AI pollution tracing method and system based on combination of laser radar and unmanned aerial vehicle

Abstract

The invention provides an AI pollution traceability method based on combination of a laser radar and an unmanned aerial vehicle, the AI pollution traceability method is applied to a pollution traceability system, the AI pollution traceability system comprises a laser radar device, a cloud server, an unmanned aerial vehicle and a ground client, and the laser radar device, the unmanned aerial vehicle and the ground client are all connected with the cloud server. Comprising the steps that the laser radar equipment acquires laser point cloud data of a to-be-monitored area and coordinate information corresponding to hotspot cloud data in a hotspot map generated by the laser point cloud data; the laser radar equipment sends coordinate information corresponding to hotspot cloud data in a hotspot map generated by the laser point cloud data to a cloud server through a network signal; the high-definition image and thermal imaging information are intelligently recognized through an AI algorithm, the relation between the atmospheric pollutants and a co-occurrence target thereof is speculated according to whether the co-occurrence target exists at the edge of the atmospheric pollutants in the high-definition image or not, and then whether the thermal forming characteristics of the atmospheric pollutants and the co-occurrence target thereof in the high-definition image conform to the principle or not is analyzed.

Description

Technical Field

[0001] This invention patent relates to the technical field of pollution source tracing, and particularly relates to an AI pollution source tracing method and system based on the combined use of lidar and unmanned aerial vehicle (UAV). Background Art

[0002] Atmospheric pollution source tracing refers to tracking and determining the sources, transmission paths, and influence ranges of atmospheric pollutants through various technical means and methods, providing a scientific basis for atmospheric pollution control and environmental supervision. Nowadays, atmospheric pollution source tracing integrates multi-source information such as satellite remote sensing, ground monitoring, and meteorological data, transforms from planar monitoring to three-dimensional monitoring, combines technologies such as lidar to achieve all-round monitoring of atmospheric pollution, and combines AI and machine learning algorithms to achieve automatic identification and tracing of pollution sources; Chinese Patent Publication No. CN119165116A discloses an atmospheric pollution source tracing system and method based on an unmanned aerial vehicle, including: an airborne atmospheric monitoring module for obtaining atmospheric data; a ground station system for monitoring whether the flight state of the unmanned aerial vehicle is normal, sending status information to the unmanned aerial vehicle control platform, and receiving and storing the atmospheric data; the airborne atmospheric monitoring module is loaded on the unmanned aerial vehicle, and the unmanned aerial vehicle is controlled to fly through the unmanned aerial vehicle control platform, and the unmanned aerial vehicle control platform controls the unmanned aerial vehicle in real time according to the status information; a data processing module for preprocessing the atmospheric data; a source tracing analysis module for performing source tracing analysis on the preprocessed atmospheric data to obtain the source location of the pollution source. Accurate pollution source positioning and pollutant diffusion simulation provide a scientific basis for the formulation and implementation of environmental policies, and contribute to more effectively controlling and reducing atmospheric pollution; In the above technology, the ground station system can monitor the flight state of the unmanned aerial vehicle in real time, and timely feedback information to the unmanned aerial vehicle control platform, ensuring the safety of the flight mission and the continuity of data collection. However, when the unmanned aerial vehicle performs visual data collection, it has strong singularity, only speculates on the pollution source through the spatial relationship, has poor logic, and limits the accuracy of atmospheric pollution source tracing. Therefore, there is an urgent need for an AI pollution source tracing method and system based on the combined use of lidar and unmanned aerial vehicle. Summary of the Invention

[0003] Aiming at the defects in the prior art, this invention patent provides an AI pollution source tracing method and system based on the combined use of lidar and unmanned aerial vehicle to improve the accuracy of atmospheric pollution source tracing.

[0004] According to the first aspect of the embodiments of the present disclosure, a preferred embodiment of this invention patent provides an AI pollution source tracing method based on the combined use of lidar and unmanned aerial vehicle, which is applied to a pollution source tracing system, including a lidar device, a cloud server, an unmanned aerial vehicle, and a ground client. The lidar device, the unmanned aerial vehicle, and the ground client are all connected to the cloud server. The method includes: The lidar device acquires the lidar point cloud data of the area to be monitored and the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data; The lidar device sends the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data to the cloud server through a network signal; The cloud server controls the drone to collect at least two visual information of the area to be monitored according to the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data, wherein the visual information includes high-definition image information and thermal imaging information; The cloud server at least checks the recognition results of the high-definition image information and the thermal imaging information, and preferably transmits the recognition results that conform to each visual information to the ground client.

[0005] In one embodiment, the cloud server at least checks the recognition results of the high-definition image information and the thermal imaging information, and preferably transmits the recognition results that conform to each visual information to the ground client, including: Using the underlying logic formula of YOLOv8 to extract the edge feature information in the high-definition image information and establish a recognition target set, wherein the recognition target set includes air pollution recognition targets and general recognition targets, and the edge feature information extraction steps are: First, by introducing more skip connections and Split operations, the gradient flow is enhanced and the calculation amount is reduced, and its code logic formula is expressed as: Among them, the Split operation divides the input feature map into two parts, one part directly passes through multiple Bottleneck layers, and the other part serves as a Shortcut connection, and finally the features are fused through a convolutional layer; Then, multi-scale feature fusion is achieved by concatenating the maximum pooling results of different scales, and its code logic formula is expressed as: ; Obtain the general recognition targets on the edge of the air pollution recognition target in the high-definition image recognition target set as the high-definition image inference target, and infer the first recognition result set based on the air pollution recognition target and the high-definition image inference target set; Fuse the high-definition image information and the thermal imaging information to obtain the thermal imaging information features of any target in the air pollution recognition target and its edge general recognition targets; If all the thermal imaging information features match the preset features of the air pollution recognition target and its edge general recognition targets, directly output the first recognition result set; If the thermal imaging information features of the air pollution identification target and the ordinary identification target on its edge do not match their preset features, then modify the air pollution identification target and the high-definition image inference target, and infer a second identification result set based on the new set for output.

[0006] In one embodiment, the lidar device acquires lidar point cloud data of the area to be monitored and the coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the lidar point cloud data, including: The lidar device performs a 360° scan in at least one plane, with a single scan angle of 2° and a scan radius of 6 km. A total of 180 scan intervals are divided in one scan plane to obtain a hot spot map generated by the lidar point cloud data; If the percentage of the difference between the particulate matter reflection detection frequency detected by the lidar device and the particulate matter reflection standard frequency to the particulate matter reflection standard frequency is greater than the standard threshold, then the reflection area within this scan interval will be marked as the hot spot cloud data in the hot spot map; Calculate the distance between the lidar device and the hot spot cloud data based on the reflection time difference of the hot spot cloud data detected within the scan interval and the lidar wave velocity; Fuse the coordinate information of the lidar device and the hot spot cloud data relative to the lidar device with the GPS data of the lidar device to obtain the coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the lidar point cloud data.

[0007] According to the second aspect of the embodiments of the present disclosure, the present invention patent provides an AI pollution tracing system based on the combination of lidar and unmanned aerial vehicle, which is applied to a pollution tracing system and includes a lidar device, a cloud server, an unmanned aerial vehicle, and a ground client. The lidar device, the unmanned aerial vehicle, and the ground client are all connected to the cloud server. The device includes: A monitoring module for the lidar device to acquire lidar point cloud data of the area to be monitored and the coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the lidar point cloud data; A transmission module for the lidar device to send the coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the lidar point cloud data to the cloud server through a network signal; A visual acquisition module for the cloud server to control the unmanned aerial vehicle to acquire at least two visual information of the area to be monitored according to the coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the lidar point cloud data, where the visual information includes high-definition image information and thermal imaging information; An AI identification module for the cloud server to at least check the identification results of the high-definition image information and the thermal imaging information, and preferably transmit the identification results that conform to each visual information to the ground client.

[0008] In one embodiment, the AI recognition module includes: A first recognition module for extracting edge feature information from the high-definition image information using the underlying logic formula of YOLOv8 to establish a recognition target set, where the recognition target set includes air pollution recognition targets and general recognition targets. The edge feature information extraction steps are as follows: First, by introducing more skip connections and Split operations, the gradient flow is enhanced and the computational amount is reduced. Its code logic formula is expressed as: Among them, the Split operation divides the input feature map into two parts. One part directly passes through multiple Bottleneck layers, and the other part serves as a Shortcut connection. Finally, the features are fused through a convolutional layer; Then, multi-scale feature fusion is achieved by concatenating the maximum pooling results of different scales. Its code logic formula is expressed as: ; An inference module for obtaining general recognition targets on the edge of the air pollution recognition target in the high-definition image recognition target set as high-definition image inference targets, and inferring a first recognition result set based on the air pollution recognition target and the high-definition image inference target set; A second recognition module for fusing the high-definition image information with the thermal imaging information to obtain the thermal imaging information features of any target among the air pollution recognition target and its edge general recognition targets; A first output module, if all the thermal imaging information features match the preset features of the air pollution recognition target and its edge general recognition targets, directly outputs the first recognition result set; A second output module for, if the thermal imaging information features of the air pollution recognition target and its edge general recognition targets do not match their preset features, modifying the air pollution recognition target and the high-definition image inference target, and inferring a second recognition result set based on the new set for output.

[0009] In one embodiment, the monitoring module includes: A scanning module that generates a hot spot map from the laser point cloud data obtained by the lidar device; A locking module for, if the percentage of the difference between the particulate matter reflection detection frequency detected by the lidar device and the particulate matter reflection standard frequency to the particulate matter reflection standard frequency is greater than the standard threshold, marking the reflection area within this scanning interval as hot spot cloud data in the hot spot map; A calculation module for calculating the distance between the lidar device and the hot spot cloud data based on the reflection time difference of the hot spot cloud data detected within the scanning interval and the lidar wave velocity; A matching module, configured to fuse the coordinate information of the lidar device and the hotspot cloud data relative to the lidar device with the GPS data of the lidar device, so as to obtain the coordinate information corresponding to the hotspot cloud data in the hotspot map generated by the lidar point cloud data.

[0010] As can be seen from the above technical solutions, an AI pollution tracing method and system based on the combined use of lidar and unmanned aerial vehicle provided by the present invention for invention patents may include the following beneficial effects: The present disclosure performs intelligent recognition on high-definition images and thermal imaging information through AI algorithms. First, based on whether there are co-occurring targets at the edge of atmospheric pollutants in the high-definition images, the relationship between the atmospheric pollutants and their co-occurring targets is inferred. Then, it is analyzed whether the thermal forming characteristics of the atmospheric pollutants and their co-occurring targets in the high-definition images conform to common sense, so as to verify the accuracy of inferring the relationship between the atmospheric pollutants and their co-occurring targets, significantly improving the accuracy and reliability of pollution source identification, and being particularly applicable to pollution monitoring in complex industrial areas, urban environments, fire emergency responses, agricultural straw burning, industrial park leakage detection, complex terrain areas, and nighttime or low visibility conditions.

[0011] It should be understood that the above general description and the following detailed description are only exemplary and do not limit the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the specific embodiments of the present invention for invention patents, the drawings required for use in the specific embodiments or the description of the prior art will be briefly introduced below. In all the drawings, the elements or parts do not necessarily draw according to the actual scale.

[0013] Figure 1 It is a flowchart of an AI pollution tracing method based on the combined use of lidar and unmanned aerial vehicle provided by the present invention for invention patents; Figure 2 It is a flowchart of step S40 in an AI pollution tracing method based on the combined use of lidar and unmanned aerial vehicle provided by the present invention for invention patents; Figure 3 It is a flowchart of step S10 in an AI pollution tracing method based on the combined use of lidar and unmanned aerial vehicle provided by the present invention for invention patents; Figure 4 It is a block diagram of an AI pollution tracing system based on the combined use of lidar and unmanned aerial vehicle provided by the present invention for invention patents. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] The embodiments of the technical solutions of the present invention for invention patents will be described in detail below with reference to the drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention for invention patents, and thus are only examples and cannot be used to limit the protection scope of the present invention for invention patents.

[0015] Figure 1 The flowchart of an AI pollution source tracing method based on the combined use of lidar and unmanned aerial vehicle provided for this invention patent. An AI pollution source tracing method based on the combined use of lidar and unmanned aerial vehicle provided in this embodiment is applied to a pollution source tracing system, including a lidar device, a cloud server, an unmanned aerial vehicle, and a ground client. The lidar device, the unmanned aerial vehicle, and the ground client are all connected to the cloud server. As Figure 1 shown, the method includes the following steps S10 - S40: In step S10, the lidar device acquires the lidar point cloud data of the area to be monitored and the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data; In this implementation, the coordinate information of the hot spot cloud data can accurately identify the distribution area of pollutants, providing a key clue for subsequent pollution source positioning.

[0016] In step S20, the lidar device sends the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data to the cloud server through a network signal; In this implementation, by transmitting data to the cloud server in real time, the system can quickly respond to pollution incidents and notify relevant departments to take measures in a timely manner. The cloud server plans the flight path of the unmanned aerial vehicle according to the coordinate information of the hot spot cloud data to ensure that the unmanned aerial vehicle can cover all hot spot areas. Path planning needs to consider the flight speed, endurance, and time requirements of the acquisition task of the unmanned aerial vehicle.

[0017] In step S30, the cloud server controls the unmanned aerial vehicle to collect at least two visual information of the area to be monitored according to the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data. Among them, the visual information includes high-definition image information and thermal imaging information; In this implementation, the unmanned aerial vehicle has the characteristics of fast response and flexible maneuverability, and can quickly reach the hot spot area to collect high-definition images and thermal imaging information. At the same time, real-time data transmission ensures that the cloud server can obtain the latest monitoring information in a timely manner. The combination of high-definition image information and thermal imaging information provides multi-dimensional data support for pollution source tracing. High-definition images can provide visual details, while thermal imaging information can reveal temperature anomalies. The combination of the two can more comprehensively identify pollution sources.

[0018] In step S40, the cloud server at least checks the recognition results of the high-definition image information and the thermal imaging information, and preferably transmits the recognition results that conform to each visual information to the ground client; In this implementation method, the AI algorithm is used to evaluate multiple recognition results and select the result that best conforms to logic and common sense. Based on this, multi-spectral imaging technology can also be used to generate images by capturing spectral information in multiple specific bands. It can not only obtain information within the visible light range but also extend to bands such as near-infrared and short-wave infrared. By taking advantage of the different reflectivities of different substances in different bands, multi-spectral imaging can more accurately identify the target substance and can be used to analyze pollutant components, concentrations, etc.

[0019] In one embodiment, as Figure 2 shown, in step S40, the cloud server at least checks the recognition results of the high-definition image information and the thermal imaging information, and preferably transmits the recognition results that conform to each visual information to the ground client, including the following steps S41 - S45: In step S41, the edge feature information in the high-definition image information is extracted using the underlying logic formula of YOLOv8, and a recognition target set is established. Among them, the recognition target set includes air pollution recognition targets and general recognition targets. The edge feature information extraction steps are as follows: First, by introducing more skip connections and Split operations, the gradient flow is enhanced and the computational amount is reduced. Its code logic formula is expressed as: Among them, the Split operation divides the input feature map into two parts. One part directly passes through multiple Bottleneck layers, and the other part serves as a Shortcut connection. Finally, the features are fused through a convolutional layer; Then, multi-scale feature fusion is achieved by concatenating the maximum pooling results of different scales. Its code logic formula is expressed as: ; In this implementation method, the edge feature refers to the area where the brightness or color in the image changes significantly, usually corresponding to the contour of an object or the boundary between different objects. Through the edge detection algorithm, the contour of the pollutant or object can be clearly extracted, and information such as the type, size, and position of the recognition target can be judged through the contour of the pollutant or object.

[0020] In step S42, the general recognition targets on the edge of the air pollution recognition targets in the high-definition image recognition target set are obtained as the high-definition image inference targets, and the first recognition result set is inferred based on the air pollution recognition targets and the high-definition image inference target set; In this implementation method, common air pollution targets include: bare soil, dust, smoke, straw burning, etc. Common general recognition targets include: fog guns, fog gun machines, fog gun vehicles, earth-hauling vehicles, off-road vehicles, dust-proof nets, sprinklers, concrete, semi-trailers, moving cars, moving buses, etc. General recognition targets around air pollution targets are identified as high-definition image inference targets. Based on this, the absence of a dust-proof net near bare soil, the absence of a sprinkler or fog gun near dust, the presence of a moving car or a moving bus near dust, etc. all belong to the recognition results of air pollution. It should be noted that the recognition results need to be determined by means of the three-dimensional spatial relationship between air pollution targets and high-definition image inference targets, that is, at least two perspectives of top view and side view are required, which can significantly reduce the misjudgment phenomenon caused by a single perspective.

[0021] In step S43, the high-definition image information is fused with the thermal imaging information to obtain the thermal imaging information characteristics of any target in the air pollution recognition target and its edge general recognition target. In this implementation method, at the same viewing angle, the high-definition image information and the thermal imaging information can complement each other, reduce the limitations of a single information source, and can analyze the characteristics of the target more comprehensively, improving the accuracy of recognition. For example: A common air pollution target is dust, and the high-definition image inference targets are chimneys and moving cars. By analogy with the positions of the chimneys and moving cars in the high-definition image information, the temperature distribution of the chimneys and moving cars can be obtained in the thermal imaging information.

[0022] In step S44, if all the thermal imaging information characteristics match the preset characteristics of the air pollution recognition target and its edge general recognition target, the first recognition result set is directly output. In this implementation method, based on step S43, if it is obtained in the thermal imaging information that the temperatures of the dust, chimney, and moving car are all in a non-high-temperature state, the recognition result of air pollution is that there is a moving car near the dust (the non-high-temperature chimney has no logical relationship with the dust).

[0023] In step S45, if the thermal imaging information characteristics of the air pollution recognition target and its edge general recognition target do not match their preset characteristics, the air pollution recognition target and the high-definition image inference target are modified, and a second recognition result set is inferred and output according to the new set. In this implementation method, based on step S43, if it is obtained in the thermal imaging information that the temperatures of the dust and chimney are both in a high-temperature state and the temperature of the car is in a non-high-temperature state, it means that the air pollution target is misjudged. The air pollution recognition target is modified to smoke, and the recognition result of air pollution is that there is a chimney near the smoke (the non-high-temperature moving car has no logical relationship with the smoke).

[0024] In one embodiment, as Figure 3 shown, in step S10, the lidar device acquires the lidar point cloud data of the area to be monitored and the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data, including the following steps S11 - S14: In step S11, the lidar device is used to obtain the hot spot map generated from the lidar point cloud data; In this implementation, lidar is a technology that uses lasers for distance measurement and target detection. When operating, it makes a 360° scan in at least one plane, with a single scan angle of 2° and a scan radius of 6 km. A scan plane is divided into 180 scan intervals in total. The 360° scan can cover the entire horizontal direction and is scanned in at least one plane, which may involve multiple vertical planes to obtain three - dimensional information.

[0025] In step S12, if the percentage of the difference between the particulate matter reflection detection frequency detected by the lidar device and the particulate matter reflection standard frequency and the particulate matter reflection standard frequency is greater than the standard threshold, the reflection area within this scan interval will be marked as the hot spot cloud data in the hot spot map; In this implementation, when the percentage difference of the reflection area exceeds the standard threshold, this area will be marked as the hot spot cloud data in the hot spot map. The hot spot cloud data indicates that this area may be an area with a relatively high pollutant concentration and requires further attention and analysis.

[0026] In step S13, the distance between the lidar device and the hot spot cloud data is calculated according to the reflection time difference of the hot spot cloud data detected within the scan interval and the lidar wave velocity; In this implementation, the lidar device emits laser pulses towards the target area. After the laser pulses encounter the target object, they are reflected back to the lidar device. The device measures the distance to the target object by receiving the reflected light. By measuring the flight time (TOF) or phase difference of the laser pulses, the distance to the target object is calculated, and three - dimensional hot spot cloud data is generated in combination with the scan angle.

[0027] In step S14, the coordinate information of the lidar device and the hot spot cloud data relative to the lidar device is fused with the GPS data of the lidar device to obtain the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data; In this implementation, fusing the relative coordinate information of the hot spot cloud data acquired by the lidar device with the GPS data of the lidar device is a key step in achieving the precise positioning of the hot spot cloud data. This process ensures that the coordinate information of the hot spot cloud data is converted from relative coordinates to absolute geographical coordinates (such as longitude and latitude), thereby enabling the accurate positioning of the pollution source in the actual environment.

[0028] The following are the device embodiments of the present disclosure, which can be used to implement the method embodiments of the present disclosure.

[0029] Figure 4 The block diagram of an AI pollution source tracing system based on the combined use of lidar and unmanned aerial vehicle provided by this invention patent is as Figure 4 shown: An AI pollution source tracing system based on the combined use of lidar and unmanned aerial vehicle, applied to a pollution source tracing system, includes a lidar device, a cloud server, an unmanned aerial vehicle, and a ground client. The lidar device, the unmanned aerial vehicle, and the ground client are all connected to the cloud server. The device includes: A monitoring module 100, configured to obtain the lidar point cloud data of the area to be monitored by the lidar device and the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data; A transmission module 200, configured to send the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data to the cloud server through a network signal by the lidar device; A visual acquisition module 300, configured to control the unmanned aerial vehicle by the cloud server to acquire at least two visual information of the area to be monitored according to the coordinate information corresponding to the hot spot cloud data in the hot spot map generated from the lidar point cloud data, where the visual information includes high-definition image information and thermal imaging information; An AI recognition module 400, configured to check at least the recognition results of the high-definition image information and the thermal imaging information by the cloud server, and preferably transmit the recognition results that match each visual information to the ground client.

[0030] In one embodiment, as Figure 4 shown, the AI recognition module 400 includes: A first recognition module 401, configured to extract the edge feature information in the high-definition image information by using the underlying logic formula of YOLOv8, and establish a recognition target set, where the recognition target set includes air pollution recognition targets and ordinary recognition targets. The edge feature information extraction steps are as follows: First, by introducing more skip connections and Split operations, the gradient flow is enhanced and the calculation amount is reduced. Its code logic formula is expressed as: Among them, the Split operation divides the input feature map into two parts. One part directly passes through multiple Bottleneck layers, and the other part serves as a Shortcut connection. Finally, the features are fused through a convolutional layer; Then, multi-scale feature fusion is achieved by concatenating the maximum pooling results of different scales. Its code logic formula is expressed as: ; An inference module 402, configured to obtain a general recognition target on the edge of an air pollution recognition target in the high-definition image recognition target set as a high-definition image inference target, and infer a first recognition result set based on the air pollution recognition target and the high-definition image inference target set; A second recognition module 403, configured to fuse the high-definition image information with the thermal imaging information to obtain the thermal imaging information characteristics of any target among the air pollution recognition target and its edge general recognition targets; A first output module 404, if all the thermal imaging information characteristics match the preset characteristics of the air pollution recognition target and its edge general recognition targets, directly output the first recognition result set; A second output module 405, configured to, if the thermal imaging information characteristics of the air pollution recognition target and its edge general recognition targets do not match their preset characteristics, modify the air pollution recognition target and the high-definition image inference target, and infer a second recognition result set based on the new set for output.

[0031] In one embodiment, as Figure 4 shown, the monitoring module 100 includes: A scanning module 101, configured to obtain a hotspot map generated from the lidar point cloud data through the lidar device; A locking module 102, configured to, if the percentage of the difference between the particulate matter reflection detection frequency detected by the lidar device and the particulate matter reflection standard frequency to the particulate matter reflection standard frequency is greater than the standard threshold, mark the reflection area within this scanning interval as hotspot cloud data in the hotspot map; A calculation module 103, configured to calculate the distance between the lidar device and the hotspot cloud data based on the reflection time difference of the hotspot cloud data detected within the scanning interval and the lidar wave velocity; A matching module 104, configured to fuse the coordinate information of the lidar device and the hotspot cloud data relative to the lidar device with the GPS data of the lidar device to obtain the coordinate information corresponding to the hotspot cloud data in the hotspot map generated from the lidar point cloud data.

[0032] Those skilled in the art will readily conceive of other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure, which follow the general principles of the present disclosure and include known common knowledge or conventional technical means in the technical field not disclosed by the present disclosure. The specification and examples are only to be considered as exemplary, and the true scope and spirit of the present disclosure are pointed out by the following claims.

[0033] It should be understood that the present disclosure is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is only limited by the appended claims.

Claims

1. An AI pollution source tracing method based on the combination of laser radar and drone, applied to the pollution source tracing system, including laser radar equipment, cloud server, drone and ground client, the laser radar equipment, drone and ground client are all connected to the cloud server, characterized in that: The method comprises: The laser radar device obtains laser point cloud data of the area to be monitored and coordinate information corresponding to the hot spot cloud data in a hot spot map generated by the laser point cloud data; The laser radar device sends the coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the laser point cloud data to the cloud server through a network signal; The cloud server controls the drone to collect at least two visual information of the monitored area according to the coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the laser point cloud data, wherein the visual information includes high-definition image information and thermal imaging information; The cloud server at least checks the recognition results of the high-definition image information and the thermal imaging information, and selects the recognition results that are consistent with each visual information and transmits them to the ground client.

2. The method according to claim 1, characterized in that The cloud server at least checks the recognition results of the high-definition image information and the thermal imaging information, and selects the recognition results that meet the various visual information and transmits them to the ground client, including: The edge feature information in the high-definition image information is extracted using the underlying logic formula of YOLOv8 to establish a recognition target set, wherein the recognition target set includes air pollution recognition targets and common recognition targets. The edge feature information extraction steps are as follows: First, by introducing more skip-layer connections and Split operations, the gradient flow is enhanced and the amount of calculation is reduced. The code logic is expressed as follows: The Split operation divides the input feature map into two parts. One part directly passes through multiple Bottleneck layers, and the other part is connected as a shortcut. Finally, the features are fused through the convolution layer. Then, multi-scale feature fusion is achieved by concatenating the maximum pooling results of different scales. The code logic is expressed as follows: ； Acquire a common recognition target at the edge of the air pollution recognition target in the high-definition image recognition target set as a high-definition image inference target, and infer a first recognition result set based on the air pollution recognition target and the high-definition image inference target set; The high-definition image information is integrated with the thermal imaging information to obtain the thermal imaging information features of any one of the air pollution identification target and the common identification targets on its edge; If all of the thermal imaging information features match the preset features of the atmospheric pollution identification target and the common identification target at its edge, directly outputting the first identification result set; If the thermal imaging information features of the atmospheric pollution identification target and the ordinary identification targets on its edges do not match their preset features, the atmospheric pollution identification target and the high-definition image inference target are modified, and a second identification result set is inferred based on the new set for output.

3. The method according to claim 1, characterized in that The laser radar device obtains laser point cloud data of the area to be monitored and coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the laser point cloud data, including: The laser radar device is used to obtain a heat map generated by laser point cloud data; If the difference between the particle reflection detection frequency detected by the laser radar device and the particle reflection standard frequency and the percentage of the particle reflection standard frequency is greater than the standard threshold, the reflection area within this scanning interval will be marked as hot spot cloud data in the heat map; Calculate the distance between the laser radar device and the hotspot cloud data according to the reflection time difference of the hotspot cloud data detected within the scanning interval and the laser radar wave velocity; The coordinate information of the laser radar device and the hotspot cloud data relative to the laser radar device is integrated with the GPS data of the laser radar device to obtain the coordinate information corresponding to the hotspot cloud data in the hotspot map generated by the laser point cloud data.

4. An AI pollution source tracing system based on the combination of laser radar and drone, characterized in that: Applied to the pollution tracing system, including laser radar equipment, cloud server, drone and ground client, the laser radar equipment, drone and ground client are all connected to the cloud server, the device includes: A monitoring module, used for the laser radar device to obtain laser point cloud data of the area to be monitored and coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the laser point cloud data; A transmission module, used for the laser radar device to send the coordinate information corresponding to the hot spot cloud data in the hot spot map generated by the laser point cloud data to the cloud server through a network signal; A visual acquisition module, used for the cloud server to control the drone to collect at least two visual information of the monitored area according to the coordinate information corresponding to the hot spot cloud data in the heat map generated by the laser point cloud data, wherein the visual information includes high-definition image information and thermal imaging information; The AI ​​recognition module is used for the cloud server to at least check the recognition results of the high-definition image information and the thermal imaging information, and select the recognition results that meet the various visual information and transmit them to the ground client.

5. The system according to claim 4, characterized in that The AI ​​recognition module includes: The first recognition module is used to extract edge feature information from the high-definition image information using the underlying logic formula of YOLOv8, and establish a recognition target set, wherein the recognition target set includes air pollution recognition targets and common recognition targets, and the edge feature information extraction steps are: First, by introducing more skip-layer connections and Split operations, the gradient flow is enhanced and the amount of calculation is reduced. The code logic is expressed as follows: The Split operation divides the input feature map into two parts. One part directly passes through multiple Bottleneck layers, and the other part is connected as a shortcut. Finally, the features are fused through the convolution layer. Then, multi-scale feature fusion is achieved by concatenating the maximum pooling results of different scales. The code logic is expressed as follows: ； an inference module, configured to obtain a common recognition target at the edge of an air pollution recognition target in the high-definition image recognition target set as a high-definition image inference target, and infer a first recognition result set based on the air pollution recognition target and the high-definition image inference target set; A second recognition module is used to merge the high-definition image information with the thermal imaging information to obtain thermal imaging information features of any one of the air pollution recognition targets and the common recognition targets on its edges; A first output module, directly outputting a first recognition result set if all of the thermal imaging information features match the preset features of the air pollution recognition target and the common recognition target at its edge; The second output module is used to modify the atmospheric pollution identification target and the high-definition image inference target if the thermal imaging information features of the atmospheric pollution identification target and the ordinary identification targets on its edges do not match their preset features, and to infer a second identification result set based on the new set for output.

6. The system according to claim 4, characterized in that The monitoring module comprises: A scanning module, which obtains a heat map generated by laser point cloud data through the laser radar device; A locking module, for marking the reflection area within the scanning interval as hotspot cloud data in the heat map if the difference between the particle reflection detection frequency detected by the laser radar device and the particle reflection standard frequency and the percentage of the particle reflection standard frequency are greater than a standard threshold; A calculation module, used to calculate the distance between the laser radar device and the hotspot cloud data according to the reflection time difference of the hotspot cloud data detected in the scanning interval and the laser radar wave velocity; The matching module is used to fuse the coordinate information of the laser radar device and the hotspot cloud data relative to the laser radar device with the GPS data of the laser radar device to obtain the coordinate information corresponding to the hotspot cloud data in the hotspot map generated by the laser point cloud data.

Images