Radio altimeter test data anomaly detection method and system

Abstract

The invention discloses a radio altimeter test data anomaly detection method and system, and relates to the technical field of altimeter testing, and the method comprises the steps: obtaining historical test data, outputting a first analysis result, collecting real-time flight parameters, calculating dynamic statistic data according to the first analysis result, building a self-adaptive baseline, and constructing a first anomaly detection model; according to the method, the dynamic baseline is constructed, self-adaptive anomaly detection of radio altimeter test data is achieved, accuracy is improved, fusion analysis is carried out in combination with multi-mode flight parameters, and the reliability of the radio altimeter test data is improved, so that the reliability of the radio altimeter test data is improved, and the reliability of the radio altimeter test data is improved. According to the altimeter anomaly detection method, the detection comprehensiveness is enhanced, a physical verification mechanism is introduced, the anomaly detection reliability is ensured, meanwhile, alarm levels are divided according to anomaly results, real-time response is achieved, a powerful guarantee is provided for flight safety, and the intelligence and safety of altimeter anomaly detection are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of altimeter testing, and particularly to a method and system for detecting abnormal radio altimeter test data. Background Art

[0002] As an important navigation device for aircraft, the accuracy of a radio altimeter is directly related to flight safety. The altimeter measures the relative height between the aircraft and the ground or obstacles, providing critical flight information for the pilot. However, during actual flight, due to various factors such as environmental noise, signal interference, and equipment aging, the test data of the radio altimeter may be abnormal, resulting in the pilot receiving incorrect altitude information and potentially causing flight accidents.

[0003] Currently, the Chinese invention patent with the application number CN202110868184.3 discloses a method for improving the low precision of a radio altimeter, mainly including seven steps: determining that the low altitude error after the altimeter is installed is large, detecting and confirming that the altimeter works normally, testing whether the zero position error δ0 of the altimeter on the aircraft is qualified, testing whether the low altitude error δ1 of the altimeter on the aircraft is qualified, calibrating the zero point of the altimeter, readjusting the error of the altimeter and confirming that the performance of the altimeter is qualified, and judging whether the system low altitude error δ2 of the altimeter on the aircraft is qualified. If it is qualified, the process ends; if it is not qualified, it returns to the step of detecting and confirming that the altimeter works normally.

[0004] The above technology is difficult to achieve accurate abnormal detection of radio altimeter test data through dynamic baseline construction, multi-modal fusion detection, physical verification mechanism, and alarm level division, affecting the detection effect. Summary of the Invention

[0005] The technical problem solved by the present invention is that the prior art is difficult to achieve accurate abnormal detection of radio altimeter test data through dynamic baseline construction, multi-modal fusion detection, physical verification mechanism, and alarm level division, affecting the detection effect.

[0006] To solve the above technical problem, the present invention provides the following technical solutions: A method for detecting abnormal radio altimeter test data, including the following steps: Step S1, obtaining historical test data and performing spatio-temporal feature extraction and analysis, and outputting a first analysis result; Step S2, collecting real-time flight parameters according to the first analysis result and storing the real-time flight parameters in a test database; Step S3, calculating dynamic statistic data according to the first analysis result and establishing an adaptive baseline based on the dynamic statistic data; Step S4, construct the first anomaly detection model, obtain the real-time flight phase, detect anomalies and output the anomaly prediction results; Step S5, select the alarm level according to the anomaly prediction results and input it to the flight control terminal.

[0007] Preferably, the step S1 includes the following sub-steps: Step S101, retrieve historical test data from the test database, where the historical test data includes flight parameters in the takeoff phase, cruise phase and landing phase, and the flight parameters include historical altitude values, historical altitude change rates, historical signal strengths and historical environmental noises; Step S102, perform quantum wavelet decomposition on the historical test data and output the time-frequency energy matrix; Step S103, extract spatio-temporal features from the time-frequency energy matrix and calculate the feature correlation coefficient, and output the first analysis result, where the first analysis result includes the takeoff phase feature correlation coefficient, the cruise phase feature correlation coefficient and the landing phase feature correlation coefficient.

[0008] Preferably, the step S2 includes the following sub-steps: Step S201, analyze the first analysis result: If the takeoff phase feature correlation coefficient, the cruise phase feature correlation coefficient or the landing phase feature correlation coefficient is less than the preset reference correlation coefficient, it is determined as an environment-sensitive phase, and the real-time flight parameters corresponding to the takeoff phase feature correlation coefficient, the cruise phase feature correlation coefficient or the landing phase feature correlation coefficient are collected and the high-frequency sampling mode is set. The real-time flight parameters include real-time altitude values, altitude change rates, signal strengths and environmental noises; If the takeoff phase feature correlation coefficient, the cruise phase feature correlation coefficient or the landing phase feature correlation coefficient is greater than or equal to the preset reference correlation coefficient, it is determined as a stable phase, and the preset standard sampling frequency is adopted; Step S202, align the real-time flight parameters according to the time stamp and store them in the test database.

[0009] Preferably, the step S3 includes the following sub-steps: Step S301, calculate the dynamic statistic data according to the first analysis result, where the dynamic statistic data includes the sliding window mean and the standard deviation; The mathematical expression of the sliding window mean is: ; Among them, is the sliding window mean, is the sliding window length, is the real-time flight parameter at the moment, is the time point index within the window, is the current time point; The mathematical expression of the standard deviation is: ; where, is the standard deviation; Step S302: Establish an adaptive baseline based on the dynamic statistic data.

[0010] Preferably, the step S4 includes the following sub-steps: Step S401: Construct a first anomaly detection model based on the adaptive baseline, and calculate the ratio of the local density to the neighboring density of the historical parameters by using the local outlier factor algorithm. The mathematical expression of the local outlier factor algorithm is: ; where, is the local outlier factor of point , is any data point in the real-time flight parameters, is point 's neighboring set, is point 's neighbor point, is a natural number greater than 0, is the local reachability density of point , is the local reachability density of point ; Step S402: Divide the model categories according to the flight phases, and output the cruise phase model, the landing phase model, and the takeoff phase model; Step S403: Calculate the Mahalanobis distance between the current altitude value and the chronological baseline, and analyze the Mahalanobis distance based on the anomaly distance prediction classification logic. If the distance exceeds the preset Mahalanobis distance threshold, it is marked as an abnormal prediction result.

[0011] Preferably, the anomaly result distance prediction classification logic is: If the Mahalanobis distance satisfies that the Mahalanobis distance is greater than or equal to twice the Mahalanobis distance threshold, it is marked as confirmed abnormal; If the Mahalanobis distance satisfies that the Mahalanobis distance is greater than or equal to the Mahalanobis distance threshold and less than twice the Mahalanobis distance threshold, it is marked as suspected abnormal; If the Mahalanobis distance is less than the Mahalanobis distance threshold, it is marked as normal; The confirmed abnormal, suspected abnormal, and normal are combined and output as the abnormal prediction result.

[0012] Preferably, the step S5 includes the following sub-steps: Step S501: Select the alarm level according to the abnormal prediction result: If the abnormal prediction result is confirmed as abnormal, trigger an alarm signal with a first-level alarm level; If the abnormal prediction result is suspected of being abnormal, trigger an alarm signal with a second-level alarm level; If the abnormal prediction result is normal, no alarm is triggered; Step S502, combine the alarm level with the corresponding real-time flight parameters and transmit them to the flight control terminal.

[0013] Preferably, it further includes: Step S601, in combination with the aircraft dynamics model, verify whether the rate of change of altitude exceeds the performance envelope. The constraint condition is: ; Among them, is the rate of change of altitude, is the maximum airspeed, is the maximum angle of attack; Step S602, if the verification in step S601 fails, overwrite the current detection result and trigger an alarm signal with a first-level alarm level.

[0014] Preferably, the mathematical expression of the Mahalanobis distance threshold is: ; Among them, is the Mahalanobis distance threshold, is the chi-square distribution critical value.

[0015] The radio altimeter test data anomaly detection system, which is applied to the radio altimeter test data anomaly detection method described above, includes a data analysis module, a real-time acquisition module, a baseline establishment module, a model construction module, and an anomaly alarm module; The data analysis module is used to obtain historical test data and perform spatio-temporal feature extraction and analysis, and output the first analysis result; The real-time acquisition module is used to collect real-time flight parameters according to the first analysis result and store the real-time flight parameters in the test database; The baseline establishment module is used to calculate dynamic statistic data according to the first analysis result and establish an adaptive baseline based on the dynamic statistic data; The model construction module is used to construct a first anomaly detection model, obtain the real-time flight phase, match the normality and output the anomaly prediction result; The anomaly alarm module is used to select the alarm level according to the anomaly prediction result and input it to the flight control terminal.

[0016] Advantages of the present invention: By constructing a dynamic baseline, the present invention realizes adaptive anomaly detection for radio altimeter test data, improves accuracy, combines multi-modal flight parameters for fusion analysis, enhances the comprehensiveness of detection, introduces a physical verification mechanism to ensure the reliability of anomaly detection. At the same time, by dividing the alarm level according to the anomaly results, real-time response is achieved, providing a strong guarantee for flight safety and significantly improving the intelligence and safety of altimeter anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 FIG. is a flowchart of the steps of a method for detecting anomalies in radio altimeter test data provided by an embodiment of the present invention; Figure 2 FIG. is a schematic diagram of the basic process of a system for detecting anomalies in radio altimeter test data provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific embodiments of the present invention will be made with reference to the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments.

[0019] Embodiment 1, referring to Figure 1 , provides a method for detecting anomalies in radio altimeter test data, including the following steps: Step S1, obtain historical test data and perform spatio-temporal feature extraction and analysis, and output a first analysis result.

[0020] Step S2, collect real-time flight parameters according to the first analysis result, and store the real-time flight parameters in the test database.

[0021] Step S3, calculate dynamic statistic data according to the first analysis result, and establish an adaptive baseline based on the dynamic statistic data.

[0022] Step S4, construct a first anomaly detection model, obtain the real-time flight phase, match and output an anomaly prediction result.

[0023] Step S5, select an alarm level according to the anomaly prediction result and input it to the flight control terminal.

[0024] Step S1 includes the following sub-steps: Step S101, retrieve historical test data from the test database. The historical test data includes flight parameters in the takeoff phase, cruise phase, and landing phase. The flight parameters include historical altitude values, historical altitude change rates, historical signal strengths, and historical environmental noises.

[0025] Step S101 retrieves comprehensive historical test data from the test database, covering flight parameters during takeoff, cruise, and landing phases, providing detailed data support for subsequent analysis.

[0026] Step S102 performs quantum wavelet decomposition on the historical test data and outputs a time-frequency energy matrix.

[0027] Step S102 performs quantum wavelet decomposition on the historical test data, effectively extracts the time-frequency energy matrix, reveals the energy distribution of flight parameters at different frequencies and time scales, and lays a foundation for subsequent feature extraction.

[0028] Step S103 extracts spatio-temporal features from the time-frequency energy matrix and calculates feature correlation coefficients, outputting a first analysis result. The first analysis result includes takeoff-phase feature correlation coefficients, cruise-phase feature correlation coefficients, and landing-phase feature correlation coefficients.

[0029] Step S103 extracts spatio-temporal features from the time-frequency energy matrix, calculates feature correlation coefficients, and outputs a first analysis result closely related to the flight phase, providing a key basis for subsequent anomaly detection and alarm level classification.

[0030] Step S1 extracts spatio-temporal features related to the flight phase from historical test data, providing a basis for subsequent anomaly detection. Through quantum wavelet decomposition and spatio-temporal feature extraction, the variation laws of flight parameters in different phases can be deeply analyzed, thereby outputting feature correlation coefficients closely related to the takeoff, cruise, and landing phases, providing a strong basis for flight safety assessment.

[0031] Step S2 includes the following sub-steps: Step S201 analyzes the first analysis result: If the takeoff-phase feature correlation coefficient, cruise-phase feature correlation coefficient, or landing-phase feature correlation coefficient is less than a preset reference correlation coefficient, it is determined as an environment-sensitive phase, and the real-time flight parameters corresponding to the takeoff-phase feature correlation coefficient, cruise-phase feature correlation coefficient, or landing-phase feature correlation coefficient are collected and a high-frequency sampling mode is set. The real-time flight parameters include real-time altitude value, altitude change rate, signal strength, and environmental noise.

[0032] If the takeoff-phase feature correlation coefficient, cruise-phase feature correlation coefficient, or landing-phase feature correlation coefficient is greater than or equal to the preset reference correlation coefficient, it is determined as a stable phase, and a preset standard sampling frequency is adopted.

[0033] Step S201 intelligently analyzes the first analysis result and determines the environmental sensitivity of the flight phase based on the magnitude of the feature correlation coefficient. For the environmentally sensitive phase, a high-frequency sampling mode is adopted, which can more accurately capture the subtle changes in flight parameters and improve the sensitivity of anomaly detection. In the stable phase, a preset standard sampling frequency is used to balance data quality and acquisition efficiency.

[0034] In step S202, the real-time flight parameters are aligned by timestamp and stored in the test database.

[0035] Step S202 stores the real-time flight parameters in the test database after aligning them by timestamp, ensuring the timeliness and consistency of the data. This ordered data storage method not only facilitates subsequent data analysis and mining but also helps to track anomalies during the flight process, providing strong support for flight safety assessment.

[0036] Based on the first analysis result, step S2 intelligently adjusts the sampling frequency of real-time flight parameters and stores the collected data in an orderly manner. By identifying the environmentally sensitive phase and applying the high-frequency sampling mode, key changes during the flight process can be captured more precisely, providing high-quality data support for anomaly detection. At the same time, aligning and storing the real-time data by timestamp ensures the integrity and traceability of the data, providing a solid foundation for subsequent analysis.

[0037] Step S3 includes the following sub-steps: In step S301, dynamic statistic data is calculated according to the first analysis result. The dynamic statistic data includes the moving window mean and standard deviation.

[0038] The mathematical expression for the moving window mean is: ; where is the moving window mean, is the moving window length, is the real-time flight parameter at the th moment, is the time point index within the window, is the current time point.

[0039] The mathematical expression for the standard deviation is: ; where is the standard deviation.

[0040] Step S301 calculates dynamic statistic data based on the first analysis result. This step can quantify the average level and dispersion degree of flight parameters within the sliding window. The sliding window mean reflects the overall trend of flight parameters, while the standard deviation reveals the fluctuation range of the parameters. These data provide a key basis for establishing an adaptive baseline in the subsequent steps.

[0041] Step S302 establishes an adaptive baseline based on the dynamic statistic data.

[0042] Step S302 establishes an adaptive baseline based on the dynamic statistic data. This step combines the real-time state of flight parameters with historical data to construct a reference standard that can change with flight conditions. The establishment of the adaptive baseline makes anomaly detection more flexible and accurate, and can more effectively identify abnormal situations during flight, providing strong guarantee for flight safety.

[0043] Step S3 constructs an adaptive baseline using the first analysis result to more accurately monitor the real-time state of flight parameters. By calculating the dynamic statistic data, the trend and fluctuation of flight parameters over time can be captured, and then an adaptive baseline that conforms to the current flight state can be established. This step provides a more practical reference standard for subsequent anomaly detection, improving the accuracy and reliability of detection.

[0044] Step S4 includes the following sub-steps: Step S401 constructs a first anomaly detection model based on the adaptive baseline, and uses the Local Outlier Factor (LOF) algorithm to calculate the ratio of the local density to the neighboring density of historical parameters. The mathematical expression of the Local Outlier Factor algorithm is: ; where, is the local outlier factor of point , is any data point in the real-time flight parameters, is the neighboring set of point , is the neighbor point of point , is a natural number greater than 0, is the local reachability density of point , is the local reachability density of point .

[0045] Step S401 constructs the first anomaly detection model based on the adaptive baseline and calculates the ratio of the local density to the neighboring density of the historical parameters using the Local Outlier Factor (LOF) algorithm. This step provides an effective method for identifying outliers by quantifying the degree of dispersion of flight parameters in historical data. The application of the LOF algorithm enables the model to capture local changes in parameters more sensitively, improving the sensitivity of anomaly detection.

[0046] In step S402, the model categories are divided according to the flight phases, and the cruise-phase model, landing-phase model, and takeoff-phase model are output.

[0047] In step S402, the model categories are divided according to the flight phases, and the cruise-phase model, landing-phase model, and takeoff-phase model are output. This step takes into account the differences in flight parameters in different flight phases, providing a more practical classification criterion for the model. By constructing models specifically for different flight phases, the parameter characteristics of each phase can be more accurately reflected, improving the accuracy and pertinence of anomaly detection.

[0048] In step S403, the Mahalanobis distance between the current altitude value and the temporal baseline is calculated, and the Mahalanobis distance is analyzed based on the anomaly distance prediction classification logic. If the distance exceeds the preset Mahalanobis distance threshold, it is marked as an abnormal prediction result.

[0049] The anomaly result distance prediction classification logic is as follows: If the Mahalanobis distance satisfies that the Mahalanobis distance is greater than or equal to twice the Mahalanobis distance threshold, it is marked as a confirmed anomaly.

[0050] If the Mahalanobis distance satisfies that the Mahalanobis distance is greater than or equal to the Mahalanobis distance threshold and less than twice the Mahalanobis distance threshold, it is marked as a suspected anomaly.

[0051] If the Mahalanobis distance is less than the Mahalanobis distance threshold, it is marked as normal.

[0052] The confirmed anomalies, suspected anomalies, and normal results are combined and output as the anomaly prediction result.

[0053] In step S403, the Mahalanobis distance between the current altitude value and the temporal baseline is calculated, and the Mahalanobis distance is analyzed based on the anomaly distance prediction classification logic. This step realizes the precise identification of abnormal situations by quantifying the degree of difference between the current parameter value and the baseline. The application of the anomaly distance prediction classification logic enables the model to classify and mark according to the degree of anomaly, providing clear guidance for subsequent alarm responses. By combining the confirmed anomalies, suspected anomalies, and normal results and outputting them as the anomaly prediction result, the entire anomaly detection process becomes clearer and more organized.

[0054] Step S4 constructs an anomaly detection model based on an adaptive baseline, achieving precise monitoring of flight parameters at different flight stages. By adopting the Local Outlier Factor algorithm and Mahalanobis distance analysis, it can accurately identify abnormal situations during flight and classify and label them according to the degree of abnormality. This step not only improves the accuracy and efficiency of anomaly detection but also provides strong support for subsequent alarm response and flight safety assessment.

[0055] Step S5 includes the following sub-steps: Step S501, select the alarm level according to the anomaly prediction result: If the anomaly prediction result is confirmed anomaly, trigger an alarm signal with a first-level alarm level.

[0056] If the anomaly prediction result is suspected anomaly, trigger an alarm signal with a second-level alarm level.

[0057] If the anomaly prediction result is normal, no alarm is triggered.

[0058] Step S501 selects the alarm level according to the anomaly prediction result. This step accurately judges the anomaly prediction result and intelligently triggers the corresponding alarm level. When a confirmed anomaly occurs, a first-level alarm is triggered, indicating that the situation is urgent and immediate measures need to be taken; when a suspected anomaly occurs, a second-level alarm is triggered, prompting the pilot to pay attention and prepare countermeasures; when the prediction result is normal, no alarm is triggered to avoid unnecessary interference. This step ensures the accuracy and pertinence of the alarm signal and improves the efficiency of flight safety monitoring.

[0059] Step S502, merge the alarm level with the corresponding real-time flight parameters and transmit them to the flight control terminal.

[0060] Step S502 merges the alarm level with the corresponding real-time flight parameters and transmits them to the flight control terminal. This step combines the alarm information with the real-time flight parameters, providing the pilot with comprehensive flight status information. By transmitting this information to the flight control terminal, the pilot can timely understand the abnormal situations during flight and make correct judgments and decisions based on the alarm level and real-time parameters. This step realizes the timely transmission and effective utilization of information and provides strong guarantee for flight safety.

[0061] Step S5 intelligently selects the alarm level according to the anomaly prediction result, and merges the alarm level with the real-time flight parameters and transmits them to the flight control terminal, realizing the timely response and effective management of abnormal situations during flight. This step not only improves the accuracy and efficiency of flight safety monitoring, This method further includes: Step S601, combined with the aircraft dynamics model, verify whether the rate of change of altitude exceeds the performance envelope, and the constraint condition is: ; wherein, is the rate of change of altitude, is the maximum airspeed, is the maximum angle of attack.

[0062] Step S601 combines the aircraft dynamics model to verify whether the rate of change of altitude exceeds the performance envelope. This step uses the aircraft dynamics model to accurately calculate and analyze flight parameters, and can accurately determine whether the rate of change of altitude is within the safe range. Through this verification process, potential flight parameter anomalies can be detected and corrected in a timely manner to ensure the safety and stability of the flight process.

[0063] In step S602, if the verification in step S601 fails, the current detection result is overwritten and an alarm signal with an alarm level of triggering a first-level alarm is triggered.

[0064] The mathematical expression of the Mahalanobis distance threshold is: ; wherein, is the Mahalanobis distance threshold, is the critical value of the chi-square distribution.

[0065] If the verification in step S601 fails, the current detection result is overwritten and an alarm signal with an alarm level of triggering a first-level alarm is triggered. In this step, in the case of verification failure, the highest-priority alarm measure is immediately taken to remind the pilot to pay attention and take emergency measures. By overwriting the current detection result and triggering a first-level alarm, it can be ensured that the pilot obtains key information in the first time, so as to effectively respond to potential flight safety risks.

[0066] This method establishes an adaptive baseline by calculating dynamic statistic data, which can reflect the changes of the flight environment and altimeter performance in real time, thereby improving the accuracy and adaptability of anomaly detection. Compared with the traditional fixed-threshold method, the dynamic baseline can better adapt to data changes in different flight stages and environmental conditions, reducing false alarms and missed alarms.

[0067] This method combines multiple flight parameters such as historical altitude values, rate of change of altitude, signal strength, and environmental noise for spatio-temporal feature extraction and analysis, realizing the fusion detection of multi-modal data. This multi-modal fusion method can more comprehensively reflect the performance state of the altimeter, improving the accuracy and reliability of anomaly detection. At the same time, by constructing the first anomaly detection model and adopting advanced technologies such as the local outlier factor algorithm, the intelligent level of anomaly detection is further improved.

[0068] This method also combines the aircraft dynamics model for physical verification. By verifying whether the rate of altitude change exceeds the performance envelope, the accuracy of the anomaly detection result is further confirmed. This physical verification mechanism not only enhances the reliability of anomaly detection but also can promptly detect and correct potential risks caused by equipment failures or data errors, providing strong guarantee for flight safety.

[0069] This method selects warning signals of different levels according to the anomaly prediction results and transmits them to the flight control end in real time, enabling the pilot to quickly understand the performance status of the altimeter and take corresponding countermeasures. This real-time response mechanism helps reduce the risk of flight accidents and improve the safety and reliability of flight.

[0070] Example 2, referring to Figure 2 , provides a radio altimeter test data anomaly detection system, including a data analysis module, a real-time acquisition module, a baseline establishment module, a model construction module, and an anomaly warning module.

[0071] The data analysis module is used to obtain historical test data and perform spatio-temporal feature extraction and analysis, and output the first analysis result.

[0072] By obtaining historical test data and conducting in-depth spatio-temporal feature extraction and analysis, the data analysis module can accurately capture the variation laws of flight parameters at different stages. The output first analysis result provides key data support and decision-making basis for subsequent modules, helping to improve the accuracy and reliability of the entire anomaly detection system.

[0073] The real-time acquisition module is used to collect real-time flight parameters according to the first analysis result and store the real-time flight parameters in the test database.

[0074] Based on the first analysis result output by the data analysis module, the real-time acquisition module can intelligently collect real-time flight parameters and store them in the test database. This step ensures the timeliness and integrity of the data, providing a rich and accurate data source for subsequent anomaly detection.

[0075] The baseline establishment module is used to calculate dynamic statistic data according to the first analysis result and establish an adaptive baseline based on the dynamic statistic data.

[0076] The baseline establishment module calculates dynamic statistic data using the first analysis result and establishes an adaptive baseline based on these data. The adaptive baseline can be adjusted according to the change of flight conditions, thus more accurately reflecting the current flight state and providing a reference standard close to the actual situation for subsequent anomaly detection.

[0077] The model construction module is used to construct the first anomaly detection model, obtain the real-time flight stage, match the anomaly and output the anomaly prediction result.

[0078] The model construction module constructs the first anomaly detection model, can obtain real-time flight phase information, match the real-time flight parameters with the model, and output the anomaly prediction result. By adopting advanced algorithms and models, this module can achieve real-time monitoring of flight parameters and anomaly identification, providing strong guarantee for flight safety.

[0079] The anomaly warning module is used to select the warning level according to the anomaly prediction result and input it to the flight control terminal.

[0080] The anomaly warning module intelligently selects the warning level according to the anomaly prediction result output by the model construction module and inputs the warning information to the flight control terminal. Through timely and accurate warnings, this module can remind the pilot to pay attention to the anomalies during the flight and take corresponding measures to ensure flight safety. At the same time, this module can also provide strong support for subsequent flight safety assessment and data analysis.

[0081] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. Among them, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Read-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, magnetic disk or optical disk. These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device, and the instruction device realizes the functions specified in one process Figure 1 one process or multiple processes and / or boxes Figure 1 specified in one box or multiple boxes.

[0082] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for detecting abnormal radio altimeter test data, characterized in that It includes the following steps: Step S1: Obtain historical test data, perform spatio-temporal feature extraction and analysis, and output the first analysis result; Step S2: Collect real-time flight parameters according to the first analysis result, and store the real-time flight parameters in the test database; Step S3: Calculate dynamic statistic data according to the first analysis result, and establish an adaptive baseline based on the dynamic statistic data; Step S4: Construct a first anomaly detection model, obtain the real-time flight phase, and output an anomaly prediction result; Step S5: Select an alarm level according to the anomaly prediction result and input it to the flight control terminal.

2. The abnormal detection method for radio altimeter test data according to claim 1, wherein The said Step S1 includes the following sub-steps: Step S101: Retrieve historical test data from the test database. The historical test data includes flight parameters in the takeoff phase, cruise phase, and landing phase. The flight parameters include historical altitude values, historical altitude change rates, historical signal strengths, and historical environmental noises; Step S102: Perform quantization wavelet decomposition on the historical test data, and output a time-frequency energy matrix; Step S103: Perform spatio-temporal feature extraction on the time-frequency energy matrix and calculate the feature correlation coefficient, and output the first analysis result. The first analysis result includes the takeoff phase feature correlation coefficient, the cruise phase feature correlation coefficient, and the landing phase feature correlation coefficient.

3. The radio altimeter test data anomaly detection method according to claim 2, characterized in that, The said Step S2 includes the following sub-steps: Step S201: Analyze the first analysis result: If the takeoff phase feature correlation coefficient, the cruise phase feature correlation coefficient, or the landing phase feature correlation coefficient is less than a preset reference correlation coefficient, it is determined as an environment-sensitive phase, and the real-time flight parameters corresponding to the takeoff phase feature correlation coefficient, the cruise phase feature correlation coefficient, or the landing phase feature correlation coefficient are collected and a high-frequency sampling mode is set. The real-time flight parameters include real-time altitude values, altitude change rates, signal strengths, and environmental noises; If the takeoff phase feature correlation coefficient, the cruise phase feature correlation coefficient, or the landing phase feature correlation coefficient is greater than or equal to the preset reference correlation coefficient, it is determined as a stable phase, and a preset standard sampling frequency is adopted; Step S202: Align the real-time flight parameters according to the time stamp and store them in the test database.

4. The radio altimeter test data abnormal detection method according to claim 3, wherein The said Step S3 includes the following sub-steps: Step S301: Calculate dynamic statistic data according to the first analysis result. The dynamic statistic data includes a sliding window mean and a standard deviation; The mathematical expression of the sliding window mean is: ； Among them, is the moving window mean, is the moving window length, is the real-time flight parameter at time is the time point index within the window, is the current time point; The mathematical expression of the standard deviation is: ； Among them, is the standard deviation; Step S302: Establish an adaptive baseline based on the dynamic statistic data.

5. The abnormal detection method for radio altimeter test data according to claim 4, characterized in that, The said Step S4 includes the following sub-steps: Step S401: Construct a first anomaly detection model based on the adaptive baseline, and use the local outlier factor algorithm to calculate the ratio of the local density to the neighboring density of the historical parameters. The mathematical expression of the local outlier factor algorithm is: ； Among them, is the local outlier factor of the point , is any data point in the real-time flight parameters, is the proximity set of the point , is the neighbor point of the point , is a natural number greater than 0, is the local reachability density of the point , is the local reachability density of the point . Step S402: Divide the model categories according to the flight phase, and output a cruise phase model, a landing phase model, and a takeoff phase model; Step S403: Calculate the Mahalanobis distance between the current altitude value and the time-series baseline, analyze the Mahalanobis distance based on the anomaly distance prediction classification logic. If the distance exceeds a preset Mahalanobis distance threshold, it is marked as an anomaly prediction result.

6. The abnormal detection method for radio altimeter test data according to claim 5, wherein The anomaly distance prediction classification logic is: If the Mahalanobis distance satisfies that the Mahalanobis distance is greater than or equal to twice the Mahalanobis distance threshold, it is marked as confirmed abnormal; If the Mahalanobis distance satisfies that the Mahalanobis distance is greater than or equal to the Mahalanobis distance threshold and less than twice the Mahalanobis distance threshold, it is marked as suspected abnormal; If the Mahalanobis distance is less than the Mahalanobis distance threshold, it is marked as normal; The confirmed abnormal, suspected abnormal and normal are combined and output as the abnormal prediction result.

7. The abnormal detection method for radio altimeter test data according to claim 6, characterized in that The step S5 includes the following sub-steps: Step S501, select the alarm level according to the abnormal prediction result: If the abnormal prediction result is confirmed abnormal, trigger an alarm signal with the alarm level of first-level alarm; If the abnormal prediction result is suspected abnormal, trigger an alarm signal with the alarm level of triggering second-level alarm; If the abnormal prediction result is normal, no alarm is triggered; Step S502, combine the alarm level with the corresponding real-time flight parameters and transmit them to the flight control end.

8. The method for detecting abnormal radio altimeter test data according to claim 7, wherein It also includes: Step S601, combined with the aircraft dynamics model, verify whether the rate of change of altitude exceeds the performance envelope, and the constraint condition is: ； wherein, is the rate of change of altitude, is the maximum airspeed, is the maximum angle of attack; Step S602, if the verification in step S601 fails, overwrite the current detection result and trigger an alarm signal with the alarm level of triggering first-level alarm.

9. The abnormal detection method for radio altimeter test data according to claim 8, wherein The mathematical expression of the Mahalanobis distance threshold is: ； Among them, is the Mahalanobis distance threshold, is the critical value of the chi-square distribution.

10. A radio altimeter test data anomaly detection system, which is applied to the radio altimeter test data anomaly detection method according to any one of claims 1-9, and is characterized in that, It includes a data analysis module, a real-time acquisition module, a baseline establishment module, a model construction module and an abnormal alarm module; The data analysis module is used to obtain historical test data and perform spatio-temporal feature extraction and analysis, and output the first analysis result; The real-time acquisition module is used to collect real-time flight parameters according to the first analysis result and store the real-time flight parameters in the test database; The baseline establishment module is used to calculate dynamic statistic data according to the first analysis result and establish an adaptive baseline based on the dynamic statistic data; The model construction module is used to construct the first abnormal detection model, obtain the real-time flight phase, match the abnormality and output the abnormal prediction result; The abnormal alarm module is used to select the alarm level according to the abnormal prediction result and input it to the flight control end.

Images