Underground garage fire intelligent early warning method and system based on multi-source perception

Abstract

The invention discloses an underground garage fire intelligent early warning method and system based on multi-source perception, and belongs to the technical field of intelligent safety monitoring, and the method comprises the steps: collecting the attribute information of a vehicle entering a garage, so as to generate an early warning scheme; obtaining corresponding detection data based on a corresponding sensing source according to the early warning scheme; and performing accident prediction according to the detection data, and executing a corresponding fireproof plan. Attribute information of a vehicle entering a garage is collected to generate an early warning scheme, a reasonable scheme can be formulated in advance, and preparation is made to reduce the risk and cost of subsequent processing; according to the early warning scheme, corresponding detection data is obtained based on a corresponding sensing source, and the monitoring efficiency can be improved through a targeted detection mode; and accident prediction is performed according to the detection data, and a corresponding fireproof plan is executed, so that the prevention cost of the overall operation cost can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent safety monitoring, and in particular to an intelligent early warning method and system for underground garage fires based on multi-source perception. Background Art

[0002] Garages are important places for vehicles to be stored. On the one hand, they can protect vehicles, and on the other hand, they can relieve the congestion of urban roads. Among them, underground garages are valued and have been built in large quantities because they are close to residential areas and can be built in multiple floors to accommodate more vehicles.

[0003] However, because underground garages are built underground, timely rescue efforts are difficult in the event of a fire. Furthermore, due to cost constraints, it's difficult to deploy a sufficient amount of surveillance equipment to monitor every corner. Furthermore, even if a large number of surveillance devices were available, existing technologies lack a suitable monitoring, detection, and response mechanism. Summary of the Invention

[0004] The present invention proposes an intelligent early warning method and system for underground garage fires based on multi-source perception to solve the problem of insufficient monitoring, detection and processing capabilities in the existing technology.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] An intelligent early warning method for underground garage fires based on multi-source perception includes: collecting attribute information of vehicles entering the garage to generate an early warning plan; obtaining corresponding detection data based on the early warning plan and corresponding sensing sources; predicting accidents based on the detection data and executing the corresponding fire prevention plan.

[0007] Furthermore, the attribute information includes vehicle model, license plate, age, real-time driver, and motion parameters; correspondingly, the attribute information of vehicles entering the garage is collected to generate an early warning plan, including: collecting accident data from public information to extract accident factors based on the vehicle model, and determining the features to be monitored based on the accident factors; using the license plate as a tracking clue for the vehicle; obtaining the age of the vehicle through public information and personal information to decide whether to focus on the vehicle; identifying the real-time driver to determine the degree of risk based on personal information and behavioral information; and pre-planning the control route based on the motion parameters and the layout of the underground garage.

[0008] Furthermore, the perception source includes equipment and personnel; the detection data includes visual data, radar data, temperature data, humidity data, and light intensity data; correspondingly, the license plate, the vehicle model, the driver, and the motion parameters are identified through a camera; or, at least one of the corresponding license plate information, vehicle model information, driver information, and motion parameter information is obtained through communication interaction between the vehicle and the site; the license plate, the vehicle model, and the driver are identified by personnel, and the corresponding license plate information, vehicle model information, and driver information are uploaded; and at least one of the visual data, the radar data, the temperature data, the humidity data, and the light intensity data is obtained through a sensor group.

[0009] Furthermore, the method of determining accident data from public information based on the vehicle model to extract accident factors includes: identifying the vehicle model information based on the vehicle logo and vehicle appearance, searching for accident records based on accident-related keywords based on online public information, and searching for corresponding official analysis results based on the accident records; identifying the accident phenomenon, accident cause and accident prevention measures from the official analysis results based on text analysis to extract the accident factors; correspondingly, determining the features to be monitored based on the accident factors includes: determining the type of data acquisition equipment and selecting the corresponding image processing algorithm based on the accident phenomenon; correspondingly, tracking the vehicle based on the license plate to select the data acquisition equipment at the corresponding position and the corresponding working time.

[0010] Furthermore, obtaining the vehicle age through public information and personal information includes: obtaining the vehicle age of the corresponding vehicle from the management entity based on the license plate information; obtaining the vehicle age of the corresponding vehicle from the vehicle manager; correspondingly, deciding whether to focus on the vehicle includes: processing the accident factors based on weights to obtain an accident risk formula; calculating a corresponding risk value using the vehicle age and the accident risk formula, and deciding whether to focus on the vehicle based on the risk value.

[0011] Furthermore, an artificial intelligence model is set up to process the accident factors to form an accident prediction model; the accident prediction model is continuously updated by obtaining real-time accident records; and the vehicle is a new energy vehicle.

[0012] Furthermore, based on the accident factors, visual data of public information and / or accident simulation visual data are collected to obtain original samples; the original samples are classified based on the accident factors to form training samples, and an accident simulation model is established based on the training samples; based on the accident simulation model and the detection data, accident simulation data is output to perform accident prediction.

[0013] Furthermore, the execution of the corresponding fire prevention plan includes: determining the risk value of the current target based on the accident risk formula, determining the detection data collection plan based on the risk value, and collecting corresponding real-time detection data; executing the corresponding early warning plan based on the real-time detection data; the early warning plan includes further obtaining the detection data, on-site manual inspection, calling safety equipment, notifying the responsible department and executing route control.

[0014] Furthermore, the real-time driver identification is used to determine the risk level based on personal information and behavioral information, including: comparing the driver's past driving records in the garage, and determining the risk level based on the difference in driving parameters; the control route is pre-planned based on the motion parameters and the layout of the underground garage, including: if the risk level of the vehicle is greater than a threshold, it is marked as a dangerous vehicle, and the control route is pre-planned based on the motion parameters and the layout of the underground garage to isolate the dangerous vehicle from other vehicles; route control is performed by closing doors, issuing sound and light reminders, and communicating with vehicles.

[0015] An intelligent early warning system for underground garage fires based on multi-source sensing, including:

[0016] The first module collects attribute information of vehicles entering the garage to generate early warning plans;

[0017] The second module acquires corresponding detection data based on the corresponding sensing source according to the early warning plan;

[0018] The third module predicts accidents based on the detection data and implements corresponding fire prevention plans.

[0019] Due to the adoption of the above technical solution, the present invention has the following beneficial effects:

[0020] 1. The present invention generates an early warning plan by collecting attribute information of vehicles entering the garage, so that a reasonable plan can be formulated in advance and preparations can be made to reduce the risks and costs of subsequent processing; according to the early warning plan, corresponding detection data can be obtained based on the corresponding perception source, and the efficiency of monitoring can be improved through targeted detection; accident prediction is carried out based on the detection data, and the corresponding fire prevention plan is implemented, which can reduce the overall operating cost and prevention cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a flow chart of an intelligent early warning method for underground garage fire based on multi-source perception proposed by the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0023] like Figure 1 The method shown is an intelligent early warning method for underground garage fires based on multi-source perception, including: S1, collecting attribute information of vehicles entering the garage to generate an early warning plan; S2, according to the early warning plan, based on the corresponding perception source, to obtain corresponding detection data; S3, based on the detection data, predicting accidents and executing the corresponding fire prevention plan.

[0024] Garages are buildings, primarily consisting of non-combustible concrete and metal structures. Other components include decorative materials, auxiliary structures like electrical circuits, and temporary storage. Combustible components, due to past accidents, are a particular concern in on-site safety management and are therefore controllable.

[0025] Currently, the most dangerous element in garages is vehicles. Vehicles move by converting chemical energy into kinetic and electrical energy. Therefore, in the event of an accident, vehicles are the most vulnerable to fire, and the resulting fire is particularly dangerous. Vehicles are a key target for preventing garage disasters and accidents. By collecting attribute information about vehicles entering the garage and generating early warning plans based on this information, we can provide appropriate warning plans based on the most dangerous elements. This allows for the development of appropriate plans and preparations in advance to reduce the risks and costs of subsequent handling.

[0026] The early warning scheme needs to rely on corresponding sensing sources to maximize the use of the garage's existing detection methods to improve detection efficiency. At the same time, targeted detection can improve monitoring efficiency.

[0027] It is best to prevent accidents. If that is not possible, it is best to detect them in time. If that is still not possible, it is best to have sufficient practical means to reduce losses. Accident prediction based on the detection data and implementation of the corresponding fire prevention plan can reduce the overall operating cost and prevention cost. The fire prevention plan includes dispatching personnel, pre-starting firefighting equipment, notifying the fire department, etc.

[0028] The attribute information includes vehicle model, license plate, age, real-time driver, and motion parameters; correspondingly, the attribute information of vehicles entering the garage is collected to generate an early warning plan, including: collecting accident data from public information to extract accident factors based on the vehicle model, and determining the features to be monitored based on the accident factors; using the license plate as a tracking clue for the vehicle; obtaining the age of the vehicle through public information and personal information to decide whether to focus on the vehicle; identifying the real-time driver to determine the degree of risk based on personal information and behavioral information; and pre-planning the control route based on the motion parameters and the layout of the underground garage.

[0029] Vehicle safety is difficult to assess without actual functional inspections. However, vehicles are mature industrial products, and their quality is guaranteed. Accordingly, if an anomaly or accident occurs, the cause can be properly analyzed and classified. The resulting classification is attribute information, including vehicle model, license plate, age, real-time driver, and motion parameters. The vehicle model corresponds to the vehicle's structural design, with different structural designs representing varying degrees of risk. The license plate is used to avoid data collection and analysis conflicts when the same type of vehicle appears simultaneously. Vehicle age, similar to the vehicle model, represents the probability of structural failure. The real-time driver indicates risks beyond the vehicle's structure. Motion parameters are used for subsequent tracking and collision risk analysis.

[0030] Accident data is collected from public sources based on the vehicle model to extract accident factors. Features to be monitored are then determined based on these factors. This is based on the principle that news reports or government announcements inform the public of accident-related information. Accident factors, such as the accident name, time of occurrence, location, cause, and handling results, can be extracted from this publicly available information. This information can be obtained through technical means, such as confirming the time of occurrence using a clock or the location using location tracking. This information provides a general understanding of factors that require attention, such as whether certain environments or time periods are prone to accidents. This information can be used as a reference to determine which factors should be focused on. Vehicles are mobile, and using license plates as tracking clues can help identify sources of danger, facilitating subsequent monitoring, analysis, and action. The older a vehicle is, the more likely it is to develop structural abnormalities, thus requiring greater attention. The same vehicle may be driven by different people. Different personalities can negatively impact the vehicle's performance while driving, leading to other issues, such as smoking and leaving the engine on. Risk levels can be determined by using personal and behavioral information to predict risk. The means to prevent the expansion of accidents is to place the risk source in a controllable position, such as near fire-fighting equipment, or in an isolated corner to avoid harming other things. Based on the movement parameters and the layout of the underground garage, pre-planning of the control route is carried out. On the one hand, the risk vehicle itself can be guided, and on the other hand, the probability of harm to other vehicles or personnel can be reduced by allowing other vehicles / people to avoid the risk vehicle.

[0031] The perception source includes equipment and personnel; the detection data includes visual data, radar data, temperature data, humidity data, and light intensity data; correspondingly, the license plate, the vehicle model, the driver, and the motion parameters are identified through a camera; or, at least one of the corresponding license plate information, vehicle model information, driver information, and motion parameter information is obtained through communication interaction between the vehicle and the site; the license plate, the vehicle model, and the driver are identified by personnel, and the corresponding license plate information, vehicle model information, and driver information are uploaded; at least one of the visual data, the radar data, the temperature data, the humidity data, and the light intensity data is obtained through a sensor group.

[0032] As mentioned above, underground garages may not have comprehensive surveillance equipment due to cost or design considerations. However, human assistance can improve monitoring efficiency and reduce risk. Specifically, detection data includes visual data, radar data, temperature data, humidity data, and light intensity data. These data are fire-related factors that require attention, such as the appearance of flames (visual), unusual parking, location, or behavior (radar), temperature data (fire symptoms), humidity data (fire environment), and light intensity data (fire symptoms).

[0033] As for communication between personnel and the site, data can be uploaded through a mobile phone app, conversations can be conducted through intercoms, calls can be made via mobile phones, and indirect communication can be achieved through various switches / alarm buttons set on the site. This dual combination of people and equipment can effectively carry out safety management.

[0034] The method of determining accident data from public information based on the vehicle model to extract accident factors includes: identifying the vehicle model information based on the vehicle logo and vehicle appearance, searching for accident records based on accident-related keywords based on online public information, and searching for corresponding official analysis results based on the accident records; identifying the accident phenomenon, accident cause and accident prevention measures from the official analysis results based on text analysis to extract the accident factors; correspondingly, determining the features to be monitored based on the accident factors includes: determining the type of data acquisition equipment and selecting the corresponding image processing algorithm based on the accident phenomenon; correspondingly, tracking the vehicle based on the license plate to select the data acquisition equipment at the corresponding position and the corresponding working time.

[0035] The appearance of a vehicle, or its shape, is not significantly altered unless it has undergone specific modifications, and the image recognition template can be used to identify the vehicle model. The vehicle logo is a highly recognizable element, and general vehicle model information is also associated with the vehicle logo. Combining this with the vehicle logo can further improve recognition efficiency. Vehicle model information is used to describe the vehicle's specific information, including its structure, appearance, and technical specifications. Accident records are obtained by searching for accident-related keywords in publicly available online data. These keywords can include spontaneous combustion, collision combustion, arson, and other abnormal combustion.

[0036] Based on the accident's location, time, location, and vehicle information, accident records and corresponding official analysis results are obtained. The accident phenomena, causes, and preventative measures are identified from these official analysis results. These factors are then refined and refined to identify key details of the accident phenomena, causes, and preventative measures. Key details include technical names and corresponding logical relationships. The technical names represent the phenomena that occurred, while the logical relationships represent the causes and sequence of these phenomena. For example, a front-end collision leads to an engine fire, a bottom collision leads to a battery fire, or human error leads to a collision fire. Determining the type of data acquisition equipment and selecting the appropriate image processing algorithm based on the accident phenomena can improve recognition efficiency, thereby reducing accident risks and losses. Tracking vehicles based on license plates to select data acquisition equipment at the corresponding locations and operating times can improve early warning and recognition efficiency, thereby reducing accident risks and losses.

[0037] The obtaining of the vehicle age through public information and personal information includes: obtaining the age of the corresponding vehicle from a management entity based on license plate information; obtaining the age of the corresponding vehicle from a vehicle manager; correspondingly, deciding whether to focus on the vehicle includes: processing the accident factors based on weights to obtain an accident risk formula; calculating a corresponding risk value using the vehicle age and the accident risk formula, and deciding whether to focus on the vehicle based on the risk value.

[0038] Generally speaking, experienced drivers are less likely to cause accidents. However, other types of drivers can also cause accidents, such as those who are careless behind the wheel of someone else's car or those who are unfamiliar with the car. Some car models may be too large for small parking lots, while others may be unsuitable for high-temperature parking lots. Accident factors vary greatly, so it's difficult to generalize. Without careful analysis, we can only roughly calculate the risk. Specifically, we assign weights to different accident factors, or to different combinations of factors. Then, based on actual accident data, we can develop a formula for the accident risk.

[0039] Commercial products generally have a shelf life. The older the vehicle, the greater the risk and the more attention it requires.

[0040] An artificial intelligence model is set up to process the accident factors to form an accident prediction model; the accident prediction model is continuously updated by obtaining real-time accident records; the vehicle is a new energy vehicle.

[0041] In the general local data processing model, the algorithm update requires personnel to go to the site for modulation, and the stored processing algorithm is updated slowly; generally, companies are able to collect too few samples, resulting in low accuracy of algorithm processing.

[0042] By utilizing AI intelligent models available on the market, efficient data processing capabilities can be achieved to form accident prediction models. Specifically, by obtaining real-time accident records, the accident prediction model can be continuously updated. In this way, the needs of garages can be met under cost constraints, and the algorithm can continuously follow up on actual vehicle models and make corresponding improvements, which is particularly suitable for new energy vehicles that are updated quickly. If the battery of a new energy vehicle burns, it will generate high heat and at high speed, which requires special attention.

[0043] Based on the accident factors, visual data from public materials and / or accident simulation visual data are collected to obtain original samples; based on the accident factors, the original samples are classified to form training samples, and an accident simulation model is established based on the training samples; based on the accident simulation model and the detection data, accident simulation data is output to perform accident prediction.

[0044] The widespread use of surveillance cameras allows for the acquisition of a wealth of visual data from accidents. Automakers also conduct their own accident simulations and generate corresponding visual simulation data. By using publicly available visual data and / or accident simulation data, we can generate raw samples. This large amount of sample data improves the efficiency of subsequent model training.

[0045] The original samples are classified based on the accident factors to form training samples, and an accident simulation model is established based on the training samples. This can be combined with existing virtual (simulation) and real (accident) data to improve the depth and authenticity of the training samples and improve the efficiency and accuracy of the training model.

[0046] Based on the accident simulation model and the detection data, accident simulation data is output to perform accident prediction, which can facilitate preventive work to reduce risks.

[0047] The execution of the corresponding fire prevention plan includes: determining the risk value of the current target based on the accident risk formula, determining the detection data collection plan based on the risk value, and collecting corresponding real-time detection data; based on the real-time detection data, executing the corresponding early warning plan; the early warning plan includes at least one of further obtaining the detection data, on-site manual inspection, calling safety equipment, notifying the responsible department and executing route control.

[0048] Modern data processing equipment has high setup and operating costs. To reduce costs or improve data processing efficiency, a data collection plan is determined based on the risk value, and corresponding real-time data is collected. This reduces data collection costs while improving data processing efficiency. For example, data from areas with high risk values ​​is collected and processed in real time, or backup equipment in the corresponding area is activated to obtain more data.

[0049] Early warning plans are used to reduce risks, specifically including: improving monitoring efficiency by further acquiring the detection data (adding other data or repeating acquisition); on-site manual inspections to immediately handle anomalies, or discovering problems that are difficult for equipment to detect through close inspections; improving firefighting efficiency by reducing the time it takes for safety equipment to be deployed when anomalies actually occur; notifying responsible departments to allow personnel to prepare in advance so that they can be dispatched immediately when anomalies actually occur; and implementing route control to keep other vehicles away from the area to avoid harming innocent people.

[0050] The identifying of the real-time driver to determine the risk level based on personal information and behavioral information includes: comparing the driver's past driving records in the garage, and determining the risk level based on the difference in driving parameters; pre-planning of the control route based on the motion parameters and the layout of the underground garage includes: if the risk level of the vehicle is greater than a threshold, marking it as a dangerous vehicle, and pre-planning of the control route based on the motion parameters and the layout of the underground garage to isolate the dangerous vehicle from other vehicles; and performing route control by closing doors, issuing sound and light reminders, and communicating with vehicles.

[0051] After all, vehicles are used by people, and abnormal situations will always occur, such as the driver being in a bad mood or in poor health, etc. These abnormalities will be reflected in the driving parameters. The degree of risk is determined based on the difference in driving parameters. Specifically, the greater and more drastic the changes in driving parameters, the higher the degree of abnormality and the greater the risk. According to the on-site settings, that is, based on the movement parameters and the layout of the underground garage, the control route is pre-planned, including: if the risk level of the vehicle is greater than the threshold, it is marked as a dangerous vehicle. Based on the movement parameters and the layout of the underground garage, the control route is pre-planned to isolate the dangerous vehicle from other vehicles; the risk can be reduced by closing the door, issuing sound and light reminders, and communicating with the vehicle to control the route.

[0052] An intelligent early warning system for underground garage fires based on multi-source sensing, including:

[0053] The first module collects attribute information of vehicles entering the garage to generate early warning plans;

[0054] The second module acquires corresponding detection data based on the corresponding sensing source according to the early warning plan;

[0055] The third module predicts accidents based on the detection data and implements corresponding fire prevention plans.

[0056] The above description is a detailed description of the preferred embodiments of the present invention, but the embodiments are not intended to limit the scope of the patent application of the present invention. Any equivalent changes or modifications completed under the technical spirit suggested by the present invention should fall within the patent scope covered by the present invention.

Claims

1. An intelligent early warning method for underground garage fire based on multi-source perception, characterized in that: include: Collect attribute information of vehicles entering the garage to generate early warning plans; According to the early warning scheme, based on the corresponding sensing source, corresponding detection data is obtained; Accident prediction is performed based on the detection data, and corresponding fire prevention plans are implemented.

2. The intelligent early warning method for underground garage fire based on multi-source perception according to claim 1 is characterized in that: The attribute information includes vehicle model, license plate, vehicle age, real-time driver, and movement parameters; Correspondingly, the collection of attribute information of vehicles entering the garage to generate an early warning plan includes: Based on the vehicle model, accident data is collected from public information to extract accident factors, and features to be monitored are determined based on the accident factors; Using the license plate as a tracking clue for the vehicle; Obtain the vehicle age through public and personal information to decide whether to focus on the vehicle; identifying the real-time driver to determine risk based on personal information and behavioral information; Based on the movement parameters and the layout of the underground garage, a control route is pre-planned.

3. The intelligent early warning method for underground garage fire based on multi-source perception according to claim 2 is characterized in that: The sensing sources include equipment and personnel; the detection data include visual data, radar data, temperature data, humidity data, and light intensity data; Correspondingly, the license plate, the vehicle model, the driver and the motion parameters are identified through a camera; Alternatively, at least one of corresponding license plate information, vehicle model information, driver information, and motion parameter information is obtained through communication interaction between the vehicle and the site; The personnel identify the license plate, the vehicle model, and the driver, and upload the corresponding license plate information, vehicle model information, and driver information; At least one of the visual data, the radar data, the temperature data, the humidity data, and the light intensity data is acquired through a sensor group.

4. The intelligent early warning method for underground garage fire based on multi-source perception according to claim 3 is characterized in that: Determining accident data from public data based on the vehicle model to extract accident factors includes: Based on the vehicle logo and appearance, the vehicle model information is obtained by identification. Based on the publicly available information on the Internet, accident-related keywords are searched to obtain accident records. The corresponding official analysis results are found based on the accident records. Based on text analysis, the accident phenomenon, accident cause and accident prevention measures are identified from the official analysis results to extract the accident factors; Correspondingly, determining the features to be monitored based on the accident factors includes: According to the accident phenomenon, determine the type of data acquisition equipment and select the corresponding image processing algorithm; Correspondingly, the vehicle is tracked according to the license plate to select the data collection device at the corresponding position and the corresponding working time.

5. The intelligent early warning method for underground garage fire based on multi-source perception according to claim 4 is characterized in that: The vehicle age is obtained through public information and personal information, including: Based on the license plate information, the age of the corresponding vehicle is obtained from the management entity; the age of the corresponding vehicle is learned from the vehicle manager; Correspondingly, the process of determining whether to focus on the vehicle includes: Processing the accident factors based on weights to obtain an accident risk formula; The corresponding risk value is calculated based on the vehicle age and the accident risk formula, and a decision is made based on the risk value whether to focus on the vehicle.

6. The intelligent early warning method for underground garage fire based on multi-source perception according to claim 5 is characterized in that: Setting an artificial intelligence model to process the accident factors to form an accident prediction model; By acquiring real-time accident records, the accident prediction model is continuously updated; The vehicle is a new energy vehicle.

7. The intelligent early warning method for underground garage fire based on multi-source perception according to claim 6 is characterized in that: Based on the accident factors, visual data from public sources and / or accident simulation visual data are collected to obtain an original sample; classifying the original samples based on the accident factors to form training samples, and establishing an accident simulation model based on the training samples; Based on the accident simulation model and the detection data, accident simulation data is output to perform accident prediction.

8. The intelligent early warning method for underground garage fire based on multi-source perception according to claim 7 is characterized in that: The execution of the corresponding fire prevention plan includes: Based on the accident risk formula, determine the risk value of the current target, based on the risk value, determine the detection data collection plan, and collect corresponding real-time detection data; Based on the real-time detection data, executing the corresponding early warning plan; The early warning plan includes at least one of further obtaining the detection data, on-site manual inspection, calling safety equipment, notifying the responsible department and executing route control.

9. The intelligent early warning method for underground garage fire based on multi-source perception according to claim 8 is characterized in that: The real-time driver identification to determine the risk level based on personal information and behavioral information includes: Comparing the driver's past driving records in the garage, and determining the risk level based on the difference in driving parameters; The pre-planning of the control route based on the movement parameters and the layout of the underground garage includes: If the risk level of a vehicle is greater than a threshold, it is marked as a dangerous vehicle. Based on the movement parameters and the layout of the underground garage, a control route is pre-planned to isolate the dangerous vehicle from other vehicles. Route control is carried out by closing doors, issuing sound and light reminders, and communicating with vehicles.

10. An intelligent early warning system for underground garage fire based on multi-source perception, characterized in that: include: The first module collects attribute information of vehicles entering the garage to generate early warning plans; The second module acquires corresponding detection data based on the corresponding sensing source according to the early warning plan; The third module predicts accidents based on the detection data and implements corresponding fire prevention plans.

Images