Airborne navigation data cleaning method based on multi-source sensing data

Abstract

The present application relates to the technical field of electric digital data processing, in particular to an airborne navigation data cleaning method based on multi-source sensing data, comprising: collecting parameter data of airborne navigation at multiple historical time nodes and constructing as parameter sequence, and constructing a multi-scale window with any parameter data in the parameter sequence as the center; using an improved local outlier factor algorithm to calculate the abnormal score of the any parameter data under any scale window, and determining the parameter data with the mean value of the abnormal scores of the any parameter data under all scale windows greater than a set threshold as abnormal data, and cleaning the abnormal data. The present application solves the problem of misjudgment or omission in the data cleaning process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric digital data processing. More particularly, the present application relates to an airborne navigation data cleaning method based on multi-source sensing data. BACKGROUND

[0002] In the field of modern aerospace, unmanned aerial vehicle system and intelligent aircraft navigation control, the airborne navigation system has become a core component to ensure flight safety and perform high-precision tasks. With the continuous improvement of the intelligence level of flight platforms, the airborne navigation system usually integrates multiple sensors, including but not limited to global positioning system (GPS), barometric altimeter, inertial measurement unit (IMU), magnetometer, etc. These sensors collect multi-dimensional navigation parameters such as longitude, latitude, altitude, three-axis acceleration, three-axis angular velocity, etc. at a high frequency, and continuously output a large amount of time series data during flight. While providing rich information for flight control and path planning, such multi-source heterogeneous, high-dimensional and high-frequency navigation data also has problems such as frequent noise interference, frequent local mutation anomalies, and complex spatio-temporal characteristics, which can easily have a negative impact on downstream data fusion, state estimation and control strategies, and even may cause flight deviation, attitude estimation distortion and other safety hazards. Therefore, how to efficiently and accurately clean and identify anomalies in airborne navigation data during flight has become an important issue that needs to be addressed in navigation data preprocessing technology.

[0003] At present, the academic and engineering communities have proposed a variety of anomaly detection methods based on statistical analysis, time series modeling, clustering analysis and machine learning for navigation data cleaning. Among them, the local outlier factor algorithm has become an important tool for detecting navigation data anomalies due to its excellent performance in handling high-dimensional data and local density differences. The local outlier factor algorithm compares the local density relationship between a sample and its neighbor samples to identify data with a local density significantly lower than that of its neighbors, thereby determining it as an outlier.

[0004] However, the existing LOF algorithm has several key technical bottlenecks when applied to navigation data processing, especially in terms of parameter selection. Its performance is highly dependent on the setting of the neighborhood size value. A too small value will cause the algorithm to pay excessive attention to local minor disturbances, easily misjudging noise or normal local fluctuations as anomalies, reducing detection accuracy; while a too large value will flatten the local real changes, resulting in a sluggish response to sudden anomalies, and even missing detection. In a dynamic flight environment, the fluctuation characteristics of navigation parameters differ greatly at different times. A fixed value cannot balance the recognition effect in different scenarios, leading to misjudgment or missed detection problems in the data cleaning process. SUMMARY

[0005] To solve the misjudgment or missed judgment problem in the data cleaning process proposed in the background art, the present application provides the following solutions.

[0006] The present application provides an airborne navigation data cleaning method based on multi-source sensing data, comprising: collecting parameter data of airborne navigation at multiple historical time nodes and constructing as parameter sequence, and constructing a multi-scale window with any parameter data in the parameter sequence as the center; using an improved local outlier factor algorithm to calculate the abnormal score of the any parameter data under any scale window, and determining the parameter data whose average abnormal score under all scale windows is greater than a set threshold as abnormal data, and cleaning the abnormal data; wherein the improved local outlier factor algorithm comprises The value is positively correlated with the initial value, and is inversely correlated with the fluctuation degree within the scale window where the any parameter data is located. The fluctuation degree is the difference between the abnormal degree under the scale window where the parameter data is located and the abnormal degree of the parameter sequence.

[0007] The above technical solution fully considers the change characteristics of navigation parameter data under different time scales by introducing a multi-scale window mechanism, effectively improving the recognition ability of local anomalies. By constructing a multi-scale window and calculating the abnormal score under each scale, the outlying behavior of data at different time granularities can be fully captured, avoiding the missed detection or misjudgment problem that may occur under a single scale.

[0008] Further, the value is: , The initial value is The fluctuation degree within the scale window where the parameter data is located is The floor function is.

[0009] The above technical solution dynamically combines the value of the clustering parameter with the local fluctuation degree, realizes adaptive adjustment of the neighborhood scale, and significantly improves the adaptability and robustness of the abnormal detection algorithm under different data states. Specifically, when the data is in a stable state with small fluctuations, the algorithm tends to use a larger neighborhood range to enhance the recognition ability of subtle outliers. In an unstable state with large fluctuations, the neighborhood range is automatically reduced, thereby reducing the misjudgment caused by local dramatic changes.

[0010] Further, the abnormal degree under the scale window where the parameter data is located is: , The first data under the scale window is a mean value of all data under the scale window, a number of data under the scale window, a mean value of all data in the difference sequence corresponding to the parameter sequence.

[0011] The technical solution introduces an abnormality degree calculation method based on the deviation degree of data in the sliding window and the overall change level, effectively enhancing the sensitivity and discrimination of local abnormal fluctuations. Specifically, by measuring the deviation degree between each data in the window and the mean value, and combining the overall change level after the difference of the entire parameter sequence as an adjustment factor, the evaluation of abnormality degree not only reflects the local data distribution characteristics, but also takes into account the global change trend. This design that combines local deviation and global dynamics can more accurately identify non-typical fluctuations at specific moments, especially when dealing with small mutations caused by external interference or sensor errors during flight, significantly improving the sensitivity and stability of abnormality detection, providing a more reliable basis for subsequent data cleaning and state recognition.

[0012] Further, it also includes calculating the distance between any two parameter data using the local outlier factor algorithm , , the Euclidean distance between the first parameter data and the first parameter data, the time interval sequence number between the first parameter data and the first parameter data.

[0013] The above technical solution introduces spatial difference and time interval sequence number when calculating the distance between any two parameter data, constructing a distance measurement method with more spatio-temporal correlation, thereby improving the perception ability of abnormality detection on time sequence characteristics; traditional methods often ignore the behavior pattern differences that may be caused by collection time differences, while the above technical solution incorporates time factors into distance calculation, which not only can more accurately distinguish between short-term mutations and long-term trend changes, but also effectively suppresses the misjudgment problem caused by too large time span.

[0014] Further, the parameter data is longitude, latitude, altitude, three-axis acceleration or three-axis angular velocity on board.

[0015] Further, the difference sequence is obtained by first-order difference of the parameter sequence.

[0016] Further, the altitude on board is collected by using a barometric altimeter, the longitude and latitude data on board are collected by using GPS, and the three-axis acceleration and three-axis angular velocity on board are collected by using IMU.

[0017] Further, the multi-scale window is constructed, including: taking any parameter data in the parameter sequence as the center, selecting different window lengths respectively, and symmetrically expanding a corresponding number of adjacent data points to the left and right of the any parameter data to form a plurality of sliding windows with different scales centered on the any parameter data, so as to obtain the constructed multi-scale window.

[0018] The technical solution described above constructs a plurality of symmetric sliding windows with different lengths around any data point in the parameter sequence, thereby forming multi-scale observation angles, effectively enhancing the joint perception ability of local features and overall trends of data; through the sliding windows with different scales, sudden phenomena in the short term and slow changes in the medium and long term can be captured at the same time, so that the detection of abnormal points is more comprehensive and accurate, and the deviation risk caused by the fixed window scale is reduced.

[0019] Further, when the parameter data is determined to be abnormal data, the three-point smoothing method is used to correct the parameter data.

[0020] Further, the length of the multi-scale window includes 3, 5 and 7.

[0021] The beneficial effects of the present application are:

[0022] The present application can comprehensively evaluate the abnormal characteristics of data from multiple time scales by fusing multi-source sensing data to construct a parameter sequence, combining a multi-scale sliding window and an improved local outlier factor algorithm, and improves the robustness and accuracy of abnormal detection. The mechanism of adaptively adjusting the neighborhood range according to the fluctuation degree is introduced, so that abnormal data can be flexibly identified under different noise levels or dynamic changing environments, and the adaptability of the algorithm to sudden disturbances and complex flight states is enhanced. At the same time, by introducing the joint distance calculation method of time and space, the description effect of the evolution law of time series data is improved, and the misjudgment caused by pure numerical anomaly is avoided. For the identified abnormal data, the three-point smoothing method is used for correction, which effectively removes the abnormality while retaining the trend characteristics of the original data, significantly improves the continuity and stability of the data, and provides a more reliable input basis for subsequent navigation calculation and state evaluation. Overall, the present application has high adaptability, high precision and good engineering realizability, and can meet the strict requirements of onboard navigation system for high-quality data input in a variable environment. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a flowchart schematically showing a multi-source sensing data-based onboard navigation data cleaning method according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] Multi-source sensing data-based onboard navigation data cleaning method embodiment.

[0025] As Figure 1 shown, the multi-source sensor data based on-board navigation data cleaning method flow chart of the embodiment of the present application comprises the following steps:

[0026] S1: Collect parameter data of on-board navigation at multiple historical time nodes and construct as parameter sequence, and construct multi-scale window with any parameter data in the parameter sequence as the center.

[0027] In one embodiment, in order to realize fine analysis and multi-dimensional dynamic perception of the aircraft state, the collected parameter data includes but is not limited to multi-dimensional time series information such as longitude, latitude, altitude, three-axis acceleration and three-axis angular velocity of the on-board aircraft. These parameter data are continuously sampled based on multiple historical time nodes, thereby forming a multi-dimensional parameter sequence with time correlation and spatial continuity;

[0028] Specifically, the on-board altitude data is collected in real time by a high-precision barometric altimeter, which can effectively capture the slight height change of the aircraft in the vertical direction, and has high sensitivity and resolution; the longitude and latitude data are obtained by GPS, which ensures high-precision positioning of the aircraft position within the range, especially in dynamic flight trajectory reconstruction and track matching; the three-axis acceleration and three-axis angular velocity data are collected by an inertial measurement unit (IMU), thereby reflecting the linear motion state and rotational attitude change of the aircraft in three spatial axes in real time;

[0029] The multi-scale window is constructed, including: taking any parameter data in the parameter sequence as the center, selecting different window lengths respectively, and symmetrically expanding a corresponding number of adjacent data points to the left and right of the any parameter data, forming a plurality of sliding windows with different scales with the any parameter data as the center, thereby obtaining the constructed multi-scale window, and the length of the multi-scale window includes 3, 5 and 7. Through the multi-scale window construction mechanism, feature information can be extracted in parallel on multiple time scales, and the dual sensitivity to short-term local changes and long-term overall trends is improved.

[0030] S2: Calculate the anomaly score of the any parameter data under any scale window by using the improved local outlier factor algorithm.

[0031] In one embodiment, the improved local outlier factor algorithm includes value, the value is: , is the initial value, is the fluctuation degree of the parameter data in the scale window, is a floor function.

[0032] By introducing a neighbor number adjustment mechanism adaptively correlated with the degree of local volatility, the improved local outlier factor algorithm dynamically adjusts the neighborhood range, thereby enhancing the flexibility and accuracy of outlier identification under different volatility environments. When local volatility is large, the algorithm automatically reduces the neighborhood range to enhance sensitivity to local abrupt changes; while in areas with smaller volatility, it maintains a larger neighborhood to enhance the stability and robustness of the judgment. This mechanism effectively balances local sensitivity and global consistency, improving the adaptability of anomaly detection, and is particularly suitable for complex time-series data scenarios with significant dynamic changes.

[0033] The degree of fluctuation is the difference between the degree of anomaly of the parameter data within the scale window and the degree of anomaly of the parameter sequence. By calculating the difference between the degree of anomaly of the parameter data within the local scale window and the overall degree of anomaly of the entire parameter sequence, the relative fluctuation of the current data point in the local environment is effectively measured.

[0034] Anomaly level within the scale window of the parameter data for: , For the first scale window One data point, The mean of all data within the stated scale window. The number of data points within the scale window. This is the mean of all data in the difference sequence corresponding to the parameter sequence. The difference sequence is obtained by performing a first-order difference on the parameter sequence.

[0035] By evaluating the deviation of parameter data from the mean within a specific scale window and combining this with the overall average change level of parameter sequence differences, an anomaly measurement method that balances local volatility and global dynamic changes is constructed. This method not only accurately identifies the degree of anomaly in local data within its neighborhood but also effectively incorporates the moderating effect of the overall trend of parameter sequence changes, thereby avoiding misjudgments caused by short-term noise or minor disturbances and improving the robustness and sensitivity of anomaly detection. Furthermore, this measurement method possesses good versatility and scalability, applicable to different types of multidimensional time-series data, contributing to the construction of more stable and efficient flight status monitoring and early warning systems.

[0036] Furthermore, it also includes calculating the distance between any two parameter data points in the local outlier factor algorithm. , , For the first The parameter data and the first Euclidean distance of each parameter data For the first The parameter data and the first a time interval sequence number of parameter data collection.

[0037] The technical solution comprehensively considers the spatial difference of the data itself and the interval of the collection time when calculating the distance between any two parameter data, effectively introduces a spatiotemporal coupling measurement mechanism. Compared with the traditional method based only on the difference between data values, the evolution law and correlation characteristics of the parameter data in the time sequence can be more accurately reflected, and the misjudgment caused by ignoring the time factor can be avoided.

[0038] S3: determining the parameter data whose mean of the anomaly scores of the parameter data under all scale windows is greater than a set threshold as abnormal data, and cleaning the abnormal data.

[0039] In one embodiment, in order to improve the comprehensive judgment ability of multi-scale abnormal behavior and further improve the quality stability of the parameter data as a whole and the reliability of subsequent processing, a judgment and cleaning mechanism based on multi-scale anomaly scores is proposed. The mechanism first performs statistical analysis on the anomaly scores of any parameter data under all preset scale windows, and obtains the mean of the anomaly scores of the parameter data under each scale. If the mean exceeds the system set abnormal judgment threshold, it can be preliminarily determined that the parameter data shows significant deviation behavior under multiple time scales, and has high abnormal credibility. In this way, misjudgment caused by single-scale local fluctuation or accidental abnormality can be avoided, and the identified abnormal data points have stronger global consistency and time sequence reliability.

[0040] Based on the above determination result, in order to further eliminate the influence of abnormal data on the overall sequence stationarity and subsequent modeling effect, an abnormal data cleaning mechanism is introduced. In the cleaning process, for the data points determined to be abnormal, a three-point smoothing method is used for numerical correction, that is, the values of the adjacent two data points before and after the data point are weighted and averaged with the value of the data point itself, so as to generate a new value more consistent with the local trend. The three-point smoothing method is a classic and efficient sliding mean correction strategy, and its main advantage is that it can not only effectively suppress short-time noise and sudden abnormality in the data, but also can largely retain the local trend characteristics of the original data, avoiding information loss caused by excessive smoothing.

[0041] The introduction of the above cleaning strategy significantly enhances the robustness of the data processing chain. On the one hand, while ensuring the reliability of the abnormal detection result, the continuity and stability of the data are further improved; on the other hand, by retaining the original structural characteristics of the data, higher quality input data is provided for subsequent state analysis, prediction modeling and behavior recognition, and the accuracy, robustness and generalization ability of the overall system are improved.

[0042] The scheme of the present application realizes accurate identification and effective repair of multi-source heterogeneous, fluctuating and locally mutated conditions in airborne navigation data by constructing parameter sequences through fusion of multi-source sensing data and introducing adaptive anomaly detection and data cleaning mechanism under multi-scale sliding window. The improved local outlier factor algorithm is used to dynamically adjust the neighborhood range according to the local fluctuation degree, thereby improving the sensitivity and discrimination accuracy of the algorithm to abnormal points in complex flight states. At the same time, the abnormal score mean under the multi-scale window is combined for anomaly judgment, thereby enhancing the robustness to sudden anomalies and periodic disturbances. Further, a comprehensive measurement method of fusing time interval and data distance is used to realize similarity evaluation more in line with the time sequence characteristics of flight data. The three-point smoothing method is used for cleaning of abnormal points, which takes into account local trend preservation and mutation value suppression, thereby effectively improving the continuity, reliability and engineering usability of the overall data. The above scheme is particularly suitable for data preprocessing and anomaly detection scenarios in high dynamic environments, and provides a high-quality and stable data basis for subsequent navigation, state estimation and flight control modeling.

[0043] In the description of the present specification, the meaning of "a plurality of", "several" is at least two, such as two, three or more, etc., unless otherwise explicitly specified.

[0044] Although the present specification has shown and described several embodiments of the present application, it will be apparent to those skilled in the art that such embodiments are provided only by way of example. Those skilled in the art will think of many changes, changes and alternatives without departing from the idea and spirit of the present application. It should be understood that various alternatives to the embodiments of the present application described herein can be employed in practicing the present application.

Claims

1. An airborne navigation data cleaning method based on multi-source sensing data, characterized in that, The method comprises the following steps: Collecting parameter data of airborne navigation at multiple historical time nodes and constructing parameter sequences, and constructing multi-scale windows with any parameter data in the parameter sequences as the center, which comprises the following steps: selecting different window lengths with any parameter data in the parameter sequences as the center, respectively, and symmetrically expanding a corresponding number of adjacent data points to the left and right of the any parameter data to form a plurality of sliding windows with different scales and with the any parameter data as the center, so as to obtain the constructed multi-scale windows; Calculating the abnormal score of the any parameter data under any scale window by using an improved local outlier factor algorithm, determining the parameter data whose average abnormal score under all scale windows is greater than a set threshold as abnormal data, and cleaning the abnormal data; wherein the improved local outlier factor algorithm comprises a value, said value is positively correlated with an initial value and negatively correlated with a degree of fluctuation within a scale window in which said any parameter data lies. The fluctuation degree is the difference between the abnormal degree of the parameter data under the scale window and the abnormal degree of the parameter sequence; Also included is calculating the distance between any two parameter data in the local outlier factor algorithm , , is the Euclidean distance between the first parameter data and the first parameter data, is the time interval sequence number of the first parameter data and the first parameter data acquisition.

2. The method for cleaning the airborne navigation data based on multi-source sensing data according to claim 1, characterized in that, The value is: , is the initial value, is the degree of fluctuation within the scale window in which the parameter data lies, is a floor function.

3. The method for cleaning the airborne navigation data based on multi-source sensing data according to claim 1, wherein, Anomaly level within the scale window of the parameter data for: , For the first scale window One data point, The mean of all data within the stated scale window. The number of data points within the scale window. It is the mean of all data in the difference sequence corresponding to the parameter sequence.

4. The method for cleaning the airborne navigation data based on multi-source sensing data according to claim 1, wherein, The parameter data is longitude, latitude, height, three-axis acceleration or three-axis angular velocity onboard.

5. The method for cleaning the airborne navigation data based on multi-source sensing data according to claim 3, characterized in that, The difference sequence is obtained by first-order difference of the parameter sequence.

6. The method for cleaning the airborne navigation data based on multi-source sensing data according to claim 4, characterized in that, The height onboard is collected by using a barometric altimeter, the longitude and latitude onboard are collected by using GPS, and the three-axis acceleration and three-axis angular velocity onboard are collected by using IMU.

7. The method for cleaning the airborne navigation data based on multi-source sensing data according to claim 1, characterized in that, The cleaning of the abnormal data comprises the following steps: when the parameter data is determined as abnormal data, the parameter data is modified by using a three-point smoothing method.

8. The method for cleaning the airborne navigation data based on multi-source sensing data according to claim 1, wherein, The length of the multi-scale window comprises 3, 5 and 7.

Images