An AI-based logistics inspection drone

Abstract

The present invention relates to the technical field of unmanned aerial vehicles, and particularly to a logistics inspection unmanned aerial vehicle based on artificial intelligence, comprising: an information collection module; an inspection optimization module for responding to inspection conditions to determine an inspection optimization method; an inspection change module for responding to setting conditions to determine an inspection change method; an inspection adjustment module for determining an inspection adjustment method according to the category of the inspection area; an inspection review module for collecting and storing inspection images and infrared thermal imaging images, and determining whether early warning is required for each inspection area according to the danger trigger coefficient of the inspection images and the thermal imaging coefficient of the infrared thermal imaging images; the present invention can improve the inspection efficiency of the unmanned aerial vehicle.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and particularly to a logistics inspection unmanned aerial vehicle based on artificial intelligence. Background Art

[0002] Under the background of the rapid development of the current logistics and warehousing management industry, due to a large number of production raw materials and product equipment in the logistics warehouse, the safety prevention measures in the warehouse area cannot be ignored. The use of logistics inspection unmanned aerial vehicles can reduce the workload of staff inspection operations and lower the inspection cost. However, during the inspection process of the unmanned aerial vehicle, there are disadvantages such as the complex logistics warehouse environment and the limited battery life of the unmanned aerial vehicle, resulting in poor inspection efficiency of the unmanned aerial vehicle. Therefore, how to improve the inspection efficiency according to the actual state of the unmanned aerial vehicle is a technical problem that needs to be solved urgently by those skilled in the art.

[0003] Chinese Patent Publication No. CN113313852A discloses an unmanned aerial vehicle inspection system, including: a plurality of position transmitters, an unmanned aerial vehicle, and a charging station. Each position transmitter is adapted to be installed on a meter or electrical equipment for transmitting a position signal and an equipment identifier, and one meter or electrical equipment corresponds to one equipment identifier; the unmanned aerial vehicle is used for receiving the position signal and flying to the meter or electrical equipment corresponding to the equipment identifier, and collecting the dial image of the meter or detecting the temperature of the electrical equipment; the charging station includes a charging platform for carrying the unmanned aerial vehicle, a charging sensor for detecting the landing signal of the unmanned aerial vehicle, and a charging module for charging the unmanned aerial vehicle. It can be seen that the above technical solution has the following problems: The inspection is only based on the position signal, the inspection method is single, and it is impossible to adaptively select different inspection methods according to the actual state of the unmanned aerial vehicle when there are many position signals, resulting in poor inspection efficiency. Summary of the Invention

[0004] Therefore, the present invention provides a logistics inspection unmanned aerial vehicle based on artificial intelligence to overcome the problems in the prior art that the inspection method is single, it is impossible to adaptively select different inspection methods according to the actual state of the unmanned aerial vehicle, and the inspection efficiency is poor.

[0005] To achieve the above object, the present invention provides a logistics inspection unmanned aerial vehicle based on artificial intelligence, including:

[0006] An information collection module for obtaining the basic information of the logistics warehouse and the flight information of the unmanned aerial vehicle;

[0007] An inspection optimization module connected to the information collection module for determining an inspection optimization method in response to an inspection condition, where the inspection optimization method is to determine an inspection change method according to a set condition, or to determine an inspection adjustment method according to the inspection area category;

[0008] The inspection change module is respectively connected to the information collection module and the inspection optimization module, and is used to respond to the set conditions to determine the inspection change method. The inspection change method is to determine the preferred selection method according to the change conditions, or to determine whether to conduct inspections on each uninspected area according to the information matching coefficient;

[0009] The preferred selection method is to determine the inspection priority coefficient of each uninspected area according to the information matching coefficient, or to determine the inspection priority coefficient of the pre-inspection area corresponding to each associated combination according to the flight impact coefficient;

[0010] The inspection adjustment module is respectively connected to the information collection module and the inspection optimization module, and is used to determine the inspection adjustment method according to the inspection area category. The inspection adjustment method is to determine the direction setting method of each collection point according to the adjustment conditions or to set the collection direction of each collection point to the vertical direction;

[0011] The inspection area category is determined according to the regional layout complexity and the cargo stacking coefficient, and the direction setting method is to determine the collection direction according to the associated direction group or the direction evaluation coefficient;

[0012] The inspection review module is respectively connected to the inspection change module and the inspection adjustment module, and is used to collect and store inspection images and infrared thermal imaging images, and determine whether early warnings are required for each inspection area according to the danger trigger coefficient of the inspection images and the thermal imaging coefficient of the infrared thermal imaging images.

[0013] Further, the inspection optimization module responds to the inspection conditions to determine the inspection optimization method, where,

[0014] The inspection conditions responded by the inspection optimization module are that the inspection area influence coefficient is less than the preset inspection area influence coefficient or the endurance reference value is less than the preset endurance reference value, and it is determined that the inspection optimization method is to determine the inspection change method according to the set conditions;

[0015] The inspection conditions responded by the inspection optimization module are that the inspection area influence coefficient is greater than or equal to the preset inspection area influence coefficient and the endurance reference value is greater than or equal to the preset endurance reference value, and it is determined that the inspection optimization method is to determine the inspection adjustment method according to the inspection area category.

[0016] Further, the inspection change module responds to the set conditions to determine the inspection change method, where,

[0017] The set conditions responded by the inspection change module are that the inspection information degree is greater than or equal to the preset inspection information degree or the interaction correlation degree is greater than or equal to the preset interaction correlation degree, and it is determined that the inspection change method is to determine the preferred selection method according to the change conditions;

[0018] The setting conditions for the inspection change module to respond are that the inspection information degree is less than the preset inspection information degree and the interaction correlation degree is less than the preset interaction correlation degree. It is determined that the inspection change method is to determine whether to conduct inspections on each uninspected area according to the information matching coefficient.

[0019] Further, the confirmation method of the inspection information degree is as follows:

[0020] If the inspection misalignment value is greater than or equal to the preset inspection misalignment value, the inspection information degree is determined according to the trajectory characterization value and the inspection coefficient;

[0021] If the inspection misalignment value is less than the preset inspection misalignment value, the inspection information degree is determined according to the inspection danger threshold and the inspection proportion.

[0022] Further, the inspection change module responds to the change conditions to determine the priority selection method. Among them,

[0023] The change condition for the inspection change module to respond is that the trajectory smoothness is greater than or equal to the preset trajectory smoothness or the regional correlation coefficient is less than the preset regional correlation coefficient. It is determined that the priority selection method is to determine the inspection priority coefficient of each uninspected area according to the information matching coefficient;

[0024] The change condition for the inspection change module to respond is that the trajectory smoothness is less than the preset trajectory smoothness and the regional correlation coefficient is greater than or equal to the preset regional correlation coefficient. It is determined that the priority selection method is to determine the inspection priority coefficient of the pre-inspection area corresponding to each associated combination according to the flight influence coefficient;

[0025] The inspection priority coefficient of a single uninspected area has a negative correlation with the information matching coefficient of this uninspected area;

[0026] The inspection priority coefficient of a single pre-inspection area has a positive correlation with the flight influence coefficient corresponding to this pre-inspection area.

[0027] Further, the confirmation method of the associated combination is as follows:

[0028] If the cross coefficient is greater than or equal to the preset cross coefficient, the associated combination is determined according to the cross similarity and the regional balance coefficient;

[0029] If the cross coefficient is less than the preset cross coefficient, the associated combination is determined according to the regional similarity.

[0030] Further, the inspection adjustment module determines the inspection adjustment method according to the inspection area category. Among them,

[0031] For the first-class inspection area, the inspection adjustment method is to determine the direction setting method of each collection point according to the adjustment conditions;

[0032] For the second - type inspection area, the inspection adjustment method is to set the collection direction of each collection point to the vertical direction.

[0033] Furthermore, the categories of the inspection areas include:

[0034] The first - type inspection area where the regional layout complexity is greater than or equal to the preset regional layout complexity or the goods stacking coefficient is greater than or equal to the preset goods stacking coefficient;

[0035] The second - type inspection area where the regional layout complexity is less than the preset regional layout complexity and the goods stacking coefficient is less than the preset goods stacking coefficient.

[0036] Furthermore, the inspection adjustment module responds to the adjustment conditions to determine the direction setting method of each collection point. Among them,

[0037] For a single collection point,

[0038] When the adjustment condition that the inspection adjustment module responds to is that the cross - correlation coefficient is greater than or equal to the preset cross - correlation coefficient or the influence threshold is less than the preset influence threshold, it is determined that the direction setting method is to determine the collection direction according to the associated direction group;

[0039] When the adjustment condition that the inspection adjustment module responds to is that the cross - correlation coefficient is less than the preset cross - correlation coefficient and the influence threshold is greater than or equal to the preset influence threshold, it is determined that the direction setting method is to determine the collection direction according to the direction evaluation coefficient.

[0040] Furthermore, the inspection review module determines whether each inspection area needs to give an alarm according to the danger trigger coefficient of the inspection image and the thermal imaging coefficient of the infrared thermal imaging image. Among them,

[0041] For a single inspection area,

[0042] If the danger trigger coefficient is greater than or equal to the preset danger trigger coefficient or the thermal imaging coefficient is greater than or equal to the preset thermal imaging coefficient, an alarm is sent to the user terminal;

[0043] If the danger trigger coefficient is less than the preset danger trigger coefficient and the thermal imaging coefficient is less than the preset thermal imaging coefficient, there is no need to send an alarm to the user terminal.

[0044] Compared with the prior art, the beneficial effects of the present invention are as follows. In the technical solution of the present invention, the inspection optimization module responds to the inspection conditions to determine the inspection optimization method. The inspection conditions effectively reflect the inspection difficulty of the logistics warehouse and the actual endurance state of the unmanned aerial vehicle. Then, according to the inspection conditions, different inspection optimization methods are adaptively selected, and the inspection optimization method can be adjusted in real time according to the actual situation, making the selection of the inspection optimization method more in line with the actual application scenario, and thus improving the inspection efficiency.

[0045] Furthermore, in the present invention, the patrol inspection change module determines the patrol inspection change method according to the set conditions, effectively reflects the correlation degree between the patrolled and unpatrolled areas of the UAV and the danger degree of the patrolled areas of the UAV through the set conditions, and then adaptively selects different patrol inspection change methods according to the set conditions, so that the selected patrol inspection change method can preferentially patrol high-risk patrol areas, can quickly identify potential safety hazards, and can improve the patrol inspection efficiency.

[0046] Furthermore, in the present invention, the patrol inspection adjustment module determines the patrol inspection area category according to the regional layout complexity and the cargo stacking coefficient, effectively reflects the cargo state of the patrol inspection area through the regional layout complexity and the cargo stacking coefficient, and then adaptively selects different patrol inspection adjustment methods according to the patrol inspection area category, so that the selected patrol inspection adjustment method can improve the accuracy of the images collected by the UAV, and further improve the patrol inspection efficiency and the accuracy of the patrol inspection of the patrol inspection area.

[0047] Furthermore, in the present invention, the direction setting method of each acquisition point is determined according to the adjustment conditions, the correlation degree of each acquisition direction of the patrol inspection area is effectively reflected through the adjustment conditions, and then different direction setting methods are adaptively selected according to the adjustment conditions, so that the selected direction setting method can collect more comprehensive and accurate data, and further improve the accuracy of the patrol inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is the module connection diagram of the logistics patrol inspection UAV based on artificial intelligence of the present invention;

[0049] Figure 2 is the flowchart of determining the patrol inspection optimization method according to the patrol inspection conditions of the present invention;

[0050] Figure 3 is the flowchart of determining the patrol inspection change method according to the set conditions of the present invention;

[0051] Figure 4 is the flowchart of determining the preferred selection method according to the change conditions of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] In order to make the objectives and advantages of the present invention clearer, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0053] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0054] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation to the present invention.

[0055] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and defined, the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0056] Please refer to Figures 1 to 4 as shown, the present invention provides an AI-based logistics inspection drone, including:

[0057] An information acquisition module for obtaining the basic information of the logistics warehouse and the drone flight information;

[0058] An inspection optimization module connected to the information acquisition module for responding to inspection conditions to determine the inspection optimization method. The inspection optimization method is to determine the inspection change method according to the set conditions, or to determine the inspection adjustment method according to the inspection area category;

[0059] An inspection change module connected to the information acquisition module and the inspection optimization module respectively for responding to the set conditions to determine the inspection change method. The inspection change method is to determine the priority selection method according to the change conditions, or to determine whether to conduct inspections on each uninspected area according to the information matching coefficient;

[0060] The priority selection method is to determine the inspection priority coefficient of each uninspected area according to the information matching coefficient, or to determine the inspection priority coefficient of the pre-inspection area corresponding to each associated combination according to the flight influence coefficient;

[0061] An inspection adjustment module connected to the information acquisition module and the inspection optimization module respectively for determining the inspection adjustment method according to the inspection area category. The inspection adjustment method is to determine the direction setting method of each acquisition point according to the adjustment conditions or to set the acquisition direction of each acquisition point to the vertical direction;

[0062] The inspection area category is determined according to the regional layout complexity and the goods stacking coefficient, and the direction setting method is to determine the acquisition direction according to the associated direction group or the direction evaluation coefficient;

[0063] The patrol inspection review module, which is respectively connected to the patrol inspection change module and the patrol inspection adjustment module, is used to collect and store patrol inspection images and infrared thermal imaging images, and determine whether early warning is required for each patrol inspection area according to the danger trigger coefficient of the patrol inspection image and the thermal imaging coefficient of the infrared thermal imaging image.

[0064] The application scenario of the present invention is the patrol inspection of unmanned aerial vehicles in a logistics warehouse. The basic information of the logistics warehouse includes but is not limited to the positions of various goods in the logistics warehouse and the corresponding goods categories of each patrol inspection area. The goods categories include but are not limited to matches, lighters, and fireworks. The flight information of the unmanned aerial vehicle is the flight trajectory of the unmanned aerial vehicle, which is easily understood by those skilled in the art and will not be described in detail here.

[0065] The confirmation method of the patrol inspection area is to divide the logistics warehouse into several rectangular areas with the same area and equal length and width, and take the rectangular area with the goods quantity greater than the preset goods quantity as the patrol inspection area. The number of divided rectangular areas is in a positive correlation with the area of the logistics warehouse.

[0066] The initial patrol inspection method of the unmanned aerial vehicle is to conduct patrol inspections on each patrol inspection area in the order of decreasing goods quantity. And the unmanned aerial vehicle collects patrol inspection images and infrared thermal imaging images at the collection points corresponding to each patrol inspection area. The collection direction of each collection point is the vertical direction, and the vertical direction is the direction from the collection point corresponding to a single patrol inspection area to the center point of the patrol inspection area. The collection point corresponding to a single patrol inspection area is a point located 6m away from the center point of the patrol inspection area and directly above the center point of the patrol inspection area. The center point of a single patrol inspection area is the center of the circumscribed circle of the patrol inspection area. The goods quantity corresponding to a single patrol inspection area is the total number of goods in the patrol inspection area. The value of the preset goods quantity can be determined by the user according to the actual application scenario. The greater the user's demand for improving the patrol inspection accuracy of the logistics warehouse, the smaller the value of the preset goods quantity. Provide a value of the preset goods quantity, and the preset goods quantity is 100. The patrol inspection image is an image taken by a high-definition camera carried by the unmanned aerial vehicle, and the thermal infrared imaging image is an image obtained by measuring the thermal radiation of the patrol inspection area by a thermal infrared imager carried by the unmanned aerial vehicle, which is easily understood by those skilled in the art and will not be described in detail here.

[0067] In the present invention, a continuously cyclic monitoring period is set, and the data status is determined once at the end of each monitoring period. The duration of the monitoring period can be set according to the user's needs. The greater the user's demand for monitoring accuracy, the smaller the duration of the monitoring period. Provide a value of the monitoring period, and the monitoring period is 5 minutes.

[0068] In the present invention, a number of historical records are correspondingly set. Any one of the historical records records at least one of the inspection area influence coefficient, inspection information degree, interaction correlation degree, area dispersion degree, information matching coefficient, and inspection dislocation value, etc. in the historical process of UAV inspection in the logistics warehouse. And each historical record corresponds to a qualified mark, which records whether the waybill photo review process meets the user's requirements. The qualified mark can be manually recorded. It can be understood that the user can determine whether the UAV inspection process meets the requirements according to the self-set indicators. The self-set indicators can be, but are not limited to, the number of misinspections, which will not be elaborated here. Among them, the number of misinspections is the number of times of wrongly determining whether the inspection area needs to be warned.

[0069] Specifically, the inspection optimization module responds to the inspection conditions to determine the inspection optimization method. Among them,

[0070] The inspection conditions responded by the inspection optimization module are that the inspection area influence coefficient is less than the preset inspection area influence coefficient or the endurance reference value is less than the preset endurance reference value. It is determined that the inspection optimization method is to determine the inspection change method according to the set conditions;

[0071] The inspection conditions responded by the inspection optimization module are that the inspection area influence coefficient is greater than or equal to the preset inspection area influence coefficient and the endurance reference value is greater than or equal to the preset endurance reference value. It is determined that the inspection optimization method is to determine the inspection adjustment method according to the inspection area category.

[0072] Among them, the inspection conditions include the first inspection condition and the second inspection condition. The first inspection condition is that the inspection area influence coefficient is less than the preset inspection area influence coefficient or the endurance reference value is less than the preset endurance reference value. The second inspection condition is that the inspection area influence coefficient is greater than or equal to the preset inspection area influence coefficient and the endurance reference value is greater than or equal to the preset endurance reference value;

[0073] The confirmation method of the inspection area influence coefficient is

[0074] If the area dispersion degree is greater than or equal to the preset area dispersion degree, the inspection area influence coefficient is determined according to the number of inspection areas and the influence mean value. Among them, the inspection area influence coefficient = the number of inspection areas - the influence mean value;

[0075] If the reference value of the number of inspection areas is less than the preset number of inspection areas, the inspection area influence coefficient is determined according to the historical accident frequency. Among them, the inspection area influence coefficient and the historical accident frequency are in a positive correlation relationship. The inspection area influence coefficient = α × the historical accident frequency. A value of α is provided, and α is 1.1;

[0076] The number of inspection areas is the total number of inspection areas in the logistics warehouse; the impact mean is the average of the sub-impact coefficients corresponding to each inspection area in the logistics warehouse. For a single inspection area, this inspection area is denoted as the target area, and the sub-impact coefficient corresponding to the target area = the number of inspection areas adjacent to the target area + the average of the correlation coefficients corresponding to each inspection area adjacent to the target area; the confirmation method of the correlation coefficient is to denote a single impact area adjacent to the target area as the reference area, and denote the number of cargo categories existing in both the reference area and the target area as the impact mean;

[0077] The historical accident frequency is the average of the warning times corresponding to each inspection area. The warning times corresponding to a single inspection area are the number of times this inspection area needs to be warned in the historical records that can meet the user's needs;

[0078] The endurance reference value = the remaining power of the drone in the current monitoring period / the total power when the drone is fully charged;

[0079] The regional dispersion is the average of the reference distances corresponding to each inspection area. For a single inspection area, this inspection area is denoted as the target inspection area, and the other inspection areas except the target inspection area are denoted as the reference inspection areas. The shortest distance from the center point of the target inspection area to the center points of each reference inspection area is denoted as the reference distance;

[0080] The values of the preset inspection area impact coefficient, the preset endurance reference value, and the preset regional dispersion can be determined by the user according to the actual application scenario. The greater the user's demand for determining the inspection change method according to the set conditions, the smaller the values of the preset inspection area impact coefficient and the preset endurance reference value. Provide a value of the preset inspection area impact coefficient, detect the historical records of determining the inspection change method according to the set conditions, and denote the average of the inspection area impact coefficients corresponding to the historical records that can meet the user's needs as the preset inspection area impact coefficient. The preset endurance reference value is 50%. The smaller the value of the preset regional dispersion, the greater the user's demand for determining the inspection area impact coefficient according to the number of inspection areas and the impact mean. Provide a value of the preset regional dispersion, detect the historical records of determining the inspection area impact coefficient according to the historical accident frequency, and denote the average of the regional dispersions corresponding to the historical records that can meet the user's needs as the preset regional dispersion.

[0081] Specifically, the inspection change module responds to the set conditions to determine the inspection change method, where,

[0082] The set conditions to which the inspection change module responds are that the inspection information degree is greater than or equal to the preset inspection information degree or the interaction correlation degree is greater than or equal to the preset interaction correlation degree. The determined inspection change method is to determine the preferred selection method according to the change conditions;

[0083] The setting conditions for the inspection change module response are that the inspection information degree is less than the preset inspection information degree and the interaction correlation degree is less than the preset interaction correlation degree. It is determined that the inspection change method is to determine whether to perform inspections on each uninspected area according to the information matching coefficient.

[0084] Among them, the setting conditions include the first setting condition and the second setting condition. The first setting condition is that the inspection information degree is greater than or equal to the preset inspection information degree or the interaction correlation degree is greater than or equal to the preset interaction correlation degree. The second setting condition is that the inspection information degree is less than the preset inspection information degree and the interaction correlation degree is less than the preset interaction correlation degree;

[0085] The confirmation method of the interaction correlation degree is

[0086] If the similarity coefficient is greater than or equal to the preset similarity coefficient, the interaction correlation degree is determined according to the regional correlation degree. Among them, the interaction correlation degree and the regional correlation degree are in a positive correlation relationship, and the interaction correlation degree = regional correlation degree × β1, where β1 is 1;

[0087] If the similarity coefficient is less than the preset similarity coefficient, the interaction correlation degree is determined according to the similarity coefficient. Among them, the interaction correlation degree and the similarity coefficient are in a positive correlation relationship, and the interaction correlation degree = similarity coefficient × β2, where β2 is 0.8;

[0088] The inspection areas that the unmanned aerial vehicle has inspected during the current monitoring period are recorded as the inspected areas, and the inspection areas that the unmanned aerial vehicle has not inspected during the current monitoring period are recorded as the uninspected areas. The similarity coefficient = the number of cargo categories that exist in both all the inspected areas and all the uninspected areas / A. The number of all cargo categories in all the inspected areas is recorded as a1, the number of all cargo categories in all the uninspected areas is recorded as a2, and the larger value of a1 and a2 is recorded as A;

[0089] The regional correlation degree = the number of adjacent areas / the total number of uninspected areas. The adjacent areas are the uninspected areas adjacent to the inspected areas;

[0090] For the values of the preset inspection information degree, the preset interaction correlation degree, and the preset similarity coefficient, the user can determine them according to the actual application scenario. The smaller the values of the preset inspection information degree and the preset interaction correlation degree, the greater the user's need to determine the preferred selection method according to the change conditions. Provide a set of values for the preset inspection information degree and the preset interaction correlation degree. Detect the historical records of the user's determination of the preferred selection method according to the change conditions, and record the average value of the inspection information degrees corresponding to the historical records that can meet the user's needs as the preset inspection information degree, and record the average value of the interaction correlation degrees corresponding to the historical records that can meet the user's needs as the preset interaction correlation degree. The larger the value of the preset similarity coefficient, the greater the user's need to determine the interaction correlation degree according to the similarity coefficient. Provide a value for the preset similarity coefficient, and the preset similarity coefficient is 60%;

[0091] Determine whether to conduct inspections on each uninspected inspection area according to the information matching coefficient. Among them,

[0092] For a single uninspected area,

[0093] If the information matching coefficient is greater than or equal to the preset information matching coefficient, do not conduct an inspection on this uninspected area;

[0094] If the information matching coefficient is less than the preset information matching coefficient, conduct an inspection on this uninspected area;

[0095] The confirmation method of the information matching coefficient is that for a single uninspected area, mark this uninspected area as the target uninspected area, mark the inspected area closest to the target uninspected area as the target inspected area, and the information matching coefficient = the number of cargo categories that exist in both the target uninspected area and the target inspected area - the shortest distance from the center point of the target uninspected area to the center point of the target inspected area;

[0096] For the value of the preset information matching coefficient, the user can determine it according to the actual application scenario. The higher the user's inspection accuracy requirement for the inspection area, the smaller the value of the preset information matching coefficient. Provide a value of the preset information matching coefficient, detect the historical records of not conducting inspections on uninspected areas, and mark the minimum value of the information matching coefficient corresponding to the historical records that can meet the user's needs as the preset information matching coefficient.

[0097] Specifically, the confirmation method of the inspection information degree is as follows:

[0098] If the inspection displacement value is greater than or equal to the preset inspection displacement value, determine the inspection information degree according to the trajectory characterization value and the inspection coefficient;

[0099] If the inspection displacement value is less than the preset inspection displacement value, determine the inspection information degree according to the inspection danger threshold and the inspection ratio.

[0100] Among them, if the inspection displacement value is greater than or equal to the preset inspection displacement value, the inspection information degree = the trajectory characterization value × the inspection coefficient;

[0101] If the inspection displacement value is less than the preset inspection displacement value, the inspection information degree = the inspection danger threshold × the inspection ratio;

[0102] The inspection displacement value is the average value of the distance thresholds corresponding to each inspected area. For a single inspected area, mark this inspected area as the first area, mark the other inspected areas except the first area as the second area, and the distance threshold corresponding to the first area is the average value of the shortest distances from the center point of the first area to the center points of each second area;

[0103] For the value of the preset inspection misalignment value, the user can determine it according to the actual application scenario. The smaller the value of the preset inspection misalignment value, the greater the user's need to determine the inspection information degree based on the risk coefficient and the trajectory characterization value. Provide a value of the preset inspection misalignment value, detect the historical records of determining the inspection information degree based on the risk coefficient and the trajectory characterization value, and record the average value of the inspection misalignment values corresponding to the historical records that can meet the user's needs as the preset inspection misalignment value;

[0104] The trajectory characterization value is the average value of the inspection risk coefficients corresponding to each inspected area. The confirmation method of the inspection risk coefficient is as follows: for a single inspected area, record this inspected area as the target area, and the inspection risk coefficient corresponding to the target area = the risk trigger coefficient corresponding to the target area + the thermal imaging coefficient corresponding to the target area;

[0105] The inspection coefficient = a2 / a1, and the inspection ratio = (the area of the smallest rectangle that can contain all inspected areas - the area of the smallest rectangle that can contain each inspected area) / the area of the smallest rectangle that can contain each inspected area; the inspection risk threshold is the maximum value of the inspection risk coefficients corresponding to each inspected area;

[0106] The risk trigger coefficient is the average value of the sub-risk trigger coefficients of each inspection image corresponding to a single inspected area. For a single inspection image, the sub-risk trigger coefficient = the personnel reference value + the instability threshold. The personnel reference value is the total number of people in a single inspection image, and the instability threshold is the number of damaged goods in the inspection image. The damaged goods in the inspection image are determined by convolutional neural network technology, which is easy for those skilled in the art to understand and will not be elaborated here;

[0107] The thermal imaging coefficient is the maximum value of the temperature reference values corresponding to each infrared thermal imaging image corresponding to a single inspection area. The temperature reference value corresponding to a single infrared thermal imaging image is the highest value of the temperatures corresponding to each pixel value in this infrared thermal imaging image. The temperature corresponding to a single pixel point is determined by infrared thermal imaging analysis software, which is easy for those skilled in the art to understand and will not be elaborated here.

[0108] Specifically, the inspection change module responds to the change condition to determine the priority selection method, where,

[0109] The change condition that the inspection change module responds to is that the trajectory smoothness is greater than or equal to the preset trajectory smoothness or the regional correlation coefficient is less than the preset regional correlation coefficient. The determination of the priority selection method is to determine the inspection priority coefficients of each uninspected area according to the information matching coefficient;

[0110] The change conditions responded by the inspection tour change module are that the trajectory smoothness is less than the preset trajectory smoothness and the area correlation coefficient is greater than or equal to the preset area correlation coefficient. It is determined that the priority selection method is to determine the inspection priority coefficient of each associated combination corresponding to the pre-inspection area according to the flight influence coefficient;

[0111] There is a negative correlation between the inspection priority coefficient of a single uninspected area and the information matching coefficient of this uninspected area;

[0112] There is a positive correlation between the inspection priority coefficient of a single pre-inspection area and the flight influence coefficient corresponding to this pre-inspection area.

[0113] Among them, the change conditions include the first change condition and the second change condition. The first change condition is that the trajectory smoothness is greater than or equal to the preset trajectory smoothness or the area correlation coefficient is less than the preset area correlation coefficient. The second change condition is that the trajectory smoothness is less than the preset trajectory smoothness and the area correlation coefficient is greater than or equal to the preset area correlation coefficient;

[0114] The confirmation method of the trajectory smoothness is as follows: after obtaining the UAV movement trajectory, a three-dimensional rectangular coordinate system is established with the UAV position corresponding to the starting time point as the origin. The UAV positions in the UAV movement trajectories corresponding to each time point excluding the starting time point are recorded as trajectory points, and the vector angles corresponding to each trajectory point are detected. The confirmation method of the vector angle corresponding to a single trajectory point is as follows: for a trajectory point, this trajectory point is recorded as the target trajectory point, and the vector corresponding to the trajectory point adjacent to the target trajectory point and before the target trajectory point in the movement order and the vector corresponding to the target trajectory point The included angle is recorded as the vector angle. , the vectors corresponding to each trajectory point are tangent to the movement trajectory and the direction of the vector is the same as the movement direction. The trajectory smoothness is the standard deviation of the vector angles corresponding to each trajectory point; the time point is set by the user himself. A method for setting the time point is provided. In the order of the UAV inspection time from early to late, every 1 s is recorded as a time point.

[0115] The regional correlation coefficient is the average value of the sub-correlation coefficients corresponding to each un-inspected area. The method for determining the sub-correlation coefficient is as follows: for a single un-inspected area, mark this un-inspected area as the target area, and mark the other un-inspected areas except the target area as the reference areas. The sub-correlation coefficient is the maximum value among the regional similarity degrees between the target area and each reference area. For a single reference area, mark this reference area as the analysis reference area. The regional similarity degree between the analysis reference area and the target area = goods similarity degree + distribution similarity degree. The goods similarity degree = (the number of goods categories that exist in both the analysis reference area and the target area / the total number of goods categories included in the analysis reference area and the target area). The distribution similarity degree = the absolute value of the difference between the distribution thresholds corresponding to the analysis reference area and the target area / the larger value among the distribution thresholds corresponding to the analysis reference area and the target area. The distribution threshold is the average value of the goods distances corresponding to each good in a single inspection area. The goods distance corresponding to a single good is the average value of the shortest distances from this good to other goods in the inspection area where this good is located;

[0116] For the values of the preset trajectory smoothness and the preset regional correlation coefficient, users can determine them according to the actual application scenario. The smaller the value of the preset trajectory smoothness and the larger the value of the preset regional correlation coefficient, the greater the user's need to determine the inspection priority coefficient of each un-inspected area according to the information matching coefficient. Provide a set of values for the preset trajectory smoothness and the preset regional correlation coefficient, detect the historical records of determining the inspection priority coefficient of each un-inspected area according to the information matching coefficient, and record the average value of the trajectory smoothness corresponding to the historical records that can meet the user's needs as the preset trajectory smoothness, and record the average value of the regional correlation coefficient corresponding to the historical records that can meet the user's needs as the preset regional correlation coefficient;

[0117] The flight influence coefficient corresponding to a single pre-inspection area = the shortest distance from the drone to this pre-inspection area + the yaw coefficient. The yaw coefficient is the angle between the first line segment and the second ray. The first line segment is the connection line between the trajectory point corresponding to the last time point in the current monitoring period of the drone and the center point corresponding to this pre-inspection area. The second ray is a ray starting from the trajectory point corresponding to the last time point in the current monitoring period of the drone and having the same vector direction as the trajectory point;

[0118] When determining the inspection priority coefficient of each un-inspected area according to the information matching coefficient, the smaller the information matching coefficient of the un-inspected area, the more prior the inspection order;

[0119] When determining the inspection priority coefficient of the pre-inspection area corresponding to each associated combination according to the flight influence coefficient, each associated combination corresponds to a pre-inspection area. For a single associated combination, randomly select any un-inspected area as a pre-inspection area. The smaller the flight influence coefficient of the pre-inspection area, the more prior the inspection order;

[0120] It should be noted that if the pre-inspection area meets the preset conditions, warnings are sent to the user terminal for each inspection area included in the associated combination corresponding to the pre-inspection area. The preset conditions are that the danger trigger coefficient is greater than or equal to the preset danger trigger coefficient or the thermal imaging coefficient is greater than or equal to the preset thermal imaging coefficient.

[0121] Specifically, the confirmation method of the associated combination is as follows:

[0122] If the cross coefficient is greater than or equal to the preset cross coefficient, the associated combination is determined according to the cross similarity and the area balance coefficient;

[0123] If the cross coefficient is less than the preset cross coefficient, the associated combination is determined according to the area similarity.

[0124] Among them, the cross coefficient is the average value of the cross reference values corresponding to each uninspected area. For a single uninspected area, the uninspected area is denoted as the first target area, and each inspected area adjacent to the first target area is denoted as the first reference area. The cross reference value corresponding to the first target area = the number of first reference areas / the cross angle; the line connecting the center points of each first reference area and the center point of the first target area is denoted as the reference line, and the angle of the smallest included angle that can contain each reference line is denoted as the cross angle; it should be noted that if the number of first reference areas is 0 or 1, then the cross reference value = the number of first reference areas;

[0125] When determining the associated combination according to the cross similarity and the area balance coefficient or according to the area similarity, correlation analysis is performed for each uninspected area. When performing correlation analysis for a single uninspected area, the uninspected area is denoted as the first target uninspected area, and the other uninspected areas that are not included in the associated combination except the first target uninspected area are denoted as the first reference uninspected areas. The set of the first reference uninspected areas and the first target uninspected area that meet the benchmark conditions is denoted as an associated combination, and correlation analysis is continued for the uninspected areas that are not included in the associated combination until all uninspected areas are included in the associated combination;

[0126] The benchmark conditions corresponding to different confirmation methods are different. Among them,

[0127] When the confirmation method is to determine the associated combination according to the cross similarity and the area balance coefficient, the benchmark condition is that the cross similarity with the first target inspection area is greater than the preset cross similarity and the area balance coefficient is greater than the preset area balance coefficient;

[0128] When the confirmation method is to determine the associated combination according to the area similarity, the benchmark condition is that the area similarity with the first target inspection area is greater than the preset area similarity;

[0129] Cross similarity = 1 - (absolute value of the difference between the cross-reference values corresponding to the two un-inspected areas / the larger of the cross-reference values corresponding to the two un-inspected areas),

[0130] For any two un-inspected areas, the way to confirm the area balance coefficient is as follows:

[0131] If the distance reference value is greater than or equal to the preset distance reference value, then the area balance coefficient = the area of the inspected area in the analysis rectangle / the area of the analysis rectangle. Denote the smallest rectangle that can contain the two un-inspected areas as the analysis rectangle.

[0132] If the distance reference value is less than the preset distance reference value, then the area balance coefficient is the larger of the coincidence ratios corresponding to the two un-inspected areas; Denote the smallest rectangle that can contain each inspected area as the first analysis rectangle. Coincidence ratio = the area where a single un-inspected area coincides with the first analysis rectangle / the area of the un-inspected area;

[0133] The values of the preset cross coefficient, preset cross similarity, preset area balance coefficient, and preset area similarity can be determined by the user according to the actual application scenario. The larger the value of the preset cross coefficient, the greater the user's need to determine the associated combination according to the area similarity. Provide a value of the preset cross coefficient, detect the historical records of determining the associated combination according to the cross similarity and the area balance coefficient, and denote the average value of the cross coefficients corresponding to the historical records that can meet the user's needs as the preset cross coefficient. The greater the user's need to improve the inspection accuracy, the smaller the values of the preset cross similarity, preset area balance coefficient, and preset area similarity. Provide a value of the preset cross similarity, preset area balance coefficient, and preset area similarity. The preset cross similarity is 70%, the preset area balance coefficient is 50%, and the preset area similarity is 80%.

[0134] Specifically, the inspection adjustment module determines the inspection adjustment method according to the inspection area category, where

[0135] For the first type of inspection area, the inspection adjustment method is to determine the direction setting method of each collection point according to the adjustment conditions;

[0136] For the second type of inspection area, the inspection adjustment method is to set the collection direction of each collection point to the vertical direction.

[0137] Among them, set the collection direction of each collection point to the vertical direction, where the collection direction is the direction of the UAV taking inspection images and infrared thermal imaging images; the vertical direction is the direction from the collection point corresponding to a single inspection area to the center point of the inspection area.

[0138] Specifically, the inspection area category includes:

[0139] A type of inspection area where the regional layout complexity is greater than or equal to a preset regional layout complexity or the cargo stacking coefficient is greater than or equal to a preset cargo stacking coefficient;

[0140] A second type of inspection area where the regional layout complexity is less than the preset regional layout complexity and the cargo stacking coefficient is less than the preset cargo stacking coefficient.

[0141] Among them, for a single inspection area, the regional layout complexity = the area of the area where goods are stacked in the inspection area / the area of the inspection area;

[0142] The method for confirming the cargo stacking coefficient is as follows: Denote the cargo categories that appear in a single inspection area during the current monitoring period as reference categories, and denote the smallest rectangle that can contain the goods corresponding to a single reference category in the inspection area as the rectangle to be analyzed. Each reference category corresponds to a rectangle to be analyzed. Denote the average value of the overlapping ratios corresponding to each rectangle to be analyzed as the cargo stacking coefficient; For a single rectangle to be analyzed, denote the rectangle to be analyzed as the target rectangle to be analyzed. The overlapping ratio corresponding to the target rectangle to be analyzed = the area of the area where the target rectangle to be analyzed overlaps with other rectangles to be analyzed / the area of the target rectangle to be analyzed;

[0143] The values of the preset regional layout complexity and the preset cargo stacking coefficient can be determined by the user according to the actual application scenario. The smaller the value of the preset regional layout complexity and the larger the value of the preset cargo stacking coefficient, the greater the user's demand for determining the direction setting method of each acquisition point according to the adjustment conditions. Provide a set of values for the preset regional layout complexity and the preset cargo stacking coefficient. The preset regional layout complexity is 70%. Detect the historical records of determining the inspection area as the second type of area, and denote the average value of the cargo stacking coefficients corresponding to the historical records that can meet the user's needs as the preset cargo stacking coefficient.

[0144] Specifically, the inspection adjustment module responds to the adjustment conditions to determine the direction setting method of each acquisition point. Among them,

[0145] For a single acquisition point,

[0146] The adjustment condition that the inspection adjustment module responds to is that the cross-correlation coefficient is greater than or equal to a preset cross-correlation coefficient or the influence threshold is less than a preset influence threshold, and it is determined that the direction setting method is to determine the acquisition direction according to the associated direction group;

[0147] The adjustment condition that the inspection adjustment module responds to is that the cross-correlation coefficient is less than the preset cross-correlation coefficient and the influence threshold is greater than or equal to the preset influence threshold, and it is determined that the direction setting method is to determine the acquisition direction according to the direction evaluation coefficient.

[0148] Among them, the adjustment conditions include a first adjustment condition and a second adjustment condition. The first adjustment condition is that the cross-correlation coefficient is greater than or equal to a preset cross-correlation coefficient or the influence threshold is less than a preset influence threshold. The second adjustment condition is that the cross-correlation coefficient is less than the preset cross-correlation coefficient and the influence threshold is greater than or equal to the preset influence threshold;

[0149] The method for confirming the cross-correlation coefficient is as follows: for the center point of a single inspection area, the due north direction of this center point is recorded as the reference direction. Starting from the reference direction and rotating clockwise, the direction corresponding to each 30° rotation is recorded as a direction to be analyzed until 360° is rotated, and 12 directions to be analyzed can be obtained. The average value of the sub-correlation means corresponding to each direction to be analyzed is recorded as the cross-correlation coefficient;

[0150] The sub-correlation mean is the average value of the sub-correlation degrees corresponding to a direction to be analyzed and other directions to be analyzed;

[0151] For a single direction to be analyzed in a single inspection area, this inspection area is recorded as the second target inspection area, this direction to be analyzed is recorded as the target direction, the line segment starting from the center point of the second target inspection area and ending at a point on the boundary of the second target inspection area and having the same direction as the target direction is recorded as the reference line, and the area that coincides with the second target inspection area in the circle with the midpoint of the reference line as the center and the length of the reference line as the diameter is recorded as the radiation area corresponding to the target direction. It can be understood that each direction to be analyzed corresponds to a radiation area;

[0152] For any two directions to be analyzed, the area where the radiation area corresponding to one direction to be analyzed does not overlap with the radiation area corresponding to the other direction to be analyzed is recorded as the non-overlapping reference area, and each of the two directions to be analyzed corresponds to a non-overlapping reference area;

[0153] Sub-correlation degree = the area of the overlapping area between the radiation area corresponding to one direction to be analyzed and the radiation area corresponding to the other direction to be analyzed + the number of types of goods that exist in both the non-overlapping reference area corresponding to one direction to be analyzed and the non-overlapping reference area corresponding to the other direction to be analyzed;

[0154] Direction evaluation coefficient = (the number of types of goods contained in the radiation area corresponding to a single direction to be analyzed / the total number of types of goods in the inspection area where the collection point is located) + the goods density corresponding to a single direction to be analyzed;

[0155] The goods density is the average value of the distance coefficients corresponding to each good in the radiation area corresponding to a single direction to be analyzed. The distance coefficient corresponding to a single good is the average value of the shortest distances from this good to other goods in the radiation area where this good is located;

[0156] Influence threshold = regional layout complexity + goods stacking coefficient;

[0157] Determine the acquisition direction according to the associated direction group, including: performing relevant analysis on each direction to be analyzed. For a single direction to be analyzed, denote this direction to be analyzed as the target direction to be analyzed, denote the directions to be analyzed that are not included in the associated direction group except the target direction to be analyzed as the reference directions to be analyzed. Denote the set of the reference directions to be analyzed with a sub - correlation greater than the preset sub - correlation with the target direction to be analyzed and the target direction to be analyzed as an associated direction group, and continue to perform relevant analysis on the directions to be analyzed that are not included in the associated direction group until all directions to be analyzed are included in the associated direction group; each associated direction group corresponds to an analysis direction. The single analysis direction is any randomly selected direction to be analyzed in a single associated direction group. Take the direction from the drone to the center of the radiation area corresponding to the single analysis direction as an acquisition direction;

[0158] Determine the acquisition direction according to the direction evaluation coefficient. Among them, take the direction to be analyzed with a direction evaluation coefficient greater than the preset direction evaluation coefficient as the analysis direction, and take the direction from the drone to the center of the radiation area corresponding to the single analysis direction as an acquisition direction. Each analysis direction corresponds to an acquisition direction;

[0159] The values of the preset cross - correlation coefficient, the preset influence threshold, and the preset sub - correlation can be determined by the user according to the actual application scenario. The smaller the value of the preset cross - correlation coefficient and the larger the value of the preset influence threshold, the greater the user's need to determine the acquisition direction according to the associated direction group. Provide a set of values for the preset cross - correlation coefficient and the preset influence threshold. Denote the historical record of determining the acquisition direction according to the associated direction group as the reference historical record. Denote the average value of the cross - correlation coefficients corresponding to the reference historical records that can meet the user's needs as the preset cross - correlation coefficient, and denote the average value of the influence thresholds corresponding to the reference historical records that can meet the user's needs as the preset influence threshold; the greater the user's need to improve the accuracy of warehouse inspection, the larger the value of the preset sub - correlation. Provide a set of values for the preset sub - correlation. Denote the average value of the sub - correlations corresponding to the reference historical records that can meet the user's needs as the preset sub - correlation.

[0160] Specifically, the inspection review module determines whether early warning is required for each inspection area according to the danger trigger coefficient of the inspection image and the thermal imaging coefficient of the infrared thermal imaging image. Among them,

[0161] For a single inspection area,

[0162] If the danger trigger coefficient is greater than or equal to the preset danger trigger coefficient or the thermal imaging coefficient is greater than or equal to the preset thermal imaging coefficient, send an early warning to the user terminal;

[0163] If the danger trigger coefficient is less than the preset danger trigger coefficient and the thermal imaging coefficient is less than the preset thermal imaging coefficient, there is no need to send an early warning to the user terminal.

[0164] Among them, the values of the preset danger trigger coefficient and the preset thermal imaging coefficient can be determined by the user according to the actual application scenario. The greater the user's demand for reducing the occurrence of logistics warehouse accidents, the smaller the values of the preset danger trigger coefficient and the preset thermal imaging coefficient. Provide the values of the preset danger trigger coefficient and the preset thermal imaging coefficient, detect the historical records of sending early warnings to the user side, and record the average value of the danger trigger coefficients corresponding to the historical records that can meet the user's needs as the preset danger trigger coefficient, and record the average value of the thermal imaging coefficients corresponding to the historical records that can meet the user's needs as the preset thermal imaging coefficient.

[0165] So far, the technical solution of the present invention has been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or replacements to the relevant technical features, and the technical solutions after these changes or replacements will all fall within the protection scope of the present invention.

[0166] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An AI-based logistics inspection drone, characterized in that, Including: An information acquisition module, used to obtain the basic information of the logistics warehouse and the UAV flight information; An inspection optimization module, connected to the information acquisition module, used to respond to inspection conditions to determine the inspection optimization method. The inspection optimization method is to determine the inspection change method according to the set conditions, or to determine the inspection adjustment method according to the inspection area category; An inspection change module, respectively connected to the information acquisition module and the inspection optimization module, used to respond to the set conditions to determine the inspection change method. The inspection change method is to determine the priority selection method according to the change conditions, or to determine whether to conduct inspections on each uninspected area according to the information matching coefficient; The priority selection method is to determine the inspection priority coefficient of each uninspected area according to the information matching coefficient, or to determine the inspection priority coefficient of the pre-inspection area corresponding to each associated combination according to the flight influence coefficient; An inspection adjustment module, respectively connected to the information acquisition module and the inspection optimization module, used to determine the inspection adjustment method according to the inspection area category. The inspection adjustment method is to determine the direction setting method of each acquisition point according to the adjustment conditions or to set the acquisition direction of each acquisition point to the vertical direction; The inspection area category is determined according to the regional layout complexity and the cargo stacking coefficient. The direction setting method is to determine the acquisition direction according to the associated direction group or the direction evaluation coefficient; An inspection review module, respectively connected to the inspection change module and the inspection adjustment module, used to collect and store inspection images and infrared thermal imaging images, and determine whether early warnings are required for each inspection area according to the danger trigger coefficient of the inspection image and the thermal imaging coefficient of the infrared thermal imaging image; The inspection conditions include a first inspection condition and a second inspection condition. The first inspection condition is that the inspection area influence coefficient is less than the preset inspection area influence coefficient or the endurance reference value is less than the preset endurance reference value. The second inspection condition is that the inspection area influence coefficient is greater than or equal to the preset inspection area influence coefficient and the endurance reference value is greater than or equal to the preset endurance reference value; The change conditions include a first change condition and a second change condition. The first change condition is that the trajectory smoothness is greater than or equal to the preset trajectory smoothness or the regional correlation coefficient is less than the preset regional correlation coefficient. The second change condition is that the trajectory smoothness is less than the preset trajectory smoothness and the regional correlation coefficient is greater than or equal to the preset regional correlation coefficient; The adjustment conditions include a first adjustment condition and a second adjustment condition. The first adjustment condition is that the cross-correlation coefficient is greater than or equal to the preset cross-correlation coefficient or the influence threshold is less than the preset influence threshold. The second adjustment condition is that the cross-correlation coefficient is less than the preset cross-correlation coefficient and the influence threshold is greater than or equal to the preset influence threshold; The confirmation method of the information matching coefficient is as follows: for a single uninspected area, record this uninspected area as the target uninspected area, and record the inspected area closest to the target uninspected area as the target inspected area. Information matching coefficient = the number of cargo categories that exist in both the target uninspected area and the target inspected area - the shortest distance from the center point of the target uninspected area to the center point of the target inspected area; Flight impact coefficient corresponding to a single pre-inspection area = Shortest distance from the UAV to the pre-inspection area + Yaw coefficient; For a single inspection area, Area layout complexity = Area of the area where goods are stacked in the inspection area / Area of the inspection area; The method for confirming the goods stacking coefficient is as follows: Denote the types of goods that appear in a single inspection area during the current monitoring period as reference types. Denote the smallest rectangle that can contain the goods corresponding to a single reference type in the inspection area as the rectangle to be analyzed. Each reference type corresponds to a rectangle to be analyzed. Denote the average value of the overlapping ratios corresponding to each rectangle to be analyzed as the goods stacking coefficient; The method for confirming the associated direction group is as follows: Conduct relevant analysis for each direction to be analyzed. For a single direction to be analyzed, denote the direction to be analyzed as the target direction to be analyzed, denote the directions to be analyzed that have not been recorded in the associated direction group except the target direction to be collected and analyzed as the reference directions to be analyzed. Denote the set of the reference directions to be analyzed whose sub-correlation degree with the target direction to be analyzed is greater than the preset sub-correlation degree and the target direction to be analyzed as an associated direction group, and continue to conduct relevant analysis for the directions to be analyzed that have not been recorded in the associated direction group until all directions to be analyzed are recorded in the associated direction group; Direction evaluation coefficient = (Number of types of goods contained in the radiation area corresponding to a single direction to be analyzed / Total number of types of goods in the inspection area where the collection point is located) + Goods density corresponding to a single direction to be analyzed; The danger trigger coefficient is the average value of the sub-danger trigger coefficients of each inspection image corresponding to a single inspected area. For a single inspection image, Sub-danger trigger coefficient = Personnel reference value + Instability threshold. The personnel reference value is the total number of people in a single inspection image, and the instability threshold is the number of damaged goods in the inspection image; The thermal imaging coefficient is the maximum value among the temperature reference values corresponding to each infrared thermal imaging image corresponding to a single inspection area; The method for confirming the said associated combination is as follows, If the cross coefficient is greater than or equal to the preset cross coefficient, determine the associated combination according to the cross similarity and the area balance coefficient; If the cross coefficient is less than the preset cross coefficient, determine the associated combination according to the area similarity; The cross coefficient is the average value of the cross reference values corresponding to each uninspected area. For a single uninspected area, denote the uninspected area as the first target area, and denote each inspected area adjacent to the first target area as the first reference area. The cross reference value corresponding to the first target area = Number of the first reference areas / Cross angle; Denote the line connecting the center points of each first reference area and the center point of the first target area as the reference line, and denote the angle of the smallest included angle that can contain each reference line as the cross angle; Cross similarity = 1 - (Absolute value of the difference between the cross reference values corresponding to two uninspected areas / Larger value among the cross reference values corresponding to two uninspected areas); For any two uninspected areas, the method for confirming the area balance coefficient is as follows, If the distance reference value is greater than or equal to the preset distance reference value, then Area balance coefficient = Area of the inspected area in the analysis rectangle / Area of the analysis rectangle. Denote the smallest rectangle that can contain two uninspected areas as the analysis rectangle, If the distance reference value is less than the preset distance reference value, the regional balance coefficient is the larger value of the coincidence ratios corresponding to the two un-inspected regions. Regional similarity = Goods similarity + Distribution similarity. Goods similarity = (the number of goods categories existing in both the analysis reference area and the target area / the total number of goods categories included in the analysis reference area and the target area). Distribution similarity = the absolute value of the difference between the distribution thresholds corresponding to the analysis reference area and the target area / the larger value of the distribution thresholds corresponding to the analysis reference area and the target area. The distribution threshold is the average of the goods distances corresponding to each good in a single inspection area. The goods distance corresponding to a single good is the average of the shortest distances from this good to other goods in the inspection area where this good is located.

2. The logistics inspection UAV based on artificial intelligence according to claim 1, wherein, The inspection optimization module responds to the inspection conditions to determine the inspection optimization method, where the inspection conditions responded by the inspection optimization module are that the inspection area influence coefficient is less than the preset inspection area influence coefficient or the battery life reference value is less than the preset battery life reference value, and it is determined that the inspection optimization method is to determine the inspection change method according to the set conditions; the inspection conditions responded by the inspection optimization module are that the inspection area influence coefficient is greater than or equal to the preset inspection area influence coefficient and the battery life reference value is greater than or equal to the preset battery life reference value, and it is determined that the inspection optimization method is to determine the inspection adjustment method according to the inspection area category; the confirmation method of the inspection area influence coefficient is if the regional dispersion degree is greater than or equal to the preset regional dispersion degree, the inspection area influence coefficient is determined according to the number of inspection areas and the influence mean value, where the inspection area influence coefficient = the number of inspection areas - the influence mean value; if the reference value of the number of inspection areas is less than the preset number of inspection areas, the inspection area influence coefficient is determined according to the historical accident frequency, where the inspection area influence coefficient and the historical accident frequency are in a positive correlation, and the inspection area influence coefficient = α × historical accident frequency. A value of α is provided, and α is 1.1; Battery life reference value = the remaining power of the drone in the current monitoring period / the total power when the drone is fully charged.

3. The logistics inspection UAV based on artificial intelligence according to claim 2, wherein, The inspection change module responds to the set conditions to determine the inspection change method, where the set conditions responded by the inspection change module are that the inspection information degree is greater than or equal to the preset inspection information degree or the interaction correlation degree is greater than or equal to the preset interaction correlation degree, and it is determined that the inspection change method is to determine the preferred selection method according to the change conditions; the set conditions responded by the inspection change module are that the inspection information degree is less than the preset inspection information degree and the interaction correlation degree is less than the preset interaction correlation degree, and it is determined that the inspection change method is to determine whether to conduct inspections on each un-inspected area according to the information matching coefficient; the confirmation method of the interaction correlation degree is if the similarity coefficient is greater than or equal to the preset similarity coefficient, the interaction correlation degree is determined according to the regional correlation degree, where the interaction correlation degree and the regional correlation degree are in a positive correlation, and the interaction correlation degree = regional correlation degree × β1, and β1 is 1; if the similarity coefficient is less than the preset similarity coefficient, the interaction correlation degree is determined according to the similarity coefficient, where the interaction correlation degree and the similarity coefficient are in a positive correlation, and the interaction correlation degree = similarity coefficient × β2, and β2 is 0.

8.

4. The logistics inspection UAV based on artificial intelligence according to claim 3, characterized in that, The confirmation method of the inspection information degree is If the inspection displacement value is greater than or equal to the preset inspection displacement value, the inspection information degree is determined according to the trajectory characterization value and the inspection coefficient; If the inspection displacement value is less than the preset inspection displacement value, the inspection information degree is determined according to the inspection danger threshold and the inspection ratio; The inspection displacement value is the average of the distance thresholds corresponding to each inspected area. For a single inspected area, this inspected area is denoted as the first area, and the other inspected areas except the first area are denoted as the second area. The distance threshold corresponding to the first area is the average of the shortest distances from the center point of the first area to the center points of each second area; The trajectory characterization value is the average of the inspection danger coefficients corresponding to each inspected area. The confirmation method of the inspection danger coefficient is that for a single inspected area, this inspected area is denoted as the target area, and the inspection danger coefficient corresponding to the target area = the danger trigger coefficient corresponding to the target area + the thermal imaging coefficient corresponding to the target area; The inspection coefficient = the number of all cargo categories in all uninspected areas / the number of all cargo categories in all inspected areas; The inspection danger threshold is the maximum value of the inspection danger coefficients corresponding to each inspected area; The inspection ratio = (the area of the smallest rectangle that can contain all inspection areas - the area of the smallest rectangle that can contain each inspected area) / the area of the smallest rectangle that can contain each inspected area.

5. The logistics inspection UAV based on artificial intelligence according to claim 3, characterized in that, The inspection change module responds to the change condition to determine the preferred selection method, where The change condition responded by the inspection change module is that the trajectory smoothness is greater than or equal to the preset trajectory smoothness or the area correlation coefficient is less than the preset area correlation coefficient, and the preferred selection method is determined according to the information matching coefficient to determine the inspection priority coefficient of each uninspected area; The change condition responded by the inspection change module is that the trajectory smoothness is less than the preset trajectory smoothness and the area correlation coefficient is greater than or equal to the preset area correlation coefficient, and the preferred selection method is determined according to the flight influence coefficient to determine the inspection priority coefficient of the pre-inspected area corresponding to each associated combination; The inspection priority coefficient of a single uninspected area is negatively correlated with the information matching coefficient of this uninspected area; The inspection priority coefficient of a single pre-inspected area is positively correlated with the flight influence coefficient corresponding to this pre-inspected area; The trajectory smoothness is the standard deviation of the vector angles corresponding to each trajectory point; The area correlation coefficient is the average of the sub-correlation coefficients corresponding to each uninspected area. The confirmation method of the sub-correlation coefficient is that for a single uninspected area, this uninspected area is denoted as the target area, and the other uninspected areas except the target area are denoted as the reference areas. The sub-correlation coefficient is the maximum value of the area similarities between the target area and each reference area.

6. The logistics inspection UAV based on artificial intelligence according to claim 5, characterized in that The inspection adjustment module determines the inspection adjustment method according to the inspection area category, where For the first type of inspection area, the inspection adjustment method is to determine the direction setting method of each collection point according to the adjustment condition; For the second type of inspection area, the inspection adjustment method is to set the collection direction of each collection point to the vertical direction.

7. The logistics inspection drone based on artificial intelligence according to claim 6, wherein The inspection area category includes: The first type of inspection area where the regional layout complexity is greater than or equal to the preset regional layout complexity or the cargo stacking coefficient is greater than or equal to the preset cargo stacking coefficient; The second - class inspection area where the regional layout complexity is less than the preset regional layout complexity and the cargo stacking coefficient is less than the preset cargo stacking coefficient.

8. The logistics inspection UAV based on artificial intelligence according to claim 6, characterized in that The inspection adjustment module responds to the adjustment conditions to determine the direction setting method of each collection point. Among them, For a single collection point, When the adjustment condition responded by the inspection adjustment module is that the cross - correlation coefficient is greater than or equal to the preset cross - correlation coefficient or the influence threshold is less than the preset influence threshold, it is determined that the direction setting method is to determine the collection direction according to the associated direction group; When the adjustment condition responded by the inspection adjustment module is that the cross - correlation coefficient is less than the preset cross - correlation coefficient and the influence threshold is greater than or equal to the preset influence threshold, it is determined that the direction setting method is to determine the collection direction according to the direction evaluation coefficient; The method for confirming the cross - correlation coefficient is as follows: For the center point of a single inspection area, the due - north direction of this center point is recorded as the reference direction. Starting from the reference direction and rotating clockwise, the direction corresponding to each 30° rotation is recorded as a direction to be analyzed until 360° is rotated, and 12 directions to be analyzed can be obtained. The average value of the sub - correlation means corresponding to each direction to be analyzed is recorded as the cross - correlation coefficient; Influence threshold = regional layout complexity + cargo stacking coefficient.

9. The logistics inspection UAV based on artificial intelligence according to claim 8, wherein The inspection review module determines whether each inspection area needs to give an early warning according to the danger trigger coefficient of the inspection image and the thermal imaging coefficient of the infrared thermal imaging image. Among them, For a single inspection area, If the danger trigger coefficient is greater than or equal to the preset danger trigger coefficient or the thermal imaging coefficient is greater than or equal to the preset thermal imaging coefficient, an early warning is sent to the user terminal; If the danger trigger coefficient is less than the preset danger trigger coefficient and the thermal imaging coefficient is less than the preset thermal imaging coefficient, there is no need to send an early warning to the user terminal.

Images