Unmanned aerial vehicle navigation positioning method and system based on big data

Abstract

The invention discloses an unmanned aerial vehicle navigation positioning method and system based on big data, and the method comprises the steps: obtaining the position data, sensor data and environment data of an unmanned aerial vehicle, constructing a real-time map according to the position data and the sensor data through employing an SLAM algorithm, and carrying out the real-time positioning; carrying out sub-region division on the real-time map, and obtaining a risk score by adopting clustering analysis; obtaining path time between the sub-regions according to the environment data and the risk score, establishing a fitness function according to the risk score and the path time, and establishing a path planning model based on the fitness function; obtaining offset data according to the sensor data, establishing a path compensation strategy according to the offset data, and performing path correction; learning the path compensation strategy and the path planning model to obtain an unmanned aerial vehicle navigation model, and optimizing the unmanned aerial vehicle navigation model through a sine and cosine optimization algorithm. The method is suitable for various unmanned aerial vehicle application scenes, and wide application of the unmanned aerial vehicle technology in a complex environment is promoted.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle navigation and positioning technology, and in particular to a unmanned aerial vehicle navigation and positioning method based on big data and a system thereof. Background Art

[0002] With the rapid development of drone technology, it has been widely used in many fields, such as aerial photography, logistics distribution, agricultural plant protection, etc. However, drones face many challenges during flight, among which the accuracy and reliability of navigation and positioning are crucial issues.

[0003] Traditional UAV navigation and positioning methods mainly rely on the Global Positioning System (GPS), but in some complex environments, such as urban high-rise areas, mountainous areas, and areas with strong electromagnetic interference, GPS signals may be blocked, interfered with, or reflected, resulting in inaccurate or even invalid positioning. In addition, existing navigation and positioning methods often do not fully consider the impact of environmental factors on UAV flight during path planning, such as meteorological conditions (wind direction, wind speed, etc.), obstacle distribution, and electromagnetic interference, which may cause UAVs to face higher risks during flight, such as extended flight time, increased energy consumption, and even collision accidents. Therefore, it is necessary to develop a more accurate, reliable, and complex environment-adaptable UAV navigation and positioning method and system. Summary of the invention

[0004] The purpose of the present invention is to provide a UAV navigation and positioning method and system based on big data.

[0005] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0006] The present invention comprises the following steps:

[0007] Acquire the position data, sensor data and environmental data of the drone, and use the SLAM algorithm to build a real-time map and perform real-time positioning according to the position data and the sensor data;

[0008] Dividing the real-time map into sub-areas and obtaining risk scores by cluster analysis;

[0009] Obtaining the path time between the sub-areas according to the environmental data and the risk score, establishing a fitness function according to the risk score and the path time, and establishing a path planning model based on the fitness function;

[0010] Obtaining offset data according to the sensor data, establishing a path compensation strategy according to the offset data and performing path correction;

[0011] The path compensation strategy and the path planning model are learned to obtain a UAV navigation model, and the UAV navigation model is optimized by a sine-cosine optimization algorithm.

[0012] Furthermore, the method for obtaining the position data, sensor data and environmental data of the drone includes:

[0013] The drone position obtained by GPS, GLONASS and Galileo is fused through the pseudo-range algorithm to obtain the position data. The sensors include data from three-dimensional cameras, barometers, gyroscopes, accelerometers and inertial measurement units. The sensor data includes attitude data, speed data, acceleration data, air pressure data, and image data. The environmental data includes meteorological data and electromagnetic interference data. The meteorological data includes wind direction, wind speed and air pressure. The electromagnetic interference data includes the intensity and frequency of electromagnetic interference.

[0014] Furthermore, a method for constructing a real-time map and performing real-time positioning based on the position data and the sensor data using a SLAM algorithm includes:

[0015] The GIS system is used to obtain the GIS map of the area around the drone based on the location data. The SLAM algorithm is used to build an obstacle model based on the real-time monitoring data of the three-dimensional camera. A real-time map is built based on the GIS map to obtain the real-time position of the drone in the real-time map.

[0016] Furthermore, the real-time map is divided into sub-areas and a method of obtaining a risk score by cluster analysis includes:

[0017] The real-time map distinguishes the flyable area and the non-flyable area, and divides the flyable area into cubic sub-areas.

[0018] Data points are established based on the surrounding obstacle density, wind speed, electromagnetic interference data and air pressure of the sub-area, and the SOM algorithm is used to cluster the data points to form risk areas:

[0019] Randomly select k data points as the initial weight vectors of the competitive layer neurons. One competitive layer neuron represents a cluster center. Calculate the similarity between the data point and the weight vector of each competitive layer neuron, and assign the data point to the cluster of neurons with the highest similarity.

[0020] The calculation formula for similarity is:

[0021]

[0022] d(h,e) is the similarity between h and e, h and e are two r-dimensional vectors, ||h(w κ )|| and||e(w κ )|| are the weighted vector lengths of vector h and vector e, respectively.κ represents the component of vector h in the κth dimension, w κ is the weight coefficient of the κth dimension, and a neighborhood with the neuron as the center and σ(0) as the initial neighborhood radius is drawn.

[0023] Dynamic topological neighborhood function in the competition layer:

[0024]

[0025] Among them, h gi (t) is the dynamic topological neighborhood function of neuron i and neuron g at the tth iteration, d ig (t) is the distance between neuron i and neuron g at the t-th iteration, is the similarity between neurons i and g, σ(t) is the neighborhood radius at the t-th iteration, t max is the maximum number of iterations, δ(t) is the influencing factor that changes with time, ξ is the initial value of the impact factor, ζ is the adjustment coefficient of the impact factor,

[0026] Update of the neuron's weight vector:

[0027] w ij (t+1)=w ij (t)+η(t)·h gi (t)·(x j -w ij (t))-γ·w ij (t)

[0028] Among them, γ is the strength of the regularization term, w ij (t) and w ij (t+1) is the j-th dimension component of the weight vector of the ith neuron at the t-th and t+1-th iterations, η(t) is the learning rate of the t-th iteration, x j is the j-th dimension component of the data point,

[0029] After reaching the maximum number of iterations, the clustering result is output, and the risk score of each risk area is the average value of the weighted vector length of the data points therein.

[0030] Furthermore, the method for obtaining the path time between the sub-areas according to the environmental data and the risk score includes:

[0031] Path time function:

[0032]

[0033] Among them, T cb is the path time between the center of sub-area c and the center of sub-area b, w Pis the impact weight of risk score, P c and P b are the risk scores of the risk areas corresponding to sub-area c and sub-area b, respectively, S c and S b are the parts of the path length between the centers of sub-areas c and b in sub-areas c and b, V is the speed of the drone, and V f (c) and V f (b) are the wind speeds in sub-area c and sub-area b, and are the clockwise angles between the wind direction in sub-area c and sub-area b and the forward direction of the UAV, ω f is the weight of the impact of wind speed on the UAV speed.

[0034] Furthermore, the method of establishing a fitness function according to the risk score and the path time includes:

[0035] Fitness function:

[0036]

[0037] Where f(L) is the fitness function of path L, c∈L means that sub-region c is in the sub-region set of path L, sub-region c is adjacent to sub-region b, cb∈L means that sub-region c to b is a sub-path in path L, W cb is the energy consumption of subpath cb, w T and w W are the weights of path time and energy consumption,

[0038] The formula for calculating energy consumption is:

[0039]

[0040] Among them, m and m l are the weight of the drone and the payload, φ is the acceleration of gravity, ΔH is the height change, and are the air pressures in sub-areas c and b, respectively, is the gas constant, and are the temperatures of sub-areas c and b respectively, and A is the windward area of ​​the UAV.

[0041] Furthermore, the method for establishing a path planning model based on the fitness function includes:

[0042] Each ant moves on the directed graph composed of nodes consisting of sub-regions and leaves pheromones, initializing the parameters of the ant colony algorithm. The probability of ant transfer between nodes is:

[0043]

[0044] Among them is The probability that the kth ant moves from node u to node v at time s is, and are the weights of pheromone and heuristic function respectively, is the index of the node in N(u), N(u) is the set of neighbor nodes of node u, From node u to node The heuristic function is From node u to node The distance between the center of sub-region u and the sub-region The path length between centers, For sub-region The risk score of

[0045] The pheromone update formula is:

[0046]

[0047] Among them, μ uv (s) and μ uv (s+1) represents the amount of pheromone on path uv at time s and time s+1, α is the pheromone volatility coefficient, Z uv is the fitness of path uv, is the fitness function value of path uv,

[0048] When the pheromone update amount is lower than the update threshold, the best path corresponding to the strongest pheromone signal is output as the planned path.

[0049] Re-plan the route when real-time map and environmental data change.

[0050] Furthermore, the method of obtaining the offset data according to the sensor data, establishing a path compensation strategy according to the offset data and performing path correction includes:

[0051] According to the real-time position, attitude data and speed data, the deviation between the actual flight path of the UAV and the planned path is obtained as the offset data, including position deviation, attitude deviation and speed deviation. The state equation and observation equation of the UAV are established based on the offset data through the Kalman filtering algorithm, and the state of the UAV is continuously estimated. Based on the state estimation and the offset of the planned path, the path compensation strategy is reversely applied to adjust the flight direction and speed of the UAV, perform path correction, and reduce the offset of the state estimation and the planned path.

[0052] Furthermore, a method for obtaining a UAV navigation model by learning a path compensation strategy and a path planning model and optimizing the UAV navigation model by a sine-cosine optimization algorithm includes:

[0053] The UAV navigation model is trained using a reinforcement learning algorithm. The state space is the real-time position, attitude data, risk score, path time, speed data, and environmental data of the UAV. The action space is to adjust the flight direction and speed of the UAV. The reward function is:

[0054]

[0055] Among them, k P , k T , k W and k θ are the risk score reward coefficient, path time reward coefficient, energy consumption reward coefficient and path stability reward coefficient, respectively. expected and T actrual The path time of the planned path of the path planning model and the actual path time of the reinforcement learning algorithm are respectively expected and W actrual are the energy consumption of the planned path and the actual energy consumption of the reinforcement learning algorithm, Δθ is the deviation rate between the path direction and the path planning direction of the reinforcement learning algorithm, and θ stabel is the deviation rate threshold of the path direction;

[0056] The inverse of the objective function value of the drone navigation model is used as the fitness value. The position of each search individual represents a set of hyperparameter values. The search individual position update formula is:

[0057]

[0058] in, and are the positions of the ith search individual at the t+1th and tth iterations, respectively, and X * is the global optimal position, is the historical optimal position searched by the i-th search individual, ω t is the nonlinear weight factor for the tth iteration, is the control parameter of the amplitude of the sine function, t is the current iteration number, t max is the maximum number of iterations, r1 is the control parameter, r1=a(1-t / t max ), a is a constant greater than 1, is the adjustment parameter of the frequency of the sine function and the cosine function, r2∈(0,2π) is a random weight that obeys a uniform distribution, r3∈(0,1) is a random number that obeys a uniform distribution and is used to select the individual position update strategy,

[0059] When the average fitness value of the historical optimal position searched by all search individuals has not changed for four consecutive iterations, the search is considered to be stagnant. At this time, the Levy flight random walk strategy is used to update the position of the search individual. The random walk step length formula of Levy flight is:

[0060]

[0061] Where s is the random walk step length, u and v are parameters that obey the normal distribution, Γ(·) is the gamma function, β is the exponential parameter that controls the shape of the Levy distribution, β∈(0,2), β=1.5, σ u is the standard deviation of u, σ v is the standard deviation of v;

[0062] The random walk strategy based on Levy flight is:

[0063]

[0064] Where S randn is a random quantity that obeys the standard normal distribution, α∈[-1,1] is the proportional factor, α=0.9,

[0065] When the maximum number of iterations is reached, the hyperparameter value corresponding to the global optimal position is output.

[0066] A UAV navigation and positioning system based on big data, characterized by comprising:

[0067] The acquisition module obtains the location data, sensor data and environmental data of the UAV, divides the flight area, and uses the GIS system to obtain the spatial information of the flight area;

[0068] A computing module, using a SLAM algorithm to construct a real-time map according to the position data and the sensor data and perform real-time positioning, divide the real-time map into sub-areas, use cluster analysis to obtain a risk score, obtain a path time between the sub-areas according to the environmental data and the risk score, establish a fitness function according to the risk score and the path time, establish a path planning model based on the fitness function, obtain offset data according to the sensor data, and learn a path compensation strategy and a path planning model to obtain a UAV navigation model;

[0069] A correction module, establishing a path compensation strategy and performing path correction according to the offset data;

[0070] The optimization module optimizes the UAV navigation model through a sine-cosine optimization algorithm.

[0071] The beneficial effects of the present invention are:

[0072] The present invention realizes accurate navigation and positioning by integrating multiple technologies. It obtains and integrates data from multiple systems to improve positioning accuracy; analyzes environmental risks based on big data to optimize path planning; and uses reinforcement learning and optimization algorithms to improve navigation model performance, thereby enhancing the adaptability, safety, and efficiency of drones in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 The present invention is a flowchart of the steps of a UAV navigation and positioning method based on big data and its system. DETAILED DESCRIPTION

[0074] The present invention is further described below by means of specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0075] The present invention provides a UAV navigation and positioning method based on big data and a system thereof, comprising the following steps:

[0076] like Figure 1 As shown, in this embodiment, the following steps are included:

[0077] Acquire the position data, sensor data and environmental data of the drone, and use the SLAM algorithm to build a real-time map and perform real-time positioning according to the position data and the sensor data;

[0078] In the actual evaluation, in a certain flight, the initial position data of the drone was 116.404° longitude, 39.915° latitude, and 50 meters altitude. The altitude of the drone was calculated by the difference between the altitude of the drone and the altitude of the ground. The drone was facing due north, and the 3D camera captured in real time that there was an obstacle 50 meters ahead with a height of 100 meters and a width of 30 meters. The barometer showed that the current air pressure was 1013.25hPa. The gyroscope and accelerometer provided the attitude and acceleration information of the drone, with a pitch angle of 5°, a roll angle of 0°, a speed of 20m / s, and an acceleration of 0.5m / s. 2 ; Environmental data: The current wind direction is north wind, wind speed is 3m / s; electromagnetic interference intensity is 5V / m, frequency is 1MHz; after establishing the real-time map, it is found that the obstacle length is 50 meters, and the initial real-time position of the drone is actually longitude 116.444°, latitude 30.015°, and altitude 50 meters;

[0079] Dividing the real-time map into sub-areas and obtaining risk scores by cluster analysis;

[0080] In actual assessment, the size of each sub-area can be different or the same when the area is divided. The 50×50×50m cube area between the drone and the obstacle is divided into sub-area 1, the 50×50×50m cube area on the left side of the obstacle is divided into sub-area 2, the 50×50×50m cube area on the right side of the obstacle is divided into sub-area 3, and the 50×50×50m cube area behind the obstacle is divided into sub-area 4; the surrounding obstacle densities are: sub-areas 1, 2, 3 and 4: 0.4, 0.5, 0.4 and 0.5, and the risk scores are: sub-areas 1, 2, 3 and 4: 0.3, 0.4, 0.3 and 0.3;

[0081] Obtaining the path time between the sub-areas according to the environmental data and the risk score, establishing a fitness function according to the risk score and the path time, and establishing a path planning model based on the fitness function;

[0082] In the actual evaluation, the path time is: 8.5s for sub-area 1 to 2, 4.8s for sub-area 1 to 3, 5.3s for sub-area 2 to 4, and 8s for sub-area 3 to 4; the fitness function value is: 0.53 for path 1-2, 0.8 for path 1-3, 0.73 for path 2-4, 0.6 for path 3-4, 1.36 for path 1-2-4, and 1.5 for path 1-3-4; the planned path of the path planning model is path 1-3-4;

[0083] Obtaining offset data according to the sensor data, establishing a path compensation strategy according to the offset data and performing path correction;

[0084] In the actual evaluation, the position deviation between the actual flight path of the drone and the planned path was 2 meters to the west, and the speed deviation was -2m / s. The path compensation strategy was to adjust the flight direction of the drone to the northeast and the acceleration to 1m / s for 2s.

[0085] The path compensation strategy and the path planning model are learned to obtain a UAV navigation model, and the UAV navigation model is optimized by a sine-cosine optimization algorithm.

[0086] In this embodiment, the method for obtaining the position data, sensor data and environmental data of the drone includes:

[0087] The drone position obtained by GPS, GLONASS and Galileo is fused through the pseudo-range algorithm to obtain the position data. The sensors include data from three-dimensional cameras, barometers, gyroscopes, accelerometers and inertial measurement units. The sensor data includes attitude data, speed data, acceleration data, air pressure data, and image data. The environmental data includes meteorological data and electromagnetic interference data. The meteorological data includes wind direction, wind speed and air pressure. The electromagnetic interference data includes the intensity and frequency of electromagnetic interference.

[0088] In this embodiment, a method for constructing a real-time map and performing real-time positioning based on the position data and the sensor data using a SLAM algorithm includes:

[0089] The GIS system is used to obtain the GIS map of the area around the drone based on the location data. The SLAM algorithm is used to build an obstacle model based on the real-time monitoring data of the three-dimensional camera. A real-time map is built based on the GIS map to obtain the real-time position of the drone in the real-time map.

[0090] In this embodiment, the method of dividing the real-time map into sub-areas and obtaining risk scores by cluster analysis includes:

[0091] According to the real-time map, the flyable area and the non-flyable area are distinguished. The non-flyable area includes the no-fly zone, the area beyond the flight height limit of the drone, and obstacles. The flyable area is divided into cubic sub-areas. The side length of each sub-area is greater than or equal to 3 times the maximum length of the drone and less than or equal to 500 meters. The part of the flyable area that is smaller than the smallest sub-area is removed and regarded as the non-flyable area.

[0092] Data points are established based on the surrounding obstacle density, wind speed, electromagnetic interference data and air pressure of the sub-area. The surrounding obstacle density is the obstacle volume ratio in the impact area of ​​3 times the maximum length of the drone around the sub-area. The SOM algorithm is used to cluster the data points to form a risk area:

[0093] Randomly select k data points as the initial weight vectors of the competitive layer neurons. One competitive layer neuron represents a cluster center. Calculate the similarity between the data point and the weight vector of each competitive layer neuron, and assign the data point to the cluster of neurons with the highest similarity.

[0094] The calculation formula for similarity is:

[0095]

[0096] d(h,e) is the similarity between h and e, h and e are two r-dimensional vectors, ||h(w κ )|| and||e(w κ )|| are the weighted vector lengths of vector h and vector e, respectively. κ represents the component of vector h in the κth dimension, w κ is the weight coefficient of the κth dimension, and a neighborhood with the neuron as the center and σ(0) as the initial neighborhood radius is drawn.

[0097] Dynamic topological neighborhood function in the competition layer:

[0098]

[0099] Among them, h gi (t) is the dynamic topological neighborhood function of neuron i and neuron g at the tth iteration, d ig (t) is the distance between neuron i and neuron g at the t-th iteration, is the similarity between neurons i and g, σ(t) is the neighborhood radius at the t-th iteration, t max is the maximum number of iterations, δ(t) is the influencing factor that changes with time, ξ is the initial value of the impact factor, ζ is the adjustment coefficient of the impact factor,

[0100] Update of the neuron's weight vector:

[0101] w ij (t+1)=w ij (t)+η(t)·h gi (t)·(x j -w ij (t))-γ·w ij (t)

[0102] Among them, γ is the strength of the regularization term, w ij (t) and w ij (t+1) is the j-th dimension component of the weight vector of the ith neuron at the t-th and t+1-th iterations, η(t) is the learning rate of the t-th iteration, x j is the j-th dimension component of the data point,

[0103] After reaching the maximum number of iterations, the clustering result is output, and the risk score of each risk area is the average value of the weighted vector length of the data points therein.

[0104] In this embodiment, the method for obtaining the path time between the sub-areas according to the environmental data and the risk score includes:

[0105] Path time function:

[0106]

[0107] Among them, T cb is the path time between the center of sub-area c and the center of sub-area b, w P is the impact weight of risk score, P c and P b are the risk scores of the risk areas corresponding to sub-area c and sub-area b, respectively, S c and S b are the parts of the path length between the centers of sub-areas c and b in sub-areas c and b, V is the speed of the drone, and V f (c) and V f(b) are the wind speeds in sub-area c and sub-area b, and are the clockwise angles between the wind direction in sub-area c and sub-area b and the forward direction of the UAV, ω f is the weight of the impact of wind speed on the UAV speed.

[0108] In this embodiment, the method of establishing a fitness function according to the risk score and the path time includes:

[0109] Fitness function:

[0110]

[0111] Where f(L) is the fitness function of path L, c∈L means that sub-region c is in the sub-region set of path L, sub-region c is adjacent to sub-region b, cb∈L means that sub-region c to b is a sub-path in path L, W cb is the energy consumption of subpath cb, w T and w W are the weights of path time and energy consumption respectively. The energy consumption is calculated by:

[0112]

[0113] Among them, m and m l are the weight of the drone and the payload, φ is the acceleration of gravity, ΔH is the height change, and are the air pressures in sub-areas c and b, respectively, is the gas constant, and are the temperatures of sub-areas c and b respectively, and A is the windward area of ​​the UAV.

[0114] In this embodiment, the method for establishing a path planning model based on the fitness function includes:

[0115] Each ant moves on the directed graph composed of nodes consisting of sub-regions and leaves pheromones, initializing the parameters of the ant colony algorithm. The probability of ant transfer between nodes is:

[0116]

[0117] Among them is The probability that the kth ant moves from node u to node v at time s is, and are the weights of pheromone and heuristic function respectively, is the index of the node in N(u), N(u) is the set of neighbor nodes of node u, From node u to node The heuristic function is From node u to node The distance between the center of sub-region u and the sub-region The path length between centers, For sub-region The risk score of

[0118] The pheromone update formula is:

[0119]

[0120] Among them, μ uv (s) and μ uv (s+1) represents the amount of pheromone on path uv at time s and time s+1, α is the pheromone volatility coefficient, Z uv is the fitness of path uv, is the fitness function value of path uv,

[0121] When the pheromone update amount is lower than the update threshold, the best path corresponding to the strongest pheromone signal is output as the planned path.

[0122] Re-plan the route when real-time map and environmental data change.

[0123] In this embodiment, the method of obtaining the offset data according to the sensor data, establishing a path compensation strategy according to the offset data and performing path correction includes:

[0124] According to the real-time position, attitude data and speed data, the deviation between the actual flight path of the UAV and the planned path is obtained as the offset data, including position deviation, attitude deviation and speed deviation. The state equation and observation equation of the UAV are established based on the offset data through the Kalman filtering algorithm, and the state of the UAV is continuously estimated. Based on the state estimation and the offset of the planned path, the path compensation strategy is reversely applied to adjust the flight direction and speed of the UAV, perform path correction, and reduce the offset of the state estimation and the planned path.

[0125] In this embodiment, a method for obtaining a UAV navigation model by learning a path compensation strategy and a path planning model and optimizing the UAV navigation model by a sine-cosine optimization algorithm includes:

[0126] The UAV navigation model is trained using a reinforcement learning algorithm. The state space is the real-time position, attitude data, risk score, path time, speed data, and environmental data of the UAV. The action space is to adjust the flight direction and speed of the UAV. The reward function is:

[0127]

[0128] Among them, k P , kT , k W and k θ are the risk score reward coefficient, path time reward coefficient, energy consumption reward coefficient and path stability reward coefficient, respectively. expected and T actrual The path time of the planned path of the path planning model and the actual path time of the reinforcement learning algorithm are respectively expected and W actrual are the energy consumption of the planned path and the actual energy consumption of the reinforcement learning algorithm, Δθ is the deviation rate between the path direction and the path planning direction of the reinforcement learning algorithm, and θ stabel is the deviation rate threshold of the path direction;

[0129] The inverse of the objective function value of the drone navigation model is used as the fitness value. The position of each search individual represents a set of hyperparameter values. The search individual position update formula is:

[0130]

[0131] in, and are the positions of the ith search individual at the t+1th and tth iterations, respectively, and X * is the global optimal position, is the historical optimal position searched by the i-th search individual, ω t is the nonlinear weight factor of the tth iteration, which is used to adjust the dependence of the search individual on its current position information during the population iteration process. is the control parameter of the amplitude of the sine function, t is the current iteration number, t max is the maximum number of iterations, r1 is the control parameter, r1=a(1-t / t max ), a is a constant greater than 1, is the adjustment parameter of the frequency of the sine function and the cosine function, r2∈(0,2π) is a random weight that obeys a uniform distribution, r3∈(0,1) is a random number that obeys a uniform distribution and is used to select the individual position update strategy,

[0132] When the average fitness value of the historical optimal position searched by all search individuals has not changed for four consecutive iterations, the search is considered to be stagnant. At this time, the Levy flight random walk strategy is used to update the position of the search individual. The random walk step length formula of Levy flight is:

[0133]

[0134] Where s is the random walk step length, u and v are parameters that obey the normal distribution, Γ(·) is the gamma function, β is the exponential parameter that controls the shape of the Levy distribution, β∈(0,2), β=1.5, σ u is the standard deviation of u, σ v is the standard deviation of v;

[0135] The random walk strategy based on Levy flight is:

[0136]

[0137] Where S randn is a random quantity that obeys the standard normal distribution, α∈[-1,1] is the proportional factor, α=0.9,

[0138] When the maximum number of iterations is reached, the hyperparameter value corresponding to the global optimal position is output.

[0139] A UAV navigation and positioning system based on big data, characterized by comprising:

[0140] The acquisition module obtains the location data, sensor data and environmental data of the UAV, divides the flight area, and uses the GIS system to obtain the spatial information of the flight area;

[0141] A computing module, using a SLAM algorithm to construct a real-time map according to the position data and the sensor data and perform real-time positioning, divide the real-time map into sub-areas, use cluster analysis to obtain a risk score, obtain a path time between the sub-areas according to the environmental data and the risk score, establish a fitness function according to the risk score and the path time, establish a path planning model based on the fitness function, obtain offset data according to the sensor data, and learn a path compensation strategy and a path planning model to obtain a UAV navigation model;

[0142] A correction module, establishing a path compensation strategy and performing path correction according to the offset data;

[0143] The optimization module optimizes the UAV navigation model through a sine-cosine optimization algorithm.

[0144] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A UAV navigation and positioning method based on big data, characterized in that: The following steps are involved: Acquire the position data, sensor data and environmental data of the drone, and use the SLAM algorithm to build a real-time map and perform real-time positioning according to the position data and the sensor data; Dividing the real-time map into sub-areas and obtaining risk scores by cluster analysis; Obtaining the path time between the sub-areas according to the environmental data and the risk score, establishing a fitness function according to the risk score and the path time, and establishing a path planning model based on the fitness function; Obtaining offset data according to the sensor data, establishing a path compensation strategy according to the offset data and performing path correction; The path compensation strategy and the path planning model are learned to obtain a UAV navigation model, and the UAV navigation model is optimized by a sine-cosine optimization algorithm.

2. The method for unmanned aerial vehicle navigation and positioning based on big data according to claim 1, characterized in that: Methods for obtaining the drone's location data, sensor data, and environmental data include: The drone position obtained by GPS, GLONASS and Galileo is fused through the pseudo-range algorithm to obtain the position data. The sensors include data from three-dimensional cameras, barometers, gyroscopes, accelerometers and inertial measurement units. The sensor data includes attitude data, speed data, acceleration data, air pressure data, and image data. The environmental data includes meteorological data and electromagnetic interference data. The meteorological data includes wind direction, wind speed and air pressure. The electromagnetic interference data includes the intensity and frequency of electromagnetic interference.

3. The method for unmanned aerial vehicle navigation and positioning based on big data according to claim 1, characterized in that: The method of using a SLAM algorithm to construct a real-time map and perform real-time positioning according to the position data and the sensor data comprises: The GIS system is used to obtain the GIS map of the area around the drone based on the location data. The SLAM algorithm is used to build an obstacle model based on the real-time monitoring data of the three-dimensional camera. A real-time map is built based on the GIS map to obtain the real-time position of the drone in the real-time map.

4. The method for unmanned aerial vehicle navigation and positioning based on big data according to claim 1, characterized in that: The method of dividing the real-time map into sub-areas and obtaining risk scores by cluster analysis includes: The real-time map distinguishes the flyable area and the non-flyable area, and divides the flyable area into cubic sub-areas. Data points are established based on the surrounding obstacle density, wind speed, electromagnetic interference data and air pressure of the sub-area, and the SOM algorithm is used to cluster the data points to form risk areas: Randomly select k data points as the initial weight vectors of the competitive layer neurons. One competitive layer neuron represents a cluster center. Calculate the similarity between the data point and the weight vector of each competitive layer neuron, and assign the data point to the cluster of neurons with the highest similarity. The calculation formula for similarity is: d(h,e) is the similarity between h and e, h and e are two r-dimensional vectors, ||h(w κ )|| and||e(w κ )|| are the weighted vector lengths of vector h and vector e, respectively. κ represents the component of vector h in the κth dimension, w κ is the weight coefficient of the κth dimension, and a neighborhood with the neuron as the center and σ(0) as the initial neighborhood radius is drawn. Dynamic topological neighborhood function in the competition layer: Among them, h gi (t) is the dynamic topological neighborhood function of neuron i and neuron g at the tth iteration, d ig (t) is the distance between neuron i and neuron g at the t-th iteration, is the similarity between neurons i and g, σ(t) is the neighborhood radius at the t-th iteration, t max is the maximum number of iterations, δ(t) is the influencing factor that changes with time, ξ is the initial value of the impact factor, ζ is the adjustment coefficient of the impact factor, Update of the neuron's weight vector: w ij (t+1)=w ij (t)+η(t)·h gi (t)·(x j -w ij (t))-γ·w ij (t) Among them, γ is the strength of the regularization term, w ij (t) and w ij (t+1) is the j-th dimension component of the weight vector of the ith neuron at the t-th and t+1-th iterations, η(t) is the learning rate of the t-th iteration, x j is the j-th dimension component of the data point, After reaching the maximum number of iterations, the clustering result is output, and the risk score of each risk area is the average value of the weighted vector length of the data points therein.

5. The method for unmanned aerial vehicle navigation and positioning based on big data according to claim 1, characterized in that: The method for obtaining the path time between the sub-areas according to the environmental data and the risk score comprises: Path time function: Among them, T cb is the path time between the center of sub-area c and the center of sub-area b, w P is the impact weight of risk score, P c and P b are the risk scores of the risk areas corresponding to sub-area c and sub-area b, respectively, S c and S b are the parts of the path length between the centers of sub-areas c and b in sub-areas c and b, V is the speed of the drone, and V f (c) and V f (b) are the wind speeds in sub-area c and sub-area b, and are the clockwise angles between the wind direction in sub-area c and sub-area b and the forward direction of the UAV, ω f is the weight of the impact of wind speed on the UAV speed.

6. The method for unmanned aerial vehicle navigation and positioning based on big data according to claim 1, characterized in that: The method of establishing a fitness function according to the risk score and the path time comprises: Fitness function: Where f(L) is the fitness function of path L, c∈L means that sub-region c is in the sub-region set of path L, sub-region c is adjacent to sub-region b, cb∈L means that sub-region c to b is a sub-path in path L, W cb is the energy consumption of subpath cb, w T and w W are the weights of path time and energy consumption, The formula for calculating energy consumption is: Among them, m and m l are the weight of the drone and the payload, φ is the acceleration of gravity, ΔH is the height change, and are the air pressures in sub-areas c and b, respectively, is the gas constant, and are the temperatures of sub-areas c and b respectively, and A is the windward area of ​​the UAV.

7. The method for unmanned aerial vehicle navigation and positioning based on big data according to claim 1, characterized in that: The method for establishing a path planning model based on the fitness function includes: Each ant moves on the directed graph composed of nodes consisting of sub-regions and leaves pheromones, initializing the parameters of the ant colony algorithm. The probability of ant transfer between nodes is: Among them is The probability that the kth ant moves from node u to node v at time s is, and are the weights of pheromone and heuristic function respectively, is the index of the node in N(u), N(u) is the set of neighbor nodes of node u, From node u to node The heuristic function is From node u to node The distance between the center of sub-region u and the sub-region The path length between centers, For sub-region The risk score of The pheromone update formula is: Among them, μ uv (s) and μ uv (s+1) represents the amount of pheromone on path uv at time s and time s+1, α is the pheromone volatility coefficient, Z uv is the fitness of path uv, is the fitness function value of path uv, When the pheromone update amount is lower than the update threshold, the best path corresponding to the strongest pheromone signal is output as the planned path. Re-plan the route when real-time map and environmental data change.

8. The method for unmanned aerial vehicle navigation and positioning based on big data according to claim 1, characterized in that: The method of obtaining offset data according to the sensor data, establishing a path compensation strategy according to the offset data and performing path correction includes: According to the real-time position, attitude data and speed data, the deviation between the actual flight path of the UAV and the planned path is obtained as the offset data, including position deviation, attitude deviation and speed deviation. The state equation and observation equation of the UAV are established based on the offset data through the Kalman filtering algorithm, and the state of the UAV is continuously estimated. Based on the state estimation and the offset of the planned path, the path compensation strategy is reversely applied to adjust the flight direction and speed of the UAV, perform path correction, and reduce the offset of the state estimation and the planned path.

9. The method for unmanned aerial vehicle navigation and positioning based on big data according to claim 1, characterized in that: The method of obtaining a UAV navigation model by learning a path compensation strategy and a path planning model and optimizing the UAV navigation model by a sine-cosine optimization algorithm comprises: The UAV navigation model is trained using a reinforcement learning algorithm. The state space is the real-time position, attitude data, risk score, path time, speed data, and environmental data of the UAV. The action space is to adjust the flight direction and speed of the UAV. The reward function is: Among them, k P , k T , k W and k θ are the risk score reward coefficient, path time reward coefficient, energy consumption reward coefficient and path stability reward coefficient, respectively. expected and T actrual The path time of the planned path of the path planning model and the actual path time of the reinforcement learning algorithm are respectively expected and W actrual are the energy consumption of the planned path and the actual energy consumption of the reinforcement learning algorithm, Δθ is the deviation rate between the path direction and the path planning direction of the reinforcement learning algorithm, and θ stabel is the deviation rate threshold of the path direction; The inverse of the objective function value of the drone navigation model is used as the fitness value. The position of each search individual represents a set of hyperparameter values. The search individual position update formula is: in, and are the positions of the ith search individual at the t+1th and tth iterations, respectively, and X * is the global optimal position, is the historical optimal position searched by the i-th search individual, ω t is the nonlinear weight factor for the tth iteration, is the control parameter of the amplitude of the sine function, t is the current iteration number, t max is the maximum number of iterations, r1 is the control parameter, r1=a(1-t / t max ), a is a constant greater than 1, is the adjustment parameter of the frequency of the sine function and the cosine function, r2∈(0,2π) is a random weight that obeys a uniform distribution, r3∈(0,1) is a random number that obeys a uniform distribution and is used to select the individual position update strategy, When the average fitness value of the historical optimal position searched by all search individuals has not changed for four consecutive iterations, the search is considered to be stagnant. At this time, the Levy flight random walk strategy is used to update the position of the search individual. The random walk step length formula of Levy flight is: Where s is the random walk step length, u and v are parameters that obey the normal distribution, Γ(·) is the gamma function, β is the exponential parameter that controls the shape of the Levy distribution, β∈(0,2), β=1.5, σ u is the standard deviation of u, σ v is the standard deviation of v; The random walk strategy based on Levy flight is: Where S randn is a random quantity that obeys the standard normal distribution, α∈[-1,1] is the proportional factor, and α=0.9 is taken. When the maximum number of iterations is reached, the hyperparameter value corresponding to the global optimal position is output.

10. A UAV navigation and positioning system based on big data, used to execute the UAV navigation and positioning method based on big data according to any one of claims 1 to 9, characterized in that: include: The acquisition module obtains the location data, sensor data and environmental data of the UAV, divides the flight area, and uses the GIS system to obtain the spatial information of the flight area; A computing module, using a SLAM algorithm to construct a real-time map according to the position data and the sensor data and perform real-time positioning, divide the real-time map into sub-areas, use cluster analysis to obtain a risk score, obtain a path time between the sub-areas according to the environmental data and the risk score, establish a fitness function according to the risk score and the path time, establish a path planning model based on the fitness function, obtain offset data according to the sensor data, and learn a path compensation strategy and a path planning model to obtain a UAV navigation model; A correction module, establishing a path compensation strategy and performing path correction according to the offset data; The optimization module optimizes the UAV navigation model through a sine-cosine optimization algorithm.

Images