A Wildfire Identification and Inspection Method Based on Dynamic Path Planning

Abstract

This invention relates to a wildfire identification and inspection method based on dynamic path planning. Firstly, environmental information is monitored in real time using environmental detectors. During the inspection process, the drone can retrieve monitoring information via short-range communication, ensuring that on-site environmental data is acquired. This allows for real-time monitoring of the inspection area, avoiding coverage issues caused by signal limitations. Secondly, environmental information can be analyzed to plan future inspections, allowing for timely increases in the number of inspections and dynamic adjustments to the frequency of inspections in high-risk locations. Furthermore, the drone is controlled using an intelligent identification and communication algorithm, preventing loss of control due to remote control signal risks that could hinder inspection work. Finally, the drone uses on-site environmental detectors as landmarks, enabling inspections in complex terrain.

Description

Technical Field

[0001] This invention relates to the field of power grid inspection technology, and more specifically, to a wildfire identification and inspection method based on dynamic path planning. Background Technology

[0002] Currently, with the increasing demand for electricity in various regions, the safety of the power grid has become a key concern. Due to the optimization and upgrading of new energy power generation technologies, many power plants are located in remote areas due to their high environmental requirements. This results in the need to lay power grid facilities for hundreds or even thousands of kilometers. However, if the power grid facilities experience aging or abnormalities, it will affect the entire power grid. Since the remoteness and the need for high-altitude operations make manual inspections inconvenient, drones are now being widely used for inspections. Wildfire identification and inspection are high-risk inspection subjects, and signal coverage in most areas is problematic, making drone-based wildfire identification challenging. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide a wildfire identification and inspection method based on dynamic path planning.

[0004] To solve the above-mentioned technical problems, the technical solution of the present invention is: a wildfire identification and inspection method based on dynamic path planning, comprising: Step S1: Configure the initial commands on the inspection drone; Step S2: Control the inspection drone to move to the inspection position according to the initial command. Each inspection position corresponds to an environmental monitor. Step S3: The inspection drone obtains flight instructions and on-site inspection information from the environmental monitor, and flies to the next inspection location according to the flight instructions. Step S4: Repeat step S3 until the inspection drone reaches the parking platform; Step S5: The inspection drone uploads the on-site inspection information to the controller of the parking platform; Step S6: The controller uploads the field detection information to the management backend; Step S7: The management backend processes the on-site detection information through a preset anomaly analysis strategy to generate the fire risk level for each inspection location; Step S8: Generate new initial instructions based on the distribution of fire risk levels, and return to step S1.

[0005] Furthermore, each environmental monitor is pre-configured with a dynamic routing table, which stores several flight commands. Each flight command is indexed by the dynamic code of the previous inspection position, and the dynamic code has a time change factor to make the dynamic code change. Both step S2 and step S3 include obtaining dynamic codes from an environmental monitor; Step S3 also includes sending the dynamic code obtained from the previous inspection location to the environmental monitor to obtain the corresponding flight command.

[0006] Furthermore, step S8 also includes generating a flight path planning network through a preset path network planning strategy, wherein the flight path planning network includes a combination of several flight planning paths; Step S1 includes generating path update information based on the flight planning path and configuring it on the corresponding inspection drone; Step S3 also includes the inspection drone sending path update information to the corresponding environmental monitor, and the environmental monitor updating the corresponding dynamic routing table when it receives the path update information. The path network planning strategy includes Step A1: Obtain historical fire sample information; Step A2: Compare the correlation between the information of each historical fire sample and the fire situation in the pre-constructed virtual model of the mountain in the inspection area. ,in, For fire-related information, For the first The relevant values ​​of each fire-related item For the first The preset weights for each fire-related item This represents the total number of items related to the fire situation. Step A3: Calculate the fire risk value of the inspection area. ,in, This represents the fire risk value. For the first The correlation between historical fire sample information and fire situation information For the first The fire spread value in a historical fire sample information, wherein the fire spread value reflects the spread rate of the fire in the historical fire. For the first The fire point concealment value is a historical fire incident sample information, which reflects the degree of concealment of fire points in historical fire incidents. The total number of historical fire incident sample information; Step A4: Retrieve the corresponding inspection task from the pre-built inspection task database in the background according to the fire risk value. The inspection task includes several sets of inspection task parameters and inspection triggering conditions corresponding to the inspection task parameters. Step A5: Determine whether the inspection triggering conditions are met by analyzing the distribution of fire risk levels in order to retrieve the corresponding inspection task parameters; Step A6: Input the inspection task parameters into the mountain virtual model to generate the corresponding flight path planning network.

[0007] Furthermore, it also includes a pre-constructed environmental information simulation model of the inspection area, which is used to generate detection simulation information for each inspection location in real time based on the environmental information; the static factors of the environmental information simulation model include vegetation cover factor, terrain factor, and geographical location factor; the environmental information includes climate data, sunshine data, and time data; the on-site detection information includes humidity data and temperature data; The anomaly analysis strategy includes a difference analysis sub-strategy, which is used to calculate the environmental anomaly difference between field detection information and detection simulation information. ,in, The environmental anomaly difference, This refers to the temperature change waveform from the field detection information. To detect the temperature change waveform in the simulation information, This is a waveform representing humidity changes from on-site monitoring data. To detect the humidity change waveform in the simulation information, As a preset temperature difference weight, The fire risk level and the environmental anomaly difference are positively correlated with the preset humidity difference weight.

[0008] Furthermore, the inspection drone is equipped with a thermal imager, and the controller has a pre-configured image recognition algorithm. During flight, the inspection drone uses the thermal imager to capture thermal images. Step S5 also includes the inspection drone uploading thermal imaging images to the controller of the parking platform; Step S6 also includes the controller using an image recognition algorithm to identify fire anomalies in the thermal imaging image and mark the corresponding anomalies to generate thermal imaging anomaly information, and the controller uploads the thermal imaging anomaly information to the management backend. The anomaly analysis strategy includes an anomaly retrieval sub-strategy. This sub-strategy retrieves corresponding anomaly values ​​from a pre-built type information table based on the anomaly type of fire anomaly points in the thermal imaging anomaly information, and calculates the type anomaly value corresponding to each inspection location based on the anomaly location. , For the type of abnormal value of the inspection location, For the first Identifying outliers at individual fire anomaly points For the first The distance values ​​between each fire anomaly point and the inspection location. This represents the total number of abnormal fire locations.

[0009] Furthermore, step S4 also includes the following: the inspection drone is configured with a safe flight value. If the inspection drone fails to establish communication with the environmental monitor at the corresponding location when its flight distance reaches the safe flight value, the inspection drone is controlled to execute a preset rapid return-to-home strategy. The rapid return-to-home strategy includes... Step B1: Control the inspection drone to return to the previous inspection position according to the flight path of the inspection drone; Step B2: Send a return-to-home command to the corresponding environmental monitor. The environmental monitor is pre-configured with return-to-home route information. When the environmental monitor receives the return-to-home command, it outputs the corresponding return-to-home route information to the inspection drone. Step B3: The inspection drone flies to the next inspection location based on the return route information; Step B4: Repeat step B2 until the inspection drone reaches the parking platform; Step B5: The inspection drone sends the last flight command it received to the parking platform; Step B6: The controller generates the loss of contact anomaly information for the corresponding inspection location based on the received flight command and uploads it to the management backend; The fire risk level and the abnormal information about being out of contact are positively correlated.

[0010] Furthermore, the shutdown platform also includes a capture signal transmitter, which transmits capture signals in real time; Step B3 also includes the process of the inspection drone flying directly to the corresponding parking platform when it receives any capture signal, and then proceeding to step B5.

[0011] Furthermore, the anomaly analysis strategy includes a trend difference sub-strategy, which includes... Step C1: Calculate the image sampling temperature value of the inspection location using thermal imaging images; Step C2: Calculate the deviation between the sampled temperature value and the temperature data from the environmental monitor at the same time to generate a sampling confidence value; Step C3: Calculate the difference trend value of the inspection sampling locations. ,in, This represents the difference trend value. The sampling confidence value; The fire risk level and the difference trend value are positively correlated.

[0012] Furthermore, step S8 includes a pre-configured inspection instruction table, which stores several different initial instructions. Each initial instruction corresponds to an inspection trigger condition for each inspection position. When the inspection trigger condition corresponding to an initial instruction is met, the selection weight of the corresponding initial instruction is increased by one cumulative unit. An initial instruction is randomly selected as a new initial instruction based on the selection weight, and the selected initial instruction is deducted by one cumulative unit.

[0013] The main technical advantages of this invention are reflected in the following aspects: First, environmental information is monitored in real time through an environmental monitor. During the inspection process, the drone can retrieve monitoring information via short-range communication, ensuring that on-site environmental data can be acquired. On the one hand, this allows for real-time monitoring of the inspection area, avoiding the problem of incomplete coverage due to signal limitations. On the other hand, environmental information can be analyzed for future inspection planning, allowing for timely increases in the number of inspections and dynamic adjustments to the inspection frequency for risky locations. Furthermore, the drone's control is based on an intelligent communication algorithm, preventing the drone from losing control due to remote control signal risks, thus avoiding difficulties in completing the inspection work. Moreover, the drone uses the on-site environmental monitor as a landmark, enabling inspection work in complex terrain. Attached Figure Description

[0014] Figure 1 The flowchart of a wildfire identification and inspection method based on dynamic path planning is presented in this invention. Detailed Implementation

[0015] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, so that the technical solution of the present invention can be more easily understood and mastered.

[0016] A wildfire identification and inspection method based on dynamic path planning, including Step S1: Configure the initial command on the inspection drone. The purpose of this step is for the drone to fly according to the initial command. The initial command includes not only the drone's position, but also only records the absolute position of the flight. The specific flight path is constrained, such as obstacle avoidance and detour. These actions can be completed by the drone based on image information. Artificial intelligence technology is used to train and provide feedback on the drone's image information to achieve the absolute position. It should be noted that the drone is allowed to have a certain deviation in finding the position, as long as short-distance transmission can be completed.

[0017] Step S2: Control the inspection drone to move to the inspection location according to the initial command. Each inspection location corresponds to an environmental monitor. This environmental monitor is pre-configured and includes a solar panel, a temperature monitoring module, a humidity monitoring module, and a short-range communication module. This allows for real-time detection of temperature and humidity, which is then transmitted to the corresponding drone when it arrives. The solar panel provides power to the device. With the popularization of power generation technology, the environmental monitor can be equipped with image acquisition capabilities. However, at present, it is preferable to use a monitoring module with lower energy consumption to ensure the service life of the environmental monitor.

[0018] Step S3: The inspection drone obtains flight instructions and on-site detection information from the environmental monitor and flies to the next inspection location according to the flight instructions. After reaching the target inspection location, the inspection drone establishes communication via short-range communication and then sends a corresponding request. Upon receiving the request, the environmental monitor sends the actual detection information collected during the process to the inspection drone to reduce its storage space occupation. Simultaneously, the inspection drone can bring back the most recent detection data. On the other hand, each environmental monitor is pre-configured with a dynamic routing table, which stores several flight instructions. Each flight instruction is indexed by the dynamic code of the previous inspection location. The dynamic code has a time-varying factor to change the dynamic code. Through the dynamic routing table and flight instructions, the absolute position of the next flight can be determined. This allows for position correction of the drone with each flight, improving flight accuracy and ensuring that the drone does not experience significant positional deviations due to long-distance flight, thus preventing the drone from failing to complete the inspection smoothly. Steps S2 and S3 both involve obtaining dynamic codes from the environmental monitor. The dynamic code is a verification information; to improve data security and avoid penalties, a dynamic code that adjusts over time is edited. Step S3 also includes sending the dynamic code obtained from the previous inspection location to the environmental monitor to obtain the corresponding flight command.

[0019] Step S4: Repeat step S3 until the inspection drone reaches the parking platform; continuously repeating step S3 will control the drone to reach the parking platform. Step S4 also includes the following: the inspection drone is configured with a safe flight value. If the inspection drone fails to establish communication with the corresponding environmental monitor when its flight distance reaches the safe flight value, the drone will execute a preset fast return-to-home strategy. If the inspection drone cannot find the target object, there are generally two possibilities: first, the drone's long flight time has caused a large positional deviation, indicating complex terrain, significant terrain changes, or human interference with the drone; second, there is a problem with the equipment feedback, such as a power outage or damage to the detector. In these cases, the drone needs to perform a fast return-to-home strategy. Specifically, the fast return-to-home strategy includes... Step B1: Control the inspection drone to return to the previous inspection position according to the flight path of the inspection drone; Step B2: Send a return-to-home command to the corresponding environmental monitor. The environmental monitor is pre-configured with return-to-home route information. When the environmental monitor receives the return-to-home command, it outputs the corresponding return-to-home route information to the inspection drone. The route provided by the return-to-home route information is different from the inspection route, and it is directed to return to home as quickly and safely as possible.

[0020] Step B3: The inspection drone flies to the next inspection location according to the return route information; preferably, the parking platform also includes a capture signal transmitter, which sends capture signals in real time; Step B3 further includes the process where, upon receiving any capture signal, the inspection drone flies directly to the corresponding parking platform and proceeds to step B5. This method of capturing and guiding the drone's flight allows for a rapid return to its home base.

[0021] Step B4: Repeat step B2 until the inspection drone reaches the parking platform; Step B5: The inspection drone sends the last flight command it received to the parking platform; Step B6: The controller generates the loss of contact anomaly information for the corresponding inspection location based on the received flight command and uploads it to the management backend; The fire risk level and the missing contact anomaly information are positively correlated. After a rapid return to base, the fire risk level of the corresponding inspection location is increased based on the missing contact anomaly information, thereby raising the inspection requirements and weight.

[0022] Step S5: The inspection drone uploads the on-site inspection information to the controller of the parking platform; Step S6: The controller uploads the field detection information to the management backend. The drone can upload data for unified analysis. The advantage of unified analysis is that different drones may perform multiple inspections of the same location, resulting in the same inspection location being located at different drones. This allows for data aggregation and better analysis. The controller can pre-transmit data to the management backend via wired communication, relying on the management backend server and database for analysis, which is more accurate and reliable.

[0023] Step S7: The management backend processes the on-site detection information through a preset anomaly analysis strategy to generate the fire risk level for each inspection location; The anomaly analysis and processing strategy includes a difference analysis sub-strategy, which is used to calculate the environmental anomaly difference between field detection information and detection simulation information. ,in, The environmental anomaly difference, This refers to the temperature change waveform from the field detection information. To detect the temperature change waveform in the simulation information, This is a waveform representing humidity changes from on-site monitoring data. To detect the humidity change waveform in the simulation information, As a preset temperature difference weight, With a preset humidity difference weight, the fire risk level and the environmental anomaly difference are positively correlated. Specifically, an environmental information simulation model of the inspection area is pre-constructed. This model is used to generate real-time detection simulation information for each inspection location based on environmental information. The static factors of the environmental information simulation model include vegetation cover factor, terrain factor, and geographical location factor. The environmental information includes climate data, sunshine data, and time data. The on-site detection information includes humidity data and temperature data. First, a theoretical humidity and theoretical temperature output based on each inspection location is constructed according to the simulation model. By inputting static data, such as vegetation cover factor reflecting vegetation cover, terrain factor reflecting terrain, and geographical factor reflecting location, the theoretical climate and theoretical temperature of each point are analyzed through a big data model. By inputting different climates, sunshine, and times, the corresponding temperature and humidity can be obtained, thereby simulating the actual situation. The simulation results provide a basis for calculating the environmental difference value. The difference is calculated by integration, and the corresponding definite integral range is the corresponding time period reflected by the on-site detection information. If the theoretical value deviates significantly from the actual value, it indicates that the actual value may be abnormal, such as a fire point or extreme dryness due to some reason. This would increase the corresponding fire anomaly value for that point.

[0024] The anomaly analysis strategy includes an anomaly retrieval sub-strategy. This sub-strategy retrieves corresponding anomaly values ​​from a pre-built type information table based on the anomaly type of fire anomaly points in the thermal imaging anomaly information, and calculates the type anomaly value corresponding to each inspection location based on the anomaly location. , For the type of abnormal value of the inspection location, For the first Identifying outliers at individual fire anomaly points For the first The distance values ​​between each fire anomaly point and the inspection location. This represents the total number of fire anomalies. Specifically, this is achieved through the following: the inspection drone is equipped with a thermal imager, the controller has a pre-configured image recognition algorithm, and the inspection drone captures thermal images during flight using the thermal imager. Step S5 also includes the inspection drone uploading thermal imaging images to the controller of the parking platform; Step S6 further includes the controller using an image recognition algorithm to identify fire anomalies in the thermal imaging image and mark the corresponding anomaly locations to generate thermal imaging anomaly information. The controller uploads the thermal imaging anomaly information to the management backend. The image recognition algorithm may specifically include facial recognition, fire point recognition, and other recognition algorithms, which are disclosed in the prior art and will not be elaborated here. Different anomaly values ​​are assigned according to different situations, and then the type anomaly value of each inspection location is calculated based on the anomaly value. The closer the distance, the higher the type anomaly value, and the higher the corresponding fire risk level.

[0025] The anomaly analysis strategy includes a trend difference sub-strategy, which includes... Step C1: Calculate the image sampling temperature value of the inspection location using thermal imaging images; Step C2: Calculate the deviation between the sampled temperature value and the temperature data from the environmental monitor at the same time to generate a sampling confidence value; Step C3: Calculate the difference trend value of the inspection sampling locations. ,in, This represents the difference trend value. The sampling confidence value; The fire risk level and the difference trend value are positively correlated. The difference trend value is calculated by analyzing the difference trend. For example, if the temperature rises rapidly in a certain area and the temperature data collected at that location is of high reliability, it indicates that patrols in that area need to be strengthened.

[0026] The fire risk level can be a weighted sum of the above values.

[0027] Step S8: Generate new initial instructions based on the distribution of fire risk levels, and return to step S1.

[0028] Step S8 further includes generating a flight path planning network using a preset path network planning strategy. This flight path planning network comprises a combination of several flight path planning routes. The purpose of this step is to reconfigure the pointing relationships of inspection locations. For example, seasonal changes, increased tourist traffic, or natural weather events necessitate a comprehensive adjustment of the entire inspection path planning. This adjustment is achieved by updating the corresponding dynamic routing table. Specifically, artificial intelligence is used to determine the risk level. A risk level is generated based on external data, and then the frequency, number of inspections, and scope requirements are generated according to the risk level. Based on these requirements, the flight path planning network can be regenerated to configure corresponding flight path planning routes according to different needs.

[0029] Step S1 includes generating path update information based on the flight planning path and configuring it on the corresponding inspection drone; Step S3 also includes the inspection drone sending path update information to the corresponding environmental monitor, and the environmental monitor updating the corresponding dynamic routing table when it receives the path update information. The path network planning strategy includes Step A1: Obtain historical fire sample information; Step A2: Compare the correlation between the information of each historical fire sample and the fire situation in the pre-constructed virtual model of the mountain in the inspection area. ,in, For fire-related information, For the first The relevant values ​​of each fire-related item For the first The preset weights for each fire-related item This represents the total number of items related to the fire situation. Step A3: Calculate the fire risk value of the inspection area. ,in, This represents the fire risk value. For the first The correlation between historical fire sample information and fire situation information For the first The fire spread value in a historical fire sample information, wherein the fire spread value reflects the spread rate of the fire in the historical fire. For the first The fire point concealment value is a historical fire incident sample information, which reflects the degree of concealment of fire points in historical fire incidents. This represents the total number of historical fire incident samples. By analyzing historical data, a list of inspection strategies for each inspection area can be obtained. Through analyzing historical data, a recognition model is built and learned to obtain corresponding fire concealment values ​​and fire spread values, enabling pre-judgment of fire situations and the execution of corresponding tasks.

[0030] Step A4: Retrieve the corresponding inspection task from the pre-built inspection task database in the background according to the fire risk value. The inspection task includes several sets of inspection task parameters and inspection triggering conditions corresponding to the inspection task parameters. The task parameters of each inspection task correspond to the inspection frequency, the concentration of the inspection path, and the inspection requirements for different risks.

[0031] Step A5: Determine whether the inspection triggering conditions are met by analyzing the distribution of fire risk levels, and retrieve the corresponding inspection task parameters. By configuring the inspection task parameters, the inspection requirements can be improved, and the reliability of the inspection can be guaranteed.

[0032] Step A6: Input the inspection task parameters into the virtual mountain model to generate the corresponding flight path planning network. Since the inspection locations are known, the corresponding flight path planning network can be obtained simply by inputting the parameters.

[0033] Preferably, step S8 includes a pre-configured inspection instruction table. This table stores several different initial instructions, each corresponding to an inspection location with a specific inspection trigger condition. When the inspection trigger condition for an initial instruction is met, the selection weight of that initial instruction increases by one cumulative unit. An initial instruction is then randomly selected as the new initial instruction based on its selection weight, and the selected initial instruction is deducted by one cumulative unit. This configuration ensures that each initial instruction selection corresponds to a reliable inspection path. The task ends once the cumulative number of inspections at that location exceeds the cumulative unit. By increasing the inspection level, the probability of selecting the corresponding initial instruction is improved, allowing for dynamic and real-time adjustment of the inspection strategy to enhance inspection reliability.

[0034] Of course, the above are just typical examples of the present invention. In addition, the present invention may have many other specific embodiments. All technical solutions formed by equivalent substitution or equivalent transformation fall within the scope of protection claimed by the present invention.

Claims

1. A wildfire identification and inspection method based on dynamic path planning, characterized in that: include Step S1: Configure the initial commands on the inspection drone; Step S2: Control the inspection drone to move to the inspection position according to the initial command. Each inspection position corresponds to an environmental monitor. Step S3: The inspection drone obtains flight instructions and on-site inspection information from the environmental monitor, and flies to the next inspection location according to the flight instructions. Step S4: Repeat step S3 until the inspection drone reaches the parking platform; Step S5: The inspection drone uploads the on-site inspection information to the controller of the parking platform; Step S6: The controller uploads the field detection information to the management backend; Step S7: The management backend processes the on-site detection information through a preset anomaly analysis strategy to generate the fire risk level for each inspection location; Step S8: Generate new initial instructions based on the distribution of fire risk levels, and return to step S1; Each environmental monitor is pre-configured with a dynamic routing table, which stores several flight commands. Each flight command is indexed by the dynamic code of the previous inspection position, and the dynamic code has a time change factor to make the dynamic code change. Both step S2 and step S3 include obtaining dynamic codes from an environmental monitor; Step S3 also includes sending the dynamic code obtained from the previous inspection location to the environmental monitor in order to obtain the corresponding flight command. Step S8 further includes generating a flight path planning network through a preset path network planning strategy, wherein the flight path planning network includes a combination of several flight planning paths. Step S1 includes generating path update information based on the flight planning path and configuring it on the corresponding inspection drone; Step S3 also includes the inspection drone sending path update information to the corresponding environmental monitor, and the environmental monitor updating the corresponding dynamic routing table when it receives the path update information. The path network planning strategy includes Step A1: Obtain historical fire sample information; Step A2: Compare the correlation between the information of each historical fire sample and the fire situation in the pre-constructed virtual model of the mountain in the inspection area. ,in, For fire-related information, For the first The relevant values ​​of each fire-related item For the first The preset weights for each fire-related item This represents the total number of items related to the fire situation. Step A3: Calculate the fire risk value of the inspection area. ,in, This represents the fire risk value. For the first The correlation between historical fire sample information and fire situation information For the first The fire spread value in a historical fire sample information, wherein the fire spread value reflects the spread rate of the fire in the historical fire. For the first The fire point concealment value is a historical fire incident sample information, which reflects the degree of concealment of fire points in historical fire incidents. The total number of historical fire incident sample information; Step A4: Retrieve the corresponding inspection task from the pre-built inspection task database in the background according to the fire risk value. The inspection task includes several sets of inspection task parameters and inspection triggering conditions corresponding to the inspection task parameters. Step A5: Determine whether the inspection triggering conditions are met by analyzing the distribution of fire risk levels in order to retrieve the corresponding inspection task parameters; Step A6: Input the inspection task parameters into the mountain virtual model to generate the corresponding flight path planning network; It also includes a pre-built environmental information simulation model of the inspection area, which is used to generate detection simulation information for each inspection location in real time based on the environmental information; the static factors of the environmental information simulation model include vegetation cover factor, terrain factor and geographical location factor; the environmental information includes climate data, sunshine data and time data; the on-site detection information includes humidity data and temperature data; The anomaly analysis strategy includes a difference analysis sub-strategy, which is used to calculate the environmental anomaly difference between field detection information and detection simulation information. ,in, The environmental anomaly difference, This refers to the temperature change waveform from the field detection information. To detect the temperature change waveform in the simulation information, This is a waveform representing humidity changes from on-site monitoring data. To detect the humidity change waveform in the simulation information, As a preset temperature difference weight, The fire risk level and the environmental anomaly difference are positively correlated with the preset humidity difference weight.

2. The wildfire identification and inspection method based on dynamic path planning as described in claim 1, characterized in that: The inspection drone is equipped with a thermal imager, and the controller has a pre-configured image recognition algorithm. During flight, the inspection drone uses the thermal imager to capture thermal images. Step S5 also includes the inspection drone uploading thermal imaging images to the controller of the parking platform; Step S6 also includes the controller using an image recognition algorithm to identify fire anomalies in the thermal imaging image and mark the corresponding anomalies to generate thermal imaging anomaly information, and the controller uploads the thermal imaging anomaly information to the management backend. The anomaly analysis strategy includes an anomaly retrieval sub-strategy, which involves retrieving corresponding identification anomaly values ​​from a pre-built type information table based on the anomaly type of fire anomaly points in thermal imaging anomaly information, and calculating the type anomaly value corresponding to each inspection location based on the anomaly location. , For the type of outlier value of the inspection location, For the first Identifying outliers at individual fire anomaly points For the first The distance values ​​between each fire anomaly point and the inspection location. This represents the total number of abnormal fire locations.

3. The wildfire identification and inspection method based on dynamic path planning as described in claim 1, characterized in that: Step S4 also includes the following: the inspection drone is configured with a safe flight value. If the inspection drone fails to establish communication with the environmental monitor at the corresponding location when its flight distance reaches the safe flight value, the inspection drone is controlled to execute a preset rapid return-to-home strategy. The rapid return-to-home strategy includes... Step B1: Control the inspection drone to return to the previous inspection position according to the flight path of the inspection drone; Step B2: Send a return-to-home command to the corresponding environmental monitor. The environmental monitor is pre-configured with return-to-home route information. When the environmental monitor receives the return-to-home command, it outputs the corresponding return-to-home route information to the inspection drone. Step B3: The inspection drone flies to the next inspection location based on the return route information; Step B4: Repeat step B2 until the inspection drone reaches the parking platform; Step B5: The inspection drone sends the last flight command it received to the parking platform; Step B6: The controller generates the loss of contact anomaly information for the corresponding inspection location based on the received flight command and uploads it to the management backend; The fire risk level and the abnormal information about being out of contact are positively correlated.

4. The wildfire identification and inspection method based on dynamic path planning as described in claim 3, characterized in that: The shutdown platform also includes a capture signal transmitter, which sends capture signals in real time; Step B3 also includes the process of the inspection drone flying directly to the corresponding parking platform when it receives any capture signal, and then proceeding to step B5.

5. A wildfire identification and inspection method based on dynamic path planning as described in claim 2, characterized in that: The anomaly analysis strategy includes a trend difference sub-strategy, which includes... Step C1: Calculate the image sampling temperature value of the inspection location using thermal imaging images; Step C2: Calculate the deviation between the sampled temperature value and the temperature data from the environmental monitor at the same time to generate a sampling confidence value; Step C3: Calculate the difference trend value of the inspection sampling locations. ,in, This represents the difference trend value. The sampling confidence value; The fire risk level and the difference trend value are positively correlated.

6. The wildfire identification and inspection method based on dynamic path planning as described in claim 1, characterized in that: Step S8 includes a pre-configured inspection instruction table, which stores several different initial instructions. Each initial instruction corresponds to an inspection trigger condition for each inspection position. When the inspection trigger condition corresponding to an initial instruction is met, the selection weight of the corresponding initial instruction is increased by one cumulative unit. An initial instruction is randomly selected as a new initial instruction based on the selection weight, and the selected initial instruction is deducted by one cumulative unit.

Images