Unmanned aerial vehicle patrol route optimization method fusing weather conditions and power consumption prediction

Abstract

The invention discloses an unmanned aerial vehicle patrol route optimization method fusing weather conditions and power consumption prediction, and the method comprises the steps: (1), according to the historical flight data of a patrol unmanned aerial vehicle, combining a random forest and a linear regression method; establishing an unmanned aerial vehicle power consumption prediction model considering the coupling influence of environmental factors and task factors, and drawing a time-power-power curved surface; (2) inputting a patrol object digital map, a required patrol point set, a selectable patrol point set, unmanned aerial vehicle performance data, a route starting point and a route ending point, and obtaining an initial flight route without considering real-time weather conditions; and (3) optimizing the initial flight route in combination with the real-time weather condition and the time-power-electric quantity curve, and when the predicted consumed electric quantity is greater than the acceptable consumed electric quantity, sequentially giving up the selectable patrol points according to the importance sequence and re-drawing the route until the predicted consumed electric quantity is less than or equal to the acceptable consumed electric quantity. The method has a remarkable optimization effect on precise patrol task execution of the unmanned aerial vehicle.

Description

Technical Field

[0001] The present application relates to the field of drone control technology, and in particular to a method for dynamically optimizing drone inspection routes based on weather data fusion and power consumption prediction. Background Art

[0002] Drone path planning refers to planning the flight path of a drone from the starting point to the end point, or from one mission point to another. The path must meet constraints such as safe flight, power consumption, and mission completion. Usually, the "optimal" in the optimal path means minimizing the flight distance of the drone, thereby shortening the flight time and reducing consumption. In addition, when planning the path, it is also necessary to comprehensively consider the drone's maneuverability (such as dynamic responsiveness to obstacle avoidance) while meeting feasibility constraints such as flight time, flight altitude, power consumption, turning rate, and climb angle.

[0003] At present, the particle swarm optimization method is mostly used to search for the optimal path for the path planning of UAVs. It can converge to the global optimal solution with a high probability, has a faster calculation speed and better global search ability, but for functions with multiple local extreme points, it is easy to fall into the local optimum, resulting in premature convergence; there are also hybrid optimization algorithms formed by combining the advantages of other algorithms for path planning, but most of the proposed hybrid algorithms are based on the population evolution of one algorithm, and then apply another algorithm or a single operator, without considering improving the algorithm search ability and diversity; therefore, the algorithms used in the above methods have not better improved the accuracy and efficiency of UAV path planning. Therefore, how to safely and efficiently plan the path of UAVs in complex environments has become an urgent problem to be solved. Summary of the invention

[0004] Purpose of the invention: In response to the shortcomings of the prior art, this patent proposes a method for optimizing drone inspection routes that integrates weather conditions and power consumption forecasts. Through the analysis and prediction of weather data and power consumption, the safest inspection route is formed based on the optimal energy consumption and the maximum inspection range. By applying this patent, under limited power conditions, it is possible to more accurately predict power consumption based on weather conditions, maximize the inspection range, minimize inspection time, optimize resource allocation, prevent drones from being affected by weather and power consumption, reduce the risk of crashes, save electricity costs, and improve inspection efficiency.

[0005] Technical solution: A method for optimizing the inspection route of a drone by integrating weather conditions and power consumption prediction, comprising the following steps: S1. Based on the historical flight data of patrol drones, combined with random forest and linear regression methods, a drone power consumption prediction model considering the coupling influence of environmental factors and task factors is established, and the drone power consumption prediction model is used to draw the time-power-power surface; S2. Input the digital map of the inspection object, the set of mandatory inspection points, the set of optional inspection points, the performance data of the UAV, the starting point and the end point of the route, and use the EGO-Planner algorithm to obtain the initial flight route without considering the real-time weather conditions; S3. The initial flight route is optimized based on the real-time weather conditions and the time-power-electricity curve. When the predicted power consumption is greater than the acceptable power consumption, the optional inspection points are abandoned in order of importance and the route is re-planned until the predicted power consumption is less than or equal to the acceptable power consumption.

[0006] Furthermore, the step 1 comprises the following steps: S11. Obtain the flight mission data set from the historical flight database, including the following parameters: UAV blade rotation speed; east-west speed, with negative in the west and positive in the east; north-south speed, with negative in the south and positive in the west; vertical ascent and descent speed, with negative in descent and positive in ascent; and aircraft rotation angular velocity.

[0007] S12. Obtain an environmental factor data set from a historical flight database, including the following parameters: weather conditions; air humidity; Celsius temperature; air pressure; wind speed in the east-west direction, with negative values ​​in the west and positive values ​​in the east; wind speed in the north-south direction, with negative values ​​in the south and positive values ​​in the west.

[0008] S13. Combine the data in S11 and S12 with the corresponding power factors to form a training set and a test set, and randomly divide them in a ratio of 8:2. Train the random forest and linear regression models, and run the test set to obtain their respective mean square errors. , take the smaller one as the "universal power factor model".

[0009] S14. In addition, a new "special power factor model" is retrained for the following eight situations: Strong wind: The sum of wind speed vectors is greater than or equal to , The value is 10 m / s or more; high speed: the sum of the UAV velocity vector is greater than or equal to , The value is 10 m / s; Fast forward: the angular velocity of the drone is greater than or equal to , The value is 30rad / s; High humidity: humidity is higher than or equal to , The value is 70% or higher; High temperature: the temperature is greater than or equal to , The value is 30℃; Low temperature: the temperature is less than or equal to , The value is 0℃; High pressure: the air pressure is higher than or equal to , The value is 105kPa; low pressure: the air pressure is lower than or equal to , The value of is 70kPa; as above, both random forest and linear regression models are trained, and the mean square error of eight cases is obtained by running the test set , in each case taking the mean square error The smaller one is used as the "special power factor model" for that situation.

[0010] S15. When using the power factor model, determine whether it is a special case based on the real-time weather data input externally. If it is not a special case, call it directly; if it is a special case, call the "special power factor model" or the "general power factor model", and select the best one based on the specific mean square error results of the two power factor models to obtain the predicted power factor.

[0011] S16, using the three parameters of time, power factor and battery power to draw a time-power-power surface through a formula; , Assume the battery power percentage is , the maximum capacity marked on the battery nameplate is , the maximum output power is , the power factor is , time is .

[0012] S17, real machine verification time-power-electricity curve, measure the actual voltage and current, calculate the actual power factor , and at the same time, based on the actual flight mission data, the power factor model in S15 is used to predict , the mean square error of the two Less than Considered as a power factor model and time-power-electricity surface available ( e ∈ [5,50] ), otherwise return to S13 and S14 to increase the training scale and training steps of the power factor model.

[0013] Furthermore, the “power factor” appearing in the above step S13 and thereafter is specifically explained as follows: The basic power system of a typical electric propeller drone consists of batteries, ESCs, motors, and propellers, of which the motors and propellers form the output end, and the batteries and ESCs form the input end. It is extremely difficult to accurately and in real time calculate the output power of a drone. We can start from the input end and calculate the power consumption of the drone according to the electrical formula P=U*I. U is the battery output voltage, and I is the battery output current. By installing sensors or a BMS system, the U and I of the drone can be monitored in real time, and the actual output power P of the battery can be calculated: , To facilitate the drawing of time-power-power curves and the matching of different models of drones and batteries, a power factor can be set , range [0,100], and is positively correlated with the UAV battery output power P.

[0014] When P=0, =0;

[0015] when , =100;

[0016] When 0<P<P max , = , Output power for drone batteries, is the measured output voltage of the drone battery, is the measured output current of the drone battery, is the power factor, is the maximum output power of the drone battery, is the maximum output voltage of the drone battery, It is the maximum output current of the drone battery.

[0017] Furthermore, the step 2 comprises the following steps: S21. Input the digital map of the inspection object, mark all required inspection points, mark all optional inspection points and the importance weight of each, and input the starting point and end point of the route.

[0018] S22. Input drone data, including available battery power, maximum turning radius, maximum flight speed, etc.

[0019] S23. Use the EGO-Planner algorithm to establish the optimal flight trajectory without considering the real-time weather impact, hereinafter referred to as the algorithm route. At the same time, the UAV performance constraints are imposed on the route, requiring the planned UAV mission to be operational and within the range of various UAV performance parameters.

[0020] S24, marking the algorithm route, dividing the route into several segments, where the segment division is based on the track between two adjacent mandatory inspection points; each segment is also divided into several minimum segments, where the minimum segment division is based on the segment between two adjacent optional inspection points.

[0021] S25. Record the drone mission information for each minimum flight segment, including the drone's estimated flight time, flight action, and flight speed.

[0022] S26. Record the drone mission information for each inspection point, including the time when the drone is expected to arrive at the inspection point, the remaining power information, and the heading and speed information.

[0023] Furthermore, the step 3 comprises the following steps: S31. At the starting point and when passing each necessary inspection point, the total estimated flight time of the next flight segment is calculated based on the estimated flight time of each minimum segment in S25.

[0024] S32. According to the flight actions of the next flight segment, combined with the international standard atmosphere, no wind, clear sky, and 50% humidity, the ideal power factor is calculated through the power factor model, and the ideal power is calculated through the following formula. The expected power consumption percentage is calculated based on this ideal power and the total expected flight time in S31.

[0025] = , , in, is the ideal power factor, is the ideal output power, is the maximum output power of the drone battery, is the total estimated flight time, is the total battery capacity of the drone, The estimated power consumption percentage.

[0026] S33. Calculate using a power factor model based on the flight actions of the next flight segment and the actual weather data detected by the drone.

[0027] S34. According to the time-power-electricity surface in S16, query the time-electricity curve corresponding to the actual power factor in the surface, calculate the remaining electric power starting from the current electric power, the percentage of the expected electric power consumption in S32, and the acceptable discharge time length in the middle.

[0028] S35. Compare the acceptable discharge time length in S34 with the total expected flight time in S31. If the acceptable discharge time length is less than the total expected flight time, it is considered that the flight segment has insufficient power. Search for the selected optional waypoints in the flight segment, deselect some optional waypoints according to the order of importance of the inspection points, and rebuild a new flight segment so that the consumption of the new flight segment matches the expectation.

[0029] Beneficial effects: In the present invention, by using a random forest model and a linear regression model, the existing data are integrated and calculated to obtain a power factor model required for different flight missions under different environmental factors. The relationship between the three data of time, power factor, and power is presented through the established time-power-power surface. Input the digital map of the inspection object, the set of required inspection points, the set of optional inspection points and their respective importance weights, the drone data, the starting point and the end point of the route, and use the EGO-Planner algorithm to calculate the algorithm inspection route with the relatively shortest time, the least power consumption, the highest inspection coverage, the minimum consumable loss and the highest safety margin. Compare the predicted power consumption with the actual power consumption during flight. When the actual consumption is too much, the unnecessary inspection points are abandoned one by one in the flight segment to ensure that the drone inspection plan completes the inspection task with high quality and low risk. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flow chart of the method according to the present invention. DETAILED DESCRIPTION

[0031] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0032] Figure 1 The flowchart of the method for optimizing the inspection route of a UAV by integrating weather conditions and power consumption prediction is shown in the present invention. The present invention dynamically optimizes the initial flight route based on weather data fusion and power consumption prediction, and mainly includes the following steps: S1. Based on the historical flight data of patrol drones, combined with random forest and linear regression methods, a drone power consumption prediction model considering the coupling influence of environmental factors and task factors is established, and the drone power consumption prediction model is used to draw the time-power-power surface; The S1 comprises the following steps: S11. Obtain the flight mission data set from the historical flight database, including the following parameters: UAV blade speed; east-west speed, with west being negative and east being positive; north-south speed, with south being negative and west being positive; vertical ascent and descent speed, with descent being negative and ascent being positive; and aircraft rotation angular velocity.

[0033] S12. Obtain an environmental factor data set from a historical flight database, including the following parameters: weather conditions; air humidity; Celsius temperature; air pressure; wind speed in the east-west direction, with negative values ​​in the west and positive values ​​in the east; wind speed in the north-south direction, with negative values ​​in the south and positive values ​​in the west.

[0034] S13. Combine the data in S11 and S12 with the corresponding power factors to form a training set and a test set, and randomly divide them in a ratio of 8:2. Train the random forest and linear regression models, and run the test set to obtain their respective mean square errors. , take the smaller one as the "universal power factor model".

[0035] S14. In addition, a new "special power factor model" is retrained for the following eight situations: Strong wind: The sum of wind speed vectors is greater than or equal to , The value is 10 m / s or more; high speed: the sum of the UAV velocity vector is greater than or equal to , The value is 10 m / s; Fast forward: the angular velocity of the drone is greater than or equal to , The value is 30rad / s; High humidity: humidity is higher than or equal to , The value is 70% or higher; High temperature: the temperature is greater than or equal to , The value is 30℃; Low temperature: the temperature is less than or equal to , The value is 0℃; High pressure: the air pressure is higher than or equal to , The value is 105kPa; low pressure: the air pressure is lower than or equal to , The value of is 70kPa; as above, both random forest and linear regression models are trained, and the mean square error of eight cases is obtained by running the test set , in each case taking the mean square error The smaller one is used as the "special power factor model" for that situation.

[0036] S15. When using the power factor model, determine whether it is a special case based on the real-time weather data input externally. If it is not a special case, call it directly; if it is a special case, call the "special power factor model" or the "general power factor model", and select the best one based on the specific mean square error results of the two power factor models to obtain the predicted power factor.

[0037] S16, using the three parameters of time, power factor and battery power to draw a time-power-power surface through a formula; , Assume the battery power percentage is , the maximum capacity marked on the battery nameplate is , the maximum output power is , the power factor is , time is .

[0038] S17, real machine verification time-power-electricity curve, measure the actual voltage and current, calculate the actual power factor , and at the same time, based on the actual flight mission data, the power factor model in S15 is used to predict , the mean square error of the two Less than Considered as a power factor model and time-power-electricity surface available ( e ∈ [5,50] ), otherwise return to S13 and S14 to increase the training scale and training steps of the power factor model.

[0039] Furthermore, the “power factor” appearing in the above step S13 and thereafter is specifically explained as follows:

[0040] The basic power system of a typical electric propeller drone consists of batteries, ESCs, motors, and propellers. The motors and propellers form the output end, while the batteries and ESCs form the input end. It is extremely difficult to accurately and in real time calculate the output power of a drone. We can start from the input end and calculate the output power of the drone according to the electrical formula. , calculate the power consumption of the drone. U is the battery output voltage, and I is the battery output current. By installing sensors or BMS systems, the drone U and I can be monitored in real time, and the actual battery output power P can be calculated: , To facilitate the drawing of time-power-power curves and the matching of different models of drones and batteries, a power factor can be set , range [0,100], and is positively correlated with the UAV battery output power P.

[0041] When P=0, =0; when , =100; When 0<P<P max , = , Output power for drone batteries, is the measured output voltage of the drone battery, is the measured output current of the drone battery, is the power factor, is the maximum output power of the drone battery, is the maximum output voltage of the drone battery, It is the maximum output current of the drone battery.

[0042] S2. Input the digital map of the inspection object, the set of mandatory inspection points, the set of optional inspection points, the performance data of the UAV, the starting point and the end point of the route, and use the EGO-Planner algorithm to obtain the initial flight route without considering the real-time weather conditions;

[0043] The S2 comprises the following steps: S21. Input the digital map of the inspection object, mark all required inspection points, mark all optional inspection points and the importance weight of each, and input the starting point and end point of the route.

[0044] S22. Input drone data, including available battery power, maximum turning radius, maximum flight speed, etc.

[0045] S23. Use the EGO-Planner algorithm to establish the optimal flight trajectory without considering the real-time weather impact, hereinafter referred to as the algorithm route. At the same time, the UAV performance constraints are imposed on the route, requiring the planned UAV mission to be operational and within the range of various UAV performance parameters.

[0046] S24, marking the algorithm route, dividing the route into several segments, where the segment division is based on the track between two adjacent mandatory inspection points; each segment is also divided into several minimum segments, where the minimum segment division is based on the segment between two adjacent optional inspection points.

[0047] S25. Record the drone mission information for each minimum flight segment, including the drone's estimated flight time, flight action, and flight speed.

[0048] S26. Record the drone mission information for each inspection point, including the time when the drone is expected to arrive at the inspection point, the remaining power information, and the heading and speed information.

[0049] S3. The initial flight route is optimized based on the real-time weather conditions and the time-power-electricity curve. When the predicted power consumption is greater than the acceptable power consumption, the optional inspection points are abandoned in order of importance and the route is re-planned until the predicted power consumption is less than or equal to the acceptable power consumption.

[0050] The S3 comprises the following steps: S31. At the starting point and when passing each necessary inspection point, the total estimated flight time of the next flight segment is calculated based on the estimated flight time of each minimum segment in S25.

[0051] S32. According to the flight actions of the next flight segment, combined with the international standard atmosphere, no wind, clear sky, and 50% humidity, the ideal power factor is calculated through the power factor model, and the ideal power is calculated through the following formula. The percentage of expected power consumption is calculated based on this ideal power and the total expected flight time in S31: = , , in, is the ideal power factor, is the ideal output power, is the maximum output power of the drone battery, is the total estimated flight time, is the total battery capacity of the drone, The estimated power consumption percentage.

[0052] S33. Calculate using a power factor model based on the flight actions of the next flight segment and the actual weather data detected by the drone.

[0053] S34. According to the time-power-electricity surface in S16, query the time-electricity curve corresponding to the actual power factor in the surface, calculate the remaining electric power starting from the current electric power, the percentage of the expected electric power consumption in S32, and the acceptable discharge time length in the middle.

[0054] S35. Compare the acceptable discharge time length in S34 with the total expected flight time in S31. If the acceptable discharge time length is less than the total expected flight time, it is considered that the flight segment has insufficient power. Search for the selected optional waypoints in the flight segment, deselect some optional waypoints according to the order of importance of the inspection points, and rebuild a new flight segment so that the consumption of the new flight segment matches the expectation.

[0055] The technical solution of the present invention is further illustrated by using specific embodiments below.

[0056] The present invention provides a method for optimizing the inspection route of a UAV by integrating weather conditions and power consumption prediction, based on the following assumptions:

[0057] (1) The UAV can fly at an altitude of [h1,h2], with a maximum flight speed of v, a maximum acceleration, and a maximum angular acceleration. , the battery energy of each drone is E;

[0058] (2) The minimum elevation between the drone and the horizontal ground is h, and the three-dimensional coordinates of the drone’s position are known at any time;

[0059] (3) The effective inspection range of the drone is the area where the elevation angle between the plane of the area and the drone is greater than θ and is not blocked by obstacles;

[0060] (4) It is assumed that all drones fly autonomously along the optimized path without manual control and land at the destination after completing the mission;

[0061] (5) Assume that the impact of the drone’s take-off and landing process on its flight time is ignored.

[0062] (6) It is assumed that the UAV flies at a constant speed between each sub-route point until it is required to accelerate or decelerate, and then makes a constant speed turn toward different sub-route points.

[0063] A method for optimizing a drone inspection route by integrating weather conditions and power consumption prediction in this embodiment includes the following steps:

[0064] Step S1: Obtain flight action data set, environmental influencing factors and corresponding actual power factor consumption from the historical flight database. The specific parameters are shown in the following table:

[0065] Table 1. Flight action dataset: Parameter name Numeric Types Numerical range Numerical Unit Remark Rotational speed Int [0,6000] Rpm Take the dji-E5000 system as an example X speed Float [-30,30] m / s The speed in the east-west direction is negative in the west and positive in the east. Y speed Float [-30,30] m / s The speed in the north-south direction is negative in the south and positive in the west. Z speed Float [-20,20] m / s Vertical lift speed, negative for lowering and positive for raising Angular velocity Float [-500,500] Degree / s Aircraft rotation angular velocity

[0066] Table 2. Environmental impact factor dataset: Parameter name Numeric Types Numerical range Numerical Unit Remark Weather String / / Please refer to the local weather station for the latest information Humidity Float [0,100] % Air humidity Temperature Float [-10,50] ℃ Celsius Pressure Float [0,150] kPa Air pressure X airspeed Float [-50,50] m / s Wind speed in the east-west direction is negative in the west and positive in the east Y airspeed Float [-50,50] m / s Wind speed in north-south direction, negative in south and positive in west

[0067] Combine the historical data in the database with the corresponding power factor to form a training set and a test set, which are randomly divided in a ratio of 8:2. Train the random forest and linear regression models, and run the test set to obtain their respective mean square errors. , and take the smaller one as the general model.

[0068] In addition, special case models are trained separately for strong winds, high speed, fast rotation, high humidity, high temperature, low temperature, high pressure, and low pressure. Similarly, random forest and linear regression models are trained, and the mean square error of each is obtained by running the test set. , take the smaller one as the prediction model for this situation. The specific situation and corresponding thresholds are shown in the table below.

[0069] Table 3. Special power factor model corresponding threshold table:

[0070] Integrate the above information and fill in Table 4 for subsequent inquiries:

[0071] Table 4. Special power factor model example data: Model Name General Model Gale Model High-speed model Fast Forward Model High humidity model High temperature model Low temperature model High pressure model Low pressure model Mean Square Error (MSE) 10.0 4.5 7.4 2.6 8.5 12.1 10.6 4.1 2.1

[0072] Step S2: Draw the time-power-electricity surface according to the formula, and verify the accuracy of the time-power-electricity surface through actual flight.

[0073] Step S3: Input the digital map of the inspection object, the set of required inspection points, the set of optional inspection points, the performance data of the UAV, the starting point and the end point of the route, and use the EGO-Planner algorithm to obtain the initial flight route without considering the real-time weather conditions;

[0074] Step S4: Optimize the initial flight route in combination with the real-time weather conditions and the time-power-electricity curve. When the predicted power consumption is greater than the acceptable power consumption, abandon the optional inspection points in order of importance and re-set the route until the predicted power consumption is less than or equal to the acceptable power consumption.

[0075] The above content is only a further detailed description of the present invention through a specific embodiment, and the present invention cannot be considered to be limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A method for optimizing the inspection route of a UAV by integrating weather conditions and power consumption prediction, characterized in that: The method comprises the following steps: S1. Based on the historical flight data of patrol drones, combined with random forest and linear regression methods, a drone power consumption prediction model considering the coupling influence of environmental factors and task factors is established, and the drone power consumption prediction model is used to draw the time-power-power surface; S2. Input the digital map of the inspection object, the set of mandatory inspection points, the set of optional inspection points, the performance data of the UAV, the starting point and the end point of the route, and use the EGO-Planner algorithm to obtain the initial flight route without considering the real-time weather conditions; S3. Optimize the initial flight route obtained in step S2 by combining the real-time weather conditions and the time-power-electricity curve obtained in step S1. When the predicted power consumption is greater than the acceptable power consumption, abandon the optional inspection points in order of importance and re-set the route until the predicted power consumption is less than or equal to the acceptable power consumption.

2. The method for optimizing the inspection route of a UAV by integrating weather conditions and power consumption prediction according to claim 1, characterized in that: The UAV power consumption prediction model described in step S1 is to train the UAV power factor model to achieve the purpose of predicting power consumption, which specifically includes the following sub-steps: S111. Combine the historical flight data and environmental factor data of the UAV in the database with the corresponding power factor to form a training set and a test set, randomly divide them in a ratio of 8:2, train the random forest and linear regression models, and run the test set to obtain their respective mean square errors. , take the smaller one as the "universal power factor model"; S112. In addition, a new "special power factor model" is retrained for the following eight situations: Strong wind: The sum of the wind speed vector is greater than or equal to , The value is 10 m / s or more; high speed: the sum of the UAV velocity vector is greater than or equal to , The value is 10 m / s; Fast forward: The angular velocity of the drone is greater than or equal to , The value is 30rad / s; High humidity: humidity is higher than or equal to , The value is 70% or more; High temperature: temperature greater than or equal to , The value is 30℃; Low temperature: the temperature is less than or equal to , The value is 0℃; High pressure: the air pressure is higher than or equal to , The value of is 105kPa; Low pressure: air pressure less than or equal to , The value of is 70kPa; as above, both random forest and linear regression models are trained, and the mean square error of eight cases is obtained by running the test set , in each case taking the mean square error The smaller one is used as the "special power factor model" for that situation; S113. When using the power factor model, determine whether it belongs to the eight special cases listed in step S112 according to the real-time weather data input externally. If not, directly call the "general power factor model"; if it belongs to the eight special cases listed in step S112, call the "general power factor model" and the "special power factor model" in this case, and calculate the specific mean square error of the two power factor models called, and the mean square error The smaller one is executed preferentially to obtain the predicted power factor.

3. The method for optimizing the inspection route of a UAV by integrating weather conditions and power consumption prediction according to claim 1, characterized in that: Drawing the time-power-electricity curve surface in step S1 includes the following sub-steps: S121. Install the battery output power measurement module to measure the actual output current and voltage of the drone and set the power factor , [0,100], is positively correlated with the drone battery output power P, due to ,therefore, When P=0, =0; when , =100; When 0<P<P max , = , in, Output power for drone batteries, The battery output power measurement module measures the drone battery output voltage. The battery output power measurement module measures the drone battery output current. is the power factor, is the maximum output power of the drone battery, is the maximum output voltage of the drone battery, The maximum output current of the drone battery; S122, set the real-time percentage of the drone battery power to , the maximum capacity marked on the battery nameplate is , the maximum output power is , the power factor is , time is , the time , Power Factor and the real-time battery power of the drone, and a three-dimensional surface graph is drawn using the following formula; ， S123, actual machine verification step S122 draws the time-power-electricity curve: measure the actual drone voltage and current, and calculate the actual power factor At the same time, according to the actual flight mission data, the UAV power consumption prediction model in step S1 is used to predict , the mean square error of the two Less than Considered as a power factor model and time-power-electricity surface available ( ), otherwise return to step S1 to increase the power factor model training scale and training steps.

4. The method for optimizing the inspection route of a UAV by integrating weather conditions and power consumption prediction according to claim 1, characterized in that: In the step S2, a digital map of the inspection object, a set of required inspection points, a set of optional inspection points, UAV performance data, a route start point and a route end point are input, and an initial flight route without considering real-time weather conditions is obtained using the EGO-Planner algorithm, which is called an algorithm route. The algorithm route includes several flight segments, where a flight segment is a track between two adjacent required inspection points; each flight segment includes several minimum flight segments, where a minimum flight segment is a flight segment between two adjacent optional inspection points; the UAV mission information in each minimum flight segment includes the UAV's estimated flight time, flight action, and flight speed; each inspection point includes the time information when the UAV is expected to arrive at the inspection point, the remaining power information, and the heading and speed information; at the same time, UAV performance constraints are imposed on each flight segment, requiring that the planned UAV mission is actually operable and within the range of various performance parameters of the UAV.

5. The method for optimizing the inspection route of an unmanned aerial vehicle by integrating weather conditions and power consumption prediction according to claim 4 is characterized in that: In step 3, the initial flight route obtained in step S2 is optimized by combining the real-time weather conditions and the time-power-electricity curve obtained in step S1. The specific execution method is as follows: According to the algorithmic route obtained by the EGO-Planner algorithm, each segment of the algorithmic route is optimized: for the curves with the same power consumption range and the same power factor consumption on the surface, the maximum acceptable discharge time and the expected power consumption time are compared with the curves. If the maximum acceptable discharge time is less than the expected power consumption time, it is considered that the segment has insufficient power; if the segment has insufficient power, search for the selected optional waypoints in the segment, deselect some optional waypoints according to the order of importance of the inspection points, and rebuild the new segment so that the consumption of the new segment matches the expectation.

Images