CN114358663B - Artificial intelligence-based electric automobile post-fire accident comprehensive judgment method - Google Patents
Artificial intelligence-based electric automobile post-fire accident comprehensive judgment method Download PDFInfo
- Publication number
- CN114358663B CN114358663B CN202210262071.3A CN202210262071A CN114358663B CN 114358663 B CN114358663 B CN 114358663B CN 202210262071 A CN202210262071 A CN 202210262071A CN 114358663 B CN114358663 B CN 114358663B
- Authority
- CN
- China
- Prior art keywords
- area
- fire
- vehicle
- bms
- automobile
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention belongs to the technical field of new energy automobile fire investigation, and particularly relates to an artificial intelligence-based electric automobile post-fire accident comprehensive judgment method; firstly establishing an original database under a normal state, then judging whether the BMS of a burning vehicle is burnt or not, if not, investigating data recorded in the BMS, and finally determining a burning part, if the BMS is burnt, establishing a fact database of an accident vehicle, comparing the real-time database of the accident vehicle with the original database, then judging a suspected burning area according to information such as the paint color change degree of the accident vehicle, the position of burning or extinguishing parts and the like, then judging the burning area in the suspected burning area according to the paint color change and the existing parts with the highest melting point, and finally judging the burning type according to the position of the burning area.
Description
Technical Field
The invention belongs to the technical field of new energy automobile safety, and particularly relates to an artificial intelligence-based electric automobile post-fire accident comprehensive judgment method.
Background
According to incomplete statistics, the number of car burning accidents reported by media in the period from 1 to 12 months in 2020 is 124, wherein the number of accidents in 7 months, 8 months and 9 months accounts for 49% of all-year accidents. In general, possible causes and features include: 1. the vehicle base number becomes large. 2. From the statistics of accidents, the vehicle types with accidents are often concentrated on several vehicle types. Even if the safety of a new battery is greatly improved, the existing high-risk vehicle models still have a thunder area, and the hidden danger of the vehicles can not be eliminated until the vehicles gradually quit from use. 3. At present, no complete battery fault defect monitoring, discovering and recalling processing mechanism is formed in the industry.
After a fire accident occurs to the electric automobile, in order to investigate the cause of the fire accident, the fire part is generally determined, and then the cause of the fire is analyzed according to the damage condition of parts of the fire part and the working principle of related parts; according to the findings of the investigation of multiple accidents of a new energy fire vehicle by the identification center of the Xihua traffic judicial, the existing method for determining the fire position mainly depends on the experience of investigators, particularly under the conditions of relatively complex field environment and relatively serious vehicle burning loss, the fire position and the fire spreading trend cannot be determined quickly, and no method for quickly judging the fire position exists; for most of current electric automobiles, a power battery is located below a chassis, the traditional investigation mode is more used for fuel vehicles, the traditional fire investigation process is also greatly different from that of a new energy electric vehicle, and no investigation method capable of quickly judging whether the power battery is on fire exists.
Disclosure of Invention
The invention aims to provide an artificial intelligence-based electric automobile post-fire accident comprehensive judgment method.
In order to realize the purpose of the invention, the technical scheme adopted by the invention is as follows: 1. an artificial intelligence-based electric vehicle post-fire accident comprehensive judgment method is characterized by comprising the following steps: the method comprises the following steps:
s1, establishing a normal-state electric automobile original database;
s2, collecting and establishing a real-time database of the burnt vehicles;
s3, searching a BMS battery management system of the burning vehicle and detecting whether the BMS is burnt or not, if the BMS is burnt or cannot be electrified to operate, entering S4, and if the BMS is not burnt and can be electrified to operate, entering S5;
s4, carrying out field investigation on the burning vehicles, comparing the original database of S1 with the real-time database of S2, and judging the ignition part by combining the final investigation result;
s5, consulting data information stored in the BMS, judging the type of fire, if the type of fire is the type of fire of the power battery, entering S6, and if the type of fire is not the type of fire of the power battery, entering S4;
s6, determining the specific ignition position of the power battery according to the temperature change value of each battery temperature utilization device of the electric vehicle recorded by the BMS;
the method for establishing the normal-state electric vehicle original database in the step S1 comprises the following steps:
s11, establishing a parameterized model of the electric automobile in a normal state in three-dimensional model software by using a laser scanner or a camera type scanner;
s12, extracting colors of a paint layer on the electric automobile body in a normal state and determining the color number of the paint layer in an image recognition or manual recognition mode, implanting the color number of the paint layer into the parameterized model of S11 to form an original database, wherein one parameterized model corresponds to one paint layer color number or a plurality of paint layer color numbers;
s13, dividing the automobile into a left half area and a right half area in a parameterized model by using the vertical plane of the longitudinal center of the automobile;
dividing the automobile into an engine compartment area, a front row area of a passenger compartment, a rear row area of the passenger compartment and a luggage compartment area along the front-rear direction of the automobile;
along the vertical direction of the automobile, dividing an area above the lower edge of the automobile window glass into an upper area, dividing an area below a lower beam of an automobile door into a bottom area, and dividing an area between the upper area and the bottom area into a middle area;
the method for establishing the burning loss vehicle real-time database in the S2 comprises the following steps:
s21, shooting side views, top views, front views and rear views of the two sides of the outer surface of the vehicle body of the accident vehicle respectively, then lifting the vehicle body, shooting a bottom view of the chassis, and forming a first photo group;
s22, shooting pictures of a plurality of angles in an engine compartment, a passenger compartment and a luggage compartment of a vehicle body of the accident vehicle to form a second picture group;
in the step S4, the method for comparing the original database of S1 with the real-time database of S2 includes:
s41, extracting the paint layer color number stored in S12 by using an image deep learning and identification algorithm, respectively identifying the first photo group and the second photo group, and finding out the region with the changed paint layer color;
s42, implanting the area determined in the step S41 into the area defined in the step S13, and finding out one or two areas with the largest ratio of the area of the change area of the paint layer in the area to the total area of the paint layer in the area from the area defined in the step S13 as the suspected ignition area.
Preferably, the step S4 further includes a component damage degree determination method, where the component damage degree determination method includes:
in the step S41, finding out parts that are burned or lost in the fire by the three-dimensional model of the burned vehicle, marking the area where the positions of the burned or lost parts are located in the parameterized model established in S1 as a burned part area, extracting the melting points of the parts in all the burned part areas, and marking the burned part area where the part with the highest melting point is located as a part high-temperature area;
in the step S42, if the suspected ignition area and the high-temperature part area are the same area, the area is an ignition area, and the process proceeds to step S44;
if not, marking the suspected ignition area and the high-temperature area of the part as areas to be determined, and entering S43;
s43, the investigator judges whether the area to be determined is a fire area or not and inputs the result into a real-time database of the burnt vehicle, if the area to be determined is the fire area, the operation enters S44, and if the area to be determined is not the fire area, the operation is ended and the result is output, and the fire area cannot be determined temporarily;
s44, the investigator judges the position of the ignition area, if the ignition area is positioned at the bottom of the vehicle, the process goes to S45, and if the ignition area is not positioned at the bottom, the process goes to S46;
s45, the investigator judges whether the power battery is on fire, if so, the investigator outputs a result that the power battery is on fire, and if not, the investigator enters S46;
and S46, searching combustible parts in original parts in the area in the original database, and outputting a suspected ignition type according to the combustion type of the combustible parts.
Preferably, in the step S46, the combustible part includes an electrical harness, a liquid storage tank, and a high-temperature component, and the combustion type of the combustible part includes an electrical harness short circuit, a liquid storage tank leakage, and a high-temperature component ignition.
Preferably, in the step S5, the determination process when referring to the data information stored in the BMS and determining the type of fire is as follows:
first, the BMS is detached from the burnt vehicle, the BMS is connected to the data terminal in a communication manner, data information recorded in the BMS is read by the data terminal, abnormal data information is found from the read data, it is determined whether the data information is recorded by the temperature utilization device based on the abnormal data information, and the process proceeds to S6 if the data information is recorded by the temperature utilization device, and the process proceeds to S4 if the data information is not recorded by the temperature utilization device.
Preferably, in the step S2, a three-dimensional model of the burned vehicle is created by using a three-dimensional laser scanning modeling or oblique photography modeling method.
The invention has the following beneficial effects:
the method of the invention firstly establishes an original database under the normal state of the vehicle, then judges whether the BMS of the burnt vehicle is burnt or not, if not, investigates the data recorded in the BMS, finally determines the ignition part, if the BMS is burnt or not, establishes a fact database of the accident vehicle, compares the real-time database of the accident vehicle with the original database, then judges a suspected ignition area according to the information of the paint color change degree, the position of the burnt or lost part and the like of the accident vehicle, then judges the ignition area according to the paint color change and the part with the highest melting point in the suspected ignition area, and finally judges the ignition type according to the position of the ignition area, thereby realizing the purpose of quickly determining the ignition area by the process, and changing the traditional manual judgment mode into the mode of combining manual judgment with the deep image learning identification, the efficiency of judging the fire position of the vehicle in the fire accident is greatly improved, the requirements on corresponding professionals are reduced, and the method is particularly suitable for the field of traffic accident fire investigation.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a state diagram of an electric vehicle after a combustion event;
FIG. 3 is a state diagram of the engine compartment of FIG. 2;
FIG. 4 is a front row view of the passenger compartment of FIG. 2;
FIG. 5 is a rear row view of the passenger compartment of FIG. 2;
FIG. 6 is a state diagram of the chassis burnout condition of FIG. 2;
FIG. 7 is a power electric layout of FIG. 2.
Detailed Description
The present invention will be described in detail and with reference to preferred embodiments thereof, but the present invention is not limited thereto.
Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
In the description of the present invention, it should be noted that the terms "first", "second", "third", etc. are used only for distinguishing the description, and are not intended to indicate or imply relative importance.
The terms "upper", "lower", "left", "right", "inner", "outer", and the like, refer to orientations or positional relationships based on orientations or positional relationships illustrated in the drawings or orientations and positional relationships that are conventionally used in the practice of the products of the present invention, and are used for convenience in describing and simplifying the invention, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore, should not be construed as limiting the invention.
Furthermore, the terms "vertical" and the like do not require absolute perpendicularity between the components, but may be slightly inclined. Such as "vertical" merely means that the direction is relatively more vertical and does not mean that the structure must be perfectly vertical, but may be slightly inclined.
In the description of the present invention, it is also to be noted that the terms "disposed," "mounted," "connected," and the like are to be construed broadly unless otherwise specifically stated or limited. For example, the connection can be fixed, detachable or integrated; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
Examples
As shown in fig. 1, an artificial intelligence-based method for comprehensively judging an electric vehicle after a fire accident includes the following steps:
s1, establishing a normal-state electric automobile original database;
s2, collecting and establishing a real-time database of the burnt vehicles;
s3, searching a BMS battery management system of the burning vehicle and detecting whether the BMS is burnt or not, if the BMS is burnt or cannot be electrified to operate, entering S4, and if the BMS is not burnt and can be electrified to operate, entering S5;
s4, carrying out field investigation on the burning vehicles, comparing the original database of S1 with the real-time database of S2, and judging the ignition part by combining the final investigation result;
s5, consulting data information stored in the BMS, judging the type of fire, if the type of fire is the type of fire of the power battery, entering S6, and if the type of fire is not the type of fire of the power battery, entering S4;
s6, determining the specific ignition position of the power battery according to the temperature change value of each battery temperature utilization device of the electric vehicle recorded by the BMS;
in this step, it should be noted that the data information in the BMS is recorded, but not limited to, the following information: position information of each temperature utilization device, a number of each temperature utilization device, data information monitored by each temperature utilization device during operation, and the like.
When consulting the data information that BMS stored and judging the type of catching fire, concrete judgement process is as follows: first, the BMS is detached from the burnt vehicle, the BMS is connected to a data terminal (e.g., a laptop, a PC, a mobile phone, etc.) in a communication manner, data information recorded in the BMS is read by the data terminal, abnormal data information is found from the read data, it is determined whether the data information is recorded by the temperature utilization device based on the abnormal data information, and the process proceeds to S6 if the data information is recorded by the temperature utilization device, or to S4 if the data information is not recorded by the temperature utilization device. Through the process, when the engine ignition part is searched and judged, the judgment can be carried out only through the personal experience of the investigator instead of the conventional technology, so that the accuracy and the authority of the result are improved, the workload of the investigator is reduced, the investigation efficiency is effectively improved, and the time for obtaining the investigation result is shortened.
And S6, determining the specific ignition position of the power battery 1 according to the temperature change value of each battery temperature utilization device of the electric vehicle recorded by the BMS.
In this step, when it is determined that the cause of ignition is the ignition of the power battery 1 in step S5, the position of the temperature utilization device in which the abnormality information is described is found from the BMS, and the specific ignition position of the power battery 1 is determined.
The step S1 includes the following steps:
s11, establishing a parameterized model of the electric automobile in a normal state in three-dimensional model software by using a laser scanner or a camera type scanner;
s12, extracting colors of the painting layer on the electric automobile body in a normal state, determining the color number of the painting layer, implanting the color number of the painting layer into the parameterized model of S11 to form an original database, wherein one parameterized model corresponds to one painting layer color number or a plurality of painting layer color numbers;
s13, dividing the automobile into a left half area and a right half area in a parameterized model by using the vertical plane of the longitudinal center of the automobile;
dividing the automobile into an engine compartment area, a front row area of a passenger compartment, a rear row area of the passenger compartment and a luggage compartment area along the front-rear direction of the automobile;
along the vertical direction of the automobile, the area above the lower edge of the automobile window glass is divided into an upper area, the area below the lower side beam of the automobile door is divided into a bottom area, and the area between the upper area and the bottom area is divided into a middle area.
In order to more clearly and clearly illustrate the method, as a further optimization of the method, the creation method of the parameterized model can be modeled by adopting a traditional three-dimensional software mode, and can also be directly modeled by a scanning mode by using a handheld laser scanner. During modeling, only the surface contour of the part is modeled, the internal structure of the part is not obtained, after the contour creation of each model is completed, the material properties of each part are given, the melting point of each material is given, and the color information of each part extracted from the vehicle body is given to each corresponding part in the model (for example, if the color of the front wheel hub of the vehicle is silver, the color of the front wheel hub in the model is also silver).
It should be noted that, during modeling, the three-dimensional spatial information of the created spatial model needs to correspond to the three-dimensional spatial information of each model of the actual vehicle (specific examples are: the coordinates of the starting point of the center of the left front wheel of the actual vehicle are (0, 0, 0), the coordinates of the center of the left front wheel in the three-dimensional model are also (0, 0, 0), the method has the advantages that terminal data information of the real-time database can be rapidly matched with the original database when the real-time database is compared with the original database in the later period, and further, the deviation of the final judgment result caused by inaccurate matching of the original database can be avoided.
It should be further explained that, in this embodiment, the purpose of partitioning the established model is to: after a fire accident, the burning degree of the automobile is completely uncertain, that is, after the fire accident occurs, the space coordinate information of automobile parts (electrical lines, tires, batteries and the like) which are easy to burn due to the burning of the parts of the electric automobile is lost, and the condition of inaccurate matching occurs when the real-time database is matched with the original database due to the loss of the information. The automobile is partitioned (and each partition comprises at least one positioning coordinate point), so that the real-time database of the accident automobile can be quickly matched with the original database in a partition matching manner by the partition matching method in the later process of judging the ignition position of the automobile, and the defect of inaccurate matching caused by burning loss or loss of parts in a fire is effectively overcome.
It should be particularly clear and explained that the data information that needs to be matched in the data matching described in this embodiment includes, but is not limited to, the following information: the space coordinate data of each part, the burning loss or loss degree of each part, the paint color change degree of each part and the like.
The real-time database in the step S2 is established by the following method:
s21, shooting side views, top views, front views and rear views of the two sides of the outer surface of the vehicle body of the accident vehicle respectively, then lifting the vehicle body, shooting a bottom view of the chassis, and forming a first photo group;
s22, shooting pictures of a plurality of angles in an engine compartment, a passenger compartment and a luggage compartment of a vehicle body of the accident vehicle to form a second picture group;
in the step S4, the method for comparing the original database of S1 with the real-time database of S2 includes:
s41, extracting the paint layer color number stored in S12 by using an image deep learning and identification algorithm, respectively identifying the first photo group and the second photo group, and finding out the region with the changed paint layer color;
s42, implanting the area determined in the step S41 into the area defined in the step S13, and finding out one or two areas with the largest ratio of the area of the change area of the paint layer in the area to the total area of the paint layer in the area in the step S13 as a suspected ignition area;
in order to explain the present invention in more detail, in the present embodiment, since the starting time points of the subsequent investigation of the automobile fire location are different, and after the fire accident occurs in each engine compartment, the investigation time of the fire accident may be immediately after the fire accident occurs or may be a long time after the fire accident occurs. The paint color change degree of each part in the automobile after a fire accident and the time of the part far away from the fire accident are changed in a mathematical relationship, and the paint color change degrees of each part at different time points are completely different, so that more complete data support needs to be provided for later investigation results. In the actual investigation process, when an automobile subjected to an actual fire accident is subjected to image acquisition, the images of the same part at the same angle cannot completely display information such as color change degree and paint color change degree of the part subjected to the fire accident, so that the aim of displaying the state of the same part subjected to the fire accident through multi-angle and multi-azimuth images is fulfilled.
In addition, the purpose of determining the time period of the vehicle on fire can be achieved by acquiring images of the same part in different time periods, and the specific determination method comprises the following steps: firstly, editing each image of each part collected in the first photo group and the second photo group, wherein the information contained in each image comprises information such as image collection time, the type of equipment used for collecting the image, the collection angle of the image and the like. And then, training by using a CNN convolutional neural network algorithm, wherein the main training content is to identify and distinguish paint color values of all parts in all collected images, implant the obtained paint color values into the parameterized model established in S1 after obtaining the paint color values, train the first photo group and the second photo group by using the same method, and cover the parameterized model established in S1 to form a new burning loss vehicle model. And then, calling the CNN convolutional neural network algorithm again to compare the burning loss vehicle model with the parameterized model in the original database, finding out the area of paint color change, and further obtaining the paint color change degree of each part. And finally, finding and defining a suspected fire area according to the paint color change degree.
It can be further stated that, as a preferred embodiment, in this embodiment, the training of the CNN convolutional neural network algorithm may be implemented based on tensoflow, and the training process is as follows:
firstly, realizing a neural network structure or a machine learning algorithm function;
secondly, defining an error function;
thirdly, defining a training function;
fourthly, defining an algorithm evaluation function;
and fifthly, realizing a complete program in the sess by using the functions.
Various functions adopted in the training process can be realized by adopting related research achievements in the technical field of the existing artificial intelligent image recognition, and the realization processes of the related functions and programs adopted in the training process are not the invention points of the invention, so that the detailed description is omitted.
In addition, the method for establishing the real-time database of the burnt vehicles can also be realized by adopting a modeling method. Modeling methods that may be used include, but are not limited to, three-dimensional laser scanning modeling, oblique photography modeling, and the like. The advantage of using this modeling method is that a burn-out vehicle model can be created quickly.
As a further optimization of the above solution, the step S4 includes the following steps:
in step S4, the method further includes determining a damaged degree of the component, where the method for determining the damaged degree of the component includes:
in the step S41, finding out parts that are burned or lost in the fire by the three-dimensional model of the burned vehicle, marking the area where the positions of the burned or lost parts are located in the parameterized model established in S1 as a burned part area, extracting the melting points of the parts in all the burned part areas, and marking the burned part area where the part with the highest melting point is located as a part high-temperature area;
in the step S42, if the suspected ignition area and the high-temperature part area are the same area, the area is an ignition area, and the process proceeds to step S44;
if not, marking the suspected ignition area and the high-temperature area of the part as areas to be determined, and entering S43;
s43, the investigator judges whether the area to be determined is a fire area or not and inputs the result into a real-time database of the burnt vehicle, if the area to be determined is the fire area, the operation enters S44, and if the area to be determined is not the fire area, the operation is ended and the result is output, and the fire area cannot be determined temporarily;
s44, the investigator judges the position of the ignition area, if the ignition area is positioned at the bottom of the vehicle, the process goes to S45, and if the ignition area is not positioned at the bottom, the process goes to S46;
s45, the investigator judges whether the power battery is on fire, if so, the investigator outputs a result that the power battery is on fire, and if not, the investigator enters S46;
and S46, searching combustible parts in original parts in the area in the original database, and outputting a suspected ignition type according to the combustion type of the combustible parts.
In order to further clarify and clarify the present invention, the following detailed description of the invention is given in connection with the practical case: as shown in fig. 2-7, the accident is investigated on a new energy fire vehicle at a certain time aiming at the west China traffic justice accreditation center, wherein fig. 2 and 3 are photographs of the appearance of the burnt part of the vehicle, fig. 4 and 5 are photographs of the interior decoration of the burnt part of the vehicle, fig. 6 is a photograph of the bottom part of an engine compartment after the burning of the vehicle, and fig. 7 is a schematic layout view of an electric battery of the vehicle; after the photos of fig. 2 and 3 and the photos with other appearances are shot, namely, the process of acquiring data images outside the automobile is realized, the photos are named as a first photo group, the images inside the automobile are acquired by using the same method, the photos are named as a second photo group, deep learning and recognition are simultaneously performed on the acquired first photo group and second photo group by using a CNN (convolutional neural network) algorithm, paint colors in the photo groups are extracted, the extracted paint color information is implanted into a parameterized model established by S1 to form a real-time database after the paint colors are extracted, the CNN convolutional neural network algorithm is called again, the parameterized model and the real-time model in the original database are compared by using the CNN convolutional neural network algorithm, an area with large paint color change degree is found from the CNN convolutional neural network algorithm, and the area is defined as a suspected ignition area.
In addition, the invention can also provide another judgment method, specifically. After the first photo group acquisition is carried out on the outer contour of the burning-out vehicle, the second photo group acquisition is carried out on the inner part, the acquisition is completed, the paint colors of an automobile engine cabin, the front row of a passenger cabin, the rear row of the passenger cabin and a ground plate are checked, the actual paint colors are checked and compared with the paint colors of the same position of the automobile in a normal state, and an area with the largest paint color change proportion is found out.
Compared with the other areas, the proportion of the color change area in the paint layer (the left front door inner plate, the left front seat, the left front A column and the left half part of the instrument desk framework) in the left area of the front row of the passenger compartment of the accident vehicle to the total area of the paint layer is about 80%, the middle area on the left side of the passenger compartment is determined to be a suspected area, and the power battery can be excluded from being on fire according to the graph 7. If the paint color change area is less or not obvious, the suspected ignition area is a power battery, and then the final ignition area is determined according to the corresponding battery temperature and the position of the device and the burning loss degree of the battery.
According to the scheme, an original database in a normal state is established, whether the BMS of a burning vehicle is burnt or not is judged, if the BMS of the burning vehicle is not burnt, data recorded in the BMS are investigated, a burning part is finally determined, if the BMS is burnt or not, a fact database of an accident vehicle is established, a real-time database of the accident vehicle is compared with the original database, a suspected burning area is judged according to information such as the paint color change degree of the accident vehicle, the position of the burning or extinguishing part and the like, then the burning area is judged according to the paint color change and the part with the highest melting point in the suspected burning area, the burning type is finally judged according to the position of the burning area, and the purpose of quickly determining the burning area is achieved through the process.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (7)
1. An artificial intelligence-based electric vehicle post-fire accident comprehensive judgment method is characterized by comprising the following steps: the method comprises the following steps:
s1, establishing a normal-state electric automobile original database;
s2, collecting and establishing a real-time database of the burnt vehicles;
s3, searching a BMS battery management system of the burning vehicle and detecting whether the BMS is burnt or not, if the BMS is burnt or cannot be electrified to operate, entering S4, and if the BMS is not burnt and can be electrified to operate, entering S5;
s4, carrying out field investigation on the burning vehicles, comparing the original database of S1 with the real-time database of S2, and judging the ignition part by combining the final investigation result;
s5, consulting data information stored in the BMS, judging the type of fire, if the type of fire is the type of fire of the power battery, entering S6, and if the type of fire is not the type of fire of the power battery, entering S4;
s6, determining the specific ignition position of the power battery according to the temperature change value of each battery temperature utilization device of the electric vehicle recorded by the BMS;
the method for establishing the normal-state electric vehicle original database in the step S1 comprises the following steps:
s11, establishing a parameterized model of the electric automobile in a normal state in three-dimensional model software by using a laser scanner or a camera type scanner;
s12, extracting colors of a paint layer on the electric automobile body in a normal state and determining the color number of the paint layer in an image recognition or manual recognition mode, implanting the color number of the paint layer into the parameterized model of S11 to form an original database, wherein one parameterized model corresponds to one paint layer color number or a plurality of paint layer color numbers;
s13, dividing the automobile into a left half area and a right half area by using the vertical plane of the longitudinal center of the automobile in the parameterized model;
dividing the automobile into an engine compartment area, a front row area of a passenger compartment, a rear row area of the passenger compartment and a luggage compartment area along the front-rear direction of the automobile;
along the vertical direction of the automobile, dividing an area above the lower edge of the automobile window glass into an upper area, dividing an area below a lower beam of an automobile door into a bottom area, and dividing an area between the upper area and the bottom area into a middle area;
the method for establishing the burning loss vehicle real-time database in the S2 comprises the following steps:
s21, shooting side views, top views, front views and rear views of the two sides of the outer surface of the vehicle body of the accident vehicle respectively, then lifting the vehicle body, shooting a bottom view of the chassis, and forming a first photo group;
s22, shooting pictures of a plurality of angles in an engine compartment, a passenger compartment and a luggage compartment of a vehicle body of the accident vehicle to form a second picture group;
in the step S4, the method for comparing the original database of S1 with the real-time database of S2 includes:
s41, extracting the paint layer color number stored in S12 by using an image deep learning and identification algorithm, respectively identifying the first photo group and the second photo group, and finding out the region with the changed paint layer color;
s42, implanting the area determined in the step S41 into the area defined in the step S13, and finding out one or two areas with the largest ratio of the area of the change area of the paint layer in the area to the total area of the paint layer in the area from the area defined in the step S13 as the suspected ignition area.
2. The artificial intelligence based electric vehicle post-fire accident comprehensive judgment method according to claim 1, characterized in that:
in step S4, the method further includes determining a damaged degree of the component, where the method for determining the damaged degree of the component includes:
in the step S41, finding out parts that are burned or lost in the fire of the three-dimensional model of the burned vehicle, marking the area where the positions of the burned or lost parts are located in the parameterized model established in S1 as a burned part area, extracting the melting points of the parts in all the burned part areas, and marking the burned part area where the part with the highest melting point is located as a part high-temperature area;
in the step S42, if the suspected ignition area and the high-temperature part area are the same area, the area is an ignition area, and the process proceeds to step S44;
if not, marking the suspected ignition area and the high-temperature area of the part as areas to be determined, and entering S43;
s43, the investigator judges whether the area to be determined is a fire area or not and inputs the result into a real-time database of the burnt vehicle, if the area to be determined is the fire area, the operation enters S44, and if the area to be determined is not the fire area, the operation is ended and the result is output, and the fire area cannot be determined temporarily;
s44, the investigator judges the position of the ignition area, if the ignition area is positioned at the bottom of the vehicle, the process goes to S45, and if the ignition area is not positioned at the bottom, the process goes to S46;
s45, the investigator judges whether the power battery is on fire, if so, the investigator outputs a result that the power battery is on fire, and if not, the investigator enters S46;
and S46, searching combustible parts in original parts in the area in the original database, and outputting a suspected ignition type according to the combustion type of the combustible parts.
3. The artificial intelligence based electric vehicle post-fire accident comprehensive judgment method according to claim 2, characterized in that: in the step S46, the combustible parts include an electrical harness, a liquid storage tank, and a high-temperature component, and the combustion type of the combustible parts includes an electrical harness short circuit, a liquid storage tank leakage, and a high-temperature component ignition.
4. The artificial intelligence based electric vehicle post-fire accident comprehensive judgment method according to claim 1, characterized in that: in the step S5, the determination process when referring to the data information stored in the BMS and determining the type of fire is as follows:
first, the BMS is detached from the burnt vehicle, the BMS is connected to the data terminal in a communication manner, data information recorded in the BMS is read by the data terminal, abnormal data information is found from the read data, it is determined whether the data information is recorded by the temperature utilization device based on the abnormal data information, and the process proceeds to S6 if the data information is recorded by the temperature utilization device, and the process proceeds to S4 if the data information is not recorded by the temperature utilization device.
5. The artificial intelligence based electric vehicle post-fire accident comprehensive judgment method according to claim 1, characterized in that: in the step S2, a three-dimensional model of the burned vehicle is also created by using a three-dimensional laser scanning modeling or oblique photography modeling method.
6. The artificial intelligence based electric vehicle post-fire accident comprehensive judgment method according to claim 1, characterized in that: in the step S11, when modeling the parametric model of the automobile, the ratio of the model to the actual part is 1: 1, modeling; an automobile parameterization model comprises a plurality of positioning coordinate points, and each positioning coordinate point is located on a metal part which is not easy to burn and lose deformation in a combustion experiment.
7. The artificial intelligence based post-fire accident comprehensive judgment method for electric vehicles according to claim 1, characterized in that: in the step S12, a CNN convolutional neural network algorithm is used for training, the paint color values of all parts in all collected images are identified and distinguished, and the obtained paint color values are implanted into the parameterized model established in S1 after the paint color values are obtained; in the step S41, the CNN convolutional neural network algorithm is called again, the first photo group and the second photo group are trained, and the parameterized model established in S1 is covered to form a new burn-out vehicle model; in the step S42, the burning vehicle model is compared with the parameterized model in the original database to find out the area of paint color change, and the paint color change degree of each part is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210262071.3A CN114358663B (en) | 2022-03-17 | 2022-03-17 | Artificial intelligence-based electric automobile post-fire accident comprehensive judgment method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210262071.3A CN114358663B (en) | 2022-03-17 | 2022-03-17 | Artificial intelligence-based electric automobile post-fire accident comprehensive judgment method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114358663A CN114358663A (en) | 2022-04-15 |
| CN114358663B true CN114358663B (en) | 2022-05-13 |
Family
ID=81095032
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210262071.3A Active CN114358663B (en) | 2022-03-17 | 2022-03-17 | Artificial intelligence-based electric automobile post-fire accident comprehensive judgment method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114358663B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114495626B (en) * | 2022-04-18 | 2022-07-05 | 天津市杰联机电设备安装技术有限公司 | Fire rescue VR training system |
| CN115752586A (en) * | 2022-11-22 | 2023-03-07 | 广州能源检测研究院 | New energy automobile fire reason analysis and judgment method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013122624A (en) * | 2011-12-09 | 2013-06-20 | Mitsubishi Electric Corp | Vehicle fire detection device |
| CN109598700A (en) * | 2018-10-16 | 2019-04-09 | 天津大学 | Using the incipient fire detection method of convolutional neural networks |
| KR102097120B1 (en) * | 2018-12-31 | 2020-04-09 | 주식회사 애자일소다 | System and method for automatically determining the degree of breakdown by vehicle section based on deep running |
| CN113643495A (en) * | 2021-07-29 | 2021-11-12 | 宜宾市消防救援支队 | Intelligent auxiliary analysis system and method for fire cause investigation |
-
2022
- 2022-03-17 CN CN202210262071.3A patent/CN114358663B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013122624A (en) * | 2011-12-09 | 2013-06-20 | Mitsubishi Electric Corp | Vehicle fire detection device |
| CN109598700A (en) * | 2018-10-16 | 2019-04-09 | 天津大学 | Using the incipient fire detection method of convolutional neural networks |
| KR102097120B1 (en) * | 2018-12-31 | 2020-04-09 | 주식회사 애자일소다 | System and method for automatically determining the degree of breakdown by vehicle section based on deep running |
| CN113643495A (en) * | 2021-07-29 | 2021-11-12 | 宜宾市消防救援支队 | Intelligent auxiliary analysis system and method for fire cause investigation |
Non-Patent Citations (5)
| Title |
|---|
| Real‑time video fre/smoke detection based on CNN in antifre surveillance systems;Sergio Saponara 等;《Journal of Real-Time Image Processing 》;20201110;第75702-75713页 * |
| 基于深度调查的电动汽车火灾原因分析技术;林烨 等;《消防科学与技术》;20210131;第40卷(第1期);第145-148页 * |
| 浅析新能源电动汽车火灾调查方法;张玉斌 等;《消防科学与技术》;20201015;第1456-1458页 * |
| 电动汽车电池安全事故分析与研究现状;陈泽宇 等;《机械工程学报》;20191231;第55卷(第24期);第93-104、116页 * |
| 长管拖车气瓶火烧损伤分析及拖车气瓶火烧损伤分析及安全评定;董红磊 等;《监察与检验》;20140330;第30卷(第3期);第23-30页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114358663A (en) | 2022-04-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114358663B (en) | Artificial intelligence-based electric automobile post-fire accident comprehensive judgment method | |
| CN109816024B (en) | Real-time vehicle logo detection method based on multi-scale feature fusion and DCNN | |
| CN108319907A (en) | A kind of vehicle identification method, device and storage medium | |
| CN113191459B (en) | Road-side laser radar-based in-transit target classification method | |
| CN109615572B (en) | Personnel intimacy degree analysis method and system based on big data | |
| CN103824037B (en) | Vehicle anti-tracking alarm device | |
| CN110619277A (en) | Multi-community intelligent deployment and control method and system | |
| CN111401188B (en) | Traffic police gesture recognition method based on human body key point characteristics | |
| CN108171708A (en) | Car damage identification method and system | |
| US11669947B2 (en) | Machine learning assisted image analysis | |
| CN110348505B (en) | Vehicle color classification model training method and device and vehicle color identification method | |
| CN109300127A (en) | Defect inspection method, device, computer equipment and storage medium | |
| CN111444840A (en) | Automatic detection method and system for fake-licensed vehicle | |
| CN112699801A (en) | Fire identification method and system based on video image | |
| CN111310844B (en) | Vehicle identification model construction method and device and identification method and device | |
| CN113360523A (en) | Multiple screening system based on vehicle big data | |
| CN113986893A (en) | Active early warning system and method based on generator car risk identification | |
| CN110175622B (en) | Vehicle component identification method and system based on convolutional neural network of symbiotic relationship | |
| CN111723695A (en) | Improved Yolov 3-based driver key sub-area identification and positioning method | |
| CN114723938B (en) | Method for quickly determining fire position of engine compartment during burning of vehicle in traffic accident | |
| CN113869122A (en) | Distribution network engineering reinforced control method | |
| CN113554610A (en) | Photovoltaic module operation state detection method and application device thereof | |
| CN111818269A (en) | Identification method and identification system of vehicle-mounted certificate | |
| CN116153099A (en) | Intelligent license plate recognition device for road supervision and recognition system thereof | |
| CN116740549B (en) | Vehicle part identification method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |