A park drone safety linkage inspection method and system

Abstract

The present invention relates to the technical field of drone inspection, and discloses a park drone safety linkage inspection method and system. By acquiring the work task lists of several work units in the target park, constructing an operation feature set and predicting the fire occurrence probability and fire correlation operation unit ranking table of the work unit of each work task in each minimum monitoring cycle within the target monitoring period, and then analyzing the inspection capability of each inspection drone, considering the fire monitoring frequency demand and the work units associated with the target work units that may cause fire, the inspection drone and the operation video surveillance are controlled to perform joint fire monitoring of the image and temperature of the target park within the target monitoring period, thereby improving the accuracy and reliability of the park fire monitoring, avoiding the problem of low comprehensiveness of the park fire monitoring due to insufficient inspection drone resources that may occur during the peak period of fire, and significantly improving the intelligence and automation of the park safety monitoring.

Description

Technical Field

[0001] The present invention relates to the field of drone inspection technology, and in particular to a park drone safety linkage inspection method and system. Background Art

[0002] Park safety monitoring, especially fire monitoring, is crucial for ensuring the safety of life and property, as well as the stable operation of the park. Prompt or advanced detection of park fires can significantly reduce casualties and economic losses, effectively maintain park operations, improve emergency response efficiency, and optimize rescue strategies. As parks develop in a large-scale and intelligent manner, a comprehensive fire monitoring system has become not only a safety assurance tool but also a crucial component of modern park management.

[0003] Currently, fire monitoring within large industrial parks primarily relies on video surveillance (which offers a wide coverage area and can accommodate other monitoring needs, resulting in a low overall cost). Sensor monitoring is typically limited to key areas, with warnings targeted at operational units, making it less valuable for overall park safety management. Traditional video surveillance for fire monitoring in large industrial parks suffers from the large number of images and limited network bandwidth required to receive these data from the monitoring center, hindering effective and comprehensive monitoring. Using drones equipped with temperature monitoring devices for coordinated park inspections can improve the flexibility of fire monitoring within large industrial parks while effectively addressing the limitations of data transmission for these images. However, in practice, limited drone resources and the varying operational tasks and environments of different operational units within the park can lead to varying fire probabilities. Consequently, different fire monitoring frequencies are required for different operational units. During peak fire seasons, drone resources may be insufficient to meet the monitoring needs of the entire park, resulting in insufficient comprehensive monitoring.

[0004] Therefore, how to improve the accuracy and reliability of park fire monitoring, avoid the problem of low comprehensiveness of park fire monitoring caused by insufficient patrol drone resources during peak fire periods, and significantly improve the intelligence and automation of park safety monitoring is a technical problem that needs to be solved urgently. Summary of the Invention

[0005] The present invention provides a park drone safety linkage inspection method and system, aiming to solve at least one of the above technical problems.

[0006] To achieve the above objectives, the present invention provides a method for park drone safety linkage inspection, comprising the following steps:

[0007] Obtaining a list of work tasks for a number of work units within a target park; wherein the list of work tasks stores the work content of each work task according to an estimated work period;

[0008] Extracting the work content and estimated work period of each work task, using the work content to determine a first work feature associated with the work content for each work task, using the estimated work period to determine a second work feature associated with the work environment for each work task, and constructing a work feature set;

[0009] Based on the operation feature set, predict the fire occurrence probability and fire correlation operation unit ranking table of the operation unit of each operation task in each minimum monitoring cycle within the target monitoring period;

[0010] Query the inspection drone call information of the target park during the target monitoring period, analyze the inspection capability of each inspection drone, consider the preset mapping relationship between the fire occurrence probability and inspection frequency requirements of the work unit of each operation task and the fire correlation operation unit ranking table, and control the inspection drone and operation video surveillance to perform joint fire monitoring of the target park's image and temperature during the target monitoring period;

[0011] Receive the work task update information uploaded by each work unit and the work environment update information issued by the work environment database, and update the fire monitoring strategy of the inspection drone and work video surveillance.

[0012] Optionally, the steps of obtaining the operation task lists of several operation units in the target park include:

[0013] Obtaining the operation task information of the target monitoring period sent by each operation unit in the target park; wherein the operation task information includes the operation content and estimated operation period of each operation task;

[0014] The work content and estimated work period of each work task in the work task information are extracted and summarized, and stored in a pre-built work task list according to the estimated work period.

[0015] Optionally, the steps of using the work content to determine a first work characteristic associated with the work content for each work task, and using the estimated work period to determine a second work characteristic associated with the work environment for each work task, specifically include:

[0016] Extract keywords from the work content of each work task, calculate similarity between the extracted keywords and several work features associated with the work content, and use the work features with similarity exceeding a preset similarity threshold as the first work feature associated with the work content of each work task;

[0017] Access the operating environment database of the target park, match a plurality of operating environment data corresponding to the estimated operating period of each operating task, and extract a second operating feature of each operating task associated with the operating environment from the plurality of operating environment data.

[0018] Optionally, the first operation feature associated with the operation content includes at least one of: operation heat source feature, electricity usage feature, combustible material feature, combustion-supporting material feature, operation behavior feature, material reaction feature, energy superposition feature, and environmental space change feature.

[0019] Optionally, the second operating characteristic associated with the operating environment includes at least one of a temperature characteristic, a humidity characteristic, a wind speed characteristic, and a wind force characteristic.

[0020] Optionally, based on the operation feature set, the step of predicting the fire occurrence probability and fire correlation operation unit ranking table of each operation task in each minimum monitoring cycle within the target monitoring period specifically includes:

[0021] Based on the operation feature set and the estimated operation period and location of each operation task, a feature vector and an operation unit association matrix of each operation unit are constructed, and a three-dimensional composite feature tensor is generated by fusion as a fire prediction sample for each operation task in the target park;

[0022] An initial convolutional neural network model was constructed, consisting of a convolutional layer, a pooling layer, a fully connected layer, a fire probability prediction branch, and a relevance ranking branch. Using fire training samples based on historical fire accidents, a joint loss function was constructed to balance the two tasks of fire probability prediction and relevance ranking, completing the training of the initial convolutional neural network model and obtaining a fire prediction model.

[0023] The fire prediction samples are input into the fire prediction model to obtain the fire occurrence probability and the operation unit correlation of each operation unit in each minimum monitoring cycle within the target monitoring period output by the fire prediction model, and a fire correlation operation unit ranking table is generated according to the operation unit correlation.

[0024] Optionally, query the inspection drone call information of the target park during the target monitoring period and analyze the inspection capability steps of each inspection drone, including:

[0025] Query the patrol drone call information of the target park during the target monitoring period; wherein the patrol drone call information includes the idle time period and patrol drone identification of each candidate patrol drone;

[0026] Based on the idle period, candidate inspection drones that are idle during the target monitoring period are screened out as inspection drones, and the inspection speed of the inspection drones is extracted as the inspection capability.

[0027] Optionally, considering the preset mapping relationship between the fire occurrence probability and inspection frequency requirement of the work unit of each work task and the fire correlation work unit ranking table, the inspection drone and the work video surveillance are controlled to perform the joint fire monitoring steps of the image and temperature of the target park within the target monitoring period, specifically including:

[0028] Determine the fire inspection frequency requirement for each operating unit in each minimum monitoring cycle based on the fire occurrence probability of each operating unit in each minimum monitoring cycle during the target monitoring period;

[0029] Based on the fire inspection frequency requirement and the inspection speed of each inspection drone, the fire correlation operation unit ranking table of each operation unit in each minimum monitoring cycle within the target monitoring period is considered. After each inspection drone determines the inspection route, the interval time for each inspection drone to perform inspections on two adjacent operation units is greater than the shortest flight time determined by the inspection speed of the inspection drone and the distance between the two adjacent operation units as the first constraint condition. The interval time for each operation unit to be inspected and monitored by two adjacent inspection drones in each minimum monitoring cycle within the target monitoring period, and the interval time for two adjacent operation video surveillance reports are The second constraint condition is that the interval between the inspection monitoring and the reporting of the operation video surveillance by the adjacent inspection drones is less than the interval corresponding to the fire inspection frequency requirement of the operation unit in each minimum monitoring cycle. The optimization goal is to minimize the total amount of operation video surveillance data received by the monitoring center within the unit monitoring time from the operation units using operation video surveillance and the associated operation units whose fire correlation exceeds the preset correlation threshold. The fire inspection route of each inspection drone and the monitoring video reporting time point of each operation unit in the target park within the target monitoring period are optimized to generate a fire monitoring strategy for inspection drones and operation video surveillance;

[0030] Based on the inspection sub-strategy in the fire monitoring strategy, the inspection drone is controlled to perform temperature fire monitoring along the inspection route based on the temperature monitoring device on board. Based on the monitoring sub-strategy in the fire monitoring strategy, the monitoring center is controlled to call the operation video monitoring equipment of the corresponding operation unit to report the collected monitoring images, so as to realize the joint fire monitoring of the image and temperature of the target park.

[0031] Optionally, receive the updated work task information uploaded by each work unit and the updated work environment information issued by the work environment database, and update the fire monitoring strategy of the inspection drone and work video surveillance, specifically including:

[0032] Receive the updated work task information uploaded by each work unit and the updated work environment information issued by the work environment database, and determine whether the fire monitoring strategy can be regenerated without changing the fire inspection route of the inspection drone;

[0033] If so, based on the fire inspection route of the patrol drone, the monitoring video reporting time point of each operating unit in the fire monitoring strategy is regenerated; if not, the fire inspection route of each patrol drone and the monitoring video reporting time point of each operating unit in the fire monitoring strategy are regenerated.

[0034] In addition, in order to achieve the above-mentioned purpose, the present invention also provides a park drone safety linkage inspection system, comprising:

[0035] An acquisition module is used to obtain a list of work tasks of several work units in a target park; wherein the list of work tasks stores the work content of each work task according to the estimated work period;

[0036] An extraction module is configured to extract the work content and estimated work period of each work task, determine a first work feature associated with the work content for each work task using the work content, determine a second work feature associated with the work environment for each work task using the estimated work period, and construct a work feature set;

[0037] A prediction module, configured to predict, based on the operation feature set, the probability of fire occurrence and a ranking table of fire correlation operation units of each operation task in each minimum monitoring cycle within a target monitoring period;

[0038] The control module is used to query the inspection drone call information of the target park during the target monitoring period, analyze the inspection capability of each inspection drone, consider the preset mapping relationship between the fire occurrence probability and inspection frequency requirements of the work unit of each operation task and the fire correlation operation unit ranking table, and control the inspection drone and operation video surveillance to perform joint fire monitoring of the target park's image and temperature during the target monitoring period;

[0039] The update module is used to receive the work task update information uploaded by each work unit and the work environment update information issued by the work environment database, and update the fire monitoring strategy of the inspection drone and work video surveillance.

[0040] The beneficial effects of the present invention are: a method and system for park drone safety linkage inspection is proposed, by obtaining the work task list of several work units in the target park, constructing an operation feature set and predicting the fire occurrence probability and fire correlation operation unit ranking table of the work unit of each work task in each minimum monitoring cycle within the target monitoring period, and then by analyzing the inspection capability of each inspection drone, considering the fire monitoring frequency demand and the work units associated with the target work units where fire may occur, controlling the inspection drone and the operation video monitoring to perform joint fire monitoring of the image and temperature of the target park within the target monitoring period, thereby improving the accuracy and reliability of park fire monitoring, avoiding the problem of low comprehensiveness of park fire monitoring caused by insufficient inspection drone resources that may occur during the peak period of fire, and significantly improving the intelligence and automation of park safety monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of a method for linking safety inspections of campus drones according to an embodiment of the present invention;

[0042] Figure 2 This is a schematic diagram of the structure of the park drone safety linkage inspection system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0044] The embodiment of the present invention provides a method for the joint inspection of a park drone safety. Figure 1 , Figure 1 The figure is a flow chart of the method for joint safety inspection of a campus by drones according to an embodiment of the present invention.

[0045] In this embodiment, a method for linking safety inspections of a campus with drones includes the following steps:

[0046] S1: Obtain a list of work tasks of several work units in a target park; wherein the list of work tasks stores the work content of each work task according to the estimated work period;

[0047] S2: Extracting the work content and estimated work period of each work task, using the work content to determine a first work feature associated with the work content for each work task, and using the estimated work period to determine a second work feature associated with the work environment for each work task, thereby constructing a work feature set;

[0048] S3: Based on the operation feature set, predict the fire occurrence probability and fire correlation operation unit ranking table of each operation task operation unit in each minimum monitoring cycle within the target monitoring period;

[0049] S4: Query the inspection drone call information of the target park during the target monitoring period, analyze the inspection capability of each inspection drone, consider the preset mapping relationship between the fire occurrence probability and inspection frequency requirements of the work unit of each operation task and the fire correlation operation unit ranking table, and control the inspection drone and operation video surveillance to perform joint fire monitoring of the target park's image and temperature during the target monitoring period;

[0050] S5: Receive the work task update information uploaded by each work unit and the work environment update information issued by the work environment database, and update the fire monitoring strategy of the inspection drone and work video surveillance.

[0051] It should be noted that currently, fire monitoring within large industrial parks primarily relies on video surveillance. (Video surveillance offers a wide coverage area, can accommodate other monitoring needs, and has a low overall cost. Sensor monitoring is typically only deployed in key areas, and the primary warning mechanism is targeted at work units, making it less relevant for overall park safety management.) Given the large number of images and limited network bandwidth required to receive these data, traditional video surveillance in large industrial parks is limited in its ability to achieve effective and comprehensive monitoring. Using drones equipped with temperature monitoring devices to conduct coordinated inspections of industrial parks can improve the flexibility of fire monitoring within large industrial parks while effectively addressing the data transmission limitations of these images. However, in practice, limited drone resources and the diverse operational tasks and environments of different work units within the park can lead to varying fire probabilities. Consequently, different fire monitoring frequencies are required for different work units. During peak fire seasons, drone resources may not be sufficient to monitor the entire park, resulting in insufficient comprehensive monitoring.

[0052] In order to solve the above problems, this embodiment obtains the work task list of several work units in the target park, constructs an operation feature set and predicts the fire occurrence probability and fire correlation operation unit ranking table of the work unit of each work task in each minimum monitoring cycle within the target monitoring period, and then analyzes the inspection capability of each patrol drone, considers the fire monitoring frequency requirements and the work units associated with the target work units where fire may occur, controls the patrol drones and operation video surveillance to perform joint fire monitoring of the image and temperature of the target park within the target monitoring period, improves the accuracy and reliability of park fire monitoring, avoids the problem of low comprehensiveness of park fire monitoring due to insufficient patrol drone resources that may occur during the peak period of fire, and significantly improves the intelligence and automation of park safety monitoring.

[0053] In a preferred embodiment, the step of obtaining the work task lists of several work units in the target park specifically includes:

[0054] S11: Acquire operation task information of the target monitoring period sent by each operation unit in the target park; wherein the operation task information includes the operation content and estimated operation period of each operation task;

[0055] S12: Extract and summarize the work content and estimated work period of each work task in the work task information, and store them in a pre-built work task list according to the estimated work period.

[0056] In this embodiment, the operation task information about the target monitoring period sent by each operation unit in the target park is obtained, and then the information recorded in the operation task information is used.

[0057] In a preferred embodiment, the steps of determining a first operation characteristic associated with the operation content of each operation task using the operation content, and determining a second operation characteristic associated with the operation environment of each operation task using the estimated operation period, specifically include:

[0058] S21: extracting keywords from the work content of each work task, calculating similarity between the extracted keywords and a number of work features associated with the work content, and using work features with similarities exceeding a preset similarity threshold as the first work feature associated with the work content of each work task;

[0059] S22: Accessing the operating environment database of the target park, matching a plurality of operating environment data corresponding to the estimated operating period of each operating task, and extracting a second operating feature associated with the operating environment for each operating task from the plurality of operating environment data.

[0060] In this embodiment, keywords are extracted from the work content of each work task, and the first work feature of each work task is determined by using similarity calculation. The work environment data of the work environment database is matched with the estimated work time period of each work task to determine the first work feature of each work task, thereby constructing the final work feature set.

[0061] In actual applications, the first operation feature associated with the operation content includes: operation heat source feature, electricity usage feature, combustible material feature, combustion-supporting material feature, operation behavior feature, material reaction feature, energy superposition feature, and at least one of the environmental space change features.

[0062] In practical applications, the second operating characteristic associated with the operating environment includes at least one of a temperature characteristic, a humidity characteristic, a wind speed characteristic, and a wind force characteristic.

[0063] It should be noted that the first operation feature is a feature associated with the operation content, which is configured to lead to an increased probability of fire. The second operation feature is a feature associated with the operation environment, which is configured to provide a fire environment.

[0064] In a preferred embodiment, based on the operation feature set, the step of predicting the fire occurrence probability and fire correlation operation unit ranking table of each operation unit of each operation task in each minimum monitoring cycle within the target monitoring period specifically includes:

[0065] S31: Based on the operation feature set and the estimated operation period and location of each operation unit of each operation task, construct a feature vector and an operation unit association matrix for each operation unit, and fuse them to generate a three-dimensional composite feature tensor as a fire prediction sample for each operation task in the target park;

[0066] S32: Build an initial convolutional neural network model consisting of a convolutional layer, a pooling layer, a fully connected layer, a fire probability prediction branch, and a relevance ranking branch. Using fire training samples based on historical fire accidents, the model is trained by balancing the fire probability prediction and relevance ranking tasks using a constructed joint loss function to obtain a fire prediction model.

[0067] S33: Input the fire prediction sample into the fire prediction model, obtain the fire occurrence probability and the operation unit correlation of each operation unit in each minimum monitoring cycle within the target monitoring period output by the fire prediction model, and generate a fire correlation operation unit ranking table according to the operation unit correlation.

[0068] In this embodiment, by structured processing of data such as the working period, working location and working feature set of each working unit, the working period can be encoded as a time series vector, the working location is represented by a geographic coordinate vector, and the working feature set is directly converted into a numerical vector, thereby constructing the feature vector of each working unit; then, by constructing an operating unit association matrix, each element in the matrix is ​​encoded as a representation of the association status of the two working units, such as whether they are working at the same time, whether they are in adjacent working areas, whether the working characteristics of the two working units meet the requirements of fire-related coordination, etc.; finally, the feature vector of a single working unit is fused with the operating unit association matrix to form a three-dimensional tensor.

[0069] After this, an initial convolutional neural network model was constructed, consisting of convolutional layers, pooling layers, fully connected layers, a fire probability prediction branch, and a relevance ranking branch. In the fire probability prediction branch, a single-neuron output layer was connected after the fully connected layer, using a linear activation function to output the fire probability for each work unit. This task was considered a regression problem, with a mean squared error as the loss function. In the relevance ranking branch, multiple neuron output layers were connected, with the number of neurons equal to the number of work units minus their own value. A softmax activation function was used to output the probability distribution of the fire relevance between each work unit and the other work units. This task was considered a multi-classification problem, with a cross-entropy loss function. In this way, a joint loss function was constructed, combining loss functions based on different weights with the cross-entropy loss function, to balance the two tasks of fire probability prediction and relevance ranking.

[0070] Finally, the trained fire prediction model is used to output the fire occurrence probability of each work unit through the fire probability prediction branch, and then the correlation probability of a group of other units corresponding to each work unit is output through the correlation sorting branch. A fire correlation work unit sorting table is constructed to determine the work units associated with the target work units where fire may occur, which is used to guide the subsequent generation of drone inspection routes and the reporting control of work video monitoring.

[0071] In a preferred embodiment, querying the inspection drone call information of the target park during the target monitoring period and analyzing the inspection capability of each inspection drone specifically includes:

[0072] S41: Querying patrol drone call information for the target park during the target monitoring period; wherein the patrol drone call information includes the idle period and patrol drone identifier of each candidate patrol drone;

[0073] S42: Based on the idle period, candidate inspection drones that are idle during the target monitoring period are screened out as inspection drones, and the inspection speed of the inspection drones is extracted as the inspection capability.

[0074] On this basis, considering the preset mapping relationship between the fire occurrence probability and inspection frequency requirements of each work unit of each work task and the fire correlation work unit ranking table, the inspection drone and the work video surveillance are controlled to perform the joint fire monitoring steps of the target park's image and temperature within the target monitoring period, specifically including:

[0075] S43: Determine the fire inspection frequency requirement of each operating unit in each minimum monitoring period according to the fire occurrence probability of each operating unit of each operating task in each minimum monitoring period within the target monitoring period;

[0076] S44: Based on the fire inspection frequency requirement and the inspection speed of each inspection drone, the fire correlation operation unit ranking table of each operation unit in each minimum monitoring cycle within the target monitoring period is considered. After each inspection drone determines the inspection route, the interval time for each inspection drone to perform inspections on two adjacent operation units is greater than the inspection speed of the inspection drone and the shortest flight time determined by the distance between the two adjacent operation units as the first constraint condition. The interval time for each operation unit to be inspected and monitored by two adjacent inspection drones in each minimum monitoring cycle within the target monitoring period, and the interval time between two adjacent operation video surveillance reports are The second constraint condition is that the interval between the inspection monitoring and the reporting of the operation video surveillance by the adjacent inspection drones is less than the interval corresponding to the fire inspection frequency requirement of the operation unit in each minimum monitoring cycle. The optimization goal is to minimize the total amount of operation video surveillance data reported by the associated operation units whose fire correlation exceeds the preset correlation threshold and received by the monitoring center within the unit monitoring time. The fire inspection route of each inspection drone and the monitoring video reporting time point of each operation unit in the target park during the target monitoring period are optimized to generate a fire monitoring strategy for inspection drones and operation video surveillance.

[0077] S45: Based on the inspection sub-strategy in the fire monitoring strategy, the inspection drone is controlled to perform temperature fire monitoring along the inspection route based on the temperature monitoring device on board. Based on the monitoring sub-strategy in the fire monitoring strategy, the monitoring center is controlled to call the operation video monitoring equipment of the corresponding operation unit to report the collected monitoring images, so as to realize the joint fire monitoring of the image and temperature of the target park.

[0078] In this embodiment, by analyzing the inspection capability of each inspection drone, considering the fire monitoring frequency requirement determined based on the probability of fire occurrence and the work units associated with the target work units where fire may occur determined by the fire correlation work unit ranking table, a constraint condition set is constructed based on the fire monitoring frequency requirement and the inspection capability of the inspection drone, and the optimization goal is to minimize the amount of video surveillance transmission data of the work units associated with the target work units. The fire inspection route of each inspection drone and the monitoring video reporting time point of each work unit are solved, and a fire monitoring strategy for the inspection drone and the work video surveillance is generated. In this way, the inspection drone and the work video surveillance are controlled to perform joint fire monitoring of the image and temperature of the target park within the target monitoring period. Through the multi-dimensional joint fire monitoring of the image and temperature, while meeting the fire monitoring frequency requirement, the accuracy and reliability of the park fire monitoring are improved, and the problem of low comprehensiveness of the park fire monitoring caused by insufficient inspection drone resources that may occur during the peak period of fire is avoided, which significantly improves the intelligence and automation of the park safety monitoring.

[0079] In a preferred embodiment, the steps of receiving the updated work task information uploaded by each work unit and the updated work environment information issued by the work environment database and updating the fire monitoring strategy of the inspection drone and the work video surveillance include:

[0080] S51: receiving the updated operation task information uploaded by each operation unit and the updated operation environment information issued by the operation environment database, and determining whether a fire monitoring strategy can be regenerated without changing the fire inspection route of the inspection drone;

[0081] S52: If yes, based on the fire inspection route of the inspection drone, regenerate the monitoring video reporting time point of each operating unit in the fire monitoring strategy; if not, regenerate the fire inspection route of each inspection drone and the monitoring video reporting time point of each operating unit in the fire monitoring strategy.

[0082] In this embodiment, taking into account possible changes in work tasks and work environments, the fire monitoring strategy needs to be updated in real time. When updating, the first consideration is to regenerate the monitoring sub-strategy in the fire monitoring strategy without changing the fire inspection route of the inspection drone, thereby reducing the steps of re-entering the inspection route for the inspection drone due to changes in the inspection route, thereby improving the simplicity of operation.

[0083] Reference Figure 2 , Figure 2 This is a schematic diagram of the structure of the park drone safety linkage inspection system according to an embodiment of the present invention.

[0084] like Figure 2 As shown, the park drone safety linkage inspection system proposed in the embodiment of the present invention includes:

[0085] The acquisition module 10 is used to obtain a list of work tasks of several work units in the target park; wherein the list of work tasks stores the work content of each work task according to the estimated work period;

[0086] An extraction module 20 is configured to extract the work content and estimated work period of each work task, determine a first work feature associated with the work content for each work task using the work content, and determine a second work feature associated with the work environment for each work task using the estimated work period, thereby constructing a work feature set;

[0087] A prediction module 30 is configured to predict, based on the operation feature set, the probability of fire occurrence and a ranking table of fire correlation operation units of each operation task in each minimum monitoring cycle within a target monitoring period;

[0088] The control module 40 is configured to query the inspection drone call information of the target park during the target monitoring period, analyze the inspection capability of each inspection drone, consider the preset mapping relationship between the fire occurrence probability and inspection frequency requirement of the work unit of each operation task and the fire correlation operation unit ranking table, and control the inspection drone and the operation video surveillance to perform joint fire monitoring of the image and temperature of the target park during the target monitoring period;

[0089] The updating module 50 is used to receive the operation task update information uploaded by each operation unit and the operation environment update information sent by the operation environment database, and update the fire monitoring strategy of the inspection drone and the operation video monitoring.

[0090] Other embodiments or specific implementation methods of the park drone safety linkage inspection system of the present invention can refer to the above-mentioned method embodiments and will not be repeated here.

[0091] It should be understood that, in the description of this specification, reference to terms such as "one embodiment," "another embodiment," "other embodiments," or "first to Nth embodiments" means that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples.

[0092] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or system comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or system. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or system comprising the element.

[0093] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for linking safety inspection of a park with drones, characterized in that: The following steps are involved: Obtaining a list of work tasks for a number of work units within a target park; wherein the list of work tasks stores the work content of each work task according to an estimated work period; Extracting the work content and estimated work period of each work task, using the work content to determine a first work feature associated with the work content for each work task, using the estimated work period to determine a second work feature associated with the work environment for each work task, and constructing a work feature set; Based on the operation feature set, a fire occurrence probability and a fire correlation operation unit ranking table of each operation task operation unit in each minimum monitoring cycle within the target monitoring period are predicted; specifically, the table includes: Based on the operation feature set and the estimated operation period and location of each operation task, a feature vector and an operation unit association matrix of each operation unit are constructed, and a three-dimensional composite feature tensor is generated by fusion as a fire prediction sample for each operation task in the target park; An initial convolutional neural network model was constructed, consisting of a convolutional layer, a pooling layer, a fully connected layer, a fire probability prediction branch, and a relevance ranking branch. Using fire training samples based on historical fire accidents, a joint loss function was constructed to balance the two tasks of fire probability prediction and relevance ranking, completing the training of the initial convolutional neural network model and obtaining a fire prediction model. Inputting the fire prediction samples into a fire prediction model, obtaining the fire occurrence probability and the operation unit correlation of each operation unit in each minimum monitoring cycle within a target monitoring period output by the fire prediction model, and generating a fire correlation operation unit ranking table according to the operation unit correlation; Query the inspection drone call information of the target park during the target monitoring period, analyze the inspection capability of each inspection drone, consider the preset mapping relationship between the fire occurrence probability and inspection frequency requirements of the work unit of each operation task and the fire correlation operation unit ranking table, and control the inspection drone and operation video surveillance to perform joint fire monitoring of the target park's image and temperature during the target monitoring period; Receive the work task update information uploaded by each work unit and the work environment update information issued by the work environment database, and update the fire monitoring strategy of the inspection drone and work video surveillance.

2. The park drone safety linkage inspection method according to claim 1, characterized in that: The steps for obtaining the work task lists of several work units in the target park include: Obtaining the operation task information of the target monitoring period sent by each operation unit in the target park; wherein the operation task information includes the operation content and estimated operation period of each operation task; The work content and estimated work period of each work task in the work task information are extracted and summarized, and stored in a pre-built work task list according to the estimated work period.

3. The park drone safety linkage inspection method according to claim 1, characterized in that: The steps of determining a first operation feature associated with the operation content of each operation task using the operation content, and determining a second operation feature associated with the operation environment of each operation task using the estimated operation time period, specifically include: Extract keywords from the work content of each work task, calculate similarity between the extracted keywords and several work features associated with the work content, and use the work features with similarity exceeding a preset similarity threshold as the first work feature associated with the work content of each work task; Access the operating environment database of the target park, match a plurality of operating environment data corresponding to the estimated operating period of each operating task, and extract a second operating feature of each operating task associated with the operating environment from the plurality of operating environment data.

4. The park drone safety linkage inspection method according to claim 3, characterized in that: The first operation feature associated with the operation content includes: at least one of: operation heat source feature, electricity usage feature, combustible material feature, combustion-supporting material feature, operation behavior feature, material reaction feature, energy superposition feature, and environmental space change feature.

5. The park drone safety linkage inspection method according to claim 3, characterized in that: The second operating characteristic associated with the operating environment includes at least one of a temperature characteristic, a humidity characteristic, a wind speed characteristic, and a wind force characteristic.

6. The park drone safety linkage inspection method according to claim 1, characterized in that: Query the inspection drone call information of the target park during the target monitoring period, and analyze the inspection capability steps of each inspection drone, including: Query the patrol drone call information of the target park during the target monitoring period; wherein the patrol drone call information includes the idle time period and patrol drone identification of each candidate patrol drone; Based on the idle period, candidate inspection drones that are idle during the target monitoring period are screened out as inspection drones, and the inspection speed of the inspection drones is extracted as the inspection capability.

7. The park drone safety linkage inspection method according to claim 6, characterized in that: Considering the preset mapping relationship between the fire occurrence probability and inspection frequency requirements of each work unit of each work task and the fire correlation work unit ranking table, the inspection drone and the work video surveillance are controlled to perform the joint fire monitoring steps of the target park's image and temperature within the target monitoring period. Specifically, the steps include: Determine the fire inspection frequency requirement for each operating unit in each minimum monitoring cycle based on the fire occurrence probability of each operating unit in each minimum monitoring cycle during the target monitoring period; Based on the fire inspection frequency requirement and the inspection speed of each inspection drone, the fire correlation operation unit ranking table of each operation unit in each minimum monitoring cycle within the target monitoring period is considered. After each inspection drone determines the inspection route, the interval time for each inspection drone to perform inspections on two adjacent operation units is greater than the shortest flight time determined by the inspection speed of the inspection drone and the distance between the two adjacent operation units as the first constraint condition. The interval time for each operation unit to be inspected and monitored by two adjacent inspection drones in each minimum monitoring cycle within the target monitoring period, and the interval time for two adjacent operation video surveillance reports are The second constraint condition is that the interval between the inspection monitoring and the reporting of the operation video surveillance by the adjacent inspection drones is less than the interval corresponding to the fire inspection frequency requirement of the operation unit in each minimum monitoring cycle. The optimization goal is to minimize the total amount of operation video surveillance data received by the monitoring center within the unit monitoring time from the operation units using operation video surveillance and the associated operation units whose fire correlation exceeds the preset correlation threshold. The fire inspection route of each inspection drone and the monitoring video reporting time point of each operation unit in the target park within the target monitoring period are optimized to generate a fire monitoring strategy for inspection drones and operation video surveillance; Based on the inspection sub-strategy in the fire monitoring strategy, the inspection drone is controlled to perform temperature fire monitoring along the inspection route based on the temperature monitoring device on board. Based on the monitoring sub-strategy in the fire monitoring strategy, the monitoring center is controlled to call the operation video monitoring equipment of the corresponding operation unit to report the collected monitoring images, so as to realize the joint fire monitoring of the image and temperature of the target park.

8. The park drone safety linkage inspection method according to claim 1, characterized in that: Receive the updated task information uploaded by each operation unit and the updated operation environment information issued by the operation environment database, and update the fire monitoring strategy of the inspection drone and operation video surveillance. The specific steps include: Receive the updated work task information uploaded by each work unit and the updated work environment information issued by the work environment database, and determine whether the fire monitoring strategy can be regenerated without changing the fire inspection route of the inspection drone; If so, based on the fire inspection route of the patrol drone, the monitoring video reporting time point of each operating unit in the fire monitoring strategy is regenerated; if not, the fire inspection route of each patrol drone and the monitoring video reporting time point of each operating unit in the fire monitoring strategy are regenerated.

9. A park drone safety linkage inspection system, characterized by: include: An acquisition module is used to obtain a list of work tasks of several work units in a target park; wherein the list of work tasks stores the work content of each work task according to the estimated work period; An extraction module is configured to extract the work content and estimated work period of each work task, determine a first work feature associated with the work content for each work task using the work content, determine a second work feature associated with the work environment for each work task using the estimated work period, and construct a work feature set; A prediction module is used to predict the fire occurrence probability and fire correlation degree ranking table of the work unit of each work task in each minimum monitoring cycle within the target monitoring period based on the work feature set; specifically including: Based on the operation feature set and the estimated operation period and location of each operation task, a feature vector and an operation unit association matrix of each operation unit are constructed, and a three-dimensional composite feature tensor is generated by fusion as a fire prediction sample for each operation task in the target park; An initial convolutional neural network model was constructed, consisting of a convolutional layer, a pooling layer, a fully connected layer, a fire probability prediction branch, and a relevance ranking branch. Using fire training samples based on historical fire accidents, a joint loss function was constructed to balance the two tasks of fire probability prediction and relevance ranking, completing the training of the initial convolutional neural network model and obtaining a fire prediction model. Inputting the fire prediction samples into a fire prediction model, obtaining the fire occurrence probability and the operation unit correlation of each operation unit in each minimum monitoring cycle within a target monitoring period output by the fire prediction model, and generating a fire correlation operation unit ranking table according to the operation unit correlation; The control module is used to query the inspection drone call information of the target park during the target monitoring period, analyze the inspection capability of each inspection drone, consider the preset mapping relationship between the fire occurrence probability and inspection frequency requirements of the work unit of each operation task and the fire correlation operation unit ranking table, and control the inspection drone and operation video surveillance to perform joint fire monitoring of the target park's image and temperature during the target monitoring period; The update module is used to receive the work task update information uploaded by each work unit and the work environment update information issued by the work environment database, and update the fire monitoring strategy of the inspection drone and work video surveillance.

Images