Flight information monitoring system and method for unmanned aerial vehicle

Abstract

The invention relates to the technical field of electronic digital data processing, in particular to a flight information monitoring system and method for an unmanned aerial vehicle, and the method comprises the steps: obtaining a plurality of data points of the unmanned aerial vehicle at different moments, each data point comprising a multi-dimensional feature parameter; parameters at all moments in the same dimension form a parameter time sequence; inputting the plurality of data points into an improved isolated forest model, outputting an abnormal score of each data point, and responding to the data points of which the abnormal scores are greater than a set abnormal threshold value as abnormal information data points; wherein the improved isolated forest model comprises a weighted path length, the weighted path length is in positive correlation with the anomaly degree and the average path length, and the anomaly degree is the product of the anomaly amplitude and the anomaly weight. The problem that the accuracy and the reliability are not high when the flight state of the unmanned aerial vehicle is subjected to anomaly detection is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic digital data processing, and more specifically, to a flight information monitoring system and method for unmanned aerial vehicles. Background Art

[0002] With the rapid development of drone technology, flight information monitoring systems have become increasingly important in the safe operation and mission execution of drones. When drones perform various tasks, they involve complex flight data, including multiple parameters such as position, speed, attitude, and communication status. Real-time monitoring and in-depth analysis of these data are of vital importance to ensuring the flight safety of drones, improving operational efficiency, reducing accident risks, and achieving mission objectives. Specifically, real-time monitoring can help operators understand the status of drones in a timely manner, ensuring that they can respond quickly during flight, thereby effectively avoiding potential dangers. At the same time, in-depth analysis can identify potential anomalies in flight data and provide a reliable basis for drone safety decisions. Through comprehensive analysis of multiple data parameters, the flight information monitoring system can more comprehensively evaluate the working status of drones and changes in the external environment, thereby optimizing route planning and mission execution strategies.

[0003] The patent application document with application publication number CN116467950A discloses a method for detecting anomalies in drone flight data based on uncertainty representation. The patent application document introduces uncertainty representation capabilities into the prediction model by quantifying and modeling the uncertainty of data and combining Bayesian theory with ensemble learning; at the same time, considering that flight data is a multivariate time series with temporal and spatial correlation, it uses a long short-term memory network to establish a prediction model to more effectively capture the correlation between data.

[0004] However, although the above technical solutions have achieved certain results in improving the accuracy of UAV status monitoring, there are still problems with insufficient accuracy and weak adaptability of the prediction model, which results in the accuracy and reliability of UAV flight information detection failing to reach an ideal level. Summary of the invention

[0005] In order to solve the problem raised in the above background technology that the accuracy and reliability of drone flight information detection fail to reach an ideal level, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a flight information monitoring method for a drone, comprising: obtaining multiple data points of the drone at different times, and each data point includes characteristic parameters of multiple dimensions; the parameters at all times of the same dimension constitute a parameter time series; inputting the multiple data points into an improved isolation forest model, outputting anomaly scores of each data point, and responding to data points with anomaly scores greater than a set anomaly threshold as abnormal information data points; wherein the improved isolation forest model includes a weighted path length, and the weighted path length is: ; In the formula, For the The weighted path length of the data points, For the The abnormality of a data point, For the The average path length of data points; the abnormality is the product of the abnormal amplitude and the abnormal weight, the abnormal amplitude for: ; In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The first parameter in the timing Parameters and The difference of the parameters, For the The first parameter in the time series Parameters and The difference of the parameters, To take the absolute value.

[0007] The above technical solution not only improves the sensitivity to abnormal situations by introducing an improved isolation forest model and comprehensively applying the calculation of anomaly weights, anomaly amplitudes and weighted path lengths, but also can accurately capture subtle changes in time series and ensure that potential abnormal events can be identified in a timely manner at each data point. This efficient anomaly detection mechanism greatly enhances the operating safety and reliability of drones in complex environments, making real-time monitoring and early warning more accurate, thereby improving the sensitivity and accuracy of anomaly detection.

[0008] Furthermore, the abnormal weight for: ; In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The timing of the parameters is The cosine similarity of the parameter time series, For the The first parameter in the time series The value of the parameter, For the The first parameter in the time series The value of the parameter, , To take the absolute value.

[0009] The above technical solution enhances the accuracy and flexibility of anomaly detection of multi-dimensional feature parameters by comprehensively evaluating the cosine similarity and parameter differences between time series. By considering the relationship between multiple time series, the model can more comprehensively identify potential abnormal patterns. It not only improves the sensitivity of anomaly detection, allowing the system to identify and respond to potential problems earlier, but also provides a more detailed basis for subsequent analysis and decision-making, thereby further ensuring the safety and stability of drones in complex environments.

[0010] Furthermore, the abnormal weight for: ; In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The timing of the parameters is The cosine similarity of the parameter time series, For the The first parameter in the time series The value of the parameter, For the The first parameter in the time series The value of the parameter, , To take the absolute value.

[0011] The above technical solution can more comprehensively evaluate the correlation between different time series by using the difference between cosine similarity and parameter values, thereby more accurately identifying potential abnormal patterns. The design of abnormal weights enables the model to dynamically adjust the degree of attention to different features, adapt to complex monitoring environments, and improve the sensitivity and accuracy of anomaly detection.

[0012] Furthermore, the characteristic parameters include: pitch angle, yaw angle, roll angle, horizontal speed, vertical speed, acceleration, altitude, latitude and longitude of the drone.

[0013] Furthermore, the anomaly score is: ; In the formula, For the The anomaly score of the data point, For the The weighted path length of the data points, It is the average of the sum of weighted path lengths of each data point.

[0014] The above technical solution further enhances the accuracy and sensitivity of anomaly detection of multi-dimensional characteristic parameters of drones by defining the calculation method of anomaly scores, and can more sensitively reflect the degree of abnormality of data points in the form of exponential decay, thereby helping the system to quickly identify potential abnormal events. It not only improves the accuracy of detection, but also provides a more reliable basis for real-time monitoring and subsequent analysis, ensuring the safe and stable operation of drones in complex environments.

[0015] Furthermore, it also includes data cleaning and data normalization processing on the multiple feature parameters.

[0016] The above technical solution significantly improves data quality and consistency by performing data cleaning and normalization on multi-dimensional feature parameters, thus providing a more reliable basis for subsequent anomaly detection. Data cleaning can effectively remove noise and outliers, ensuring that the data input to the model is authentic and valid, while data normalization eliminates the scale differences between different features, making each feature equally important in model training. This preprocessing step not only improves the performance of the isolation forest model, but also enhances the accuracy and stability of anomaly detection, ensuring that the system can quickly and accurately identify potential abnormal events in complex environments.

[0017] Furthermore, it also includes training the isolation forest model, specifically: inputting the training set into the pre-built isolation forest model for training, and during the training process, calculating the loss between the output prediction value and the label; adjusting the model parameters using the gradient descent method to minimize the prediction error; iteratively adjusting the parameters of the isolation forest model until the loss is less than a certain value or reaches the set number of training times, and finally obtaining a trained isolation forest model.

[0018] In a second aspect, the present invention provides a flight information monitoring system for an unmanned aerial vehicle, comprising a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a flight information monitoring method for an unmanned aerial vehicle as described above is implemented.

[0019] The beneficial effects of the present invention are: The present invention uses the improved isolation forest model to perform real-time anomaly detection on multi-dimensional feature parameters. It combines the calculation of anomaly degree, anomaly weight and anomaly score to ensure accurate monitoring of the flight status of the drone and timely identify potential safety hazards. In addition, the data quality is improved by cleaning and normalizing the data, while the training and optimized isolation forest model enhances the adaptability and accuracy of the system. This comprehensive method not only improves the operational safety of drones in complex environments, but also provides a reliable basis for subsequent data analysis and decision support, thereby promoting the development and application of intelligent drone technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 is a flowchart schematically illustrating a method for monitoring flight information of a drone according to an embodiment of the present invention; Figure 2 FIG. 4 is a block diagram schematically showing a structure of a flight information monitoring system for a UAV according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0022] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0023] An embodiment of a flight information monitoring method for an unmanned aerial vehicle.

[0024] like Figure 1 As shown, a flow chart of a flight information monitoring method for a drone according to an embodiment of the present invention includes the following steps: S1: Obtain multiple data points of the drone at different times, each data point includes characteristic parameters of multiple dimensions; the parameters at all times in the same dimension constitute a parameter time series.

[0025] In one embodiment, the characteristic parameters include the pitch angle, yaw angle, roll angle, horizontal speed, vertical speed, acceleration, altitude, latitude and longitude of the drone. For multiple data points collected at different times, data cleaning and data normalization are performed on the characteristic parameters of multiple dimensions included in the data points. Data cleaning ensures the accuracy and reliability of the data and reduces the risk of deviation in the analysis results by removing noise and outliers; while data normalization allows parameters of different dimensions to be compared under the same standard, eliminating the impact of dimensional differences on the analysis results. The combination of these two technologies improves the training effect and prediction ability of the model, thereby providing a more accurate and effective data basis for drone flight status monitoring and decision support.

[0026] Parameter time series are constructed based on parameter data of multiple dimensions processed by data cleaning and data normalization, and the parameters at all moments of the same dimension are combined into a parameter time series, so that multiple parameter time series can be obtained.

[0027] For example, taking the pitch angle of the drone as an example, assuming that at different time points ( ), the resulting pitch angle parameter sequence is [5,7,10,8].

[0028] S2: Input the multiple data points into an improved isolation forest model, and output an abnormality score of each data point, wherein the improved isolation forest model includes a weighted path length.

[0029] In one embodiment, the isolation forest model needs to be trained. First, appropriate parameters are set, such as the number of isolated trees, the maximum depth, etc., and then the model is trained through a training set. Multiple data points are selected as the training set to train the isolation forest model. Specifically, The training set is input into the pre-built isolation forest model for training. During the training process, the loss between the output prediction value and the label is calculated; the model parameters are adjusted using the gradient descent method to minimize the prediction error; the parameters of the isolation forest model are adjusted iteratively until the loss is less than a certain value or the set number of training times is reached, and finally a trained isolation forest model is obtained. The label is set manually. The multiple data points are input into the above-mentioned trained isolation forest model to obtain the abnormality score of each data point.

[0030] By analyzing multiple data points through training models, the anomaly score of each data point at a specific moment can be effectively output, providing strong support for real-time monitoring and early warning. By continuously optimizing model parameters and training data, the accuracy of detection can be further improved, providing reliable anomaly detection solutions for various application scenarios.

[0031] In one embodiment, the weighted path length is: ; In the formula, For the The weighted path length of the data points, For the The abnormality of a data point, For the The average path length of the data points is specifically: The data point is input into the pre-built isolation forest model to directly calculate and obtain the average path length of the data point.

[0032] The abnormality degree is the product of the abnormal amplitude and the abnormal weight. for: ; In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The first parameter in the time series Parameters and The difference of the parameters, For the The first parameter in the time series Parameters and The difference of the parameters, To take the absolute value.

[0033] It should be noted that in order to reflect the overall trend of change, the denominator is taken when calculating the mean. ; When the anomaly weight at the sampling point is higher and the anomaly amplitude is higher, the possibility that the drone is affected by the environment at the sampling point and there is a timing anomaly is higher, and the timing anomaly at the sampling point is higher.

[0034] In one embodiment, the anomaly weight for: ; In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The timing of the parameters is The cosine similarity of the parameter time series, For the The first parameter in the time series The value of the parameter, For the The first parameter in the time series The value of the parameter, , To take the absolute value.

[0035] It should be noted that For the The timing of the parameters is The absolute value of the cosine similarity of the first parameter sequence. When the absolute value of the cosine similarity of the two parameter sequences is higher, the possibility that the sequence changes in the two sequences represent the same flight state is higher; when the The timing of the parameters is The higher the possibility that the time series changes in the time series of a parameter represent the same flight state, the higher the possibility that the corresponding sampling point is affected by environmental changes and causes a single parameter to be abnormal, and the higher the abnormal weight of the time series at the sampling point.

[0036] In another embodiment, the anomaly weight for: ; In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The timing of the parameters is The cosine similarity of the parameter time series, For the The first parameter in the time series The value of the parameter, For the The first parameter in the time series The value of the parameter, , To take the absolute value.

[0037] In one embodiment, the anomaly score is: ; In the formula, For the The anomaly score of the data point, For the The weighted path length of the data points, It is the average of the sum of weighted path lengths of each data point.

[0038] S3: In response to a data point whose anomaly score is greater than a set anomaly threshold, the data point is an abnormal information data point.

[0039] In one embodiment, the above-set abnormal threshold value may be 0.8. Of course, it can also be adjusted and optimized according to actual needs and specific application scenarios. This flexibility enables the system to adapt to different types of data sets and environmental changes, ensuring the accuracy and effectiveness of anomaly detection.

[0040] Once the system determines that the current information data point is abnormal, it will immediately generate a corresponding warning signal. This warning mechanism is real-time and can ensure that relevant personnel or systems can respond quickly at the moment of abnormal events. There are many ways to issue warning signals, which can be selected and combined according to different usage scenarios and user needs. For example, once the drone detects an abnormal information data point during the monitoring process, the system can flash the indicator light on the drone to intuitively attract the attention of the operator. In addition, the system can also activate the buzzer to sound an alarm, providing sound prompts to people around, ensuring that even if it is not immediately visible, it can still cause sufficient vigilance.

[0041] At the same time, the early warning information can also be sent to the user terminal or management center in real time through the wireless communication module. This function greatly enhances the system's responsiveness and the timeliness of information transmission. User terminals come in a variety of forms, including mobile phone applications, SMS notifications, or centralized control panels. Users can receive alerts of abnormal information data points in a timely manner through these diverse channels, so that they can respond quickly and take corresponding countermeasures.

[0042] The solution of the present invention significantly improves the detection and processing capabilities of abnormal UAV flight data. The improved isolation forest model can accurately identify and respond to potential abnormal events. At the same time, the quality of data input is ensured through data cleaning and normalization, thereby improving the accuracy and reliability of detection. It can not only monitor the flight status of the UAV in real time, but also has the ability to train and optimize the model, ensuring continuous and efficient security protection in complex and dynamic environments, and promoting the intelligent management and application of UAVs.

[0043] An embodiment of a flight information monitoring system for a drone: like Figure 2 As shown, a structural block diagram of a flight information monitoring system for a UAV according to an embodiment of the present invention includes a processor and a memory.

[0044] The present invention also provides a flight information monitoring system for an unmanned aerial vehicle. Figure 2 As shown, the system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a flight information monitoring method for a drone according to the above-mentioned invention is implemented.

[0045] The flight information monitoring system for a UAV also includes other components familiar to those skilled in the art, such as a communication interface, and the configuration and functions of the components are known in the art, so they will not be described in detail here.

[0046] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0047] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0048] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A flight information monitoring method for an unmanned aerial vehicle, characterized in that: include: Obtain multiple data points of the drone at different times, and each data point includes characteristic parameters of multiple dimensions; The parameters at all moments in the same dimension constitute a parameter time series; Input the plurality of data points into the improved isolation forest model, output anomaly scores of the data points, and respond that the data points whose anomaly scores are greater than a set anomaly threshold are abnormal information data points; The improved isolation forest model includes a weighted path length, and the weighted path length is: ； In the formula, For the The weighted path length of the data points, For the The abnormality of a data point, For the The average path length of the data points; The abnormality degree is the product of the abnormal amplitude and the abnormal weight. for: ； In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The first parameter in the time series Parameters and The difference of the parameters, For the The first parameter in the time series Parameters and The difference of the parameters, To take the absolute value.

2. A method for monitoring flight information of a drone according to claim 1, characterized in that: The anomaly weight for: ； In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The timing of the parameters is The cosine similarity of the parameter time series, For the The first parameter in the time series The value of the parameter, For the Among the timing parameters The value of the parameter, , To take the absolute value.

3. The flight information monitoring method for a UAV according to claim 1, characterized in that: The anomaly weight for: ； In the formula, is the number of parameter time series, is the preset neighborhood radius, For the The timing of the parameters is The cosine similarity of the parameter time series, For the The first parameter in the time series The value of the parameter, For the The first parameter in the time series The value of the parameter, , To take the absolute value.

4. The flight information monitoring method for a UAV according to claim 1, characterized in that: The characteristic parameters include: the pitch angle, yaw angle, roll angle, horizontal speed, vertical speed, acceleration, altitude, latitude and longitude of the drone.

5. The flight information monitoring method for a UAV according to claim 1, characterized in that: The anomaly score is: ； In the formula, For the The anomaly score of the data point, For the The weighted path length of the data points, It is the average of the sum of weighted path lengths of each data point.

6. The flight information monitoring method for a UAV according to claim 1, characterized in that: It also includes data cleaning and data normalization processing on the characteristic parameters.

7. The flight information monitoring method for a UAV according to claim 1, characterized in that: It also includes training the isolation forest model, specifically: The training set is input into the pre-built isolation forest model for training. During the training process, the loss between the output prediction value and the label is calculated; the model parameters are adjusted using the gradient descent method to minimize the prediction error; the parameters of the isolation forest model are iteratively adjusted until the loss is less than a certain value or the set number of training times is reached, and finally a trained isolation forest model is obtained.

8. A flight information monitoring system for an unmanned aerial vehicle, characterized in that: It comprises a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a flight information monitoring method for a drone as described in any one of claims 1 to 7 is implemented.

Images