Unmanned aerial vehicle position prediction method and system based on trusted UKF-BiLSTM

Abstract

The invention discloses an unmanned aerial vehicle position prediction method and system based on trusted UKF-BiLSTM, and relates to the technical field of unmanned aerial vehicle position prediction.According to the unmanned aerial vehicle position prediction method, a nonlinear transfer function is obtained through BiLSTM during prediction, the nonlinear transfer process of unmanned aerial vehicle position change can be better described, and the unmanned aerial vehicle position prediction accuracy is improved. According to the method, the position of the unmanned aerial vehicle is tracked and predicted in combination with the UKF, and a related loss function is set according to the credibility, so that the optimization direction of the BiLSTM is always kept consistent with the credibility increasing direction, and meanwhile, the matching of the model can be considerably reflected in real time. Compared with a traditional unmanned aerial vehicle prediction algorithm, the method has better precision and interpretability.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle position prediction, and in particular to a method and system for predicting the position of an unmanned aerial vehicle based on a trusted UKF-BiLSTM. Background Art

[0002] When a drone is working, it will inevitably be affected by the external environment. Factors such as wind and cyclones will cause the drone's trajectory to deviate. Such deviations will have a certain impact on the normal flight and operation of the drone.

[0003] In the process of tracking and predicting UAVs using Kalman filtering, it is impossible to accurately model the position changes of UAVs due to the influence of various external environments. At the same time, the movement of UAVs is nonlinear and requires approximation technology to process. These will cause the position change model of the UAV to not match the actual model, seriously affecting the performance of the filter.

[0004] Most of the existing algorithms for this phenomenon use adaptive technology to estimate inaccurate model parameters. However, these estimation methods can often only estimate a single inaccurate parameter. For example, when the Sage-Husa filter estimates the covariance of process noise and measurement noise simultaneously, it will inevitably produce divergence, which greatly limits the effect of model parameter estimation. At the same time, most adaptive filters have poor interpretability and cannot accurately explain how to improve model matching from the model level, and cannot accurately quantify the matching of the model.

[0005] Nowadays, with the continuous deepening of the exploration of the sky, the prediction algorithms of drones are also developing rapidly. Accurate tracking and prediction of drones is an important guarantee for the successful completion of their tasks. Therefore, the current upsurge in research on drone position prediction is still continuing, and it is expected to further improve the prediction accuracy. Summary of the invention

[0006] In order to solve the deficiencies mentioned in the above background technology, the purpose of the present invention is to provide a UAV position prediction method and system based on trusted UKF-BiLSTM.

[0007] In the first aspect, the purpose of the present invention can be achieved by the following technical solution: a method for predicting the position of a UAV based on a trusted UKF-BiLSTM, the method comprising the following steps:

[0008] Obtain the drone location data, input the drone location data into the pre-established BiLSTM model for training, and output the drone location change model;

[0009] The observation value of the drone position is obtained, and the observation value of the drone position is input into the drone position change model to predict the drone position change, and the drone position prediction result is output.

[0010] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the process of collecting the drone position data includes:

[0011] Equipped with a drone that supports RTK technology and a ground reference station, the ground reference station serves as a fixed reference point for the RTK system.

[0012] Set up the RTK base station and connect the RTK receiver on the drone to the base station via radio or network to achieve communication between devices.

[0013] During the flight of the drone, the position data and other sensor data will be saved in the internal memory of the drone or transmitted to the ground station for storage in real time via wireless transmission.

[0014] In combination with the first aspect, in some implementations of the first aspect, the method further includes: inputting the drone position data into a pre-established BiLSTM model for training, and outputting a calculation formula for obtaining a drone position change model as follows:

[0015] X i,k|k-1 =BiLSTM(X i,k-1 )

[0016] Where, X i,k-1 is the Sigma sampling point at time k-1, X i,k|k-1 is X after nonlinear transfer i,k-1 .

[0017] In combination with the first aspect, in some implementations of the first aspect, the method further includes: a process of inputting the observed value of the drone position into the drone position change model to predict the drone position change:

[0018] Nonlinear Transfer:

[0019]

[0020] Depend on and Perform Sigma sampling to obtain X i,k-1 :

[0021]

[0022] The result of the Sigma point set after nonlinear transfer is as follows:

[0023]

[0024] Update filter gains, estimates, and error covariance:

[0025]

[0026]

[0027] In the formula, the number of sampling points is 2n+1, Compute the error covariance for the filter, To calculate the measurement error covariance, ω is the inertia weight, is the process noise covariance assumed by the filter; λ is the scale factor; h(·) is the measurement function; K is the gain matrix; is the error cross-covariance of the prior error measure.

[0028] In combination with the first aspect, in some implementations of the first aspect, the method further includes: a credibility solving process and formula are as follows:

[0029]

[0030] The confidence levels are set to:

[0031]

[0032] The loss function is set to

[0033] In the second aspect, in order to achieve the above-mentioned purpose, the present invention discloses a UAV position prediction system based on a trusted UKF-BiLSTM, comprising:

[0034] The model training module is used to obtain the drone position data, input the drone position data into the pre-established BiLSTM model for training, and output the drone position change model;

[0035] The position prediction module is used to obtain the observation value of the drone position, input the observation value of the drone position into the drone position change model to predict the drone position change, and output the drone position prediction result.

[0036] In conjunction with the second aspect, in some implementations of the second aspect, the system further includes: the process of collecting the drone position data of the model training module includes:

[0037] Equipped with a drone that supports RTK technology and a ground reference station, the ground reference station serves as a fixed reference point for the RTK system.

[0038] Set up the RTK base station and connect the RTK receiver on the drone to the base station via radio or network to achieve communication between devices.

[0039] During the flight of the drone, the position data and other sensor data will be saved in the internal memory of the drone or transmitted to the ground station for storage in real time via wireless transmission.

[0040] The model training module inputs the drone position data into the pre-established BiLSTM model for training, and the calculation formula of the output drone position change model is as follows:

[0041] X i,k|k-1 =BiLSTM(X i,k-1 )

[0042] Where, X i,k-1 is the Sigma sampling point at time k-1, X i,k|k-1 is X after nonlinear transfer i,k-1 ;

[0043] The position prediction module inputs the observed value of the drone position into the drone position change model to predict the drone position change:

[0044] Nonlinear Transfer:

[0045]

[0046] Depend on and Perform Sigma sampling to obtain X i,k-1 :

[0047]

[0048] The result of the Sigma point set after nonlinear transfer is as follows:

[0049]

[0050] Update filter gains, estimates, and error covariance:

[0051]

[0052] In the formula, the number of sampling points is 2n+1, Compute the error covariance for the filter, To calculate the measurement error covariance, ω is the inertia weight, is the process noise covariance assumed by the filter; λ is the scale factor; h(·) is the measurement function; K is the gain matrix; is the error cross-covariance of the prior error measurement;

[0053] The credibility calculation process and formula of the location prediction module are as follows:

[0054]

[0055] The confidence levels are set to:

[0056]

[0057] The loss function is set to

[0058] In another aspect of the present invention, in order to achieve the above-mentioned purpose, a terminal device is disclosed, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the memory stores a computer program capable of running on the processor, and when the processor loads and executes the computer program, the drone position prediction method based on the trusted UKF-BiLSTM as described above is adopted.

[0059] In another aspect of the present invention, in order to achieve the above-mentioned purpose, a computer-readable storage medium is disclosed, in which a computer program is stored, and it is characterized in that when the computer program is loaded and executed by a processor, the UAV position prediction method based on the trusted UKF-BiLSTM as described above is adopted.

[0060] Beneficial effects of the present invention:

[0061] The UAV position prediction method proposed in the present invention uses BiLSTM to obtain nonlinear transfer function when making predictions, which can better describe the nonlinear transfer process of UAV position change, and combines UKF to track and predict the UAV position, and sets the relevant loss function according to the credibility, so that the optimization direction of BiLSTM always keeps consistent with the credibility increase direction, and can reflect the matching of the model in a more real-time and observable manner. Compared with traditional UAV prediction algorithms, it has better accuracy and interpretability. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0063] Figure 1 It is a schematic flow chart of the method of the present invention;

[0064] Figure 2 This is the BiLSTM mechanism diagram used for the first time in the present invention;

[0065] Figure 3 This is the BiLSTM mechanism diagram used for the second time in the present invention;

[0066] Figure 4 It is a schematic diagram of the BiLSTM-UKF calculation process of the present invention;

[0067] Figure 5 It is a schematic diagram of the system structure of the present invention;

[0068] Figure 6 Schematic diagram of average error of filtering algorithm in simulation experiment of this embodiment;

[0069] Figure 7 It is a schematic diagram of the credibility of the filtering model in the simulation experiment of this embodiment. DETAILED DESCRIPTION

[0070] 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 only 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] Embodiment 1:

[0072] The following is an introduction to the relevant terms involved in the embodiments of the present application:

[0073] Unmanned aerial vehicle, abbreviated as "UAV" in English, is an unmanned aircraft controlled by radio remote control equipment and self-contained program control device, or operated completely or intermittently autonomously by an onboard computer.

[0074] Compared with manned aircraft, drones are often more suitable for tasks that are too "dull, dirty or dangerous". Drones can be divided into military and civilian applications according to their application areas. In the military, drones are divided into reconnaissance aircraft and target aircraft. In the civilian field, drones + industry applications are the real rigid demand for drones; their applications in aerial photography, agriculture, plant protection, micro selfies, express delivery, disaster relief, wildlife observation, infectious disease monitoring, surveying and mapping, news reporting, power inspections, disaster relief, film and television shooting, creating romance, etc. have greatly expanded the use of drones themselves. Developed countries are also actively expanding industry applications and developing drone technology.

[0075] like Figure 1 As shown, the UAV position prediction method based on the trusted UKF-BiLSTM is characterized in that the method includes the following steps:

[0076] S101: Obtain the drone position data, input the drone position data into a pre-established BiLSTM model for training, and output a drone position change model;

[0077] The process of collecting the drone location data includes:

[0078] Equipped with a drone that supports RTK technology and a ground reference station, the ground reference station serves as a fixed reference point for the RTK system.

[0079] Set up the RTK base station and connect the RTK receiver on the drone to the base station via radio or network to achieve communication between devices.

[0080] During the flight of the drone, the position data and other sensor data will be saved in the internal memory of the drone or transmitted to the ground station for storage in real time via wireless transmission.

[0081] like Figure 2 As shown in the figure, the drone position data is input into the pre-established BiLSTM model for training, and the calculation formula of the drone position change model is output as follows:

[0082] X i,k|k-1 =BiLSTM(X i,k-1 )

[0083] Where, X i,k-1 is the Sigma sampling point at time k-1, X i,k|k-1 is X after nonlinear transfer i,k-1 .

[0084] S102: Obtain the observation value of the drone position, input the observation value of the drone position into the drone position change model to predict the drone position change, and output the drone position prediction result.

[0085] The process of collecting observations of the drone's position includes:

[0086] like Figure 4 As shown, the process of inputting the observed value of the drone position into the drone position change model to predict the drone position change is:

[0087] Nonlinear Transfer:

[0088]

[0089] Depend on and Perform Sigma sampling to obtain X i,k-1 :

[0090]

[0091] The result of the Sigma point set after nonlinear transfer is as follows:

[0092]

[0093] Update filter gains, estimates, and error covariance:

[0094]

[0095] In the formula, the number of sampling points is 2n+1, Compute the error covariance for the filter, To calculate the measurement error covariance, ω is the inertia weight, is the process noise covariance assumed by the filter; λ is the scale factor; h(·) is the measurement function; K is the gain matrix; is the error cross-covariance of the prior error measure.

[0096] The credibility solution process and formula are as follows:

[0097]

[0098] The confidence levels are set to:

[0099]

[0100] In order to ensure that the optimization direction of the model is consistent with the direction of increasing credibility, the loss function is set to The second use of BiLSTM is as follows Figure 3 shown.

[0101] Specifically, the present invention is further described below by way of embodiments:

[0102] like Figure 6 and Figure 7 As shown in the figure, the implementation process of the UAV position prediction method of the trusted UKF-BiLSTM is as follows:

[0103] 1) Set up an RTK base station to receive the drone's position data in real time during the flight of the drone, save it in the base station, and use a computer to read the data.

[0104] 2) Use Python to write relevant codes, train the first BiLSTM network with historical data, and fit relevant functions to describe the nonlinear transfer process

[0105] 3) The second BiLSTM network is trained using historical estimates, filter calculation measurement noise covariance and prior error covariance as features. At the same time, the relevant loss function is set to obtain a more reliable actual measurement error covariance to participate in the filtering calculation, so as to achieve the purpose of adaptive adjustment of the filtering process.

[0106] 4) Combined with the real-time observations, the position of the UAV is tracked and predicted.

[0107] The experimental drone position prediction system consists of three parts: a drone with an RTK receiver, an RTK base station, and a computer. The drone experimented with the above algorithm at various altitudes, and applied the algorithm to scenarios such as drone-unmanned ship coordinated landing, with good results.

[0108] Embodiment 2: In the second aspect, as Figure 5 As shown, in order to achieve the above-mentioned purpose, the present invention discloses a UAV position prediction system based on a trusted UKF-BiLSTM, comprising:

[0109] The model training module 11 is used to obtain the drone position data, input the drone position data into a pre-established BiLSTM model for training, and output a drone position change model;

[0110] The position prediction module 12 is used to obtain the observation value of the drone position, input the observation value of the drone position into the drone position change model to predict the drone position change, and output the drone position prediction result.

[0111] In conjunction with the second aspect, in some implementations of the second aspect, the system further includes: the process of collecting the drone position data of the model training module 11 includes:

[0112] Equipped with a drone that supports RTK technology and a ground reference station, the ground reference station serves as a fixed reference point for the RTK system.

[0113] Set up the RTK base station and connect the RTK receiver on the drone to the base station via radio or network to achieve communication between devices.

[0114] During the flight of the drone, the position data and other sensor data will be saved in the internal memory of the drone or transmitted to the ground station for storage in real time via wireless transmission.

[0115] The model training module 11 inputs the drone position data into the pre-established BiLSTM model for training, and outputs the calculation formula of the drone position change model as follows:

[0116] X i,k|k-1 =BiLSTM(X i,k-1 )

[0117] Where, X i,k-1 is the Sigma sampling point at time k-1, X i,k|k-1 is X after nonlinear transfer i,k-1 ;

[0118] The position prediction module 12 inputs the observed value of the drone position into the drone position change model to predict the drone position change:

[0119] Nonlinear Transfer:

[0120]

[0121] Depend on and Perform Sigma sampling to obtain X i,k-1 :

[0122]

[0123] The result of the Sigma point set after nonlinear transfer is as follows:

[0124]

[0125] Update filter gains, estimates, and error covariance:

[0126]

[0127] In the formula, the number of sampling points is 2n+1, Compute the error covariance for the filter, To calculate the measurement error covariance, ω is the inertia weight, is the process noise covariance assumed by the filter; λ is the scale factor; h(·) is the measurement function; K is the gain matrix; is the error cross-covariance of the prior error measurement;

[0128] The credibility calculation process and formula of the location prediction module 12 are as follows:

[0129]

[0130] The confidence levels are set to:

[0131]

[0132] The loss function is set to

[0133] Based on the same inventive concept, the present invention also provides a computer device, which includes: one or more processors, and a memory for storing one or more computer programs; the program includes program instructions, and the processor is used to execute the program instructions stored in the memory. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is used to implement one or more instructions, specifically for loading and executing one or more instructions in a computer storage medium to implement the above method.

[0134] It needs to be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium, on which a computer program is stored, and the computer program is executed by a processor to execute the above method. The storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electrical, magnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0135] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0136] The above shows and describes the basic principles, main features and advantages of the present disclosure. Those skilled in the art should understand that the present disclosure is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present disclosure. Without departing from the spirit and scope of the present disclosure, the present disclosure may have various changes and improvements, and these changes and improvements fall within the scope of the present disclosure to be protected.

Claims

1. The UAV position prediction method based on the trusted UKF-BiLSTM is characterized by: The method comprises the following steps: Obtain the drone location data, input the drone location data into the pre-established BiLSTM model for training, and output the drone location change model; The observation value of the drone position is obtained, and the observation value of the drone position is input into the drone position change model to predict the drone position change, and the drone position prediction result is output.

2. The method for predicting the position of a UAV based on a reliable UKF-BiLSTM according to claim 1, characterized in that: The process of collecting the drone position data includes: Equipped with RTK-enabled drones and ground reference stations, which serve as fixed reference points for the RTK system Set up the RTK base station and connect the RTK receiver on the drone to the base station via radio or network; During the flight of the drone, the drone position data and other sensor data will be saved in the drone's internal memory or transmitted to the ground station in real time via wireless transmission for storage.

3. The method for predicting the position of a UAV based on a reliable UKF-BiLSTM according to claim 2, characterized in that: The drone position data is input into the pre-established BiLSTM model for training, and the calculation formula of the drone position change model is output as follows: X i,k|k-1 =BiLSTM(X i,k-1 ) Where, X i,k-1 is the Sigma sampling point at time k-1, X i,k|k-1 is X after nonlinear transfer i,k-1 .

4. The method for predicting the position of a UAV based on a reliable UKF-BiLSTM according to claim 1, characterized in that: The process of inputting the observed value of the drone position into the drone position change model to predict the drone position change: Nonlinear Transfer: Depend on and Perform Sigma sampling to obtain X i,k-1 : The result of the Sigma point set after nonlinear transfer is as follows: Update filter gains, estimates, and error covariance: In the formula, the number of sampling points is 2n+1, Compute the error covariance for the filter, To calculate the measurement error covariance, ω is the inertia weight, is the process noise covariance assumed by the filter; λ is the scale factor; h(·) is the measurement function; K is the gain matrix; is the error cross-covariance of the prior error measure.

5. The method for predicting the position of a UAV based on a reliable UKF-BiLSTM according to claim 4, characterized in that: The credibility solution process and formula are as follows: The confidence levels are set to: The loss function is set to 6. The UAV position prediction system based on the trusted UKF-BiLSTM is characterized by: include: The model training module is used to obtain the drone position data, input the drone position data into the pre-established BiLSTM model for training, and output the drone position change model; The position prediction module is used to obtain the observation value of the drone position, input the observation value of the drone position into the drone position change model to predict the drone position change, and output the drone position prediction result.

7. The UAV position prediction system based on the trusted UKF-BiLSTM according to claim 6 is characterized in that: The process of collecting the drone position data of the model training module includes: Equipped with RTK-enabled drones and ground reference stations, which serve as fixed reference points for the RTK system Set up the RTK base station and connect the RTK receiver on the drone to the base station via radio or network; During the flight of the drone, the drone position data and other sensor data will be saved in the drone's internal memory or transmitted to the ground station in real time via wireless transmission for storage; The model training module inputs the drone position data into the pre-established BiLSTM model for training, and the calculation formula of the output drone position change model is as follows: X i,k|k-1 =BiLSTM(X i,k-1 ) Where, X i,k-1 is the Sigma sampling point at time k-1, X i,k|k-1 is X after nonlinear transfer i,k-1 ; The position prediction module inputs the observed value of the drone position into the drone position change model to predict the drone position change: Nonlinear Transfer: Depend on and Perform Sigma sampling to obtain X i,k-1 : The result of the Sigma point set after nonlinear transfer is as follows: Update filter gains, estimates, and error covariance: In the formula, the number of sampling points is 2n+1, Compute the error covariance for the filter, To calculate the measurement error covariance, ω is the inertia weight, is the process noise covariance assumed by the filter; λ is the scale factor; h(·) is the measurement function; K is the gain matrix; is the error cross-covariance of the prior error measurement; The credibility calculation process and formula of the location prediction module are as follows: The confidence levels are set to: The loss function is set to 8. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: The memory stores a computer program that can be run on the processor. When the processor loads and executes the computer program, the drone position prediction method based on the trusted UKF-BiLSTM described in any one of claims 1 to 5 is adopted.

9. A computer-readable storage medium having a computer program stored therein, characterized in that: When the computer program is loaded and executed by the processor, the UAV position prediction method based on the trusted UKF-BiLSTM described in any one of claims 1 to 5 is adopted.

Images