JP7157889B1 - Danger information estimation method, device, program, and firefighting command system for estimating danger at fire scene by artificial intelligence - Google Patents

Abstract

A method, an apparatus, a program, and a fire command system for estimating possible danger at a fire site are provided. A machine learning method is learning data on past injuries at a fire scene 99, data on causes of injuries, data on names of injuries and illnesses, data on time factors, and data on human factors of firefighters 108. a learning step of learning a machine learning model for estimating danger information at the fire scene 99 based on the learning data including the data; [Selection drawing] Fig. 1

Description

Application of Article 30, Paragraph 2 of the Patent Law (Part 1) (1) (1) Date of publication of the website May 25, 2021 (2) Website address https://www. kasaikenkyu. org/login/ (Part 2) ▲1 ▼Date May 29, 2021 ▲2 ▼Meeting name, venue 2021 Research Presentation Meeting of the Japan Society of Fire Science (held online, lecture number A-15) (Part 3) ) ▲ 1 ▼ Website publication date November 25, 2021 ▲ 2 ▼ Website address https://www. fesc119. net/lounge/denied? url=/magazine/&%2Flounge%2Flogin%3F%2Flounge%2Fdenied%3Furl=%2Fmagazine%2F&%2Flounge%2Fdenied%3Furl=%2Fmagazine%2F (Part 4) (1) Date December 8, 2021 Date ▲2 ▼Meeting name, venue AOSFST 2021 - 12th Asia-Oceania Symposium on Fire Science and Technology (held online, lecture number Track B, Evacuation and human behavior (II))

The present invention relates to a method, apparatus, system and program for estimating danger at a fire scene.

Since firefighting activities at a fire site are dangerous, firefighters who carry out firefighting activities or rescue activities for people requiring rescue may be injured or killed in the line of duty. It is required to ensure the safety of firefighters not only from the viewpoint of health and safety of firefighters, but also in order to perform the duties stipulated by the Fire Service Law. If a firefighter is injured or killed in the line of duty during an activity, more time will be required for firefighting and rescue activities. It is also important to reduce firefighters' injuries and fatalities during their activities in order to achieve the purpose of firefighting.

As a conventional technique for preventing or reducing occupational accidents and mistakes by workers, for example, Patent Literature 1 discloses a physical condition management system for workers. The physical condition management system analyzes the information on the physical condition input by the worker using the data stored in advance in the physical condition management database, and predicts the unsafe actions of the worker that lead to the occurrence of mistakes and occupational accidents. The physical condition management database stores basic information on the character and aptitude of workers, information on past occupational accidents, environmental information, and the like.

Japanese Patent Application Laid-Open No. 2020-198070

The physical condition management system described in Patent Literature 1 is mainly intended for workers in factories and the like. Since the situation at the fire site changes from moment to moment, there is a high possibility of injuries and fatalities occurring at the fire site where firefighters carry out firefighting activities, even compared to factory work sites. In order to prevent or reduce the occurrence of firefighters being injured or killed in the line of duty, it is required to pre-estimate possible dangers at fire sites.

An object of the present invention is to provide a method, an apparatus, a program, and a fire command system for estimating possible dangers at a fire site.

The present invention for solving the above problems includes, for example, the following aspects.
(Section 1)
Based on learning data related to past injuries at fire sites, including data related to causes of injuries, data related to names of injuries and diseases, data related to time factors, and data related to human factors of firefighters, A machine learning method comprising a learning step of training a machine learning model to estimate hazard information in the field.
(Section 2)
The data related to the temporal element includes at least one of a fire detection time and an activity stage;
2. The machine learning method of claim 1, wherein the data on the human element includes at least one of age group, rank, length of service, role within the platoon, and duties of the platoon of the firefighter.
(Section 3)
2. The machine learning method according to clause 1, wherein the learning data related to the injury incident further includes data related to spatial elements of the fire scene.
(Section 4)
a firefighting dispatch data acquisition step of acquiring firefighting dispatch data relating to the fire site to which the dispatch order has been received;
Estimate danger information at the fire site where the dispatch order is received, based on the firefighting dispatch data, according to the learned machine learning model learned by the machine learning method according to any one of items 1 to 3. a danger information estimation step;
A method of estimating danger information, including
(Section 5)
Item 5. The danger information estimation method according to Item 4, wherein the firefighting dispatch data includes the time when the fire was detected and data relating to human elements of the firefighters dispatched to the fire scene.
(Section 6)
Item 6. The danger information estimation method according to Item 5, wherein the firefighting dispatch data further includes data on spatial elements of the fire scene.
(Section 7)
Based on learning data related to past injuries at fire sites, including data related to causes of injuries, data related to names of injuries and diseases, data related to time factors, and data related to human factors of firefighters, A machine learning device comprising a learning unit that learns a machine learning model for estimating danger information in the field.
(Section 8)
a firefighting dispatch data acquisition unit that acquires firefighting dispatch data relating to a fire site that received a dispatch order;
A danger information estimation unit that estimates danger information at the fire site where the dispatch command is received, based on the firefighting dispatch data, according to the learned machine learning model learned by the machine learning device according to Item 7;
A danger information estimation device.
(Section 9)
Item 8. The danger information estimation device according to Item 8;
a notification unit that notifies the danger information estimated by the danger information estimation device;
Fire command system.
(Section 10)
Item 9. A program for causing a computer to function as each part of the danger information estimation device according to Item 8.

INDUSTRIAL APPLICABILITY According to the present invention, it is possible to provide a method, apparatus, program, and fire command system for estimating danger assumed at a fire site.

1 is a diagram schematically showing a schematic configuration of a fire command system according to one embodiment of the present invention; FIG. 1 is a block diagram for explaining functions of a machine learning device according to an embodiment of the present invention; FIG. It is a block diagram for explaining the function of the estimation device and the fire command device according to one embodiment of the present invention. 4 is a flowchart for explaining a procedure for learning a machine learning model using the machine learning device according to one embodiment of the present invention; 4 is a flowchart for explaining a procedure for estimating and notifying danger information using an estimation device and a learned machine learning model according to an embodiment of the present invention;

Embodiments of the present invention will now be described in detail with reference to the accompanying drawings. In the following description and drawings, the same reference numerals denote the same or similar components, and redundant description of the same or similar components will be omitted.

[Fire Command System]
<Overview of the system>
FIG. 1 is a diagram schematically showing a schematic configuration of a fire command system according to one embodiment of the present invention.

A fire command system 100 (hereinafter simply referred to as system 100) according to an embodiment of the present invention includes a danger information estimation device 2 (hereinafter also simply referred to as estimation device 2) and a fire command device 3. The system 100 receives a fire report from the fire site 99 through the public telephone network 90 from the telephone 97 of the caller 98, etc., and issues a dispatch command to the firefighters 108 who are about to go to the fire site 99. In parallel with this, the system 100 that received the fire report generates firefighting dispatch data regarding the fire site 99 that issued the dispatch command, and based on the generated firefighting dispatch data, determines the danger at the fire site 99 that issued the dispatch command. Estimate information. The system 100 notifies the estimated danger information to the firefighter 108 who received the dispatch order. As means for notifying, for example, character display on the monitor 36 or voice guidance by the speaker 37 is performed. Dangerous information to be notified includes, for example, "members aged 50 and over pay attention to worsening chronic diseases", "members of units with people aged 50 and over pay attention to health monitoring", and "people in their 20s pay attention to heat stroke". , ``Watch out for collapsing buildings'', etc.

According to the firefighting command system 100 according to one embodiment, risk information assumed at the fire site 99 upon receiving a fire report is presumed. The estimated danger information is notified to firefighters 108 who are about to be dispatched to the fire spot 99, and the firefighters 108 are alerted to accidents involving injury or death in the line of duty. As a result, firefighters are prevented or reduced from being injured or killed in the line of duty.

Danger information is estimated by the estimation device 2 according to a machine learning model 5 that has been trained using firefighting dispatch data as input data. The machine learning model 5 is learned in advance by the machine learning device 1 . A machine learning device 1 learns a machine learning model 5 for estimating danger information at a fire site based on learning data 4 relating to past injuries at the fire site. The trained machine learning model 5 is provided to the estimating device 2 . The estimating device 2 estimates danger information at the fire scene 99 where the dispatch command is received, based on the firefighting dispatch data generated by the firefighting command device 3, according to the learned machine learning model 5 learned by the machine learning device 1. do. The fire-fighting command device 3 generates fire-fighting dispatch data relating to the fire site 99 that issued the dispatch command, based on the crew's work schedule data 61 and the crew's personnel element data 62 .

The machine learning device 1 and the estimating device 2 are directly or indirectly connected by wire or wirelessly, for example, in a manner in which data can be transmitted and received via a network 9 . The estimating device 2 and the fire command device 3 are directly or indirectly connected by wire or wirelessly in such a manner that data can be transmitted and received via a dedicated communication line 109 in this embodiment.

In this embodiment, the fire command system 100 is installed, for example, in a fire command center provided in a single municipality unit or across a plurality of municipalities. The estimation device 2 and the fire command device 3 are combined and integrated as a system 100 . The machine learning device 1 is installed, for example, in a research institute related to firefighting, which collects and accumulates various data on past injury cases at various fire sites.

<Hardware configuration>
The machine learning device 1, the estimation device 2, and the fire command device 3 can be configured using, for example, general-purpose computers, tablet PCs, smartphones, and the like. All of the machine learning device 1, the estimation device 2, and the fire command device 3 can be configured with a general-purpose computer, and some of them can also be configured with a tablet PC or a smart phone.

Each of the machine learning device 1, the estimating device 2, and the fire commanding device 3 has a hardware configuration of a processor (not shown) such as a CPU that performs data processing, and a main processor that the processor uses as a work area for data processing. It has a storage device (not shown) and auxiliary storage devices 19, 29 and 39 used for temporarily storing data. Data 41, 42, 43, 44, 45 as learning data 4, data 61, 62, machine learning model 5, machine learning program, estimation program, firefighting command program, etc. are stored in respective auxiliary storage devices 19, 29, 39. are appropriately stored.

In this embodiment, the fire command device 3 includes a monitor 36 and a speaker 37 as hardware components for notifying the firefighters 108 who have received the dispatch command of the estimated danger information.

[Machine learning device]
FIG. 2 is a block diagram for explaining the functions of the machine learning device according to one embodiment of the present invention.

The machine learning device 1 includes a learning section 11 as a functional block. The learning unit 11 can be implemented as hardware using an integrated circuit or the like. Alternatively, the learning unit 11 can be implemented as software by the processor of the machine learning device 1 reading a machine learning program into the main storage device of the machine learning device 1 and executing it.

A learning unit 11 learns a machine learning model 5 for estimating danger information at a fire site based on learning data 4 relating to past injury cases at the fire site.

<Learning data>
The learning data 4 is a data set used for learning the machine learning model 5, and is a set of data set in the input layer of the machine learning model 5 and data set in the output layer of the machine learning model 5. is. The learning data 4 are pre-stored in the auxiliary storage device 19 of the machine learning device 1 .

In this embodiment, the learning data 4 includes data 41 on causes of injuries, data 42 on names of injuries, data 43 on time factors, and data 44 on human factors of firefighters. Table 1 shows an example of the data 41 on the cause of injury and the data 42 on the name of the illness.

Table 2 shows an example of data 43 regarding temporal elements. The temporal element data 43 includes the time of fire awareness and the stage of activity at the fire scene.

Table 3 shows an example of data 44 regarding the human element. The personnel element data 44 includes the firefighter's age group, rank, tenure, role within the platoon, and platoon duties.

<Machine learning model>
In this embodiment, the machine learning model 5 is configured using an artificial neural network (ANN). The machine learning model 5 includes an input layer, an intermediate layer, and an output layer as layers constituting a neural network. In the present embodiment, the learning unit 11 sets data 43 related to temporal elements and data 44 related to human elements of firefighters in the input layer, and data 41 related to causes of injury and disease, data 42 related to names of injuries and illnesses, and data 42 related to firefighter personnel. The machine learning model 5 is learned by setting the data 44 related to the target element in the output layer. Integer label values are used to set data to the input layer and output layer. The learned machine learning model 5 is stored in the auxiliary storage device 19 of the machine learning device 1 .

In this embodiment, a feedforward neural network is used as the machine learning model 5 . The neural network of machine learning model 5 includes multiple layers in the middle layer. As a neural network learning algorithm, a back propagation method is used to adjust weight parameters (synapse weights) in the machine learning model 5 .

The learned machine learning model 5 is transmitted from the machine learning device 1 to the estimating device 2 through the network 9 and stored in the auxiliary storage device 29 of the estimating device 2 . The estimating device 2 uses the received learned machine learning model 5 to estimate danger information at the fire site 99 where the dispatch order is received.

[Estimation device and fire command device]
FIG. 3 is a block diagram for explaining functions of an estimation device and a fire command device according to one embodiment of the present invention.

The estimation device 2 includes a firefighting dispatch data acquisition unit 21 and a danger information estimation unit 22 as functional blocks. These functional blocks can be implemented as hardware using an integrated circuit or the like. Alternatively, these functional blocks can be implemented as software by the processor of the estimating device 2 reading out an estimating program into the main memory of the estimating device 2 and executing it.

The firefighting dispatch data acquisition unit 21 acquires firefighting dispatch data relating to the fire site for which the dispatch command has been received. The firefighting dispatch data is generated in the firefighting command device 3 described later and transmitted to the estimation device 2 . In this embodiment, the firefighting dispatch data includes data 62 related to the human element of the firefighters 108 dispatched to the fire scene 99 who received the fire report and issued the dispatch command, and the time when the fire report was received. including. In this embodiment, as an example, the notification time from the fire site 99 is used as the fire awareness time. As for the data items, in this embodiment, the items of the data 62 regarding the human element of the firefighter 108 are the same as the items of the data 44 regarding the human element shown in Table 3.

The danger information estimator 22 estimates danger information at the fire scene 99 where the dispatch command is received, based on the firefighting dispatch data, according to the learned machine learning model 5 learned by the machine learning device 1 . The auxiliary storage device 29 stores the learned machine learning model 5 . The danger information estimating unit 22 inputs the firefighting dispatch data to the input layer of the learned machine learning model 5, so that the cause and name of injury and disease estimated by the learned machine learning model 5 and the human elements of the firefighters and are output from the output layer. The cause of injury and disease name, and the human element of the firefighter, which are output from the output layer, represent the danger information assumed at the fire scene 99 upon receiving the fire report and receiving the dispatch order. The danger information output by the danger information estimation unit 22 is then transmitted from the estimation device 2 to the fire command device 3 .

The fire command device 3 includes a fire dispatch data generator 31 as a functional block. The firefighting dispatch data generator 31 can be implemented as hardware using an integrated circuit or the like. Alternatively, the firefighting dispatch data generator 31 can be implemented as software by the processor of the firefighting command device 3 reading a firefighting command program into the main storage device of the firefighting command device 3 and executing it.

The firefighting dispatch data generating unit 31 generates firefighting dispatch data relating to the fire site 99 where the fire report was received and the dispatch command was issued, based on the firefighter's work schedule data 61 and the firefighter's personnel element data 62 . . The firefighting dispatch data includes the time when the fire was detected when the fire report was received, and the data 62 regarding the human elements of the firefighters 108 dispatched to the fire site 99 who received the fire report and issued the dispatch command.

The auxiliary storage device 39 stores firefighter work schedule data 61 and firefighter personnel data 62 . In this embodiment, the firefighter work schedule data 61 is data relating to the work schedule of the firefighters, and the items of the data 62 relating to the human elements of the firefighters are the items of the data 44 relating to the human factors shown in Table 3. are the same.

The firefighting dispatch data generation unit 31 refers to the work schedule data 61 and extracts the firefighters 108 dispatched to the fire site 99 based on the time when the fire report is received. The firefighting dispatch data generation unit 31 transmits the extracted data 62 related to the human elements of the firefighters 108 and the data including the time when the fire was detected to the estimating device 2 as firefighting dispatch data.

When the fire command device 3 receives the danger information from the estimation device 2, it notifies the fire brigade member 108 who received the dispatch command of the danger information through the monitor 36 and the speaker 37. FIG. As a form of notifying the danger information, for example, character display on the monitor 36 or voice guidance by the speaker 37 is performed. Dangerous information to be notified includes, for example, "members aged 50 and over pay attention to worsening chronic diseases", "members of units with people aged 50 and over pay attention to health monitoring", and "people in their 20s pay attention to heat stroke". etc.

In the present embodiment, the monitor 36 and the speaker 37 are integrated with the fire command device 3, but in another embodiment, the monitor 36 and the speaker 37 are capable of data communication via a network (not shown). It may be connected and arranged separately from the fire command device 3 . For example, a tablet terminal connected to the fire command device 3 via the Internet is placed in the cockpit of, for example, a fire engine dispatched to the fire scene 99. The danger information may be notified to firefighters 108 on board the vehicle.

[Machine learning procedure]
FIG. 4 is a flowchart for explaining a procedure for learning a machine learning model using the machine learning device according to one embodiment of the present invention.

The machine learning procedure is performed using the machine learning device 1 by, for example, a researcher of a firefighting-related research institute who collects and accumulates various data on past injuries at various fire sites.

The machine learning procedure includes step S<b>1 of preparing learning data 4 used for learning of the machine learning model 5 and step S<b>2 of learning the machine learning model 5 based on the learning data 4 .

In step S1 (learning data preparation step), learning data 4 is prepared based on data relating to past injuries at various fire sites. An example of aggregating data on past injury incidents at various fire sites is shown in Table 4.

For example, research institutes related to firefighting store aggregated results of past injuries shown in Table 4 in the form of electronic data. In Table 4, the item "Causes" shown in the column "Injuries" is the cause of injury or disease, and corresponds to the data 41 regarding the cause of injury or disease. The item "Time zones of perception" is the time of perception of the temporal element, the item "Activity stages of injuries" is the activity stage of the temporal element, and these items are included in the data 43 on the temporal element. Yes. "Age" is the age group of the human element, and "Ranks" is the rank of the human element, and these items correspond to the data 44 on the human element of firefighters. The item "Injury names" is the name of an injury or disease, and corresponds to the data 42 regarding the name of an injury or disease. The item "causes" shown in the "Fatalities" column is the cause of death in the line of duty, and corresponds to the data 41 regarding the cause of injury or disease.

In step S1, learning data 4 is prepared based on these past injury case data shown in Table 4 as an example. The relationship between the data 41, 42, 43, and 44 in the learning data 4, that is, the ratio of the data to be learned as the correct answer by the learning data 4, is the actual relationship (ratio ) to reflect.

The processing in step S1 may be performed by a researcher of a research institute using spreadsheet software, for example, by referring to the electronic data related to the past injuries shown in Table 4, or by using Perl, Ruby, Python, or the like. A script or program written using a text processing language may be used by a general purpose computer to reference the electronic data relating to past injuries. The learning data 4 prepared in step S1 is stored in the auxiliary storage device 19 of the machine learning device 1 through a network (not shown) and an input device such as a keyboard and mouse (not shown).

In step S<b>2 (learning step), the learning unit 11 learns the machine learning model 5 for estimating danger information at the fire scene based on the prepared learning data 4 . The learned machine learning model 5 is stored in the auxiliary storage device 19 .

The learned machine learning model 5 is transmitted from the machine learning device 1 to the estimating device 2 through the network 9 and stored in the auxiliary storage device 29 of the estimating device 2 . The estimating device 2 uses the received learned machine learning model 5 to estimate danger information at the fire site 99 where the dispatch order is received.

[Procedure for estimation and notification]
FIG. 5 is a flowchart for explaining a procedure for estimating and notifying danger information using an estimation device and a learned machine learning model according to an embodiment of the present invention.

Danger information is estimated according to the learned machine learning model 5, based on the firefighting dispatch data. Fire fighting dispatch data is generated using the fire command device 3 , and danger information is estimated using the estimating device 2 . The procedure for generating firefighting dispatch data includes steps S3 and S4, and the procedure for estimating danger information and notifying firefighters includes steps S5 and S6.

In step S3 (report acquisition step), a report from the fire site 99 is acquired. The fire command device 3 receives the fire report from the fire spot 99 from the phone 97 of the caller 98 through the public telephone network 90 .

Firefighting dispatch data is generated in step S4 (dispatch data generating step). The firefighting dispatch data generation unit 31 of the firefighting command device 3 generates firefighting dispatch data relating to the fire scene 99 that received the fire notification and issued the dispatch command. The firefighting dispatch data is generated based on the firefighter's work schedule data 61 and the data 62 regarding the firefighter's human factors. The firefighting dispatch data generated by the firefighting command device 3 is transmitted to the estimation device 2 .

In step S5 (danger information estimation step), according to the learned machine learning model 5, the danger information is estimated based on the firefighting dispatch data. The firefighting dispatch data acquisition unit 21 acquires firefighting dispatch data relating to the fire site 99 that received the dispatch command. The danger information estimator 22 estimates danger information at the fire site 99 where the dispatch command is received, based on the acquired firefighting dispatch data, according to the learned machine learning model 5 learned by the machine learning device 1 . The danger information output by the danger information estimation unit 22 is then transmitted from the estimation device 2 to the fire command device 3 .

In step S6 (danger information notification step), the danger information is notified to the firefighter 108 who received the dispatch order. The danger information estimated in step S5 is notified to the firefighter 108 who received the dispatch command through the monitor 36 or the speaker 37 of the firefighting command device 3, for example.

As described above, according to the machine learning device, the estimation device, the machine learning method, the estimation method, and the firefighting command system according to one embodiment of the present invention, it is possible to estimate the danger assumed at the fire scene. Furthermore, according to the fire command system according to one embodiment, the estimated danger information can be notified to the firefighters dispatched to the scene of the fire.

As a result, danger information assumed at the fire site 99 where the fire notification is received is estimated in advance. Furthermore, the estimated danger information is notified to firefighters 108 who are about to be dispatched to the fire spot 99 . As a result, the firefighter 108 who is about to be dispatched to the fire site 99 is alerted to an injury accident or an accident in which the firefighter 108 is killed in the line of duty.

[Other forms]
Although the present invention has been described in terms of specific embodiments, the present invention is not limited to the embodiments described above.

In other embodiments, training data 4 may further include data 45 regarding spatial elements. Table 5 shows an example of data 45 regarding spatial elements. The data 45 regarding spatial elements includes the use classification of the building and the place of occurrence of injury or disease.

When learning the machine learning model 5 , the learning unit 11 can further set data 45 regarding spatial elements in the input layer to learn the machine learning model 5 . When estimating danger information, the danger information estimating unit 22 inputs fire dispatch data including data on spatial elements to the input layer of the learned machine learning model 5, and estimates by the learned machine learning model 5. The cause of injury and the name of the illness are output from the output layer. The machine learning model 5 is learned in consideration of spatial factors, and the cause of injury and disease name estimated by the learned machine learning model 5 are estimation results in consideration of spatial factors.

Table 6 shows an example of aggregating data on past injury cases at various fire sites in consideration of spatial factors, which is used to prepare learning data 4 in another embodiment.

In another embodiment, the learning data 4 is prepared based on these past injury incident data considering the spatial factors exemplified in Table 6. The preparation of the learning data 4 may be performed based only on the data of the past injury cases in consideration of the spatial elements illustrated in Table 6, or based on the data of the past injury cases illustrated in Table 4 and the data of the past injury cases illustrated in Table 6. It may be based on both historical injury data that takes into account the spatial factors involved. In Table 5, the usage categories are classified in detail according to the Fire Defense Law Enforcement Ordinance, but there is no need to classify the usage categories in detail, and as shown in Table 6, they may be classified as fire prevention objects. .

In the embodiment described above, the temporal element data 43 includes the time of detection of the fire and the stage of activity at the fire site, but need not include both, but at least one of the two. should be Similarly, the human element data 44 includes the firefighter's age group, rank, length of service, platoon role, and platoon duties, but includes both. It does not have to be one, it just needs to include at least one of them.

In the above embodiment, an artificial neural network is used to configure the machine learning model 5, but the machine learning model 5 is not limited to an artificial neural network. Various machine learning algorithms can be used for the machine learning model, and various artificial intelligences (AI) can be used, as long as the machine learning model can be trained using the learning data.

In the above embodiment, the machine learning device 1 and the estimation device 2 are configured as separate devices, but the machine learning device 1 and the estimation device 2 can be integrated and configured as one device.

In the above embodiment, the machine learning device 1 and the estimating device 2 are connected via the network 9 so as to be able to communicate with each other. They can be connected for data exchange via a medium.

In the above embodiment, the machine learning device 1 is realized as an integrated device, but the machine learning device 1 does not need to be an integrated device, and the processor, main storage device, auxiliary storage device 19, etc. are arranged separately. , may be connected to each other so as to be able to communicate with each other via a network. The estimating device 2 and the fire command device 3 are the same as those of the machine learning device 1 .

In the above embodiment, each functional block of the machine learning device 1 is executed by a single processor, but each functional block does not need to be executed by a single processor, and is distributed by a plurality of processors. may be performed. The estimating device 2 and the fire command device 3 are the same as those of the machine learning device 1 .

A part or all of the functional blocks of the machine learning device 1, the estimation device 2, and the fire command device 3 may be clouded in a server device (not shown) connected via the network 9.

1 Machine learning device 2 Danger information estimation device 3 Fire command device 4 Learning data 5 Machine learning model 9 Network 11 Learning unit 19 Auxiliary storage device 21 Firefighting dispatch data acquisition unit 22 Danger information estimation unit 29 Auxiliary storage device 31 Firefighting dispatch data generation unit 36 Monitor 37 Speaker 39 Auxiliary storage device 61 Work schedule data 62 Human factor data 90 Public telephone network 97 Telephone 98 Caller 99 Fire scene 100 Fire command system 108 Firefighter 109 Dedicated communication line

Claims (8)

A computer with a processor
A firefighting dispatch data acquisition step of acquiring firefighting dispatch data relating to a fire site to which a dispatch order has been received, the firefighting dispatch data including data relating to the time when the fire was detected and the human elements of the firefighters dispatched to the fire site ;
The firefighting dispatch data is input to the input layer of the trained machine learning model, and the danger information estimated from the past injuries at the fire scene, the cause of injury, the name of the injury, and the human factors of the firefighters. a danger information estimation step of estimating danger information at the fire scene where the dispatch order is received by outputting the information including from the output layer ;
including
In the machine learning model, based on data on past injuries at fire sites, data on the time when the fire was noticed and data on the human factors of firefighters are set in the input layer, and data on the cause of injury and illness name A method of estimating danger information , wherein the data and the data on the human elements of the firefighters are set in the output layer and learned in advance . 2. The hazard information estimation of claim 1, wherein the data on the human element includes at least one of age group, rank, length of service, role in the platoon, and mission of the platoon of the fire brigade. Method. The firefighting dispatch data further includes data on spatial elements of the fire scene,
2. The danger information estimation method according to claim 1 , wherein said machine learning model is learned in advance by further setting data on spatial elements of a fire scene in said input layer of said machine learning model . a firefighting dispatch data acquisition unit for acquiring firefighting dispatch data , which is firefighting dispatch data relating to a fire site to which a dispatch order has been received, and which includes data relating to the time when the fire was detected and the human elements of the firefighters dispatched to the fire scene ;
The firefighting dispatch data is input to the input layer of the trained machine learning model, and the danger information estimated from the past injuries at the fire scene, the cause of injury, the name of the injury, and the human factors of the firefighters. a danger information estimating unit for estimating danger information at the fire scene where the dispatch order is received by outputting the information including from the output layer ;
with
In the machine learning model, based on data on past injuries at fire sites, data on the time when the fire was noticed and data on the human factors of firefighters are set in the input layer, and data on the cause of injury and illness name The danger information estimating device , wherein the data and the data regarding the human elements of the firefighters are set in the output layer and learned in advance . 5. The hazard information estimation of claim 4, wherein the data on the human element includes at least one of age group, rank, length of service, role within the platoon, and mission of the platoon of the fire brigade. Device. The firefighting dispatch data further includes data on spatial elements of the fire scene,
5. The danger information estimation device according to claim 4, wherein said machine learning model is learned in advance by further setting data relating to spatial elements of a fire scene in said input layer of said machine learning model. a danger information estimation device according to any one of claims 4 to 6 ;
a notification unit that notifies the danger information estimated by the danger information estimation device;
Fire command system. A program for causing a computer to function as each part of the danger information estimation device according to any one of claims 4 to 6 .

Images