Unmanned aerial vehicle Beidou and SINS combined positioning method based on 6G signal

Abstract

The invention discloses an unmanned aerial vehicle Beidou and SINS integrated positioning method based on a 6G signal, and the method comprises the steps: obtaining the positioning information of an unmanned aerial vehicle, and the positioning information comprises Beidou positioning information and inertial navigation system positioning information; performing preliminary fusion on the positioning information to obtain preliminary estimation positioning information; obtaining fingerprint signal characteristics of the 6G signal of the unmanned aerial vehicle, and obtaining auxiliary positioning information according to the fingerprint signal characteristics; and fusing the preliminary estimation positioning information and the auxiliary positioning information through an interactive multi-model and a volume Kalman filtering model to obtain fused positioning information so as to realize combined positioning. According to the invention, the positioning precision and system stability of the unmanned aerial vehicle in a complex environment can be effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of UAV positioning, and particularly relates to a combined positioning method for Beidou and SINS of UAVs based on 6G signals. Background Art

[0002] With the continuous progress of wireless communication technology, the fifth-generation communication technology (5G) has been gradually applied to various industries and brought revolutionary changes to communication worldwide. However, 5G technology still has limitations in terms of coverage and network latency. Especially in remote areas and high-density building environments, the communication quality is affected. To solve these problems, the sixth-generation communication technology (6G) has emerged, providing higher data transmission rates, ultra-low latency, extremely high network reliability, and extensive device connection capabilities, and is expected to provide better services for fields such as mobile communication, intelligent manufacturing, and autonomous driving.

[0003] However, in complex environments, traditional GNSS (Global Navigation Satellite System) and SINS (Strapdown Inertial Navigation System) combined positioning methods may be affected by satellite signal interruption or weak signals. Therefore, using 6G signals to assist UAV positioning has become an innovative solution. Existing technologies mainly focus on using 5G signals to assist GNSS / SINS combined positioning, but there are no relevant technical solutions for combining 5G signals to assist GNSS / SINS combined positioning in the existing technologies, and the effectiveness and accuracy of UAV positioning need to be further improved. Summary of the Invention

[0004] To solve the above technical problems, the present invention proposes a combined positioning method for Beidou and SINS of UAVs based on 6G signals to solve the problems existing in the above-mentioned prior art.

[0005] To achieve the above object, the present invention provides a combined positioning method for Beidou and SINS of UAVs based on 6G signals, including:

[0006] Obtain the positioning information of the UAV, where the positioning information includes Beidou positioning information and inertial navigation system positioning information; perform preliminary fusion on the positioning information to obtain preliminary estimated positioning information;

[0007] Obtain the fingerprint signal characteristics of the 6G signal of the UAV, and obtain auxiliary positioning information according to the fingerprint signal characteristics;

[0008] Fuse the preliminary estimated positioning information and the auxiliary positioning information through an interactive multi-model and a cubature Kalman filter model to obtain fused positioning information to achieve combined positioning.

[0009] Optionally, before obtaining the positioning information of the UAV, it includes:

[0010] Deploy a 6G communication module, a Beidou receiver and an IMU on the UAV, initialize the deployed devices, and perform initialization configuration on the position, speed and attitude of the UAV.

[0011] Optionally, the positioning information is preliminarily fused by the Kalman filtering algorithm.

[0012] Optionally, the 6G fingerprint signal features include the received power of the synchronization signal and the received power of the channel state information.

[0013] Optionally, the process of obtaining the auxiliary positioning information includes:

[0014] Construct a fingerprint signal feature database, in which several samples are stored, and the samples are fingerprint signal features at known positions;

[0015] Calculate the weights of the fingerprint signal features according to the samples by the ReliefF method to obtain the fingerprint signal feature weights;

[0016] According to the fingerprint signal feature weights, calculate the distances between the samples and the fingerprint signal features by the Relief-WKNN positioning method, and according to the calculation results, obtain the weighted Euclidean distance, and perform weighted sum calculation based on the weighted Euclidean distance and the position information of the samples to obtain the auxiliary positioning information.

[0017] Optionally, before fusing the preliminary estimated positioning information and the auxiliary positioning information, it further includes:

[0018] Convert the preliminary estimated positioning information and the auxiliary positioning information to the same coordinate system.

[0019] Optionally, the process of fusing the preliminary estimated positioning information and the auxiliary positioning information includes:

[0020] Fuse the auxiliary positioning information with the Beidou positioning information to obtain position fusion information;

[0021] Fuse the preliminary estimated positioning information and the position fusion information through an interactive multiple model and a cubature Kalman filter model. Among them, each Kalman filter model in the interactive multiple model is used to filter the preliminary estimated positioning information and the position fusion information respectively and fuse the filtering results, and each Kalman filter model in the interactive multiple model is replaced by a cubature Kalman filter model. The fusion process is:

[0022]

[0023] Among them, is the weighted weight of the i-th model at the k-th moment in the interactive multiple model, Kk The Kalman gain of the volume Kalman filter model represents the state prediction at the k-th moment, z k represents the measurement matrix, i represents the corresponding model label, k represents the time step label, H k is the observation matrix.

[0024] Optionally, after obtaining the fused positioning information, it further includes:

[0025] Optimizing the flight trajectory of the UAV based on the communication quality and coverage of the 6G signal; wherein, based on the communication quality and coverage of the 6G signal, the flight trajectory of the UAV is modeled through a Markov decision process, with the optimal communication quality and task execution efficiency of the UAV as the goal, and the flight trajectory of the UAV is optimized through a deep Q network.

[0026] Optionally, after optimizing the flight trajectory of the UAV, it further includes:

[0027] Adjusting the system parameters of the UAV according to the fused positioning information and the optimized flight trajectory, where the system parameters include flight parameters, communication parameters, and navigation parameters.

[0028] Compared with the prior art, the present invention has the following advantages and technical effects:

[0029] By combining 6G signals with the Beidou and SINS integrated positioning systems, and using the high-frequency and low-latency characteristics of 6G for signal fusion, the positioning accuracy problem of traditional GNSS / SINS systems in complex environments is effectively solved. At the same time, by adopting an advanced adaptive filtering algorithm, signal interference can be effectively reduced, the stability and accuracy of the system can be improved, ensuring that the UAV can operate stably even in an environment without satellite signals, thus significantly enhancing the operation ability of the UAV in various environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The drawings constituting a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation of this application. In the drawings:

[0031] Figure 1 is the flow chart of UAV positioning and task scheduling optimization according to the embodiment of the present invention;

[0032] Figure 2 is the flow chart of multi-fingerprint feature positioning based on 6G signals according to the embodiment of the present invention;

[0033] Figure 3 is the flow chart of the IAIMM-RSACKF fusion algorithm according to the embodiment of the present invention;

[0034] Figure 4 Flowchart of the DQN-based UAV trajectory optimization algorithm according to an embodiment of the present invention. Detailed implementation manners

[0035] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0036] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0037] In the 6G environment, the collaborative work of unmanned aerial vehicles (UAVs) will play an important role in areas that cannot be covered by traditional terrestrial networks. The introduction of the 6G network effectively expands the coverage of wireless communication. However, in complex environments, traditional GNSS (Global Navigation Satellite System) and SINS (Strapdown Inertial Navigation System) combined positioning methods may be affected by satellite signal interruption or weak signals. Therefore, using 6G signals to assist UAV positioning has become an innovative solution. Existing technologies mainly focus on using 5G signals to assist GNSS / SINS combined positioning, but 6G signals have higher positioning accuracy than 5G and can provide more stable and efficient positioning in weak signal or satellite-free signal environments such as underground and indoor. The advantages of 6G signals in terms of latency, spectral efficiency, and coverage enable it to continuously support UAV positioning and navigation.

[0038] The present invention provides a combined positioning method for UAV Beidou and SINS based on 6G signals, specifically a method for improving the positioning accuracy of UAVs by fusing 6G signals, Beidou positioning, and SINS data. First, deploy and initialize the relevant UAV system, preprocess the 6G fingerprint signal, start Beidou positioning and SINS solution, and use Kalman filtering to initially fuse the two data to obtain the position, speed, and attitude information of the UAV. Secondly, use 6G fingerprint positioning technology to calculate auxiliary positioning information, and then use the IMM and CKF fusion improvement algorithms to deeply fuse multi-source data to further improve the positioning accuracy. At the same time, use the DRL algorithm to optimize the UAV flight trajectory and task scheduling according to the 6G signal communication quality and coverage. Finally, dynamically adjust the system parameters according to the fusion positioning results and task execution conditions to achieve real-time feedback optimization. This method combines the advantages of multiple technologies and can effectively improve the positioning accuracy and system stability of UAVs in complex environments.

[0039] The present invention proposes a combined positioning method for UAVs using Beidou and SINS based on 6G signals. By integrating 6G signals, Beidou positioning, and SINS, data is processed using Kalman filtering, 6G fingerprint positioning, improved IMM and CKF algorithms, as well as DRL algorithms to optimize the flight trajectory and task scheduling, and dynamically adjust system parameters, thereby significantly improving the positioning accuracy and system stability of UAVs in complex environments.

[0040] As Figure 1 shown, it presents a flowchart for optimizing UAV positioning and task scheduling based on the integration of 6G, Beidou positioning, and SINS. This method significantly improves the positioning accuracy and system stability of UAVs through multi-step fusion technology, including:

[0041] Step 1: System initialization; Deploy a UAV system equipped with a 6G communication module, Beidou receiver, and inertial measurement unit (IMU), and initialize relevant parameters;

[0042] Step 2: Data acquisition and preliminary processing; Start the Beidou positioning and strapdown inertial navigation system (SINS) solution, and fuse the data of both through Kalman filtering to preliminarily estimate the position, velocity, and attitude information of the UAV;

[0043] Step 3: 6G signal feature acquisition and preprocessing; The UAV acquires 6G fingerprint signal features, including synchronization signal received power, channel state information received power, etc., and performs preprocessing, such as removing outliers and normalization;

[0044] Step 4: Positioning information fusion; Use the fingerprint positioning technology of 6G signals to calculate auxiliary positioning information, and fuse it with Beidou and SINS data through the interactive multiple model (IMM) and cubature Kalman filter (CKF) algorithms to improve the positioning accuracy;

[0045] Step 5: Flight trajectory optimization and task scheduling; Based on the communication quality and coverage of 6G signals, optimize the flight trajectory of the UAV, and optimize the task scheduling through the Markov decision process (MDP) and deep reinforcement learning (DRL) algorithms;

[0046] Step 6: System parameter adjustment and performance optimization; Dynamically adjust system parameters according to the fusion positioning results and task execution conditions, and provide real-time feedback to optimize performance to ensure system stability and accuracy.

[0047] Specifically, the system initialization in Step 1 includes:

[0048] This stage includes deploying a drone system with a 6G communication module, a Beidou receiver, and an IMU, and performing necessary parameter initialization to ensure the normal operation of each component. In addition, the initial position, speed, and attitude information of the drone need to be configured to provide reference data for subsequent positioning and navigation tasks. Finally, a communication link between the drone and the ground control center is established to ensure the stability and real-time nature of data transmission.

[0049] Specifically, the data acquisition and preliminary processing in step 2 include: starting the Beidou positioning system to obtain the initial position information of the drone, and starting the SINS to collect the acceleration and angular velocity data of the drone through the IMU for attitude solution. Subsequently, the Kalman filtering algorithm is used to fuse the Beidou positioning data and the SINS data to preliminarily estimate the position, speed, and attitude information of the drone.

[0050] Among them, the state estimation of the Kalman filtering algorithm:

[0051]

[0052] In the formula, is the state vector of the system at time k, Z k is the measurement vector of the system at time k, K k is the Kalman gain matrix, and H is the observation matrix. is the current state estimate based on the previous estimate and model prediction. Among them, the state vector is the information related to the drone estimated inside the Kalman filtering algorithm system, and the measurement vector is the relevant observed value of the drone monitored externally.

[0053] Specifically, the data acquisition and preliminary processing in step 3 include:

[0054] (1) The drone collects 6G fingerprint signal features through the 6G communication module, including the synchronization signal received power (SS-RSRP), the channel state information received power (CSI-RSRP), etc.;

[0055] (2) Preprocess the collected 6G signal features, including removing outliers and normalization. The outliers are identified and removed by the IQR (interquartile range) method to avoid the influence of abnormal fluctuations in the data on subsequent analysis. Subsequently, the normalization method is used to linearly scale the data to the range of [0,1] to improve the accuracy and consistency of the data;

[0056] (3) Set reference points and collect the fingerprint features of the reference points. Using the fingerprint features of the reference points, construct a 6G signal feature database to store the processed 6G signal feature data for reference in fingerprint positioning;

[0057] (4) Calculate the weights of each fingerprint feature using the ReliefF algorithm to reflect their contribution to positioning.

[0058] The normalization formula in the preprocessing process is:

[0059]

[0060] In the formula, max and min represent the maximum and minimum values in the sample respectively, x is the original data, and y is the normalized data.

[0061] The formula for calculating and updating the weights of each fingerprint feature using the ReliefF algorithm is:

[0062]

[0063] In the formula, w(f i ) is the weight of fingerprint feature f i . The w(f i ) on the left side of the formula represents the weight value after this iteration update, and the w(f i ) on the right side is the weight value before this iteration update and the correction term. diff(f i , R, H j ) represents the distance between fingerprint feature f i and sample H j . m represents the number of neighbors of each sample, that is, for each sample R, m nearest neighbor samples are selected. k represents the number of selected neighbors, that is, in the algorithm, the contribution degree and weight of the feature are calculated based on k closest samples. class(R) represents the class to which sample R belongs, P(C) is the proportion of samples in class C, and diff(f i , R, M j (C)) represents the distance between feature f i and other samples in class C.

[0064] Feature distance formula:

[0065]

[0066] Among them, R(f i ) is the value of feature f i . X j (f i ) is the feature value of sample H j . max(f i ) and min(f i ) are the maximum and minimum values of feature f i respectively.

[0067] Specifically, the positioning information fusion in step four includes:

[0068] (1) Using the fingerprint positioning technology of 6G signals, by matching the 6G signal feature database, calculate the auxiliary positioning information of the UAV, and finally the 6G coordinate system conversion is the same as that of the Beidou positioning system (BDS) and SINS.

[0069] (2) Adopt the IAIMM-RSACKF algorithm to fuse the Beidou positioning, SINS navigation and 6G fingerprint positioning data. By combining the improved adaptive multi-model fusion (IMM) with the robust adaptive cubature Kalman filter (RSACKF), further optimize the positioning result.

[0070] (3) By fusing the positioning results, the position, speed and attitude information of the UAV are updated in real time, providing accurate navigation data for flight control.

[0071] As Figure 2 shown, the fingerprint positioning technology of 6G signals: Using the fingerprint positioning technology of 6G signals, first calculate the auxiliary positioning information of the UAV by matching the 6G signal feature database. The features of these 6G signals include the synchronization signal received power (SS-RSRP), the channel state information received power (CSI-RSRP), etc., and are weighted according to their contribution degrees, so as to calculate and estimate the position through the weighted Euclidean distance, providing high-precision auxiliary positioning information for the UAV.

[0072] The above fingerprint positioning technology is realized by the Relief-WKNN positioning algorithm, and the formula is:

[0073]

[0074] In the formula, f i is each fingerprint feature, MFF i is the value of the current fingerprint feature, MFF ji is the fingerprint feature value at the jth position in the database, w(f i ) is the weight of each feature.

[0075] Weighted Euclidean distance calculation:

[0076]

[0077] In the formula, d j is the distance between the target fingerprint and the jth fingerprint in the database, and K is the number of selected minimum neighboring points.

[0078] Preliminary estimation of the UAV position:

[0079]

[0080] (x j , y j ) is the data at the jth position in the database, vj is the corresponding weight value.

[0081] The conversion of the 6G coordinate system enables this information to be unified with the Beidou positioning system (BDS) and SINS, and fusion processing is carried out in the same coordinate system. Assume that the coordinates of a certain point i in the custom space rectangular coordinate system are (x, y), and its corresponding geographical coordinates are (L, B). The basic principle of the conversion is as follows:

[0082] The formula for converting the geographical coordinates (L, B) to the Web Mercator coordinate system:

[0083]

[0084] Convert the Web Mercator coordinate system (X, Y) to the space rectangular coordinate system (x, y)

[0085]

[0086] Among them, T is the conversion matrix obtained through the coordinate relationships of 3 known points.

[0087]

[0088] Figure 3 As shown, the IAIMM-RSACKF algorithm is adopted to fuse Beidou positioning, SINS navigation, and 6G fingerprint positioning data. By combining the improved adaptive multi-model fusion (IMM) with the robust adaptive cubature Kalman filter (RSACKF), the positioning result is further optimized.

[0089] Before combining data processing through the improved adaptive multi-model fusion and the robust adaptive cubature Kalman filter, the Beidou positioning data and SINS data are preliminarily fused through the data acquisition and preliminary processing in step 2 above to reduce the initial error and provide a relatively stable position information, preparing for more complex multi-source fusion in the follow-up. This fusion stage first utilizes the high dynamic characteristics of SINS and the global precise positioning ability of Beidou to make the initial position, speed, and attitude information more stable, providing high-quality input data for subsequent fusion.

[0090] At the same time, the position data in the Beidou positioning data and the 6G fingerprint positioning are also preliminarily fused, and the fusion method can choose the Kalman filter algorithm in step 2, also providing high-quality input data.

[0091] Using the data obtained from the preliminary fusion of Beidou and SINS and the data obtained from the preliminary fusion of 6G and Beidou as the data basis, subsequent further precise data processing is carried out through the combination of the above-mentioned improved adaptive multi-model fusion and the robust adaptive cubature Kalman filter.

[0092] The IAIMM-RSACKF algorithm can efficiently fuse among multi-source data (Beidou positioning, SINS navigation, and 6G fingerprint positioning), automatically adjust the weights of each model, and ensure that the system can make accurate estimates in a dynamically changing environment. Through this dual adaptive filtering mechanism, the system can dynamically select the appropriate model and adjust the model weights, enabling the positioning accuracy to remain efficient and stable under different environmental conditions. The IMM algorithm generally includes four steps: input interaction, parallel filtering, model update, and output interaction.

[0093] Among them, in the input interaction, the probability weights of the models and the final covariance at the previous moment are obtained. According to the probabilities of each model and the model transition probability matrix at the previous moment, the mixed probabilities of each model at the current moment are calculated, and the states and covariances of each sub-filter are weighted and averaged to generate a new initial state. It provides a data basis for the prediction at the current moment. In parallel filtering, it is used to predict and update the current state and covariance based on the state at the previous moment and the data in the input interaction; in the model update, the current model is updated according to the model weights at the previous moment, and the output interaction is used to output the final state, Kalman gain, and final covariance after current mixing.

[0094] The formula of the IMM algorithm is as follows:

[0095] State prediction:

[0096]

[0097] In the formula, is the state prediction of the i-th model at the (k - 1)-th moment, representing the position, velocity, attitude error, gyroscope drift, accelerometer zero bias error, etc. of the navigation system. A (i) is the state transition matrix of the i-th model, u k is the control input, mainly the acceleration and angular velocity data provided by the IMU, and also includes external information such as 6G fingerprint positioning data and BDS data, which are usually used for observation correction. B (i) represents the control matrix.

[0098] It should be noted that the control input u k is neither measurement data nor model weighting weights, but an external input for state prediction.

[0099] Covariance prediction:

[0100]

[0101] In the formula, is the covariance prediction of the i-th model at the (k - 1)-th moment, and Q (i) is the process noise.

[0102] Model weighted update:

[0103]

[0104] Wherein, is the weight of the i-th model, is the measurement update probability of the i-th model.

[0105] Final state estimation:

[0106]

[0107] Wherein, is the state estimation of the i-th model.

[0108] Kalman gain update:

[0109]

[0110] Wherein, K K is the Kalman gain, H k is the observation matrix, R k is the measurement noise covariance

[0111] Final covariance estimation P K :

[0112]

[0113] Wherein, is the covariance of the i-th model, and the final covariance is the updated covariance.

[0114] In order to further improve the performance of the improved adaptive multi-model fusion algorithm, in the filtering model of the improved adaptive multi-model fusion algorithm, the RSACKF algorithm is combined, that is, the traditional Kalman filtering model in the improved adaptive multi-model fusion is updated to the RSACKF algorithm to improve the positioning accuracy and system stability.

[0115] Based on the traditional Kalman filter, the RSACKF algorithm combines the singular value decomposition (SVD) and the robust adaptive update mechanism, enabling it to effectively handle nonlinear systems, measurement noise, and outliers, thereby improving the positioning accuracy and system stability. In the fusion of multiple positioning sources, RSACKF can adaptively adjust the Kalman filter model parameters to further optimize the system performance.

[0116] SVD decomposition:

[0117] V T (P k )P k V = S k

[0118] Wherein, V is the matrix after SVD decomposition, and S k is a diagonal matrix containing singular values, and P k is the above-mentioned final covariance.

[0119] Adaptive update of measurement noise:

[0120]

[0121] Wherein, r i is the measurement noise at each moment, and R k is the value after adaptive update of the measurement noise.

[0122] Update the measurement covariance matrix:

[0123]

[0124]

[0125] Wherein, is the updated measurement covariance, r k is the measurement value at the current moment, is the noise estimation value at the current moment.

[0126] Use the above-mentioned updated measurement covariance Calculate the Kalman gain of RSACKF That is, in the Kalman gain calculation formula, replace the original measurement covariance matrix R k with the latest to obtain the final Kalman gain K of RSACKF k .

[0127] State estimation after dual adaptive filtering fusion of the IAIMM-RSACKF algorithm:

[0128]

[0129] Wherein, is the weighted weight of the model in the IMM algorithm, K k is the Kalman gain of RSACKF, and M represents the total number of models of the interactive multiple model.

[0130] The final output of the interaction is the above-mentioned fused state estimation value. The IAIMM-RSACKF algorithm fuses the information of multiple data sources (such as 6G, Beidou, and SINS) and processes it through a filtering algorithm. The final output of the interaction is the final positioning estimation.

[0131] Specifically, the flight trajectory optimization and task scheduling in step five include:

[0132] (1) Optimize the flight trajectory of the drone based on the communication quality and coverage of 6G signals to ensure the stability of the communication link and the real-time nature of data transmission;

[0133] (2) Utilize the Markov Decision Process (MDP) and Deep Reinforcement Learning (DRL) algorithms to optimize the task scheduling of the drone, improving the efficiency and reliability of task execution;

[0134] (3) Dynamically adjust the flight trajectory and task scheduling strategy by real-time monitoring of the drone's flight status and task execution situation to adapt to complex environmental changes;

[0135] (4) Combine the high-bandwidth and low-latency characteristics of 6G signals to achieve real-time communication between the drone and the ground control center, ensuring the timeliness and accuracy of task scheduling.

[0136] As Figure 4 shown, for flight trajectory optimization and task scheduling, optimizing the drone's flight trajectory based on the communication quality and coverage of 6G signals can effectively ensure that the flight path meets communication requirements, avoiding signal attenuation and interference areas, thereby improving communication quality and reducing task latency. By using the Markov Decision Process (MDP) to model flight trajectory optimization, MDP takes into account factors such as the drone's position, speed, and communication quality, and optimizes the flight path by maximizing the cumulative reward. Combined with the Deep Q-Network (DQN), this algorithm dynamically adjusts the optimal flight strategy according to environmental changes to optimize the flight trajectory at each time step. At the same time, the DCTP algorithm further optimizes task scheduling based on DQN to ensure that the drone's flight tasks maximize the uplink throughput and improve task execution efficiency. Dynamically adjust the flight trajectory and task scheduling strategy by real-time monitoring of the drone's flight status and task execution situation to adapt to complex environmental changes. Combine the high-bandwidth and low-latency characteristics of 6G signals to achieve real-time communication between the drone and the ground control center, ensuring the timeliness and accuracy of task scheduling. The reward function and policy optimization formula in MDP are as follows:

[0137]

[0138] In the formula, R is the reward function, and the reward function measures the total benefit accumulated by the drone during the entire task execution process. The goal of optimizing the flight trajectory is to adjust the strategy to maximize R, thereby improving task completion, data transmission efficiency, or energy utilization. r t is the immediate reward at each moment t, and the immediate reward represents the return obtained by the drone after taking a specific flight action at each moment t. When the drone flies to an area with better communication quality and successfully transmits data, r t may be higher, while in the case of severe signal attenuation or task failure, r tIt may be relatively low or penalized. T is the maximum number of time slots for the task. The maximum number of time slots for the task limits that the task must be completed within T time steps. The UAV covers a specified area or completes a data transmission task within a limited time. The goal of trajectory optimization is to select the optimal path within these limited time slots to maximize the cumulative reward.

[0139] Decision optimization formula:

[0140]

[0141] In the formula, π * (s) represents the optimal policy in state s, the optimal action. That is, the optimal action is that when the UAV is in different positions or states, it selects the most appropriate action plan to ensure that the flight trajectory can not only meet the communication requirements but also optimize the task execution effect. This is an operation that represents selecting from all possible actions a ∈ A the one that can make Q π (s, a) the largest. Here, it means that the UAV will evaluate various possible flight direction or speed adjustment plans in the current state and then select an optimal one, that is, an action that can bring the greatest benefit. For example, it selects to fly in the area with the best signal or bypass the area with severe interference. Q π (s, a) is the state-action value function of this action. This function can be understood as an "evaluation score", which represents the comprehensive benefit that the UAV may obtain in the future after executing a certain action. For example, if the UAV decides to fly towards a certain area, this function will predict the impact on communication quality, energy consumption, and task completion. s ∈ S represents any state in the state space, A represents the action space, representing various operations that the UAV can execute, such as flying left, flying right, accelerating, decelerating, adjusting altitude, etc. All these possible operations constitute the action space. S represents the state space, which here refers to the current various states of the UAV, such as its geographical location, altitude, speed, 6G signal strength, remaining battery power, etc. Simply put, it is the environmental information where the UAV is currently located.

[0142] DQN update formula:

[0143]

[0144] In the formula, L(θ) is the loss function, θ is the weight of the current network, θ - is the weight of the target network, Q θ (s, α) is the Q value of taking action α in the current state, r is the reward obtained after taking action α in the current state, and γ is the discount factor.

[0145] DQN update formula in the DCTP algorithm:

[0146]

[0147] Where α is the learning rate, r is the immediate reward, γ is the discount factor, and θ is the weight of the current neural network.

[0148] The system parameter adjustment and performance optimization in Step 6 include:

[0149] Dynamically adjust the flight parameters, communication parameters, and navigation parameters of the UAV according to the fused positioning results and task execution conditions to optimize the system performance. In addition, use a real-time feedback mechanism to evaluate the effect of system parameter adjustment to ensure the stability and accuracy of the system. By introducing the Deep Q-Network (DQN) algorithm in deep reinforcement learning, a parameter optimization model based on the Markov Decision Process (MDP) is constructed to achieve dynamic and reasonable allocation of system resources. This method uses communication quality, flight energy consumption, and task completion efficiency as the reward function to train the UAV to autonomously select the optimal action according to the environmental state, thereby improving the overall operation efficiency and reliability of the system. At the same time, combined with the flight experience and historical data of the UAV, continuously optimize the system parameter adjustment strategy to improve the adaptive ability and intelligent level of the system.

[0150] The system parameter adjustment and performance optimization ensure the stability and accuracy of the system under various environmental conditions by dynamically adjusting the system parameters. First, according to the fused positioning results and task execution conditions, monitor the state and performance indicators of the system in real time, such as the accuracy of the flight path, communication quality, and task completion efficiency. The system will dynamically adjust the parameters according to this feedback data, such as the speed of flight control, the adjustment frequency of path planning, and the accuracy of data collection, etc., to optimize the performance of the UAV. By adaptively adjusting the system parameters, it can ensure efficient operation performance in complex and uncertain environments and avoid task delays or failures caused by external factors such as signal interference or environmental changes. In addition, the system will regularly evaluate the effect of parameter adjustment and further optimize the strategy according to the task execution situation to ensure the continuous stable operation and precise navigation of the UAV, thereby improving the efficiency and reliability of overall task execution.

[0151] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A combined positioning method for UAV Beidou and SINS based on 6G signals, characterized in that, Including: Obtaining the positioning information of the drone, where the positioning information includes Beidou positioning information and inertial navigation system positioning information; Performing preliminary fusion on the positioning information to obtain preliminary estimated positioning information; Obtaining the fingerprint signal characteristics of the 6G signal of the drone, and obtaining auxiliary positioning information according to the fingerprint signal characteristics; Fusing the preliminary estimated positioning information and the auxiliary positioning information through an interactive multiple model and a cubature Kalman filter model to obtain fused positioning information to achieve combined positioning.

2. The method according to claim 1, characterized in that Before obtaining the positioning information of the drone, it includes: Deploying a 6G communication module, a Beidou receiver, and an IMU on the drone, initializing the deployed devices, and initializing the configuration of the position, speed, and attitude of the drone.

3. The method according to claim 1, characterized in that Performing preliminary fusion on the positioning information through the Kalman filter algorithm.

4. The method according to claim 1, characterized in that The 6G fingerprint signal characteristics include the received power of the synchronization signal and the received power of the channel state information.

5. The method according to claim 1, characterized in that The process of obtaining the auxiliary positioning information includes: Constructing a fingerprint signal characteristic database, where a number of samples are stored in the fingerprint signal characteristic database, and the samples are fingerprint signal characteristics at known positions; Calculating the weights of the fingerprint signal characteristics according to the samples by the ReliefF method to obtain the fingerprint signal characteristic weights; According to the fingerprint signal characteristic weights, calculating the distance between the samples and the fingerprint signal characteristics by the Relief-WKNN positioning method, and according to the calculation results, obtaining the weighted Euclidean distance, and performing weighted sum calculation according to the weighted Euclidean distance and the position information of the samples to obtain the auxiliary positioning information.

6. The method according to claim 1, characterized in that Before fusing the preliminary estimated positioning information and the auxiliary positioning information, it further includes: Converting the preliminary estimated positioning information and the auxiliary positioning information to the same coordinate system.

7. The method according to claim 1, characterized in that The process of fusing the preliminary estimated positioning information and the auxiliary positioning information includes: Fusing the auxiliary positioning information with the Beidou positioning information to obtain position fusion information; Fusing the preliminary estimated positioning information and the position fusion information through an interactive multiple model and a cubature Kalman filter model, where each Kalman filter model in the interactive multiple model is used to filter the preliminary estimated positioning information and the position fusion information respectively and fuse the filtering results, and each Kalman filter model in the interactive multiple model is replaced by a cubature Kalman filter model, and the fusion process is: Among them, is the weighted weight of the i-th model at the k-th moment in the interactive multi-model, K k Kalman gain of the volume Kalman filter model, represents the state prediction at the k-th moment, z k represents the measurement matrix, i represents the corresponding model label, k represents the time step label, H k is the observation matrix.

8. The method according to claim 1, characterized in that After obtaining the fused positioning information, it further includes: Optimizing the flight trajectory of the drone based on the communication quality and coverage of the 6G signal; wherein, based on the communication quality and coverage of the 6G signal, the flight trajectory of the drone is modeled through a Markov decision process, with the optimal communication quality and task execution efficiency of the drone as the goal, and the flight trajectory of the drone is optimized through a deep Q-network.

9. The method according to claim 8, wherein After optimizing the flight trajectory of the drone, it further includes: Adjusting the system parameters of the drone according to the fused positioning information and the optimized flight trajectory, wherein the system parameters include flight parameters, communication parameters, and navigation parameters.

Images