Unmanned aerial vehicle path planning method and system based on machine learning

Abstract

The embodiment of the invention provides an unmanned aerial vehicle path planning method and system based on machine learning, and the method comprises the steps: receiving an inspection instruction sent by a meteorological server, and the inspection instruction is sent after the meteorological server determines that pollutants exceed the standard according to the monitoring information of a meteorological monitoring station; determining a target geographic area from the industrial park according to the over-standard pollutant information, the position of the meteorological monitoring station and the wind speed and direction information, and calculating the average value of the productivity weight and the emission weight of the suspicious factory to obtain the sorting weight of the suspicious factory; and calculating a patrol distance according to the sorting weight of the suspicious factories, the distance between the suspicious factories and the patrol sequence of the suspicious factories, optimizing the patrol sequence of the suspicious factories by taking the shortest patrol distance as an optimization target, and controlling the unmanned aerial vehicle to sequentially patrol each suspicious factory according to the optimal patrol sequence. The target pollution source is determined from all the suspicious factories according to the inspection result, the unmanned aerial vehicle is called at any time for evidence obtaining, and the unmanned aerial vehicle inspection time is shortened.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and system for unmanned aerial vehicle path planning based on machine learning. Background Art

[0002] As environmental protection and sustainable development receive more attention, the application of drone technology in environmental monitoring has become increasingly widespread and important.

[0003] Normally, if it is necessary to monitor the exhaust emissions of a certain industrial park, gas sensors and particulate matter sensors are installed on drones, and the pilot controls the drone to fly to the exhaust outlets of various factories in the industrial park. The gas sensors monitor the concentration of pollutants in the exhaust gas emitted from the exhaust outlets, and the particulate matter sensors monitor the concentration of particulate matter in the exhaust gas emitted from the exhaust outlets.

[0004] However, it takes a lot of time for pilots to control drones to inspect factories in industrial parks based on their personal experience, which will delay the time for evidence collection. In addition, if the factory emits emissions at night, drones cannot be used in time for evidence collection. Summary of the invention

[0005] The embodiments of the present application provide a drone path planning method and system based on machine learning, which are used to achieve the effect of calling drones for evidence collection at any time and effectively reducing drone inspection time.

[0006] Some embodiments of the present application provide a method for drone path planning based on machine learning, which is applied to a background server and includes: Receiving an inspection instruction sent by a meteorological server, wherein the inspection instruction is sent by the meteorological server after determining that pollutants exceed the standard according to monitoring information of the meteorological monitoring station, and the inspection instruction includes information on pollutants exceeding the standard monitored by the meteorological monitoring station, the location of the meteorological monitoring station, and wind speed and direction information of the meteorological monitoring station; Determine the target geographic area from the industrial parks near the meteorological monitoring stations based on the information of pollutants exceeding the standard, the location of the meteorological monitoring stations, and the wind speed and direction information; Calculating the capacity weight of the suspicious factory according to the capacity of the suspicious factory in the target geographical area, calculating the emission weight of the suspicious factory according to the historical emission data of the suspicious factory in the target geographical area, calculating the average of the capacity weight and emission weight of the suspicious factory, and obtaining the ranking weight of the suspicious factory; The patrol distance is calculated according to the ranking weight of the suspicious factories, the distance between the suspicious factories and the patrol order of the suspicious factories, and the patrol order of the suspicious factories is optimized with the shortest patrol distance as the optimization goal; The drone is controlled to inspect each suspicious factory in the optimal inspection order, and the target pollution source is determined from each suspicious factory according to the inspection results, and a stop inspection command is issued to the drone.

[0007] In the above technical solution, the meteorological server generates an inspection instruction after determining that the pollutants in the gas exceed the standard, so that the background server can promptly determine the target geographical area from the industrial park near the meteorological monitoring station based on the information of the pollutants exceeding the standard, the location of the meteorological monitoring station, and the wind speed and direction information, and plan the patrol route for each suspicious factory in the target geographical area, so that the drone can patrol according to the planned route, so that the drone can be called in time to collect evidence when exceeding the standard is found. In addition, when planning the route, the weight of each suspicious factory is set according to the historical emission data and production capacity of each suspicious factory, so that the probability of each suspicious factory emitting pollutants is taken into account when optimizing the shortest patrol distance, so that the drone can find the target pollution source in a shorter time during the patrol, reduce the patrol time, and collect evidence for the target pollution source in a timely manner.

[0008] In a possible implementation, the capacity weight of the suspicious factory is calculated according to the capacity of the suspicious factory in the target geographical area, specifically including: Perform data standardization on the production capacity of the suspicious factory to obtain the standardized production capacity of the suspicious factory; Calculate the sum of the standardized inverses of the production capacity of all suspicious factories, and use the ratio of the standardized inverses of the production capacity of the suspicious factories to the sum of the inverses of the production capacity as the production capacity weight of the suspicious factories; Accordingly, the emission weights of the suspected factories are calculated based on the historical emission data of the suspected factories in the target geographical area, including: Calculate the mean and variance of the historical emission data of the suspicious factory, calculate the sum of the inverse of the mean of the historical emission data of all the suspicious factories, and calculate the sum of the inverse of the variance of the historical emission data of all the suspicious factories; Calculating a first ratio of the inverse of the mean value of the historical emission data to the sum of the inverses of the mean values; calculating a second ratio of the inverse of the variance of the historical emission data to the sum of the inverses of the variances; The average of the first ratio and the second ratio is taken as the emission weight of the suspicious factory.

[0009] In the above technical solution, the capacity weight is calculated based on the inverse of the capacity by finding the inverse of the capacity of each suspicious factory, so that the larger the capacity, the smaller the capacity weight, and the suspicious factory with a greater possibility of pollution discharge is inspected first while taking into account the shortest inspection distance. By statistically analyzing the historical emission data, the larger the average value of the emission data, the greater the possibility of pollution discharge, the smaller the emission weight, and the larger the variance, the greater the possibility of pollution discharge, and the smaller the emission weight. In this way, the suspicious factory with a greater possibility of pollution discharge is inspected first while taking into account the shortest inspection distance.

[0010] In a possible implementation, data standardization is performed on the production capacity of the suspicious factory to obtain the standardized production capacity of the suspicious factory, specifically including: The weight of pollutants exceeding the standard generated under the completed production capacity of the suspected factory is obtained, and the weight of pollutants exceeding the standard generated is used as the standardized production capacity of the suspected factory.

[0011] In the above technical solution, since the production capacity of a suspicious factory is used to determine the possibility of pollution discharge from the suspicious factory, by converting the production capacity of all suspicious factories into the weight of pollutants exceeding the standard, the possibility of the suspicious factory discharging pollutants exceeding the standard can be accurately assessed.

[0012] In a possible implementation, the first element to the m-1th element in the solution vector are then modified according to the mth element to the Nth element in the solution vector so that the values ​​of the elements in the solution vector are different, specifically including: After replacing the mth element to the Nth element in the solution vector with the mth element to the Nth element in the replacement vector, the missing order value is determined according to the value of each element in the solution vector; If there are multiple missing order values, the ranking weight of the suspicious factory represented by the correction element is obtained, and the value of the correction element is corrected according to the ranking weight of the suspicious factory represented by the correction element; wherein, the smaller the ranking weight of the suspicious factory represented by the correction element, the smaller the value of the correction element.

[0013] In the above technical scheme, when the remaining 1st to m-1th elements in the solution vector are corrected using the replaced mth to Nth elements, the element value is corrected according to the sorting weight of the suspicious factory represented by the element to be corrected. In this way, the solution vector can be closer to the optimal value, the search process for the optimal value can be accelerated, and the calculation time of the algorithm can be shortened.

[0014] In a possible implementation, the target geographical area is determined from multiple industrial parks based on the excessive pollutant information, the location of the weather monitoring station, and the wind speed and direction information, specifically including: Obtain the relative position between the industrial parks around the weather monitoring station and the weather monitoring station; According to the wind speed and direction information and the relative position between the industrial park and the meteorological monitoring station, the industrial park with the same relative position and wind direction as the meteorological monitoring station is selected as the suspicious geographical area; The violation probability of the suspicious geographical area is calculated based on the historical emission data and excessive pollutant information of the suspicious geographical area, and the suspicious geographical area with the highest violation probability is selected from the suspicious geographical areas as the target geographical area.

[0015] In the above technical scheme, after determining that a certain meteorological monitoring station has detected pollutants exceeding the standard, the suspicious geographical area is circled based on the wind direction and speed information of the meteorological monitoring station, and then the target geographical area with the highest probability of violation is selected based on the historical emission data and the information on pollutants exceeding the standard in the suspicious geographical area. Inspections are carried out in the target geographical area, so that the pollution source can be located quickly and time for on-site evidence collection can be gained.

[0016] In a possible implementation, the drone is controlled to inspect each suspicious factory in the optimal inspection order, and a target pollution source is determined from each suspicious factory according to the inspection result, specifically including: Predict the concentration of pollutants exceeding the standard at the suspected factory based on the distance between the suspected factory and the meteorological monitoring station, wind speed and direction information, and the concentration of pollutants exceeding the standard; The monitoring concentration of pollutants exceeding the standard collected by the UAV during its inspection of the suspicious factory is obtained. When the inspected suspicious factory is determined to be a target pollution source based on the monitoring concentration and predicted concentration of pollutants exceeding the standard in the inspected suspicious factory, a stop inspection instruction is sent to the UAV.

[0017] In the above technical solution, when the drone is patrolling in the optimal patrol order, the monitored concentration of pollutants exceeding the standard in the current suspicious factory is sent to the background server in real time. The background server compares the monitored concentration of pollutants exceeding the standard with the predicted concentration. If the difference is less than the preset threshold, the target pollution source is found and the drone is controlled to stop patrolling, thereby reducing the patrol mission of the drone.

[0018] In a possible implementation, determining whether the inspected suspicious factory is a target pollution source according to the monitored concentration and predicted concentration of pollutants exceeding the standard in the inspected suspicious factory specifically includes: For the current suspicious factory, the monitored concentration of the pollutants exceeding the standard of the current suspicious factory is compared with the predicted concentration. If the single difference is less than the first preset threshold, the suspicious factory is determined to be the target pollution source; if the single difference is greater than the first preset threshold, it is determined whether the monitored concentration of the pollutants exceeding the standard is greater than the standard emission concentration. If so, the cumulative monitoring value of the pollutants exceeding the standard is calculated, and the cumulative monitoring value is compared with the predicted average value. If the difference is less than the second preset threshold, the suspicious factory corresponding to the cumulative monitoring value is the target pollution source.

[0019] In the above technical solution, when determining the target pollution source from multiple suspicious factories, the impact of emissions from a single suspicious factory is judged. If the emissions from a single suspicious factory do not meet the conditions, the cumulative detection value is calculated, considering the joint impact of multiple factories. In this way, the target pollution source can be accurately selected from the suspicious factories, reducing the risk of misjudgment and missed judgment.

[0020] Some embodiments of the present application provide a drone path planning system based on machine learning, including: A receiving module, used to receive an inspection instruction sent by a meteorological server, wherein the inspection instruction is sent by the meteorological server after determining that the pollutants exceed the standard according to the monitoring information of the meteorological monitoring station, and the inspection instruction includes the information of the pollutants exceeding the standard monitored by the meteorological monitoring station, the location of the meteorological monitoring station and the wind speed and direction information of the meteorological monitoring station; A processing module, for determining a target geographical area from an industrial park near a meteorological monitoring station according to the excessive pollutant information, the location of the meteorological monitoring station and the wind speed and direction information; Calculating the capacity weight of the suspicious factory according to the capacity of the suspicious factory in the target geographical area, calculating the emission weight of the suspicious factory according to the historical emission data of the suspicious factory in the target geographical area, calculating the average of the capacity weight and emission weight of the suspicious factory, and obtaining the ranking weight of the suspicious factory; The processing module is further used to calculate the patrol distance according to the ranking weights of the suspicious factories, the distances between the suspicious factories and the patrol order of the suspicious factories, and optimize the patrol order of the suspicious factories with the shortest patrol distance as the optimization goal; The control module is used to control the UAV to inspect each suspicious factory in turn according to the optimal inspection order, determine the target pollution source from each suspicious factory according to the inspection results, and send a stop inspection instruction to the UAV.

[0021] Some embodiments of the present application provide an electronic device, including: a memory, a processor; Memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above-mentioned embodiments and / or various possible implementations.

[0022] Some embodiments of the present application provide a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the above embodiments and / or various possible implementation methods.

[0023] Some embodiments of the present application provide a computer program product, including a computer program, which implements the above embodiments and / or various possible implementation modes when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0025] Figure 1 A schematic diagram of a scenario of a drone path planning method based on machine learning provided in some embodiments of the present application; Figure 2 A schematic diagram of a process flow of a drone path planning method based on machine learning provided in some embodiments of the present application; Figure 3 A schematic diagram of an inspection order corresponding to a solution space vector provided in some embodiments of the present application; Figure 4 A schematic diagram of replacing and correcting a solution vector provided in some embodiments of the present application; Figure 5 A schematic diagram of selecting a target geographical area from a plurality of industrial parks provided for some embodiments of the present application; Figure 6 A schematic diagram of the structure of a machine learning-based drone path planning system provided for some embodiments of the present application.

[0026] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0027] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0028] As environmental protection and sustainable development receive more attention, the application of drone technology in environmental monitoring has become increasingly widespread and important.

[0029] Normally, if it is necessary to monitor the exhaust gas emissions of a certain industrial park, gas sensors and particulate matter sensors are installed on drones, and the pilot controls the drone to fly to the exhaust gas emission outlets of various factories in the industrial park. The gas sensors monitor the concentration of pollutants in the exhaust gas emitted from the exhaust gas emission outlets, and the particulate matter sensors monitor the concentration of particulate matter in the exhaust gas emitted from the exhaust gas emission outlets.

[0030] However, it takes a lot of time for pilots to control drones to inspect factories one by one in the industrial park, which will delay the time for evidence collection. In addition, if the factory emits at night, drones cannot be used to collect evidence in time.

[0031] To solve the above problems, the present application provides a drone path planning method and system based on machine learning, which can be used to call for drone evidence collection at any time and effectively reduce the drone inspection time.

[0032] Figure 1 A schematic diagram of a scenario of a UAV path planning method based on machine learning provided in this application, such as Figure 1 As shown, the meteorological monitoring station is connected to the meteorological server, the meteorological server is connected to the background server, and the background server is connected to the drone. The meteorological monitoring stations located at each monitoring point monitor the concentration of pollutants in the atmosphere. The meteorological server determines whether the concentration of pollutants exceeds the standard. When it is determined that the concentration exceeds the standard, it generates an inspection instruction, triggering the background server to select a target geographical area from the industrial park near the monitoring point, and calculate the inspection distance according to the ranking weight of the suspicious factories in the target geographical area and the distance between the suspicious factories. The inspection distance is minimized as the optimization goal, and the order of each suspicious factory is optimized. Inspections are carried out in the optimized order, which can achieve the identification of the target pollution source with a shorter inspection time and a shorter inspection distance, and timely evidence collection of the emission behavior of the target pollution source.

[0033] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0034] Figure 2 A flow chart of the UAV path planning method based on machine learning provided in this application, such as Figure 2 As shown, the method includes: S101. Each meteorological monitoring station monitors the concentration of pollutants in the atmosphere and sends the pollutant concentration to a meteorological server.

[0035] Among them, multiple meteorological monitoring stations are planned in a certain administrative area. Each meteorological monitoring station monitors the concentration of pollutants in the atmosphere and sends the pollutant concentration to the meteorological server in real time.

[0036] Optionally, high-precision atmospheric pollutant monitoring equipment is deployed at each meteorological monitoring station. These devices can monitor the concentration of various pollutants in the atmosphere in real time, including but not limited to sulfur dioxide, nitrogen oxides, particulate matter (PM2.5, PM10), etc. The monitoring equipment collects pollutant samples in the ambient air through built-in sensors and uses advanced analytical technology to quickly determine the concentration value of pollutants.

[0037] The monitoring equipment at each meteorological monitoring station automatically collects pollutant concentration data in the atmosphere at certain time intervals (such as every hour or every half hour). The collected raw data is first preprocessed, including data cleaning, denoising and outlier detection, to ensure the accuracy and reliability of the data. The preprocessed data is formatted into a standardized format to facilitate subsequent data transmission and processing.

[0038] The pre-processed data is transmitted to the meteorological server in real time through the communication equipment (such as 4G / 5G modules, satellite communication equipment, etc.) equipped by the meteorological monitoring station. During the transmission process, the data is protected by encryption protocols to ensure the security of data transmission and prevent data leakage or tampering.

[0039] After receiving the data, the meteorological server will conduct further verification and validation to ensure the integrity and accuracy of the data. After receiving the pollutant concentration data sent by each meteorological monitoring station, the meteorological server stores it in a dedicated database for subsequent query and analysis. The database design takes into account the efficient retrieval and long-term preservation needs of the data, and adopts an optimized index structure and data compression algorithm to improve data access speed and save storage space. At the same time, the meteorological server is also responsible for data backup and recovery to ensure data security and reliability. As a data integration center, the meteorological server not only receives data from each meteorological monitoring station, but also exchanges and shares data with other related systems (such as environmental monitoring systems, emergency response systems, etc.). Through data interfaces or API services, the meteorological server can provide integrated pollutant concentration data to other systems or departments that need these data to maximize the use of data resources.

[0040] S102. The meteorological server determines whether the pollutant concentration is greater than the standard concentration. If so, it is determined that the pollutant concentration exceeds the standard. An inspection instruction is generated based on the information of pollutants exceeding the standard monitored by the meteorological monitoring station, the location of the meteorological monitoring station, and the wind speed and direction information of the meteorological monitoring station, and the inspection instruction is sent to the background server.

[0041] Among them, after receiving the pollutant concentration monitored by each meteorological monitoring station, the meteorological server compares the pollutant concentration with the standard concentration. If it is determined that the pollutant concentration is greater than the standard concentration, it is determined that the pollutant concentration exceeds the standard. The meteorological monitoring station that monitors the pollutant concentration exceeding the standard, the concentration information of the pollutant exceeding the standard, and the location of the meteorological monitoring station are recorded. After encoding the location of the meteorological monitoring station that monitors the pollutant concentration exceeding the standard, the concentration information of the pollutant exceeding the standard, and the wind speed and direction information of the meteorological monitoring station, an inspection instruction is generated and sent to the background server.

[0042] For example, if there are 10 meteorological monitoring stations in a certain administrative area, and the monitoring data of meteorological monitoring station 5 indicates that the concentration of sulfur dioxide is greater than the standard concentration, the location of meteorological monitoring station 5, the concentration of sulfur dioxide and the wind direction and speed information of meteorological monitoring station 5 are recorded.

[0043] S103. The backend server determines the target geographical area from the industrial park near the meteorological monitoring station according to the information of pollutants exceeding the standard, the location of the meteorological monitoring station, and the wind speed and direction information.

[0044] Among them, the background server receives the inspection instructions sent by the meteorological server, parses the inspection instructions to obtain the information of excessive pollutants monitored by the meteorological monitoring station, the location of the meteorological monitoring station and the wind speed and direction information of the meteorological monitoring station.

[0045] The information of pollutants exceeding the standard can eliminate geographical areas that will not produce pollutants exceeding the standard, and the wind speed and direction information can determine the direction of the meteorological monitoring station where the factory that produces pollution is located. When there are multiple industrial parks that meet the conditions, only one industrial park can be selected as the current target geographical area based on the probability of illegal emissions in the industrial park. If the target pollution source is not determined in the current target geographical area, another industrial park is selected as the next target geographical area based on the probability of illegal emissions in the industrial park.

[0046] For example: There are 9 industrial parks near the meteorological monitoring station 5, and 4 industrial parks do not emit sulfur dioxide, so these 4 industrial parks can be eliminated. Of the remaining 5 industrial parks, 2 industrial parks are located northeast of the meteorological monitoring station 10, and 3 industrial parks are located south of the meteorological monitoring station 10. When the meteorological monitoring station 10 detects that sulfur dioxide exceeds the standard, there is a northerly wind. In this case, one of the 2 industrial parks located northeast can be selected as the current target geographical area. If no target pollution source is determined in the current target geographical area, another industrial park can be selected as the next target geographical area.

[0047] S104. The backend server calculates the capacity weight of the suspicious factory according to the capacity of the suspicious factory in the target geographical area, calculates the emission weight of the suspicious factory according to the historical emission data in the target geographical area, calculates the average of the capacity weight and emission weight of the suspicious factory, and obtains the ranking weight of the suspicious factory.

[0048] Among them, the background server pre-collects information of each suspicious factory in each industrial park, such as: the production capacity and historical emission data of each suspicious factory.

[0049] Extract the production capacity and historical emission data of the suspicious factories in the target geographical area from the collected data, and calculate the production capacity weight of the suspicious factories according to the production capacity of the suspicious factories in the target geographical area. The larger the production capacity of the suspicious factories, the smaller the production capacity weight of the suspicious factories. Calculate the emission weight of the suspicious factories according to the historical emission data in the target geographical area. The larger the historical emission data, the smaller the production capacity weight of the suspicious factories.

[0050] The larger the production capacity of a suspicious factory, the greater the possibility of emitting excessive pollutants. Therefore, the production capacity weight of the suspicious factory is set to be smaller. In this way, the suspicious factory is given priority while ensuring the inspection order corresponding to the minimum inspection distance.

[0051] The larger the historical emission data of a suspicious factory, the greater the possibility of emitting pollutants exceeding the standard. Therefore, the emission weight of the suspicious factory is set to be smaller. In this way, the suspicious factory is given priority while ensuring the inspection order corresponding to the minimum inspection distance.

[0052] For each suspicious factory, after obtaining the production capacity weight and emission weight of the suspicious factory, the average value of the production capacity weight and emission weight of the suspicious factory is calculated as the ranking weight of the suspicious factory.

[0053] S105. The backend server calculates the patrol distance according to the ranking weights of the suspicious factories, the distances between the suspicious factories, and the patrol order of the suspicious factories, and optimizes the patrol order of the suspicious factories with the shortest patrol distance as the optimization goal.

[0054] Among them, according to the inspection order of each suspicious factory and the distance between each suspicious factory, the path distance of the inspection path under the inspection order of the suspicious factory is calculated, and the path distance of the inspection path is corrected using the ranking weight of the suspicious factory to obtain the inspection distance.

[0055] The inspection order of suspicious factories is continuously updated, and the inspection order with the shortest inspection distance is selected from the inspection orders of multiple suspicious factories as the optimal inspection order.

[0056] S106. The backend server controls the drone to inspect each suspicious factory in turn according to the optimal inspection order, determines the target pollution source from each suspicious factory according to the inspection results, and issues a stop inspection instruction to the drone.

[0057] After obtaining the optimal patrol order, the backend server sends the optimal patrol order to the drone, so that the drone patrols according to the optimal patrol order and returns the patrol results to the backend server. After the backend server determines that a suspicious factory is the target pollution source, it sends a stop patrol command to the drone, and the drone flies back to the airport for rest.

[0058] In the above technical solution, the meteorological server generates an inspection instruction after determining that the pollutants in the gas exceed the standard, so that the background server can promptly determine the target geographical area from the industrial park near the meteorological monitoring station based on the information of the pollutants exceeding the standard, the location of the meteorological monitoring station, and the wind speed and direction information, and plan the patrol route for each suspicious factory in the target geographical area, so that the drone can patrol according to the planned route, so that the drone can be called in time to collect evidence when exceeding the standard is found. In addition, when planning the route, the weight of each suspicious factory is set according to the historical emission data and production capacity of each suspicious factory, so that the probability of each suspicious factory emitting pollutants is taken into account when optimizing the shortest patrol distance, so that the drone can find the target pollution source in a shorter time during the patrol, reduce the patrol time, and collect evidence for the target pollution source in a timely manner.

[0059] Specifically, when the above solution is applied to the background server, the technical effects achieved are mainly reflected in the following aspects: First, timely response and efficient inspection: After determining that pollutants in the atmosphere exceed the standard, the meteorological server can quickly generate inspection instructions and send them to the background server. After receiving the instructions, the background server immediately plans the path according to the relevant information, ensuring that the drone can be called for inspection as soon as possible after the pollutant exceeds the standard, thereby improving the emergency response speed. By optimizing the inspection order of the drone and aiming at the shortest inspection distance, the drone can complete the inspection of suspicious factories in the target geographical area in the shortest time, greatly improving the inspection efficiency.

[0060] Second, accurately locate the target pollution source: The above scheme combines the capacity weight and emission weight of the suspicious factory, and assigns a ranking weight to each suspicious factory by comprehensively considering the factory's historical emission data and capacity information. When planning the route, this weight is used as part of the optimization target, so that the drone can give priority to those factories that are more likely to emit excessive pollutants during the inspection process, thereby more accurately locating the target pollution source. During the inspection process, the backend server will further confirm the target pollution source based on the real-time data sent back by the drone, combined with the comparison of the predicted concentration of excessive pollutants and the monitored concentration, to ensure the accuracy of positioning.

[0061] Third, reduce the risk of misjudgment and missed judgment: By comprehensively considering the capacity weight and emission weight of the suspicious factories, as well as real-time data analysis, the above scheme effectively reduces the risk of misjudgment and missed judgment. Even when the monitored concentration of a single suspicious factory is significantly different from the predicted concentration, the system can still compare the cumulative monitored values ​​with the predicted average values, comprehensively consider the joint impact of multiple factories, and thus more accurately determine the target pollution source.

[0062] Fourth, improve resource utilization and reduce costs: By optimizing the inspection order and reducing unnecessary flight distances, the above solution significantly reduces the energy consumption and wear of drones, prolongs the service life of drones, and thus reduces inspection costs. At the same time, since the target pollution source can be located more quickly, the flight time and inspection frequency of drones are reduced, further improving resource utilization.

[0063] In a possible implementation, S104, calculating the capacity weight of the suspicious factory according to the capacity of the suspicious factory in the target geographical area, specifically includes: S201. The backend server performs data standardization processing on the production capacity of the suspicious factory to obtain the standardized production capacity of the suspicious factory.

[0064] Among them, since each suspicious factory produces different products, has different production capacity units and different production capacity magnitudes, the production capacity of each suspicious factory is converted into data of the same dimension through data standardization processing.

[0065] In some examples, the weight of pollutants exceeding the standard generated under the production capacity of the suspicious factory is obtained, and the weight of pollutants exceeding the standard generated is used as the standardized production capacity of the suspicious factory. For example, a steel plant needs to produce 10,000 tons of steel per month, burn a tons of coal, use b kilowatt-hours of electricity, and the weight of coal burned to generate b kilowatt-hours of electricity is c tons. The weight of sulfur dioxide generated by burning a+c tons of coal is determined.

[0066] In the above technical solution, since the production capacity of a suspicious factory is used to determine the possibility of pollution discharge from the suspicious factory, the possibility of pollution discharge from the suspicious factory can be accurately assessed by converting the production capacity of all suspicious factories into the weight of pollutants exceeding the standard.

[0067] Specifically, the data standardization of the production capacity of suspicious factories, especially the conversion of production capacity into the weight of pollutants exceeding the standard, has achieved the following technical effects: First, accurately assess the possibility of pollution discharge: By standardizing the production capacity of suspicious factories, especially converting it into the weight of pollutants that exceed the standard, we can more accurately assess the possibility of each factory's pollution discharge. The above scheme directly links the production capacity of the factory with the amount of pollutants it may produce, thus providing a more objective and scientific evaluation standard. Compared with relying solely on the absolute or relative value of production capacity to evaluate the possibility of pollution discharge, this conversion method takes into account the differences in pollutant generation in the production process of different factories, making the evaluation results more accurate.

[0068] Second, optimize the path planning strategy: The possibility of pollution discharge from a factory is an important consideration in drone path planning. By converting production capacity into the weight of pollutants exceeding the standard, we can more reasonably set the weight of each suspicious factory, and then take into account the actual pollution discharge risk of each factory when optimizing the inspection order. This helps drones give priority to visiting factories with a higher possibility of pollution discharge during the inspection process, so as to locate the target pollution source in a shorter time and improve inspection efficiency.

[0069] S202. The backend server calculates the sum of the standardized inverses of the production capacity of each suspicious factory in the target geographical area, and uses the ratio of the standardized inverse of the production capacity of the suspicious factory to the sum of the inverses of the production capacity as the production capacity weight of the suspicious factory.

[0070] In the target geographical area, the reciprocal of the standardized capacity of each suspicious factory is calculated, and the reciprocals of the standardized capacity of each suspicious factory are summed to obtain the sum of the reciprocals of the capacity. Taking 8 suspicious factories in the target geographical area as an example, the capacity of the 8 suspicious factories is standardized to obtain the standardized capacity of the 8 suspicious factories, which are marked as S1, S2, ..., S7, S8 respectively.

[0071] Calculate the sum of the standardized inverses of the production capacities of the eight suspicious factories: S=1 / S1+1 / S2+1 / S3+1 / S4+……+1 / S7+1 / S8 j is a positive integer, 1≤j≤8, and the capacity weight of suspicious factory j is: αj=(1 / Sj) / S Accordingly, S104, the backend server calculates the emission weight of the suspicious factory based on the historical emission data of the suspicious factory in the target geographical area, specifically including: S301. The backend server calculates the mean and variance of the historical emission data of the suspicious factories, calculates the sum of the inverses of the mean values ​​of the historical emission data of all the suspicious factories, and calculates the sum of the inverses of the variances of the historical emission data of all the suspicious factories.

[0072] For example, there are 8 suspicious factories in the target geographical area. The historical emission data of suspicious factory 1 is obtained, and the historical emission data of suspicious factory 1 is statistically analyzed to obtain the mean and variance of the historical emission data of suspicious factory 1. The mean values ​​are marked as μ1 and δ1. The same processing method is used for the remaining 7 suspicious factories.

[0073] The backend server calculates the reciprocal of the average value of the historical emission data of the eight suspicious factories, marked as 1 / μ1, 1 / μ2, ..., 1 / μ8, and sums the reciprocals of the average values ​​of the historical emission data of the eight suspicious factories to obtain the sum of the reciprocals of the average values ​​of the historical emission data. That is: μ=1 / μ1+1 / μ2+……1 / μ7+1 / μ8 The backend server calculates the inverse of the variance of the historical emission data of the eight suspicious factories, which are marked as 1 / δ1, 1 / δ2, ..., 1 / δ7, 1 / δ8 respectively, and sums the inverse of the variance of the historical emission data of the eight suspicious factories to obtain the sum of the inverse of the variance of the historical emission data. That is: δ=1 / δ1+1 / δ2+……1 / δ7+1 / δ8 S302: The backend server calculates a first ratio of the inverse of the average value of the historical emission data to the sum of the inverses of the average value, and calculates a second ratio of the inverse of the variance of the historical emission data to the sum of the inverses of the variance.

[0074] Among them, the first ratio of suspicious factory 1 is (1 / μ1) / μ, and the second ratio of suspicious factory 1 is (1 / δ1) / δ. The first ratio of suspicious factory 2 is (1 / μ2) / μ, and the second ratio of suspicious factory 2 is (1 / δ2) / δ. ..., the first ratio of suspicious factory 8 is (1 / μ8) / μ, and the second ratio of suspicious factory 8 is (1 / δ8) / δ.

[0075] S303: The backend server uses the average of the first ratio and the second ratio as the emission weight of the suspicious factory.

[0076] Among them, the emission weight of suspicious factory 1 is (1 / μ1) / μ+(1 / δ1) / δ, the emission weight of suspicious factory 2 is (1 / μ2) / μ+(1 / δ2) / δ, ..., and the emission weight of suspicious factory 8 is (1 / μ8) / μ+(1 / δ8) / δ.

[0077] In the above technical solution, the capacity weight is calculated based on the inverse of the capacity by finding the inverse of the capacity of each suspicious factory, so that the larger the capacity, the smaller the capacity weight, and the suspicious factory with a greater possibility of pollution discharge is inspected first while taking into account the shortest inspection distance. By statistically analyzing the historical emission data, the larger the average value of the emission data, the greater the possibility of pollution discharge, the smaller the emission weight, and the larger the variance, the greater the possibility of pollution discharge, and the smaller the emission weight. In this way, the suspicious factory with a greater possibility of pollution discharge is inspected first while taking into account the shortest inspection distance.

[0078] Specifically, the above scheme achieves technical effects by calculating the capacity weight and emission weight of suspicious factories, which are mainly reflected in the following aspects: First, optimize the inspection order and improve inspection efficiency: By taking the inverse of the production capacity of each suspicious factory and calculating the production capacity weight accordingly, it is ensured that the larger the production capacity of the factory, the smaller its production capacity weight. This means that when optimizing the inspection order, the system will give priority to inspecting those factories with large production capacity and relatively high possibility of pollution discharge. This design ensures that while taking into account the shortest inspection distance, factories with high emission risks can be inspected first, thereby improving the pertinence and efficiency of the inspection. Similarly, by performing statistical analysis on historical emission data, the emission weight is calculated, so that the larger the average value and variance of the emission data, the smaller the emission weight. This also helps to give priority to factories whose historical emission data show a greater possibility of pollution discharge when optimizing the inspection order.

[0079] Second, accurately locate pollution sources to reduce misjudgments and missed judgments: By comprehensively considering the capacity weight and emission weight, this method can more accurately assess the possibility of pollution discharge for each suspicious factory. This helps drones quickly locate the real pollution source during the inspection process, reducing repeated inspections and resource waste caused by misjudgment or omission. In addition, by analyzing the data sent back by drones in real time, combined with the capacity weight and emission weight, the accuracy of the pollution source can be further verified and confirmed, improving the accuracy and effectiveness of environmental supervision.

[0080] In a possible implementation, S105, the backend server calculates the patrol distance according to the ranking weights of the suspicious factories, the distances between the suspicious factories, and the patrol order of the suspicious factories, and optimizes the patrol order of the suspicious factories with the shortest patrol distance as the optimization goal, specifically including: S401, the backend server processes each solution vector in the previous solution space according to the solution vector corresponding to the shortest inspection distance in the previous solution space to generate a solution vector in the current solution space, wherein an element of each solution vector represents the inspection order of a suspicious factory in the target geographical area.

[0081] Here, we continue to use the example of 8 suspicious factories in the target geographical area, and a solution vector X1=[y1, y2, y3, ..., y8]. y1 represents the order of suspicious factory 1, y2 represents the order of suspicious factory 2, y3 represents the order of suspicious factory 3, ..., y8 represents the order of suspicious factory 8.

[0082] The value ranges of y1 to y8 are the same, which are all integers from 1 to 8, and any two values ​​of y1 to y8 are not equal. For example, a solution vector X=[3, 8, 1, 5, 7, 2, 4, 6] means that the ranking of suspicious factory 1 is 3, the ranking of suspicious factory 2 is 8, the ranking of suspicious factory 3 is 1, the ranking of suspicious factory 4 is 5, the ranking of suspicious factory 5 is 7, the ranking of suspicious factory 6 is 2, the ranking of suspicious factory 7 is 4, and the ranking of suspicious factory 8 is 6.

[0083] The solution space contains multiple solution vectors. The background server optimizes each solution vector in the previous solution space according to the solution vector corresponding to the shortest inspection distance in the previous solution space. For example, the elements in the solution vector corresponding to the shortest inspection path are partially replaced with the elements of the solution vector in the previous solution space to generate the solution vector in the current solution space. In this way, the solution space is updated, and the inspection distance corresponding to the updated solution vector is shorter.

[0084] S402. For each solution vector, let i be traversed from 1 to (N-1), calculate the product of the distance between the suspicious factory ranked i and the suspicious factory ranked i+1 and the ranking weight of the suspicious factory ranked i+1, and obtain the corrected distance ranked i+1. Calculate the product of the distance from the factory to the suspicious factory ranked 1 and the ranking weight of the suspicious factory ranked 1 to obtain the corrected distance ranked 1. Calculate the sum of N corrected distances as the inspection distance of the solution vector. 1≤i<N.

[0085] Among them, taking the above solution vector X1=[3, 8, 1, 5, 7, 2, 4, 6] as an example, suspicious factory 3 is the suspicious factory ranked 1, suspicious factory 6 is the suspicious factory ranked 2, suspicious factory 1 is the suspicious factory ranked 3, suspicious factory 7 is the suspicious factory ranked 4, suspicious factory 4 is the suspicious factory ranked 5, suspicious factory 8 is the suspicious factory ranked 6, suspicious factory 5 is the suspicious factory ranked 7, and suspicious factory 2 is the suspicious factory ranked 8.

[0086] Figure 3 is a schematic diagram of the inspection order corresponding to a solution space vector, such as Figure 3 As shown, the distance between the airport and the suspicious factory 3 is calculated by multiplying the ranking weight of the suspicious factory 3 to obtain the corrected distance between the airport and the suspicious factory 3, and the distance between the suspicious factory 3 and the suspicious factory 6 is calculated by multiplying the ranking weight of the suspicious factory 6 to obtain the corrected distance between the suspicious factory 3 and the suspicious factory 6. The distance between the suspicious factory 6 and the suspicious factory 1 is calculated by multiplying the ranking weight of the suspicious factory 1 to obtain the corrected distance between the suspicious factory 6 and the suspicious factory 1. The distance between the suspicious factory 1 and the suspicious factory 7 is calculated by multiplying the ranking weight of the suspicious factory 7 to obtain the corrected distance between the suspicious factory 1 and the suspicious factory 7.

[0087] Calculate the product of the distance from suspicious factory 7 to suspicious factory 4 and the ranking weight of suspicious factory 4 to obtain the corrected distance from suspicious factory 7 to suspicious factory 4. Calculate the product of the distance from suspicious factory 4 to suspicious factory 8 and the ranking weight of suspicious factory 8 to obtain the corrected distance from suspicious factory 4 to suspicious factory 8. Calculate the product of the distance from suspicious factory 8 to suspicious factory 5 and the ranking weight of suspicious factory 5 to obtain the corrected distance from suspicious factory 8 to suspicious factory 5. Calculate the product of the distance from suspicious factory 5 to suspicious factory 2 and the ranking weight of suspicious factory 2 to obtain the corrected distance from suspicious factory 5 to suspicious factory 2. The sum of the above 8 corrected distances is taken as the inspection distance of the solution vector.

[0088] For any solution vector in the current solution space, the inspection distance corresponding to the solution vector is calculated in the above manner.

[0089] S403, update the number of cycles, if the number of cycles is equal to the preset cycle threshold, select the solution vector with the smallest inspection distance from the current solution space as the optimal inspection order. If the number of cycles is less than the preset cycle threshold, return to S401.

[0090] The solution space is updated once, the inspection distance of the solution vector in the updated solution space is calculated, a cycle is completed, and the number of cycles is increased by 1 to obtain the updated number of cycles.

[0091] The preset cycle threshold is used to set the total number of cycles. If the number of cycles is equal to the preset cycle threshold, the set total number of cycles has been completed, and the optimization of the solution vector is completed, then the optimal patrol order is output. If the number of cycles is less than the preset cycle threshold, the set total number of cycles has not been completed, then it is necessary to continue to optimize the solution vector, and return to S401.

[0092] In the above technical solution, an element of each solution vector in the solution space represents the inspection order of a suspicious factory in the target geographical area. The distance between the suspicious factories is corrected using the ranking weights of each suspicious factory according to the order in the solution vector. The sum of the corrected distances is used as the inspection distance of the solution vector. The solution space is optimized using the solution vector corresponding to the shortest inspection distance in the previous solution space to obtain a better solution space. Through continuous cycles, an inspection order with the goal of searching for the target pollution source at a faster speed and with a shorter path is obtained, ensuring that the drone can search for the target pollution source faster and collect evidence in a timely manner.

[0093] Specifically, the above scheme optimizes the inspection order of suspicious factories through a specific algorithm, with the shortest inspection distance as the optimization goal. The technical effects achieved are mainly reflected in the following aspects: First, improve inspection efficiency and shorten inspection time: By continuously optimizing the solution vectors in the solution space, calculating the inspection distance corresponding to each solution vector, and selecting the solution vector with the smallest inspection distance as the optimal inspection order, the above scheme ensures that the drone can inspect along the optimal path, thereby significantly improving the inspection efficiency and shortening the inspection time. During the optimization process, by comprehensively considering the sorting weights of suspicious factories, the distances between suspicious factories, and the inspection order, the above scheme can ensure that the drone can prioritize the inspection of factories with a greater possibility of pollution discharge while minimizing the flight distance and achieving efficient inspection.

[0094] Second, accurately locate the pollution source and improve the efficiency of evidence collection: By optimizing the inspection order, the above solution can ensure that the drone can locate the target pollution source in a shorter time, thereby improving the efficiency of evidence collection. This is crucial for environmental protection and supervision, and can collect evidence in a timely manner to provide strong support for subsequent processing and punishment. In addition, by analyzing the data sent back by the drone in real time, combined with the optimized inspection order, the accuracy of the pollution source can be further verified and confirmed, ensuring the effectiveness and reliability of evidence collection.

[0095] Third, reduce operating costs and improve resource utilization: By optimizing the inspection order, the above solution can reduce the flight time and distance of the drone, thereby reducing the drone's energy consumption and wear, extending the drone's service life, and reducing operating costs. At the same time, since the inspection task can be completed in a shorter time, the above solution can also improve the efficiency of drone use, reduce unnecessary waiting and idle time, and improve resource utilization.

[0096] Fourth, enhance environmental supervision capabilities and improve supervision levels: By optimizing the inspection order, the above scheme can locate pollution sources more quickly and accurately, and improve the timeliness and effectiveness of environmental supervision.

[0097] In a possible implementation, S401, the backend server processes each solution vector in the previous solution space according to the solution vector with the shortest inspection distance in the previous solution space to generate a solution vector in the current solution space, specifically including: S501: The background server selects a solution vector with the shortest inspection distance from the previous solution space as a replacement vector.

[0098] There are 10 solution vectors in the solution space, marked as X1, X2, X3, X4, X5, X6, X7, X8, X9, and X10. X4 has the shortest inspection distance, so X4 is the replacement vector.

[0099] S502. For each solution vector in the previous solution space, the background server replaces the mth to Nth elements in the solution vector with the mth to Nth elements in the replacement vector, and then uses the mth to Nth elements to correct the 1st to m-1th elements in the solution vector so that the values ​​of each element in the solution vector are different, and randomly selects the values ​​of two elements in the solution vector to swap to obtain the solution vector in the contemporary solution space.

[0100] in, Figure 4 This is a schematic diagram of replacing and correcting the solution vector. Figure 4As shown in the figure, the process of element replacement and element correction is described by taking the solution vector X1 as an example, X1=[3, 8, 1, 5, 7, 2, 4, 6], and the replacement vector X4=[5, 2, 4, 7, 8, 6, 1, 3]. N is 8, m is set to 7, and the 7th and 8th elements of X1 are replaced with the 7th and 8th elements of X4, and the intermediate solution vector X1' is obtained as X1'=[3, 8, 1, 5, 7, 2, 1, 3]. Since the values ​​of the 7th and 3rd elements in the intermediate solution vector X1' are the same, and the value of the 8th element is the same as the value of the 1st element, the intermediate solution vector X1' lacks the two order values ​​4 and 6, so it is necessary to assign the two order values ​​4 and 6 to the 1st and 3rd elements so that the values ​​of each element in the solution vector are different. Then, the values ​​of the 5th element and the 7th element are selected for replacement to obtain a solution vector X'''=[6, 8, 4, 5, 1, 2, 7, 3] in the current solution space.

[0101] The above is just the processing process for one solution vector in the previous solution space. The processing process for other solution vectors in the previous solution space is similar and will not be repeated here. After performing the above processing on all solution vectors in the previous solution space, each solution vector in the current solution space can be obtained.

[0102] In the above technical solution, the solution vector with the shortest inspection distance is selected as the replacement vector in the previous solution space, that is, the current optimal solution is selected to process each solution vector in the previous solution space, so as to ensure that the solution vector in the current solution space is better. Make m>N / 2, and select any two elements in the corrected solution vector for replacement, so as to avoid the replacement vector from having too much influence on the solution vector in the previous solution space, so that the algorithm enters the local optimum too quickly, and improve the optimization performance of the algorithm.

[0103] Specifically, the above scheme generates a solution vector in the current solution space through a specific algorithm strategy to optimize the inspection order of suspicious factories. This technical scheme has shown significant technical effects, which are mainly reflected in the following aspects: First, improve the efficiency and accuracy of optimization: By selecting the solution vector with the shortest patrol distance (i.e., the current optimal solution) from the previous solution space as the replacement vector and using it to process other solution vectors, the above scheme ensures that the solution vectors in the current solution space are relatively optimized at the starting point. This strategy can effectively reduce the blindness of the algorithm search, improve the optimization efficiency, and enable the algorithm to converge to the global optimal solution or the local optimal solution close to the global optimal solution more quickly. At the same time, since the replacement vector itself represents a better patrol order, it can be modified and mutated based on it, so that the solution vectors in the current solution space can inherit excellent genes while having a certain diversity, thereby improving the accuracy of optimization.

[0104] The second aspect is to avoid premature convergence of the algorithm: In the process of correcting the solution vector, by making m>N / 2 (N is the number of suspicious factories) and randomly selecting two elements in the corrected solution vector for swapping, the above scheme effectively avoids the algorithm from falling into the local optimal solution too early. This strategy increases the randomness and exploratory nature of the algorithm, allowing the algorithm to search in a wider solution space and reduce the risk of premature convergence. Premature convergence is one of the common problems in optimization algorithms, which will cause the algorithm to stop searching before reaching the global optimal solution. Through the strategy of the above scheme, the algorithm can overcome this problem to a certain extent and continue to find a better solution.

[0105] Third, enhance the robustness of the algorithm: The replacement, modification and mutation operations in the above scheme are all based on certain rules and strategies, rather than being completely random. This design allows the algorithm to search in a predetermined direction while maintaining a certain degree of randomness, thereby enhancing the robustness of the algorithm. Robustness refers to the stability and reliability of the algorithm's performance when faced with different inputs or environmental changes. Through the application of the above scheme, the algorithm can maintain good optimization performance under different suspicious plant distributions and inspection requirements, providing strong support for practical applications.

[0106] In a possible implementation, S502, then using the mth element to the Nth element to correct the 1st element to the m-1th element in the solution vector, specifically includes: S601. After replacing the mth element to the Nth element in the solution vector with the mth element to the Nth element in the replacement vector, determine the missing order value according to the value of each element in the solution vector.

[0107] Among them, continue to Figure 4 Taking the processing process shown as an example, after replacing the 7th and 8th elements of X1 with the 7th and 8th elements of X4, the intermediate solution vector X1' is obtained as X1'=[3, 8, 1, 5, 7, 2, 1, 3]. Since the values ​​of the 7th and 3rd elements in the intermediate solution vector X1' are the same, and the value of the 8th element is the same as the value of the 1st element, the intermediate solution vector X1' lacks the two order values ​​4 and 6, so the two order values ​​4 and 6 need to be assigned to the 1st and 3rd elements.

[0108] S602. If there are multiple missing order values, obtain the ranking weight of the suspicious factory represented by the correction element, and correct the value of the correction element according to the ranking weight of the suspicious factory represented by the correction element, so that the smaller the ranking weight of the suspicious factory corresponding to the correction element, the smaller the value of the correction element.

[0109] Among them, in the solution vector X1, the first and third elements are correction elements, the first element represents the order of suspicious factory 1, and the third element represents the order of suspicious factory 3. If the ranking weight of suspicious factory 1 is less than the ranking weight of suspicious factory 3, it means that the probability of suspicious factory 1 generating a pollution source is greater, so the order value 4 is assigned to the first element and the order value 6 is assigned to the third element. The intermediate solution vector X'' is X''=[4, 8, 6, 5, 7, 2, 1, 3].

[0110] In the above technical scheme, when the remaining 1st to m-1th elements in the solution vector are corrected using the replaced mth to Nth elements, the element value is corrected according to the sorting weight of the suspicious factory represented by the element to be corrected. In this way, the solution vector can be closer to the optimal value, the search process for the optimal value can be accelerated, and the calculation time of the algorithm can be shortened.

[0111] Specifically, the present application modifies the elements in the solution vector through specific algorithm steps to ensure the uniqueness of each element in the solution vector and further optimize the patrol order. This technical solution brings significant technical effects, which are mainly reflected in the following aspects: First, improve the diversity and quality of solution vectors: By replacing some of the corresponding elements in the solution vector with the mth to Nth elements in the replacement vector and correcting the remaining elements according to the replacement results, the above scheme ensures the uniqueness of the elements in each solution vector and avoids the occurrence of duplicate elements. This increases the diversity of the solution space and provides more possibilities for the subsequent optimization process. At the same time, the element value is corrected according to the ranking weight of the suspicious plant represented by the correction element, so that the suspicious plant with a smaller ranking weight is positioned more forward in the solution vector, which helps to give priority to these plants with a greater possibility of pollution discharge during the optimization process, thereby improving the quality of the solution vector.

[0112] Second, speed up the search process for the optimal value: During the correction process, the above scheme fully considers the ranking weights of suspicious plants. By making plants with smaller ranking weights occupy a more forward position in the solution vector, the algorithm can be guided to converge to the global optimal solution or a local optimal solution close to the global optimal solution more quickly. This significantly accelerates the search process for the optimal value and shortens the calculation time of the algorithm. In addition, since the corrected solution vector is closer to the optimal value, the algorithm can explore the solution space more efficiently in the subsequent iteration process, further improving the optimization efficiency.

[0113] Third, improve the optimization performance and stability of the algorithm: By combining the replacement and correction strategies, the above scheme improves the diversity of solution vectors while ensuring the stability of the algorithm during the optimization process. The replacement strategy introduces a new search direction for the algorithm, while the correction strategy ensures the continuity and consistency of the algorithm during the search process. This combination enables the algorithm to maintain high search efficiency and search quality when exploring the solution space, thereby improving the algorithm's optimization performance and stability.

[0114] Fourth, optimize the inspection order and improve inspection efficiency: Finally, by optimizing the order of elements in the solution vector, the above scheme can generate a more reasonable inspection order. This helps the drone to cover all suspicious factories in a shorter time and distance during the inspection process, thereby improving the inspection efficiency. At the same time, since the optimization of the inspection order is based on the ranking weight of the suspicious factories, it can ensure that the drone will give priority to factories with a greater possibility of pollution during the inspection process, thereby more effectively discovering and dealing with environmental problems.

[0115] In a possible implementation, S103, determining a target geographical area from a plurality of industrial parks near the meteorological monitoring station according to the information of pollutants exceeding the standard, the location of the meteorological monitoring station, and the wind speed and direction information, specifically includes: S701. Obtain the relative position between the industrial park surrounding the meteorological monitoring station and the meteorological monitoring station.

[0116] in, Figure 5 A schematic diagram of selecting a target geographical area from multiple industrial parks provided in some embodiments of the present application, such as Figure 5 As shown in the figure, the meteorological monitoring station O detected that the concentration of sulfur dioxide in the atmosphere exceeded the standard. From the map data stored locally or in the cloud, it was determined that there were three industrial parks near the meteorological monitoring station O, which were marked as Industrial Park A, Industrial Park B, and Industrial Park C. Industrial Park A is located to the southeast of the meteorological monitoring station O, and Industrial Parks B and Industrial Parks C are located to the northwest of the meteorological monitoring station O.

[0117] S702. According to the wind speed and direction information and the relative position between the industrial park and the meteorological monitoring station, select the industrial park with the same relative position and wind speed and direction information as the meteorological monitoring station as the suspicious geographical area.

[0118] Among them, Figure 5 In the survey, when the meteorological monitoring station O detected that the sulfur dioxide concentration exceeded the standard, there was a northwest wind. Industrial Parks B and C were located northwest of the meteorological monitoring station O, so Industrial Parks B and C were regarded as suspicious geographical areas.

[0119] S703. Calculate the violation probability of the suspicious geographical area based on the historical emission data and the information of pollutants exceeding the standard in the suspicious geographical area, and select the suspicious geographical area with the highest violation probability as the target geographical area from the suspicious geographical areas.

[0120] The historical emission data of the suspected geographical area are obtained, and the information of the pollutants exceeding the standard includes the types of pollutants exceeding the standard. The emission data of the pollutants exceeding the standard are extracted from the historical emission data, and the emission data of the pollutants exceeding the standard are statistically analyzed, and the probability of violation of the suspected geographical area is determined according to the statistical analysis results.

[0121] More specifically, the mean probability is calculated based on the average emission of pollutants exceeding the standard. The variance probability is calculated based on the variance of the emission of pollutants exceeding the standard. The average of the mean probability and the variance probability is calculated to obtain the violation probability of the suspicious geographical area.

[0122] If the average amount of pollutants exceeding the standard is large, the probability of violation is greater. If the average amount of pollutants exceeding the standard is small, the probability of violation is smaller. If the variance of pollutants exceeding the standard is large, the probability of violation is greater. If the variance of pollutants exceeding the standard is small, the probability of violation is smaller.

[0123] After calculating the violation probability for each suspicious geographical area in the above manner, the geographical area with the highest violation probability is selected from multiple suspicious geographical areas as the target geographical area.

[0124] In the above technical scheme, after determining that a certain meteorological monitoring station has detected pollutants exceeding the standard, the suspicious geographical area is circled based on the wind direction and speed information of the meteorological monitoring station, and then the target geographical area with the highest probability of violation is selected based on the historical emission data and the information on pollutants exceeding the standard in the suspicious geographical area. Inspections are carried out in the target geographical area, so that the pollution source can be located quickly and time for on-site evidence collection can be gained.

[0125] Specifically, the above solution is used to determine the target geographical area to respond to pollutant exceeding the standard event. This strategy integrates the information of pollutants exceeding the standard, the location of meteorological monitoring stations, wind speed and direction information, and historical emission data of industrial parks. The technical effects achieved by this technical solution are specifically reflected in the following aspects: First, quickly locate the pollution source: By analyzing the wind direction and speed information from the meteorological monitoring station, the above scheme can quickly identify industrial parks that may be related to the pollution source, that is, suspicious geographical areas. This method is based on the principle of atmospheric diffusion and effectively narrows the search range, making the location of the pollution source faster and more accurate. After determining the suspicious geographical area, the historical emission data and excessive pollutant information of the area are further combined to calculate the probability of violation, so as to select the area with the highest probability of violation as the target geographical area. This process further improves the accuracy of positioning and helps to quickly lock in the pollution source.

[0126] Second, improve the efficiency of on-site evidence collection: Rapidly locating the pollution source means being able to arrive at the scene in a shorter time to collect evidence. This is crucial for environmental protection and supervision, as timely evidence collection is the basis for subsequent processing and punishment. Through the above solution, relevant departments can respond quickly after the pollutant exceeds the standard, dispatch drones or other inspection equipment to the target geographical area for inspection, thereby collecting key evidence in the shortest time and providing strong support for subsequent work.

[0127] Third, optimize resource allocation: When faced with multiple industrial parks and limited regulatory resources, the above solution provides an effective strategy to optimize resource allocation. By accurately locating the target geographic area, it can ensure that regulatory resources are concentrated in the areas where illegal emissions are most likely to occur, thereby improving regulatory efficiency. At the same time, since the above solution can reduce unnecessary inspections and investigations, it also helps to reduce regulatory costs and improve resource utilization.

[0128] In a possible implementation, S106, the backend server controls the drone to inspect each suspicious factory in turn according to the optimal inspection order, and determines the target pollution source from each suspicious factory according to the inspection results, specifically including: S801. The backend server predicts the concentration of pollutants exceeding the standard at the suspected factory based on the distance between the suspected factory and the meteorological monitoring station, wind speed and direction information, and the concentration of pollutants exceeding the standard.

[0129] Among them, the diffusion table of excessive pollutants at different concentrations, different wind speeds, and different diffusion distances is obtained through experimental measurement. The distance between the suspected factory and the meteorological monitoring station, wind speed, and excessive pollutant concentration are used to query the diffusion table to obtain the predicted concentration of excessive pollutants at the suspected factory.

[0130] S802. The backend server obtains the monitoring concentration of pollutants exceeding the standard collected by the drone when inspecting the suspicious factory. When the inspected suspicious factory is determined to be a target pollution source based on the monitoring concentration and predicted concentration of pollutants exceeding the standard in the inspected suspicious factory, a stop inspection instruction is sent to the drone.

[0131] Among them, the drone patrols the suspicious factories in the target geographical area according to the optimal patrol order. When passing each suspicious factory, it collects the monitoring concentration of the pollutants exceeding the standard of the suspicious factory and sends the monitoring concentration of the pollutants exceeding the standard of the suspicious factory to the background server.

[0132] Compare the monitored concentration and predicted concentration of pollutants exceeding the standard in the suspected factory, and determine whether the suspected factory is a target pollution source based on the comparison result. If the suspected factory is a target pollution source, send a stop inspection command to the drone and the drone stops the inspection.

[0133] In the above technical solution, when the drone is patrolling in the optimal patrol order, the monitored concentration of pollutants exceeding the standard in the current suspicious factory is sent to the background server in real time. The background server compares the monitored concentration of pollutants exceeding the standard with the predicted concentration, and determines the target pollution source according to the comparison result. After finding the target pollution source, the drone is controlled to stop patrolling, thereby reducing the patrol mission volume of the drone.

[0134] Specifically, the above scheme effectively determines the target pollution source and controls the drone's patrol process by combining the optimal patrol order, predicted concentration and real-time monitoring data. The technical effects achieved are mainly reflected in the following aspects: First, improve inspection efficiency: By pre-calculating and applying the optimal inspection order, the drone can inspect each suspicious factory in a scientific and reasonable path, avoiding the time waste and resource consumption caused by disorderly inspections. This strategy significantly improves inspection efficiency and enables drones to cover more suspicious factories in a shorter time.

[0135] Second, accurately locate the target pollution source: This solution uses meteorological data and historical emission data to predict the concentration of pollutants exceeding the standard at suspicious factories, providing an important reference for drone inspections. During the inspection process, the drone collects the monitored concentration of pollutants exceeding the standard at suspicious factories in real time and compares it with the predicted concentration. When the difference between the monitored concentration and the predicted concentration is less than the preset threshold, the factory can be determined as a target pollution source. This method greatly improves the positioning accuracy of the target pollution source and reduces the possibility of misjudgment and missed judgment.

[0136] Third, reduce the amount of patrol tasks: Once the target pollution source is identified, the drone will immediately stop patrolling and return to the base, thus avoiding unnecessary patrol missions. This instant stop mechanism effectively reduces the number of patrol missions for the drone, extending the drone's endurance and service life. At the same time, it also reduces patrol costs and maintenance costs.

[0137] Fourth, enhance environmental supervision capabilities: Through the implementation of this plan, environmental regulatory authorities will be able to detect and deal with excessive emissions more quickly and accurately, thereby effectively curbing environmental pollution.

[0138] In a possible implementation, S802, determining whether the inspected suspicious factory is a target pollution source according to the monitored concentration and predicted concentration of the pollutants exceeding the standard in the inspected suspicious factory, specifically includes: S901. For the current suspicious factory, compare the monitored concentration of the pollutants exceeding the standard of the current suspicious factory with the predicted concentration. If the single difference is less than the first preset threshold, the suspicious factory is determined to be the target pollution source; if the single difference is greater than the first preset threshold, enter S902.

[0139] After receiving the inspection order, the drone will inspect each suspicious factory in turn according to the optimal inspection order. After inspecting each suspicious factory, the drone will collect the monitoring concentration of pollutants exceeding the standard in the suspicious factory and return the monitoring concentration of pollutants exceeding the standard in the suspicious factory to the backend server.

[0140] The backend server compares the monitored concentration of the pollutants exceeding the standard of the suspected factory with the predicted concentration of the pollutants exceeding the standard at the suspected factory, calculates a single difference between the monitored concentration and the predicted concentration, and if the single difference is less than a first preset threshold, determines that the suspected factory is a target pollution source. The backend server sends a stop patrol instruction to the drone, causing the drone to stop patrolling.

[0141] Taking the optimal inspection order of [6, 8, 4, 5, 1, 2, 7, 3] as an example, the drone first inspects the suspicious factory 5 and sends the monitored concentration of the pollutants exceeding the standard in the suspicious factory 5 to the background server. The background server obtains the predicted concentration of the pollutants exceeding the standard in the suspicious factory 5, calculates the single difference between the monitored concentration and the predicted concentration of the pollutants exceeding the standard in the suspicious factory 5, and if the single difference is less than the first preset threshold, the suspicious factory 5 is determined to be the target pollution source.

[0142] S902. Determine whether the monitored concentration of pollutants exceeding the standard is greater than the standard emission concentration. If so, calculate the cumulative monitored value of pollutants exceeding the standard, and compare the cumulative monitored value with the predicted average value. If the difference is less than the second preset threshold, the suspicious factory corresponding to the cumulative monitored value is the target pollution source.

[0143] The predicted average value is the average value of the predicted concentration of the suspicious factory corresponding to the cumulative monitoring value. For example, if the cumulative monitoring value is the monitoring value corresponding to suspicious factory 5 and suspicious factory 6, the predicted average value is the average value of the predicted concentration corresponding to suspicious factory 5 and suspicious factory 6. As the cumulative monitoring value increases as the drone continues to inspect more suspicious factories, the predicted average value also changes.

[0144] If the single difference value corresponding to suspicious factory 5 is greater than the first preset threshold, the standardized emission concentration of the pollutants exceeding the standard is obtained, and the monitoring value of the pollutants exceeding the standard of suspicious factory 5 is compared with the standardized emission concentration. If the monitoring value is greater than the standardized emission concentration, the cumulative monitoring value is calculated. Suspicious factory 1 is ranked first, so the cumulative monitoring value is the monitoring concentration of the pollutants exceeding the standard of suspicious factory 5. The predicted average value is the predicted concentration of the pollutants exceeding the standard of suspicious factory 5. The difference between the cumulative monitoring value and the predicted average value is greater than the second preset threshold, and suspicious factory 5 is not the target pollution source, so the inspection of suspicious factory 6 continues.

[0145] The drone continues to patrol suspicious factory 6 and sends the monitored concentration of pollutants exceeding the standard in suspicious factory 6 to the background server. The background server obtains the predicted concentration of pollutants exceeding the standard in suspicious factory 6. If the single difference value corresponding to suspicious factory 6 is greater than the first preset threshold, the standardized emission concentration of pollutants exceeding the standard is obtained, and the monitored value of pollutants exceeding the standard in suspicious factory 6 is compared with the standardized emission concentration. If the monitored value is greater than the standardized emission concentration, the cumulative monitored value is calculated. The cumulative monitored value is the sum of the monitored concentration of pollutants exceeding the standard in suspicious factory 5 and the monitored concentration of pollutants exceeding the standard in suspicious factory 6. The predicted average value is the average value of the predicted concentration of pollutants exceeding the standard in suspicious factory 5 and the predicted concentration of pollutants exceeding the standard in suspicious factory 6. If the difference between the cumulative monitored value and the predicted average value is less than the second preset threshold, suspicious factory 5 and suspicious factory 6 are both target pollution sources, and the drone does not need to continue patrolling the remaining suspicious factories.

[0146] S903: If there is no target pollution source in the target geographical area, the backend server selects other suspicious geographical areas as the target geographical areas according to the violation probability of the suspicious geographical areas.

[0147] If it is determined that there is no target pollution source in the target geographical area, it is necessary to select other suspicious geographical areas for inspection until the location of the target pollution source is obtained. Before inspecting other suspicious geographical areas, it is also necessary to obtain the optimal inspection order according to the inspection order planning method described above.

[0148] In the above technical scheme, in the above technical scheme, when determining the target pollution source from multiple suspicious factories, the impact of emissions from a single suspicious factory is judged. If the emissions from a single suspicious factory do not meet the conditions, the cumulative detection value is calculated, considering the joint impact of multiple factories. In this way, the target pollution source can be accurately selected from the suspicious factories, reducing the risk of misjudgment and missed judgment.

[0149] Specifically, the above scheme significantly improves the accuracy and reliability of the target pollution source identification by comprehensively evaluating the monitored concentrations and predicted concentrations of pollutants exceeding the standard in the suspected factories that have been inspected, and considering the joint impact of multiple factories. The specific technical effects are reflected in the following aspects: First, improve the accuracy of identifying target pollution sources: By directly comparing the monitored concentration of pollutants exceeding the standard of the current suspected factory with the predicted concentration, if the single difference is less than the first preset threshold, the factory is immediately determined to be the target pollution source. This method is direct and efficient, and can quickly lock in the source of excessive emissions. When the single difference is greater than the first preset threshold, the method does not immediately exclude the factory, but further determines whether the monitored concentration of its excessive pollutants exceeds the standard emission concentration. If it exceeds, the joint impact of multiple factories is considered, and the cumulative monitoring value is calculated and compared with the predicted average value to determine whether there are multiple factories jointly causing excessive emissions. This method effectively avoids the situation where a single factory has weak emissions but multiple factories jointly exceed the emission standard and is ignored, and improves the comprehensiveness of the determination of target pollution sources.

[0150] Second, reduce the risk of misjudgment and missed judgment: Traditional pollution source identification methods often rely on data from a single monitoring point, which is easily affected by accidental factors or local environmental influences and leads to misjudgment. The method of the present application reduces the possibility of misjudgment by comprehensively considering the data and joint influence of multiple factories. At the same time, by calculating the cumulative monitoring values ​​and comparing them with the predicted average value, the above scheme can more accurately reflect the emissions of multiple factories and reduce the risk of missed judgments due to abnormal or erroneous data from individual factories.

[0151] Figure 6 This is a schematic diagram of the structure of a drone path planning system based on machine learning provided in some embodiments of the present application. Figure 6 As shown, the UAV path planning system 600 based on machine learning provided in this embodiment includes: The receiving module 610 is used to receive the inspection instruction sent by the meteorological server, wherein the inspection instruction is sent by the meteorological server after determining that the pollutants exceed the standard according to the monitoring information of the meteorological monitoring station, and the inspection instruction includes the information of the pollutants exceeding the standard monitored by the meteorological monitoring station, the location of the meteorological monitoring station and the wind speed and direction information of the meteorological monitoring station; A processing module 620 is used to determine a target geographical area from an industrial park near a meteorological monitoring station according to the excessive pollutant information, the location of the meteorological monitoring station and the wind speed and direction information; Calculating the capacity weight of the suspicious factory according to the capacity of the suspicious factory in the target geographical area, calculating the emission weight of the suspicious factory according to the historical emission data of the suspicious factory in the target geographical area, calculating the average of the capacity weight and emission weight of the suspicious factory, and obtaining the ranking weight of the suspicious factory; The processing module 620 is further used to calculate the patrol distance according to the ranking weight of the suspicious factories, the distance between the suspicious factories and the patrol order of the suspicious factories, and optimize the patrol order of the suspicious factories with the shortest patrol distance as the optimization goal; The control module 630 is used to control the UAV to inspect each suspicious factory in turn according to the optimal inspection order, determine the target pollution source from each suspicious factory according to the inspection results, and send a stop inspection instruction to the UAV.

[0152] The background server provided in this embodiment includes: at least one processor and a memory. Optionally, the device also includes a communication component. The processor, the memory and the communication component are connected via a bus.

[0153] In a specific implementation process, at least one processor executes computer-executable instructions stored in a memory, so that at least one processor executes the above method.

[0154] The specific implementation process of the processor can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.

[0155] In the above embodiments, it should be understood that the processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the invention can be directly implemented as a hardware processor, or can be implemented by a combination of hardware and software modules in the processor.

[0156] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (NVM), such as at least one disk storage.

[0157] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of this application is not limited to only one bus or one type of bus.

[0158] The present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0159] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above method is implemented.

[0160] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0161] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.

[0162] The division of units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0163] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0164] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0165] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0166] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.

[0167] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art not disclosed by the present invention, are not limited to the precise structure described above and shown in the drawings, and may be modified and changed in various ways without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for UAV path planning based on machine learning, characterized in that: The method is applied to a backend server, and the method comprises: Receiving an inspection instruction sent by a meteorological server, wherein the inspection instruction is sent by the meteorological server after determining that pollutants exceed the standard according to monitoring information of the meteorological monitoring station, and the inspection instruction includes information on pollutants exceeding the standard monitored by the meteorological monitoring station, the location of the meteorological monitoring station, and wind speed and direction information of the meteorological monitoring station; Determine a target geographical area from an industrial park near a weather monitoring station according to the excessive pollutant information, the location of the weather monitoring station, and the wind speed and direction information; Calculating the capacity weight of the suspicious factory according to the capacity of the suspicious factory in the target geographical area, calculating the emission weight of the suspicious factory according to the historical emission data of the suspicious factory in the target geographical area, calculating the average of the capacity weight and emission weight of the suspicious factory, and obtaining the ranking weight of the suspicious factory; Calculate the patrol distance according to the ranking weights of the suspicious factories, the distances between the suspicious factories, and the patrol order of the suspicious factories, and optimize the patrol order of the suspicious factories with the shortest patrol distance as the optimization goal; The drone is controlled to inspect each suspicious factory in the optimal inspection order, and the target pollution source is determined from each suspicious factory according to the inspection results, and a stop inspection command is issued to the drone.

2. The UAV path planning method according to claim 1, characterized in that: The capacity weight of the suspicious factory is calculated based on the capacity of the suspicious factory in the target geographical area, specifically including: Performing data standardization processing on the production capacity of the suspicious factory to obtain the standardized production capacity of the suspicious factory; Calculating the sum of the standardized inverses of the production capacity of all the suspicious factories, and taking the ratio of the standardized inverse of the production capacity of the suspicious factory to the sum of the inverses of the production capacity as the production capacity weight of the suspicious factory; Accordingly, the emission weight of the suspicious factory is calculated based on the historical emission data of the suspicious factory in the target geographical area, specifically including: Calculate the mean and variance of the historical emission data of the suspected factory, calculate the sum of the inverse of the mean of the historical emission data of all the suspected factories, and calculate the sum of the inverse of the variance of the historical emission data of all the suspected factories; Calculating a first ratio of the inverse of the mean value of the historical emission data to the sum of the inverses of the mean values; calculating a second ratio of the inverse of the variance of the historical emission data to the sum of the inverses of the variances; The average value of the first ratio and the second ratio is used as the emission weight of the suspicious factory.

3. The UAV path planning method according to claim 2, characterized in that: Performing data standardization processing on the production capacity of the suspicious factory to obtain the standardized production capacity of the suspicious factory specifically includes: The weight of pollutants exceeding the standard generated when the production capacity of the suspected factory is completed is obtained, and the weight of pollutants exceeding the standard generated is used as the standardized production capacity of the suspected factory.

4. The UAV path planning method according to any one of claims 1 to 3, characterized in that: The inspection distance is calculated based on the ranking weights of suspicious factories, the distances between suspicious factories, and the inspection order of suspicious factories. The inspection order of suspicious factories is optimized with the shortest inspection distance as the optimization goal. Specifically, it includes: Processing each solution vector in the previous solution space according to the solution vector corresponding to the shortest inspection distance in the previous solution space to generate a solution vector in the current solution space, wherein an element in each solution vector represents the inspection order of a suspicious factory in the target geographical area; For each of the solution vectors, let i be traversed from 1 to (N-1), calculate the product of the distance between the suspicious factory ranked i and the suspicious factory ranked i+1 and the ranking weight of the suspicious factory ranked i+1, and obtain the corrected distance ranked i+1; calculate the product of the distance from the factory to the suspicious factory ranked 1 and the ranking weight of the suspicious factory ranked 1, and obtain the corrected distance ranked 1; calculate the sum of N corrected distances as the inspection distance of the solution vector; 1≤i≤(N-1); Update the number of cycles. If the number of cycles is equal to the preset cycle threshold, output the solution vector with the smallest inspection distance as the optimal inspection order. If the number of cycles is less than the preset cycle threshold, continue to update the solution space.

5. The UAV path planning method according to any one of claims 1 to 3, characterized in that: Determining a target geographical area from a plurality of industrial parks according to the excessive pollutant information, the location of the weather monitoring station, and the wind speed and direction information specifically includes: Obtaining the relative position between the industrial park surrounding the weather monitoring station and the weather monitoring station; According to the wind speed and direction information and the relative position between the industrial park and the weather monitoring station, select the industrial park with the same relative position and wind direction as the weather monitoring station as the suspicious geographical area; The violation probability of the suspicious geographical area is calculated based on the historical emission data and the excessive pollutant information of the suspicious geographical area, and the suspicious geographical area with the highest violation probability is selected from the suspicious geographical areas as the target geographical area.

6. The UAV path planning method according to claim 5, characterized in that: Control the drone to inspect each suspicious factory in the optimal inspection order, and determine the target pollution source from each suspicious factory based on the inspection results, including: Predicting the concentration of pollutants exceeding the standard at the suspected factory based on the distance between the suspected factory and the meteorological monitoring station, the wind speed and direction information, and the concentration of pollutants exceeding the standard; Obtain the monitoring concentration of pollutants exceeding the standard collected by the drone when inspecting the suspicious factory. When the inspected suspicious factory is determined to be a target pollution source based on the monitoring concentration and predicted concentration of pollutants exceeding the standard in the inspected suspicious factory, send a stop inspection command to the drone.

7. The UAV path planning method according to claim 6, characterized in that: Determine whether the suspected factory inspected is a target pollution source based on the monitored concentration and predicted concentration of pollutants exceeding the standard in the suspected factory inspected, including: For the current suspicious factory, the monitored concentration of the pollutants exceeding the standard of the current suspicious factory is compared with the predicted concentration. If the single difference is less than the first preset threshold, the suspicious factory is determined to be the target pollution source; if the single difference is greater than the first preset threshold, it is determined whether the monitored concentration of the pollutants exceeding the standard is greater than the standard emission concentration. If so, the cumulative monitoring value of the pollutants exceeding the standard is calculated, and the cumulative monitoring value is compared with the predicted average value. If the difference is less than the second preset threshold, the suspicious factory corresponding to the cumulative monitoring value is the target pollution source.

8. A UAV path planning system based on machine learning, characterized in that: include: A receiving module, used to receive an inspection instruction sent by a meteorological server, wherein the inspection instruction is sent by the meteorological server after determining that the pollutants exceed the standard according to the monitoring information of the meteorological monitoring station, and the inspection instruction includes the information of the pollutants exceeding the standard monitored by the meteorological monitoring station, the location of the meteorological monitoring station and the wind speed and direction information of the meteorological monitoring station; A processing module, for determining a target geographical area from an industrial park near a meteorological monitoring station according to the excessive pollutant information, the location of the meteorological monitoring station and the wind speed and direction information; Calculating the capacity weight of the suspicious factory according to the capacity of the suspicious factory in the target geographical area, calculating the emission weight of the suspicious factory according to the historical emission data of the suspicious factory in the target geographical area, calculating the average of the capacity weight and emission weight of the suspicious factory, and obtaining the ranking weight of the suspicious factory; The processing module is further used to calculate the patrol distance according to the ranking weights of the suspicious factories, the distances between the suspicious factories and the patrol order of the suspicious factories, and optimize the patrol order of the suspicious factories with the shortest patrol distance as the optimization goal; The control module is used to control the UAV to inspect each suspicious factory in turn according to the optimal inspection order, determine the target pollution source from each suspicious factory according to the inspection results, and send a stop inspection instruction to the UAV.

9. An electronic device, characterized in that: include: processor; as well as, A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images