are designed, sufficient sample data is adopted to calculate a confidence interval and confidence coefficient of the detection quantities, a strict mathematical method is used for supporting the quaternion abnormity detection problem, the defect of manually setting a parameter threshold is overcome, and the method is more convincing than an expert system which depends on experience to specify and judge weight. The multi-dimensional telemetering correlation detection technology breaks through the limited thinking of a single parameter fixed threshold, the quaternion detection of the star sensor is regarded as a whole, and the quaternion detection is not carried out on each quaternion, so that the fault phenomenon caused by abnormal correlation among the quaternions can be detected; the two quaternion detection technologies are organically combined, can provide proofings for each other, has the characteristics of strong fault tolerance and high accuracy, and a detection algorithm cannot cause false alarms due to jumping caused by one wild value and two wild values.
Based on the logic structure diagram used in the star sensor data abnormality detection method described in the embodiments corresponding to fig. 1, fig. 2, fig. 3, fig. 4, fig. 5, fig. 6, fig. 7, fig. 8, and fig. 9, embodiments of the present disclosure further provide a computer-readable storage medium, for example, a non-transitory computer-readable storage medium may be a Read Only Memory (ROM), a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like. The storage medium stores computer instructions for executing the method for detecting data abnormality of the star sensor described in the embodiment corresponding to fig. 1 and 3, which is not described herein again.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
Claims (10)
1. A star sensor data anomaly detection method is characterized by comprising the following steps:
selecting telemetry sample data to carry out pretreatment to obtain target telemetry data;
selecting a confidence interval of the detection quantity of the target telemetering data obtained by the detection quantity through confidence analysis; detecting the target telemetering data exceeding the confidence interval as abnormal data on line;
the selected detection amount is selected by a single-parameter detection amount; or, selecting the multi-dimensional telemetering associated detection quantity; the confidence interval is the confidence interval of the error detection quantity of the single-parameter model; or, a confidence interval for the correlation coefficient detection quantity.
2. The anomaly detection method of claim 1, wherein said pre-processing of telemetry sample data is to cull said telemetry sample data outliers, obtain a target telemetry data segment, and perform periodic segmentation and diversity on said target telemetry data segment.
3. The abnormality detection method according to claim 2, wherein the outlier rejection is a rejection of an outlier near a zero value or a rejection of an outlier exceeding an upper and lower limit of the telemetry data;
the elimination of the abnormal values near the zero value refers to scanning from the time starting point of the target telemetering data in at least one moving time window period, sorting the data in the moving time window according to size, obtaining a median of a sorted sequence, adding a triple standard deviation of the data in the moving time window by taking the median as a reference to obtain an upper limit interval and a lower limit interval, and eliminating the data outside the upper limit interval and the lower limit interval;
and the elimination of the abnormal values exceeding the upper limit and the lower limit of the telemetering data refers to the steps of performing statistical analysis on the quaternion telemetering data of the satellite star sensor through a computer program, determining the upper limit and the lower limit of a quaternion telemetering data sequence, and eliminating the abnormal values of the quaternion telemetering data except the upper limit and the lower limit.
4. The abnormality detection method according to claim 2, wherein the period division of the target telemetry data segment means that a time-series change period of the target telemetry data is obtained by machine learning, the target telemetry data segment is scanned from a start point to an end point of the target telemetry data segment in a moving time window, and the telemetry data is divided into periods by identifying a period of the telemetry data segment and the start point and the end point of the period by judging a maximum value or a minimum value in at least one period corresponding to monotonicity of data in the moving time window;
the diversity of the target telemetering data section means that the target telemetering data is divided into 3n groups according to a period, n is more than or equal to 10, and the 3n groups of data are randomly divided into 3 groups: the 3 groups are respectively a training set, a testing set and a detection set; the training set is used for machine learning of a time sequence model of a star sensitive quaternion, the test set is used for rechecking a quaternion model, and the quaternion model is obtained by learning of the training set; the detection set telemetering data refers to a whole-cycle array with a fault time period which is picked out in advance in at least one cycle or real-time telemetering data of a satellite; the target telemetry data is the detection set telemetry data.
5. The anomaly detection method of claim 4, wherein said detection set telemetry data is modeled by fitting to obtain a target telemetry data time series model;
the fitting model is obtained by extracting the periodic characteristics and the time sequence function characteristics of the target telemetering data by a machine learning method and adopting a computer trigonometric function fitting program.
6. The abnormal abnormality detection method according to claim 5, wherein a fitting model value of said target telemetry data time series model is a difference between said fitting model value and said target telemetry data taken for at least one complete cycle at different time intervals as a detection amount.
7. The anomaly detection method according to claim 6, wherein said detected quantity is subjected to a confidence analysis to obtain a confidence interval of the detected quantity of the target telemetry data; the confidence interval is obtained by carrying out statistical analysis on the detection quantity sequence of the N sample telemetering data to obtain the mean value and standard deviation of the detection quantity sequence, presetting the significance test level, and calculating the confidence interval of the error detection quantity, wherein the confidence interval of the error detection quantity is the confidence interval of the detection quantity of the target telemetering data, and N is more than or equal to 1.
8. The abnormality detection method according to any one of claims 1 to 7, characterized in that: detecting at least one moving time window on line, monitoring the difference value of the fitting model value and the target telemetering data, and judging the target telemetering data to be abnormal when the difference value exceeds the confidence interval range of the target telemetering data detection amount; and judging the target telemetry data to be abnormal data more than 10 times when the abnormal value is judged to be abnormal, and judging the target telemetry data to be abnormal data more than 30 times when the abnormal value is monotonically increased.
9. The anomaly detection method according to any one of claims 1 to 2, wherein the confidence interval for obtaining the detection quantity of the target telemetry data by selecting the detection quantity and performing confidence analysis is to select correlation coefficients of pairwise correlation of the quaternion telemetry data at least in one period of the same time period as the detection quantity, calculate the correlation coefficients of M samples, wherein M is more than or equal to 1, preset a significance test level according to a statistical method, and obtain the confidence interval of the quaternion correlation coefficients under the level.
10. The abnormality detection method according to claim 9, characterized in that: the confidence interval of the quaternion correlation coefficient under the level is used for judging abnormal data when the target telemetering data exceeds the confidence interval range of the quaternion correlation coefficient under the level through online detection;
the abnormal data is the quaternion sample telemetering data of at least one period in the same time period, and the correlation coefficient of the quaternion is obtained and is the target telemetering data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010361728.2A CN111611294B (en) | 2020-04-30 | 2020-04-30 | Star sensor data anomaly detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010361728.2A CN111611294B (en) | 2020-04-30 | 2020-04-30 | Star sensor data anomaly detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111611294A true CN111611294A (en) | 2020-09-01 |
| CN111611294B CN111611294B (en) | 2023-06-09 |
Family
ID=72201210
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010361728.2A Active CN111611294B (en) | 2020-04-30 | 2020-04-30 | Star sensor data anomaly detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111611294B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112465044A (en) * | 2020-12-03 | 2021-03-09 | 上海卫星工程研究所 | Satellite working condition identification and segmentation method and system based on sensitive parameter multi-evidence fusion |
| CN112801171A (en) * | 2021-01-25 | 2021-05-14 | 中国商用飞机有限责任公司北京民用飞机技术研究中心 | Sensor false alarm identification method and device, computer equipment and storage medium |
| CN113435317A (en) * | 2021-06-25 | 2021-09-24 | 中国电子科技集团公司第五十四研究所 | Unmanned aerial vehicle telemetry data correlation analysis method |
| CN114858183A (en) * | 2022-03-31 | 2022-08-05 | 北京航天控制仪器研究所 | Star sensor fault self-detection method |
| CN115617023A (en) * | 2022-12-05 | 2023-01-17 | 中国西安卫星测控中心 | Spacecraft attitude control system abnormity positioning method and device |
| CN115641916A (en) * | 2022-09-29 | 2023-01-24 | 杭州准芯生物技术有限公司 | Molecular biology detection method, computer equipment and readable storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103646167A (en) * | 2013-11-22 | 2014-03-19 | 北京空间飞行器总体设计部 | Satellite abnormal condition detection system based on telemeasuring data |
| CN104156615A (en) * | 2014-08-25 | 2014-11-19 | 哈尔滨工业大学 | Sensor test data point anomaly detection method based on LS-SVM |
| CN106374993A (en) * | 2016-08-31 | 2017-02-01 | 长光卫星技术有限公司 | Full-automatic unattended satellite telemetering monitoring platform and data processing method |
| CN108734360A (en) * | 2018-06-19 | 2018-11-02 | 哈尔滨工业大学 | One kind being based on modified ELM prediction models multidimensional telemetry intelligent interpretation method |
| CN109034191A (en) * | 2018-06-19 | 2018-12-18 | 哈尔滨工业大学 | One-dimensional telemetry exception interpretation method based on ELM |
| US20190391833A1 (en) * | 2018-06-25 | 2019-12-26 | Nutanix, Inc. | System and method for managing telemetry data and agents in a telemetry system |
| EP3623964A1 (en) * | 2018-09-14 | 2020-03-18 | Verint Americas Inc. | Framework for the automated determination of classes and anomaly detection methods for time series |
| CN110909822A (en) * | 2019-12-03 | 2020-03-24 | 中国科学院微小卫星创新研究院 | Satellite anomaly detection method based on improved Gaussian process regression model |
-
2020
- 2020-04-30 CN CN202010361728.2A patent/CN111611294B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103646167A (en) * | 2013-11-22 | 2014-03-19 | 北京空间飞行器总体设计部 | Satellite abnormal condition detection system based on telemeasuring data |
| CN104156615A (en) * | 2014-08-25 | 2014-11-19 | 哈尔滨工业大学 | Sensor test data point anomaly detection method based on LS-SVM |
| CN106374993A (en) * | 2016-08-31 | 2017-02-01 | 长光卫星技术有限公司 | Full-automatic unattended satellite telemetering monitoring platform and data processing method |
| CN108734360A (en) * | 2018-06-19 | 2018-11-02 | 哈尔滨工业大学 | One kind being based on modified ELM prediction models multidimensional telemetry intelligent interpretation method |
| CN109034191A (en) * | 2018-06-19 | 2018-12-18 | 哈尔滨工业大学 | One-dimensional telemetry exception interpretation method based on ELM |
| US20190391833A1 (en) * | 2018-06-25 | 2019-12-26 | Nutanix, Inc. | System and method for managing telemetry data and agents in a telemetry system |
| EP3623964A1 (en) * | 2018-09-14 | 2020-03-18 | Verint Americas Inc. | Framework for the automated determination of classes and anomaly detection methods for time series |
| CN110909822A (en) * | 2019-12-03 | 2020-03-24 | 中国科学院微小卫星创新研究院 | Satellite anomaly detection method based on improved Gaussian process regression model |
Non-Patent Citations (7)
| Title |
|---|
| 彭喜元 等: ""航天器遥测数据异常检测综述"", 《仪器仪表学报》 * |
| 彭喜元 等: ""航天器遥测数据异常检测综述"", 《仪器仪表学报》, 15 September 2016 (2016-09-15), pages 1929 - 1945 * |
| 杨甲森等: "基于遥测数据相关性的航天器异常检测", 《仪器仪表学报》 * |
| 杨甲森等: "基于遥测数据相关性的航天器异常检测", 《仪器仪表学报》, no. 08, 19 August 2018 (2018-08-19), pages 27 - 36 * |
| 胡迪等: "基于UKF的卫星星敏故障检测算法研究", 《西北工业大学学报》 * |
| 胡迪等: "基于UKF的卫星星敏故障检测算法研究", 《西北工业大学学报》, no. 06, 15 December 2019 (2019-12-15), pages 61 - 64 * |
| 郑晓坤等: ""基于波形特征提取的卫星姿态控制***故障监测方法"", no. 5, pages 101 - 102 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112465044A (en) * | 2020-12-03 | 2021-03-09 | 上海卫星工程研究所 | Satellite working condition identification and segmentation method and system based on sensitive parameter multi-evidence fusion |
| CN112465044B (en) * | 2020-12-03 | 2022-12-27 | 上海卫星工程研究所 | Satellite working condition identification and segmentation method and system based on sensitive parameter multi-evidence fusion |
| CN112801171A (en) * | 2021-01-25 | 2021-05-14 | 中国商用飞机有限责任公司北京民用飞机技术研究中心 | Sensor false alarm identification method and device, computer equipment and storage medium |
| CN113435317A (en) * | 2021-06-25 | 2021-09-24 | 中国电子科技集团公司第五十四研究所 | Unmanned aerial vehicle telemetry data correlation analysis method |
| CN114858183A (en) * | 2022-03-31 | 2022-08-05 | 北京航天控制仪器研究所 | Star sensor fault self-detection method |
| CN115641916A (en) * | 2022-09-29 | 2023-01-24 | 杭州准芯生物技术有限公司 | Molecular biology detection method, computer equipment and readable storage medium |
| CN115641916B (en) * | 2022-09-29 | 2024-05-03 | 杭州准芯生物技术有限公司 | Molecular biology detection method, computer equipment and readable storage medium |
| CN115617023A (en) * | 2022-12-05 | 2023-01-17 | 中国西安卫星测控中心 | Spacecraft attitude control system abnormity positioning method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111611294B (en) | 2023-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111611294B (en) | Star sensor data anomaly detection method | |
| JP5016519B2 (en) | Diagnostic system and method for predictive condition monitoring | |
| US11669080B2 (en) | Abnormality detection device, abnormality detection method, and program | |
| CN112508105B (en) | Fault detection and retrieval method for oil extraction machine | |
| CN113671917B (en) | Detection method, system and equipment for abnormal state of multi-modal industrial process | |
| US12007745B2 (en) | Apparatus for predicting equipment damage | |
| CN117270514B (en) | Production process whole-flow fault detection method based on industrial Internet of things | |
| JP2020071624A (en) | Abnormality diagnosing apparatus, abnormality diagnosing method and program | |
| CN111753877B (en) | Product quality detection method based on deep neural network migration learning | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium | |
| CN117630797A (en) | Ammeter health state detection method, system and storage medium based on working current | |
| CN116627116A (en) | Process industry fault positioning method and system and electronic equipment | |
| Bond et al. | A hybrid learning approach to prognostics and health management applied to military ground vehicles using time-series and maintenance event data | |
| CN115659271A (en) | Sensor abnormality detection method, model training method, system, device, and medium | |
| CN112069621B (en) | Method for predicting residual service life of rolling bearing based on linear reliability index | |
| CN112949816A (en) | ANN-based method and system for identifying abnormal data of hydrostatic level | |
| CN116448062B (en) | Bridge settlement deformation detection method, device, computer and storage medium | |
| CN112464146B (en) | Key subsystem based on historical telemetering data and single-machine correlation health baseline construction method | |
| Bingjie et al. | An improved similarity-based prognostics method for remaining useful life estimation of aero-engine | |
| Simon et al. | Enhancing the diagnostic performance of condition based maintenance through the fusion of sensor with maintenance data | |
| CN118311529A (en) | Disaster condition monitoring method and system by MiniSAR technology | |
| CN117831026A (en) | Food quality detection method and device based on machine learning | |
| CN118051822A (en) | Equipment running state abnormality detection method based on voiceprint recognition technology | |
| CN117806233A (en) | Intelligent monitoring system and method for numerical control machine tool | |
| CN114330515A (en) | Bridge monitoring data abnormity diagnosis and repair method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |