Priori-free satellite telemetry data mode autonomous division method

Abstract

The invention discloses a priori-free satellite telemetry data mode autonomous division method, and relates to a satellite telemetry data mode autonomous division method. The invention aims to solve the problem that an existing method is difficult to accurately divide a data mode in an unsupervised scene and limits comprehensive analysis of a complex satellite operation state. According to the method, automatic judgment and fusion of data modes are realized by utilizing a rasterization adaptive data category / mode mining technology. Through adaptive mode mining of on-orbit data, telemetry data modes in different operation states can be accurately identified, and an alarm threshold value is dynamically adjusted based on mode characteristics. In addition, screening and compression of satellite telemetry data can be optimized, data redundancy is reduced, and the real-time response capability and fault positioning efficiency of the system are improved. The invention belongs to the technical field of satellites.

Description

Technical Field

[0001] The invention relates to a method for autonomously dividing satellite telemetry data modes, belonging to the technical field of satellites. Background Art

[0002] In satellite systems, telemetry data processing is a core component. This data includes not only the satellite's navigation message and ephemeris information, but also key operational parameters such as clock corrections, ionospheric delay corrections, and atmospheric refraction corrections. This information is transmitted to user receivers via satellite signals, where it is decoded and processed to obtain precise satellite positioning information. However, with the increasing number of satellites and the diversification of navigation needs, telemetry data processing requires even greater real-time performance and accuracy.

[0003] At present, the pattern division of satellite telemetry data mainly relies on the experience of designers in the development and testing stages, and is based on manually set fixed thresholds and classification rules. However, this method has significant limitations. First, the manually set pattern division is difficult to cover all the complex working conditions of satellite in-orbit operation, resulting in some abnormal patterns not being included in the preset classification system, affecting the effective identification of faults. Secondly, static thresholds cannot adapt to the dynamic operating environment of satellites (such as temperature fluctuations, payload switching, etc.), and are prone to misjudgment or missed judgments. In addition, traditional methods often fail to fully utilize the statistical characteristics of historical in-orbit data and lack an optimization mechanism based on data self-learning, making it difficult for pattern division and threshold adjustment to adapt to the actual situation of satellite operation.

[0004] In practical applications, telemetry data often lacks prior knowledge, meaning it's impossible to predict the number or characteristics of pattern categories. This makes traditional clustering algorithms (such as K-means and DBSCAN) difficult to apply reliably. K-means requires a preset number of categories, but the complexity of satellite telemetry data makes it difficult to determine an appropriate number of categories. DBSCAN relies on density parameter settings, but data distribution varies significantly across different satellites or under different operating conditions, making fixed parameters difficult to adapt to all situations. Consequently, existing methods struggle to accurately classify data patterns in unsupervised scenarios, limiting comprehensive analysis of complex satellite operational states. Summary of the Invention

[0005] In order to solve the problem that existing methods are difficult to accurately divide data patterns in unsupervised scenarios, which limits the comprehensive analysis of complex satellite operating states, the present invention proposes a method for autonomously dividing satellite telemetry data patterns without prior knowledge.

[0006] The technical solution adopted by the present invention to solve the above problems is: the steps of the present invention include:

[0007] Step 1: When processing the data of each telemetry parameter, use the 3σ principle to eliminate outliers;

[0008] Step 2: Assign all data points to corresponding intervals and count the number of data points in each interval;

[0009] Step 3: Perform rule filtering, remove unreasonable intervals, and merge continuous valid intervals.

[0010] Furthermore, step 1 specifically includes:

[0011] Set the normal mode range of a parameter X to:

[0012] [μ-3σ, μ+3σ] (1),

[0013] In formula (1), μ represents the mean of parameter X, and σ represents the standard deviation of parameter X;

[0014]

[0015] In formulas (2) and (3), d i Represents each measurement value, if d i If the data is not in [μ-3σ, μ+3σ], it is considered an outlier and is removed.

[0016] Furthermore, for data without a fixed pattern in step 2, we first divide the intervals according to the distribution of the data, and count the number and probability of data points in each interval; let the maximum value be d max , the minimum value is d min , divide the data into num_intervalsge intervals, each interval width is:

[0017]

[0018] Each data point d i Falling into the interval [a k ,a k+1 ], where a k Indicates the starting value of the interval, and sets the number of data points in each interval to C k , then the frequency of interval k is:

[0019]

[0020] In formula (5), C k Represents the number of data points in interval k, N represents the total number of data points, and the probability of each interval reflects the concentration or frequency of occurrence of the data in that interval.

[0021] Furthermore, in step 3, after the interval division and probability statistics, the algorithm filters out unreasonable intervals through rules and retains valid data intervals; the specific steps are:

[0022] Step 301: If the frequency P of a certain interval k Indicates that there is no data point in this interval. The algorithm considers this interval to be invalid, so it is proposed. If P k =P k+i =0%, 2≤k≤4, then the interval from k+1 to k+i-1 must be merged;

[0023] Step 302: If two adjacent intervals P k and P k+1 If the frequency difference of the two intervals exceeds 45%, it is considered that the two intervals belong to different data distributions and they are separated to prevent erroneous analysis caused by merging intervals with too large probability differences.

[0024] |P k -P k+1 |>0.45;

[0025] Step 303: If P k >5%, then the interval between k-1 and k is separated, and the data in this interval is considered to be reliable, and the interval is retained; in this way, the high-frequency interval is retained to avoid losing important data features; if P k If the value is less than or equal to 5%, k-1 and k are merged. The merging process aims to unify data intervals with similar probability distributions, reduce excessive subdivisions, and ensure that the analysis results are more stable and consistent.

[0026] The beneficial effects of this invention include: It proposes a method for autonomously classifying satellite telemetry data patterns without prior knowledge, utilizing "rasterized adaptive data classification / pattern mining" technology to automatically identify and integrate data patterns. Through adaptive pattern mining of in-orbit data, this method accurately identifies telemetry data patterns under different operating conditions and dynamically adjusts alarm thresholds based on pattern characteristics. Furthermore, this method optimizes the screening and compression of satellite telemetry data, reduces data redundancy, and improves the system's real-time responsiveness and fault location efficiency.

[0027] By achieving more refined pattern division, satellite on-orbit management can precisely adjust thresholds, avoiding the misjudgments and missed detections caused by traditional static thresholds and improving fault detection accuracy. This also facilitates the construction of refined monitoring models, building adaptive monitoring models based on real-world data and improving the ability to identify abnormal conditions. In terms of operations and maintenance, this reduces the scope of troubleshooting, improves satellite maintenance efficiency, and enhances system reliability and safety.

[0028] This invention not only provides a more intelligent means of processing telemetry data for satellite systems, but also provides data-driven decision support for the long-life operation of future satellites, promoting the development of satellite intelligent operation and maintenance technology.

[0029] The present invention combines real-time statistics (mean μ, standard deviation σ) to dynamically calculate the normal mode range (3σ principle), thereby avoiding the problem of insufficient adaptability caused by a fixed threshold.

[0030] The present invention uses the "gridized adaptive pattern mining" technology to automatically divide intervals based on the distribution of telemetry data without relying on manually preset category numbers or clustering parameters.

[0031] The system of the present invention can perform pattern determination and fusion on data without specific patterns. This helps to more accurately understand the operating status of satellites, timely identify potential problems, and improve the reliability and safety of satellite systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a scatter plot of ZCS007 data of telemetry parameters of Earth orbit satellite;

[0033] Figure 2 This is a schematic diagram of the satellite data mode division results;

[0034] Figure 3 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0035] Specific implementation method 1: Figure 3 As shown in FIG, a method for autonomously dividing satellite telemetry data patterns without prior knowledge includes the following steps:

[0036] Step 1: When processing the data of each telemetry parameter, use the 3σ principle to eliminate outliers; specifically,

[0037] Set the normal mode range of a parameter X to:

[0038] [μ-3σ, μ+3σ] (1),

[0039] In formula (1), μ represents the mean of parameter X, and σ represents the standard deviation of parameter X;

[0040]

[0041] In formulas (2) and (3), d i Represents each measurement value, if d i If it is not in [μ-3σ, μ+3σ], the data is considered to be an outlier and is removed;

[0042] Step 2: Assign all data points to corresponding intervals and count the number of data points in each interval;

[0043] For data without a fixed pattern, we first divide the intervals according to the distribution of the data and count the number and probability of data points in each interval; let the maximum value be d max, the minimum value is d min , divide the data into num_intervalsge intervals, each interval width is:

[0044]

[0045] Each data point d i Falling into the interval [a k ,a k+1 ], where a k Indicates the starting value of the interval, and sets the number of data points in each interval to C k , then the frequency of interval k is:

[0046]

[0047] In formula (5), C k Represents the number of data points in interval k, N represents the total number of data points, and the probability of each interval reflects the concentration or frequency of occurrence of the data in that interval;

[0048] Step 3: Perform rule filtering, remove unreasonable intervals, and merge consecutive valid intervals;

[0049] After interval division and probability statistics, the algorithm filters out unreasonable intervals through rules and retains valid data intervals. The specific steps are:

[0050] Step 301: If the frequency P of a certain interval k Indicates that there is no data point in this interval. The algorithm considers this interval to be invalid, so it is proposed. If P k =P k+i =0%, 2≤k≤4, then the interval from k+1 to k+i-1 must be merged;

[0051] Step 302: If two adjacent intervals P k and P k+1 If the frequency difference of the two intervals exceeds 45%, it is considered that the two intervals belong to different data distributions and they are separated to prevent erroneous analysis caused by merging intervals with too large probability differences.

[0052] |P k -P k+1 |>0.45;

[0053] Step 303: If P k >5%, then the interval between k-1 and k is separated, and the data in this interval is considered to be reliable, and the interval is retained; in this way, the high-frequency interval is retained to avoid losing important data features; if P kIf the value is less than or equal to 5%, k-1 and k are merged. The merging process aims to unify data intervals with similar probability distributions, reduce excessive subdivisions, and ensure that the analysis results are more stable and consistent.

[0054] After outlier removal, interval partitioning, data filtering, and interval merging, the final result is a set of valid intervals, interval_res. Each interval contains its starting and ending values, along with their corresponding probabilities, reflecting the distribution characteristics of the data within that interval and defining a new artificial partitioning pattern. The core of this process is based on statistical principles. Through reasonable interval partitioning and probability statistics, outliers are identified and removed from the data, ultimately extracting reliable and representative data intervals.

[0055] Example

[0056] A method for autonomously dividing satellite telemetry data patterns without prior knowledge, comprising the following steps:

[0057] Step 1: Prepare satellite telemetry data;

[0058] Step 2: Parameter configuration of the autonomous satellite telemetry data pattern partitioning method without prior knowledge.

[0059] This embodiment uses the telemetry parameter ZCS007 data of a low-orbit satellite. The telemetry data points of this telemetry parameter from 00:00:00 on July 1, 2022 to 23:59:00 on July 3, 2022 are 129,481. The telemetry data scatter plot within the time range of 00:00:00 on July 1, 2022 to 00:03:42 on July 1, 2022 is as follows: Figure 1 The parameters required in the algorithm are as follows:

[0060] Number of equal parts n: 20

[0061] The division results of the telemetry parameter ZCS007 data of a low-orbit satellite are as follows: Figure 2 In the method output, we can see that the pattern differentiation of ZCS007 data is presented, which is consistent with the method technical solution.

[0062] It can be seen that the method proposed in this scheme can combine satellite telemetry data and field knowledge such as pattern settings to calculate a series of data adaptive partitioning results, intelligently distinguish pattern information, and has good satellite telemetry data pattern adaptive partitioning capabilities.

[0063] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any technician familiar with the present profession can make some changes or modifications to equivalent embodiments of equivalent changes using the technical content disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent replacement and improvement of the above embodiments made according to the technical essence of the present invention, within the spirit and principles of the present invention, without departing from the content of the technical solution of the present invention, shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for autonomously dividing satellite telemetry data patterns without prior knowledge, characterized in that: The specific steps include: Step 1: When processing the data of each telemetry parameter, use the 3σ principle to eliminate outliers; Step 2: Assign all data points to corresponding intervals and count the number of data points in each interval; Step 3: Perform rule filtering, remove unreasonable intervals, and merge continuous valid intervals.

2. The method for autonomously dividing satellite telemetry data patterns without prior knowledge according to claim 1, characterized in that: Step 1 specifically includes: Set the normal mode range of a parameter X to: [μ-3σ, μ+3σ] (1), In formula (1), μ represents the mean of parameter X, and σ represents the standard deviation of parameter X; In formulas (2) and (3), d i Represents each measurement value, if d i If the data is not in [μ-3σ, μ+3σ], it is considered an outlier and is removed.

3. The method for autonomously dividing satellite telemetry data patterns without prior knowledge according to claim 1, characterized in that: In step 2, for data without a fixed pattern, we first divide the intervals according to the distribution of the data and count the number and probability of data points in each interval; let the maximum value be d max , the minimum value is d min , divide the data into num_intervalsge intervals, each interval width is: Each data point d i Falling into the interval [a k ,a k+1 ], where a k Indicates the starting value of the interval, and sets the number of data points in each interval to C k , then the frequency of interval k is: In formula (5), C k Represents the number of data points in interval k, N represents the total number of data points, and the probability of each interval reflects the concentration or frequency of occurrence of the data in that interval.

4. The method for autonomously dividing satellite telemetry data patterns without prior knowledge according to claim 1, characterized in that: In step 3, after interval division and probability statistics, the algorithm filters out unreasonable intervals through rules and retains valid data intervals. The specific steps are: Step 301: If the frequency P of a certain interval k Indicates that there is no data point in this interval. The algorithm considers this interval to be invalid, so it is proposed. If P k =P k+i =0%, 2≤k≤4, then the interval from k+1 to k+i-1 must be merged; Step 302: If two adjacent intervals P k and P k+1 If the frequency difference of the two intervals exceeds 45%, it is considered that the two intervals belong to different data distributions and they are separated to prevent erroneous analysis caused by merging intervals with too large probability differences. |P k -P k+1 |>0.45; Step 303: If P k >5%, then the interval between k-1 and k is separated, and the data in this interval is considered to be reliable, and the interval is retained; in this way, the high-frequency interval is retained to avoid losing important data features; if P k If the value is less than or equal to 5%, k-1 and k are merged. The merging process aims to unify data intervals with similar probability distributions, reduce excessive subdivisions, and ensure that the analysis results are more stable and consistent.

Images