Deep analysis method for on-orbit data fault diagnosis result based on high-order statistics

Abstract

This invention relates to the field of on-orbit telemetry data analysis, and particularly to a method for in-depth analysis of on-orbit data fault diagnosis results based on high-order statistics. First, historical fault diagnosis conclusion data is segmented by a preset time length, and the types, durations, and switching processes of the diagnosis conclusions are statistically analyzed to form high-order statistical information. Second, different conclusion types are labeled with color bands of different lengths according to their duration, and these bands are then stitched together according to the switching process of the diagnosis conclusion types to form a changing color band diagram. Next, based on manually labeled fault tags, a diagnostic model is trained using a deep learning algorithm. Finally, the statistical analysis and color band diagram drawing process are repeated for new data, and the trained diagnostic model outputs abnormal diagnosis conclusions. This invention can effectively filter out false alarms, uncover deep fault patterns, significantly reduce the burden of manual analysis, and achieve keen early warning of early fault symptoms, thus improving the automation level and reliability of on-orbit data fault diagnosis.

Description

Technical Field

[0001] This invention relates to the field of on-orbit telemetry data analysis technology, and in particular to a method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics. Background Technology

[0002] With the rapid development of aerospace technology, the scale of spacecraft telemetry data continues to increase. The continuous expansion of downlink bandwidth supports comprehensive data acquisition and also places higher demands on the efficiency and accuracy of fault diagnosis for on-orbit data. Currently, fault diagnosis methods for spacecraft on-orbit telemetry data are mainly divided into three categories: model-based methods, knowledge-based methods, and data-based methods. Among them, knowledge-based fault diagnosis expert systems have been widely used in comprehensive testing of payload ground electrical performance and during on-orbit flight due to their excellent interpretability and real-time performance.

[0003] The core advantage of this type of fault diagnosis expert system lies in its ability to quickly provide diagnostic conclusions for each frame of telemetry data, offering real-time reference for ground monitoring personnel. However, it also has significant technical limitations: its diagnostic effectiveness is highly dependent on an accurate fault diagnosis knowledge system corresponding to the state switching process of the measured object. In the early stages of system application, due to the incomplete fault diagnosis knowledge base, a large number of false alarms without corresponding actual faults are easily generated. Even in the later stages of application with relatively complete knowledge, the telemetry signal is prone to instability due to external environmental factors such as rain attenuation and signal blockage, still leading to frequent false alarms. Frequent false alarms not only increase the workload of ground monitoring personnel but also easily lead to the neglect of potential patterns in alarm frequency and alarm patterns. These changes often contain important information about abnormal equipment status or early faults. If these are not captured and analyzed in time, small faults may gradually escalate, ultimately affecting the normal on-orbit operation of spacecraft.

[0004] Therefore, there is an urgent need for a method that can perform in-depth analysis of the fault diagnosis conclusions of fault diagnosis expert systems, and conduct in-depth mining and analysis of the fault diagnosis conclusions from a high-order statistical level, so as to reduce false alarm interference and improve the accuracy and timeliness of fault identification. Summary of the Invention

[0005] In view of this, the present invention aims to provide a method for in-depth analysis of on-orbit data fault diagnosis results based on high-order statistics, in order to solve the problems of high false alarm rate and difficulty in effectively mining early fault signs from massive diagnostic conclusions in existing knowledge-based fault diagnosis expert systems.

[0006] To achieve the above objectives, the technical solution created by this invention is implemented as follows: A method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics includes the following steps: S1: Obtain historical fault diagnosis conclusion data generated by the fault diagnosis expert system, divide the historical fault diagnosis conclusion data into segments with a preset time length, and statistically analyze the high-order statistical information within each time segment. The high-order statistical information includes the type of diagnosis conclusion, the duration of each type of diagnosis conclusion, and the switching sequence between different types of diagnosis conclusion. S2: Map the high-order statistical information within each time period into a color band diagram; wherein, assign corresponding colors and weighting values ​​to different diagnostic conclusion types, map the duration of each diagnostic conclusion type within each time period to the length of the corresponding color band in the color band diagram according to the weighting value ratio, and stitch the color bands together to form a complete color band diagram according to the switching sequence of diagnostic conclusion types. S3: Ground monitoring personnel label the color bands with the presence or absence of equipment faults based on the analysis results of historical fault diagnosis data; all labeled historical color bands are input into a neural network model for supervised deep learning to establish a diagnostic model for identifying whether the equipment status corresponding to the color band is faulty; S4: Real-time acquisition of fault diagnosis conclusion data generated by the fault diagnosis expert system. Execute steps S1 and S2 to form a new color band diagram. Use the diagnostic model to diagnose whether the new color band diagram is faulty.

[0007] Furthermore, in step S2, the length of each color band in the color band diagram is calculated according to the following formula: ; ; in, This indicates the total length of the color band diagram. In the color band diagram, the first... A colored band, In the color band diagram, the first... The length of each color band Indicates the first The weighted values ​​of the diagnostic conclusion types corresponding to each color band. Indicates the first The duration of the diagnostic conclusion type corresponding to each color band. This indicates the total number of color bands in the color chart. This represents the sum of the weighted durations of all color bands in the color band diagram.

[0008] Furthermore, in step S1, the diagnostic conclusion types include at least normal, alert, abnormal, and no conclusion.

[0009] Furthermore, the weighted values ​​of the alert and abnormal diagnosis conclusion types are both greater than the weighted values ​​of the normal and no-conclusion diagnosis conclusion types, and the weighted value of the abnormal diagnosis conclusion type is greater than the weighted value of the alert diagnosis conclusion type.

[0010] Furthermore, the correspondence between the color bands and the types of diagnostic conclusions is as follows: The gray band corresponds to the "no conclusion" diagnostic conclusion type. The green band corresponds to a normal diagnostic conclusion type; The yellow band indicates the type of diagnostic conclusion. The red band corresponds to the type of abnormal diagnostic conclusion.

[0011] Furthermore, in step S1, the preset time length is one day, one week, or one month.

[0012] Furthermore, in step S3, the neural network model is a convolutional neural network.

[0013] Compared with the prior art, the present invention can achieve the following beneficial effects: 1. This invention analyzes and visualizes high-order statistical information (such as type, duration, and switching sequence) of historical fault diagnosis conclusion data, which can effectively filter out random false alarms caused by incomplete knowledge base or instantaneous interference (such as rain attenuation), and discover truly meaningful fault modes from a higher dimension, significantly improving the credibility of the diagnosis results.

[0014] 2. This invention transforms long-term, discrete diagnostic conclusions into intuitive color-coded graphs and uses a trained diagnostic model for automatic analysis, enabling rapid and automatic diagnosis of new data. This reduces the workload of ground monitoring personnel and greatly improves the efficiency of on-orbit data monitoring and analysis.

[0015] 3. By focusing on the changing patterns of high-order characteristics such as fault frequency, duration, and switching sequence, this invention can discover slowly developing abnormal trends or regularly occurring minor anomalies that are masked by a large number of false alarms. This helps to achieve early warning of faults and buys valuable time for taking preventive maintenance measures.

[0016] 4. As a visualization tool, color-coded charts visually present abstract, time-varying diagnostic conclusions in the form of colors and lengths, enabling ground monitoring personnel to quickly grasp the overall health status and changes of a spacecraft over a period of time, facilitating retrospective analysis, status assessment, and decision support. Attached Figure Description

[0017] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments and descriptions of the invention are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 A flowchart illustrating the in-depth analysis method for on-orbit data fault diagnosis results based on higher-order statistics as described in the embodiments of the present invention; Figure 2 A logical schematic diagram of the in-depth analysis method for on-orbit data fault diagnosis results based on higher-order statistics as described in the embodiments of the present invention; Figure 3 This is a schematic diagram of a color band chart drawn based on fault diagnosis conclusion data within 24 hours, as described in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not constitute a limitation thereof.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0020] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on this invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0021] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0022] The invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] like Figure 1 and Figure 2As shown in the figure, this invention provides a method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics, including the following steps: S1: Obtain historical fault diagnosis conclusion data generated by the fault diagnosis expert system, segment the historical fault diagnosis conclusion data into a preset time length, and statistically analyze the high-order statistical information within each time period after segmentation. The high-order statistical information includes the diagnosis conclusion type, the duration of each diagnosis conclusion type, and the switching sequence between different diagnosis conclusion types.

[0024] The preset time length can be a fixed time length such as one day, one week, or one month, depending on the on-orbit data acquisition cycle of the spacecraft payload.

[0025] Diagnostic conclusion types include at least four types: normal, warning, abnormal, and no conclusion. These can be further refined according to actual needs. The following explanation uses the four types of normal, warning, abnormal, and no conclusion as examples.

[0026] Normal, Alert, and Abnormal are graded health status conclusions output by the fault diagnosis expert system after comparing telemetry data with preset thresholds. The three are classified according to the degree of deviation of telemetry parameters, equipment stability, and potential risks.

[0027] Normal: The on-orbit telemetry data is completely within the preset safe operating range, the equipment operation status meets the design expectations, there are no deviations from the expectations, and no fault-related judgment conditions are triggered.

[0028] Reminder: On-orbit telemetry data has approached but not yet exceeded the preset threshold, or has shown a brief and slight deviation from the preset threshold, but has not had a substantial impact on the normal operation of the equipment. This indicates that attention is needed, but it has not yet constituted an equipment malfunction.

[0029] Anomaly: On-orbit telemetry data has significantly exceeded the preset threshold, which has affected the normal operation of the equipment and constitutes an equipment malfunction, requiring immediate action.

[0030] No conclusion: This means that no on-orbit telemetry data was detected, such as, but not limited to, situations where the spacecraft is outside the telemetry and control area.

[0031] The severity of the "no conclusion" diagnostic conclusion type can be considered the same as that of the "normal" diagnostic conclusion type, while the severity of the "normal," "warning," and "abnormal" diagnostic conclusion types increases in that order: No conclusion = Normal < Warning < Abnormal.

[0032] Using a specific day (24 hours) as the preset time length t, the historical fault diagnosis conclusion data of a certain telemetry data is segmented, as shown in Table 1: Table 1. Fault diagnosis conclusions for a certain telemetry data over one day.

[0033] S2: Map the high-order statistical information within each time period into a color band diagram; assign corresponding colors and weighting values ​​to different diagnostic conclusion types, map the duration of each diagnostic conclusion type within each time period to the length of the corresponding color band in the color band diagram according to the weighting value ratio, and stitch the color bands together to form a complete color band diagram according to the switching sequence of diagnostic conclusion types.

[0034] Different diagnostic conclusion types are labeled with color bands of different lengths (e.g., red, green, blue, etc.) according to the duration of each diagnostic conclusion type, and then stitched together according to the switching process of the diagnostic conclusion types to form a changing color band diagram.

[0035] Let the total length of the color stripe be set as Then the first i Duration bits Length of the color band for: ;

[0036] in, Indicates the first The weighted values ​​of the diagnostic conclusion types corresponding to each color band. This represents the sum of the weighted durations of all color bands in the color band diagram. In the color band diagram, the first... A colored band, This indicates the total number of color bands in the color chart.

[0037] The weighting value of the diagnostic conclusion type increases according to the severity. The larger the weighting value, the longer the duration of the diagnostic conclusion type is mapped to the length of the color band in the color chart.

[0038] In one example of the invention, a A color-coded chart of the data in Table 1 is plotted using fixed specifications. The height is fixed at 50px, and the overall width of 400px represents 24 hours. From left to right, each color represents the fault diagnosis conclusion for each moment of the day. Gray indicates no conclusion (weighted value of 1), green indicates normal (weighted value of 1), yellow indicates warning (weighted value of 2), and red indicates abnormal (weighted value of 5). The data shown in Table 1 can then be used... Figure 3 The color band diagram shown is representative.

[0039] The color-coded chart, in an intuitive and visual format, comprehensively records the dynamic changes in fault diagnosis conclusions over a specific time period. This not only provides clear training samples for the diagnostic model but also offers a clear traceability basis for manual verification of diagnostic results. When abnormal conclusions occur, ground monitoring personnel can quickly locate the abnormal time period and corresponding diagnostic conclusion type using the color-coded chart, facilitating subsequent fault cause investigation and optimization of the diagnostic knowledge system.

[0040] This invention segments historical fault diagnosis conclusion data by a preset time length, systematically collects core information such as diagnosis conclusion type, duration, and switching process, and presents it through weighted calculation and visual color band chart. It can quickly extract the statistical patterns hidden in the diagnosis conclusion without verifying massive amounts of data frame by frame, which greatly improves the efficiency and completeness of high-order statistical information extraction and solves the problems of low efficiency and easy omission of key features in traditional manual statistical analysis.

[0041] S3: Ground monitoring personnel label the color bands with tags indicating whether equipment malfunctions exist based on the analysis results of historical fault diagnosis data; all labeled historical color bands are input into a neural network model for supervised deep learning to establish a diagnostic model for identifying whether the equipment status corresponding to the color band is faulty.

[0042] Ground monitoring personnel analyze historical fault diagnosis data to determine if there are any abnormal diagnostic conclusions, thus identifying whether a fault actually occurred on that day. This allows them to label these color band charts as training samples.

[0043] If there is no analysis of historical fault diagnosis conclusions, some obviously problematic color band diagrams can be artificially created, labeled with faults, and added to the training samples.

[0044] Neural network models can be selected from supervised neural networks such as convolutional neural networks (CNNs) to learn from labeled color band images, ultimately resulting in a diagnostic model that can determine whether a color band image is normal.

[0045] This invention overcomes the limitations of traditional fault diagnosis expert systems that only focus on single-frame data conclusions. By analyzing the dynamic changes in alarm frequency and switching patterns of diagnostic conclusions, it can keenly capture early abnormal signals that are easily overlooked manually. Through the diagnostic model's learning of historical features, it can accurately identify trend changes before a fault occurs, enabling early warning of potential faults and allowing sufficient time for spacecraft fault investigation and emergency response, significantly improving the safety and stability of spacecraft in orbit.

[0046] S4: Real-time acquisition of fault diagnosis conclusion data generated by the fault diagnosis expert system. Execute steps S1 and S2 to form a new color band diagram. Use the diagnostic model to diagnose whether the new color band diagram is faulty.

[0047] The same statistical and color band plotting process is applied to the newly collected fault diagnosis conclusion data generated by the fault diagnosis expert system to generate a new color band plot. The trained diagnostic model is then used to analyze the new color band plot to determine whether a fault has occurred.

[0048] This invention uses a diagnostic model to automatically identify and classify color band diagrams, directly outputting accurate conclusions of normal / abnormal, replacing the tedious work of manually screening invalid alarms and verifying the authenticity of faults. This effectively alleviates alarm fatigue for ground monitoring personnel, reduces the workload of manual analysis, and improves the operational efficiency of on-orbit data fault diagnosis.

[0049] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0050] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics, characterized in that, Includes the following steps: S1: Obtain historical fault diagnosis conclusion data generated by the fault diagnosis expert system, divide the historical fault diagnosis conclusion data into segments with a preset time length, and statistically analyze the high-order statistical information within each time segment. The high-order statistical information includes the type of diagnosis conclusion, the duration of each type of diagnosis conclusion, and the switching sequence between different types of diagnosis conclusion. S2: Map the high-order statistical information within each time period into a color band diagram; wherein, assign corresponding colors and weighting values ​​to different diagnostic conclusion types, map the duration of each diagnostic conclusion type within each time period to the length of the corresponding color band in the color band diagram according to the weighting value ratio, and stitch the color bands together to form a complete color band diagram according to the switching sequence of diagnostic conclusion types. S3: Ground monitoring personnel label the color bands with the presence or absence of equipment faults based on the analysis results of historical fault diagnosis data; all labeled historical color bands are input into a neural network model for supervised deep learning to establish a diagnostic model for identifying whether the equipment status corresponding to the color band is faulty; S4: Real-time acquisition of fault diagnosis conclusion data generated by the fault diagnosis expert system. Execute steps S1 and S2 to form a new color band diagram. Use the diagnostic model to diagnose whether the new color band diagram is faulty.

2. The method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics according to claim 1, characterized in that, In step S2, the length of each color band in the color band diagram is calculated according to the following formula: ； ； in, This indicates the total length of the color band diagram. In the color band diagram, the first... A colored band, In the color band diagram, the first... The length of each color band Indicates the first The weighted values ​​of the diagnostic conclusion types corresponding to each color band. Indicates the first The duration of the diagnostic conclusion type corresponding to each color band. This indicates the total number of color bands in the color chart. This represents the sum of the weighted durations of all color bands in the color band diagram.

3. The method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics according to claim 1, characterized in that, In step S1, the diagnostic conclusion types include at least normal, alert, abnormal, and no conclusion.

4. The method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics according to claim 3, characterized in that, The weighted values ​​of the alert and abnormal diagnosis conclusion types are both greater than the weighted values ​​of the normal and no conclusion diagnosis conclusion types, and the weighted value of the abnormal diagnosis conclusion type is greater than the weighted value of the alert diagnosis conclusion type.

5. The method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics according to claim 4, characterized in that, The correspondence between color bands and diagnostic conclusion types is as follows: The gray band corresponds to the "no conclusion" diagnostic conclusion type. The green band corresponds to a normal diagnostic conclusion type; The yellow band indicates the type of diagnostic conclusion. The red band corresponds to the type of abnormal diagnostic conclusion.

6. The method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics according to claim 1, characterized in that, In step S1, the preset time length is one day, one week, or one month.

7. The method for in-depth analysis of on-orbit data fault diagnosis results based on higher-order statistics according to claim 1, characterized in that, In step S3, the neural network model is a convolutional neural network.

Images