Heat stroke prevention support device, method, and program

Abstract

To accurately predict the risk of a heat stroke on which a physical state of an individual worker and a heat environment for each work place are reflected.SOLUTION: A heat stroke prevention support device performs: acquiring attribute information related to a body condition of an object person; acquiring a time series prediction value of biological information on which a work load of the object person is reflected; obtaining a work intensity level of the object person based on the time series prediction value of the biological information and the attribute information; predicting a heat index in a work place of the object person; obtaining a heat environment level of the work place based on a prediction value of the head index; predicting a heat stroke risk level of the object person based on the work intensity level and the heat environment level; and outputting information expressing a prediction result of the heat stroke risk level.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] One aspect of the present invention relates to a heat stroke countermeasure support device, method, and program used to support, for example, heat stroke countermeasures for field workers. [Background technology]

[0002] According to statistics from the Ministry of Health, Labor and Welfare, the total number of deaths and injuries due to heatstroke in the workplace has reached 4,071 over the past five years since 2021. For this reason, measures to prevent heatstroke among on-site workers in the summer have become one of the most important issues in terms of balancing employee safety and operational efficiency.

[0003] As a result, a method has been proposed that uses a heat index called Wet Bulb Globe Temperature (WBGT) as an indicator for determining the risk of heatstroke, measures the WBGT at a workplace to determine the risk of heatstroke, and then issues alerts or other warnings based on the results of that assessment.

[0004] Another method for assessing the risk of heatstroke has been proposed, in which a subject's heart rate and core body temperature are measured, and information about the subject's symptoms is added to the measurement results to assess the risk of heat stroke (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2023-000538 A Summary of the Invention [Problem to be solved by the invention]

[0006] However, methods that focus only on WBGT can simply determine the risk of heatstroke for the entire workplace, but do not take into account the hot environment at each workplace within the workplace or the physical condition of each worker, including their workload. For this reason, there is a concern that the risk of heatstroke may be underestimated or overestimated depending on the workplace or worker, resulting in the onset of heatstroke and reduced operational efficiency.

[0007] On the other hand, the method described in Patent Document 1 makes it possible to determine the risk of heat stroke by taking into account the physical condition and subjective symptoms of each individual. However, because it does not take into account the hot environment of the workplace, the subject's physical condition due to illness, etc. may be judged as a heatstroke risk as is, posing an issue with the accuracy of determining the risk of heatstroke.

[0008] This invention has been made with the above-mentioned circumstances in mind, and aims to provide a technology that enables highly accurate prediction of heatstroke risk that reflects the physical condition of each individual worker and the hot environment of each workplace. [Means for solving the problem]

[0009] In order to solve the above problems, one aspect of the heat stroke countermeasure support device or support method according to the present invention acquires attribute information related to the physical condition of a subject, obtains a time series predicted value of biological information reflecting the subject's workload, and obtains the subject's work intensity level based on the time series predicted value of the biological information and the attribute information.Furthermore, a heat index at the subject's workplace is predicted, and a hot environment level at the workplace is obtained based on the predicted heat index.Then, a heat stroke risk level of the subject is predicted based on the work intensity level and the hot environment level, and information indicating the predicted result of the heat stroke risk level is output.

[0010] According to one aspect of the present invention, the risk level of heat stroke for each worker and each work place is determined based on the work intensity level reflecting the workload and attributes of each worker and the hot environment level reflecting the heat index for each work place. This makes it possible to predict the risk level of heat stroke with a higher accuracy than when the risk level of heat stroke is determined based only on the heat index or only on the physical condition of the subject. Effect of the Invention

[0011] In other words, according to one aspect of the present invention, a technology can be provided that enables highly accurate prediction of heatstroke risk that reflects the physical condition of each worker and the hot environment of each workplace. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is a diagram for explaining an overview of a support process executed by a heat stroke countermeasure support device according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram showing an example of the configuration of a system including a heatstroke countermeasure support device according to an embodiment of the present invention. [Diagram 3] FIG. 3 is a diagram showing an example of a work location that is a work target in the system shown in FIG. [Figure 4] FIG. 4 is a block diagram showing a hardware configuration of a heat stroke countermeasure support device according to an embodiment of the present invention. [Diagram 5] FIG. 5 is a block diagram showing the software configuration of a heatstroke countermeasure support device according to an embodiment of the present invention. [Figure 6] FIG. 6 is a diagram showing an example of the configuration of a multivariate DNN prediction model used to predict WBGT at a workplace in the heatstroke countermeasure support device shown in FIG. 5. [Figure 7] FIG. 7 is a flowchart showing an example of a processing procedure and processing contents of a learning process executed by the control unit of the heat stroke countermeasure support device shown in FIG. 5 in the learning phase. [Figure 8]FIG. 8 is a flowchart showing an example of a processing procedure and processing contents of a support process executed by the control unit of the heatstroke countermeasure support device shown in FIG. 5 in the operation phase. [Figure 9] FIG. 9 is a diagram for explaining an example of a process for determining an activity intensity level that reflects a clothing level in the support process shown in FIG. [Figure 10] FIG. 10 is a diagram for explaining an example of a process for determining a heat stroke risk level in the support process shown in FIG. [Figure 11] FIG. 11 is a diagram for explaining the effect of the heat stroke risk level determination process shown in FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0014] [One embodiment] (overview) FIG. 1 is a diagram for explaining an overview of a support process executed by a heat stroke countermeasure support device according to an embodiment of the present invention.

[0015] A heatstroke countermeasure support device according to one embodiment acquires heart rate data of a worker, predicts the heart rate from the present to the future from this heart rate data, and calculates the worker's work intensity (e.g., METs value) 104 based on this heart rate prediction value 101, the worker's age 102 acquired in advance as attribute information of the worker, and the resting heart rate 103 acquired before work. Then, a clothing equipment and work intensity level 106 is determined by reflecting the level of the worker's clothing and equipment 105 declared in advance in the work intensity level determined from the worker's work intensity 104.

[0016] In parallel with this, the heat stroke prevention support device acquires WBGT data representing the heat index of a reference point, estimates the WBGT 201 of the work place to be estimated based on the acquired WBGT data of the reference point and each position information 202 of the reference point and the work place to be estimated, and calculates the hot environment level 203 of the work place based on this estimated WBGT value 201.

[0017] The heatstroke prevention support device then predicts the worker's current to future heatstroke risk level 301 based on the work intensity level 106 reflecting the clothing level and the hot environment level 203, and notifies, for example, the worker or a manager of the predicted heatstroke risk level 301.

[0018] Therefore, according to the heat stroke countermeasure support device of one embodiment, a work intensity level 104 reflecting the heart rate 101, age 102, and clothing 103 of each worker is calculated, and a hot environment level 203 of the work place is calculated from the WBGT 201 of the work place estimated based on the WBGT of a reference point. Then, based on the calculated work intensity level 106 of the worker and the hot environment level 203 of the work place, a heat stroke risk level 301 at the work place is predicted for each worker.

[0019] In other words, the heatstroke risk level for each worker and each workplace is predicted based on both the work intensity level, which reflects the workload and attributes of each worker, and the hot environment level, which reflects the heat index for each workplace. This makes it possible to predict the heatstroke risk level with a higher degree of accuracy than when the heatstroke risk level is determined based only on WBGT or only on the physical condition of the worker.

[0020] (Configuration example) (1) System FIG. 2 is a diagram showing an example of the configuration of a system including a heat stroke countermeasure support device SV according to an embodiment of the present invention.

[0021] In one embodiment of the system, a heat stroke prevention support device SV transmits data via a network NW between a worker terminal UT carried by a worker US, and is capable of acquiring WBGT data of a reference point transmitted from a reference point WBGT measuring device BM.

[0022] The reference point WBGT measuring device BM is a WBGT measuring device distributed throughout the country by the Ministry of the Environment, and has the function of measuring WBGT at reference points and transmitting the measurement results, which are reference point WBGT data.

[0023] A worker US performs a given task at an assigned work location within a business establishment. Fig. 3 shows an example of a work location to be worked on, and in this example, work locations P1 to P8 such as various tanks and shipping terminals equipped in an oil refinery plant PS are shown.

[0024] A heart rate sensor S1 is attached to the worker US to measure a heart rate, which is one piece of biometric information that reflects a workload. The heart rate sensor S1 measures the worker US's heart rate at rest and a heart rate that reflects a workload while working. Note that as the biometric information, for example, the amount of sweat measured by a sweat sensor may be used in combination.

[0025] The worker US is also fitted with a temperature and humidity sensor S2 that measures the hot environment of the work place. The temperature and humidity sensor S2 measures the WBGT of the work places P1 to P8 where the worker US works.

[0026] The worker terminal UT transmits heart rate data indicating the heart rate measured by the heart rate sensor S1 and WBGT data indicating the WBGT value measured by the temperature and humidity sensor S2 to the heat stroke countermeasure support device SV via the network NW. Note that a wearable terminal incorporating the heart rate sensor S1 and the temperature and humidity sensor S2 is suitable as the worker terminal UT.

[0027] The network NW comprises, for example, a wide area network with the Internet at its core, and an access network for accessing the wide area network. For example, a public data communication network using wired or wireless, a LAN (Local Area Network) using wired or wireless, or a CATV (Cable Television) network is used as the access network. In addition, when the system is operated within a business establishment, for example, the network NW is configured by an in-house network such as a LAN or wireless LAN.

[0028] (2) Heatstroke prevention support device SV 4 and 5 are block diagrams showing an example of a hardware configuration and a software configuration, respectively, of a heat stroke countermeasure support device SV according to an embodiment of the present invention.

[0029] The heat stroke countermeasure support device SV is configured by a server computer installed on the Web or the cloud, for example. Note that a personal computer used for management purposes in a business establishment or the like may be used as the heat stroke countermeasure support device SV.

[0030] The heatstroke prevention support device SV has a control unit 1 that uses a hardware processor such as a central processing unit (CPU), and this control unit 1 is connected to a memory unit having a program memory unit 2 and a data memory unit 3, and a communication interface (hereinafter the interface will be abbreviated as I / F) unit 4 via a bus 5.

[0031] Under the control of the control unit 1, the communication I / F unit 4 receives the heart rate data and WBGT data transmitted from the worker terminal UT, and the reference point WBGT data transmitted from the reference point WBGT measuring device BM, in accordance with a communication protocol defined by the network NW.

[0032] The program storage unit 2 is configured by combining a non-volatile memory such as a solid state drive (SSD) that can be written and read at any time as a storage medium, and a non-volatile memory such as a read only memory (ROM), and stores application programs required to execute various control processes according to an embodiment, in addition to middleware such as an operating system (OS). Hereinafter, the OS and each application program will be collectively referred to as a program.

[0033] The data storage unit 3 is, for example, a combination of a non-volatile memory such as an SSD that can be written to and read from at any time as a storage medium, and a volatile memory such as a RAM (Random Access Memory), and the storage area includes the following storage units necessary for implementing one embodiment of the present invention: a storage unit 31 that stores data representing a WBGT time series prediction model, a storage unit 32 that stores data representing the WBGT prediction model, a storage unit 33 that stores a hot environment level determination table, a storage unit 34 that stores a heatstroke risk level determination table, a storage unit 35 that stores an work intensity level determination table, and a storage unit 36 ​​that stores data representing a heart rate time series prediction model.

[0034] The above-mentioned WBGT time series prediction model, WBGT prediction model and heart rate time series prediction model use machine learning models such as neural networks, and are trained before work begins based on collections of WBGT data or collections of heart rate data previously acquired at reference points and work locations.

[0035] Figure 6 shows an example of the layer structure of the WBGT prediction model. The WBGT prediction model is, for example, a multivariate DNN (Deep Neural Network) regression model, in which hidden layers L1 and L3, batch normalization layers L2 and L4, and a dropout layer L5 are arranged in series between the input layer Lin and the output layer Lout.

[0036] In the hidden layers L1 and L3, in order to break the symmetrical structure of the weights, random initial values ​​that follow a probability distribution that matches the activation function are given to the input data (Weight Initialization (He)). In addition, in the hidden layers L1 and L3, a process is performed to prevent overlearning by adding (regularizing) a cost proportional to the square of the weight coefficient (Weight Decay (L2 Norm)).

[0037] In the batch normalization layers L2 and L4, the distribution of activations is forcibly adjusted. In the dropout layer L5, neurons in the hidden layers L1 and L3 are randomly selected and deleted during learning to prevent overlearning.

[0038] In addition, this multivariate DNN regression model has an early stopping function that stops learning before overfitting to the training data causes large errors in the validation data, thereby preventing overfitting.

[0039] The control unit 1 includes, as processing functions necessary for executing the assistance control in one embodiment of the present invention, a work location WBGT data acquisition processing unit 11, a work location WBGT time series prediction processing unit 12, a reference point WBGT data acquisition processing unit 13, a work location WBGT prediction processing unit 14, a hot environment level judgment processing unit 15, an attribute information acquisition processing unit 16, a clothing equipment information acquisition processing unit 17, a heart rate data acquisition processing unit 18, a heart rate time series prediction processing unit 19, a work intensity level judgment processing unit 20, a heatstroke risk level judgment processing unit 21, and a heatstroke risk level judgment information output processing unit 22.

[0040] Each of the processing units 11 to 22 of the control unit 1 is realized by causing a hardware processor of the control unit 1 to execute an application program stored in the program storage unit 2. Note that a part or all of the processing units 11 to 22 may be realized using hardware such as an LSI (Large Scale Integration) or an ASIC (Application Specific Integrated Circuit).

[0041] The work place WBGT data acquisition processing unit 11 acquires the work place WBGT data transmitted from the worker terminal UT via the communication I / F unit 4.

[0042] In the learning phase, the workplace WBGT time series prediction processing unit 12 uses a set of workplace WBGT data acquired over a predetermined period of time in the past to train the WBGT time series prediction model stored in the storage unit 31. Then, through this learning process, a WBGT time series prediction model is created that receives the workplace WBGT data as input and outputs a time series prediction value of WBGT for the workplace over a predetermined period from the present to the future.

[0043] In addition, in the operation phase, the workplace WBGT time series prediction processing unit 12 inputs the WBGT data of the workplace acquired at the time of prediction as an explanatory variable into the WBGT time series prediction model stored in the memory unit 31, and acquires a WBGT time series prediction value for the workplace for a predetermined period from the present to the future from the WBGT time series prediction model.

[0044] The reference point WBGT data acquisition processing unit 13 acquires the WBGT data of the reference point transmitted from the reference point WBGT measuring device BM via the communication I / F unit 4.

[0045] In the learning phase, the workplace WBGT prediction processing unit 14 trains the WBGT prediction model stored in the storage unit 32. This learning process uses a set of WBGT data of the reference point acquired during a predetermined period in the past and a set of WBGT time series predicted values ​​of the workplace output from the WBGT time series prediction model, and is performed with the WBGT data of the workplace acquired by the workplace WBGT time series prediction processing unit 12 as the correct answer label. Then, by this learning process, a WBGT prediction model is created that receives the WBGT data of the reference point as input and outputs a WBGT predicted value from the present to the future at the workplace.

[0046] In the operation phase, the workplace WBGT prediction processing unit 14 inputs the WBGT data of the reference point acquired at the time of prediction and the WBGT time series predicted value of the workplace output from the WBGT time series prediction model as explanatory variables to the WBGT prediction model stored in the storage unit 32, and obtains the WBGT predicted value from the present to the future of the workplace from the WBGT prediction model. Note that the learning process and prediction process of the WBGT prediction model may further use position information of the reference point and the workplace.

[0047] The hot environment level determination processing unit 15 refers to a hot environment level determination table stored in the storage unit 33, and determines the hot environment level corresponding to the work place WBGT predicted value obtained by the work place WBGT prediction processing unit 14.

[0048] The attribute information acquisition processing unit 16 acquires, for example, attribute information of a worker input by a manager on his / her own manager terminal (not shown) via the communication I / F unit 4. Note that the attribute information may be acquired from worker information registered in a work management server or the like.

[0049] The clothing equipment information acquisition processing unit 17 acquires, for example, a clothing equipment level input by the worker US on his / her own worker terminal UT before starting work, via the communication I / F unit 4. The clothing equipment level indicates the degree of load that the clothing and equipment worn or attached by the worker imposes on the work.

[0050] The heartbeat data acquisition processing unit 18 acquires, via the communication I / F unit 4, the worker's heartbeat data transmitted from the worker's terminal UT.

[0051] In the learning phase, the heart rate time series prediction processing unit 19 uses a set of heart rate data of the worker US acquired over a predetermined period of time in the past to train a heart rate time series prediction model stored in the storage unit 36. In this way, a heart rate time series prediction model is created that receives the worker's heart rate as an input and outputs a time series prediction value of the worker's heart rate from the present to the future.

[0052] Furthermore, in the operation phase, the heart rate time series prediction processing unit 19 inputs the heart rate extracted as a feature from the worker's heart rate data acquired at the time of prediction as an explanatory variable into the heart rate time series prediction model stored in the memory unit 36, and acquires a time series prediction value of the worker's heart rate from the present to the future from the heart rate time series prediction model.

[0053] The work intensity level determination processing unit 20 calculates a METs value representing the work intensity of the worker from the worker's age included in the attribute information, the worker's heart rate at rest, and the worker's heart rate while working output from the heart rate time series prediction model. Then, with reference to an work intensity level determination table stored in the memory unit 35, the work intensity level corresponding to the METs value is determined.

[0054] Furthermore, the work intensity level determination processing unit 20 determines a work intensity level that reflects the clothing level declared by the worker by referring to the work intensity level determination table.

[0055] The heatstroke risk level determination processing unit 21 determines the heatstroke risk level corresponding to the work intensity level reflecting the above-mentioned clothing equipment level and the hot environment level of the work location determined by the hot environment level determination processing unit 15 by referring to the heatstroke risk level determination table stored in the memory unit 34.

[0056] The heatstroke risk level determination information output processing unit 22 transmits information representing the heatstroke risk level determined by the heatstroke risk level determination processing unit 21 from the communication I / F unit 4 to a pre-set destination, such as the worker terminal UT of the worker US or the administrator terminal of the work office.

[0057] (Example of operation) Next, an example of the operation of the heat stroke countermeasure support device SV configured as above will be described.

[0058] (1) Learning Phase To operate the heatstroke prevention support device SV, it is necessary to train the WBGT time series prediction model, the WBGT prediction model (multivariate DNN regression model), and the heart rate time series prediction model in advance.

[0059] FIG. 7 is a flowchart showing an example of the processing procedure and processing contents of the learning processing executed by the control unit 1 of the heat stroke countermeasure support device SV in the learning phase.

[0060] (1-1) Preparation for study The learning of each of the above prediction models is performed for each worker and each work location. The worker US to be the prediction target wears and wears the clothes and equipment to be used for the work, and also wears the heart rate sensor S1 and the temperature and humidity sensor S2 and carries the worker terminal UT. In other words, the worker wears the same equipment as when actually performing the work.

[0061] The administrator sets the learning period and learning cycle on the administrator terminal that he or she uses. For example, the learning period is set to at least five days. The learning cycle is set to one-minute intervals. Note that, if it is assumed that the work intensity of the worker and the heat environment do not change frequently, the learning cycle may be set to any time longer than one minute, for example, 10 minutes or 15 minutes.

[0062] In steps S10 and S11, the control unit 1 of the heat stroke countermeasure support device SV receives the setting data of the learning period and the learning cycle transmitted from the manager terminal via the communication I / F unit 4, and stores it in the data storage unit 3.

[0063] (1-2) Learning each predictive model When the above preparations are completed, the worker starts work that is the same as or similar to the actual work at the assigned work location for the set learning period. Meanwhile, the system administrator inputs a learning start request on his / her own administrator terminal. The control unit 1 of the heat stroke countermeasure support device SV detects the learning start request in step S12, and thereafter executes the learning process for each prediction model as follows.

[0064] (1-2-1) Learning the WBGT time series prediction model First, in step S13, the control unit 1 receives WBGT data measured by the temperature and humidity sensor S2 attached to the worker US from the worker terminal UT under the control of the work place WBGT data acquisition processing unit 11. Then, in step S14, under the control of the work place WBGT time series prediction processing unit 12, the WBGT time series prediction model stored in the memory unit 31 is trained using the received work place WBGT data.

[0065] That is, the workplace WBGT time series prediction processing unit 12 repeatedly executes the process of learning the above-mentioned WBGT time series prediction model using the set of workplace WBGT data in steps S13 and S14 for each learning cycle counted in step S18 within the learning period counted in step S19.

[0066] In this way, a WBGT time series prediction model corresponding to the workplace is created, which inputs the WBGT data of the workplace and outputs a WBGT time series prediction value for the workplace from the present to a predetermined future period. The created trained WBGT time series prediction model data is stored in the storage unit 31.

[0067] (1-2-2) Learning the WBGT prediction model (multivariate DNN regression model) The control unit 1 of the heat stroke countermeasure support server SV executes the learning process of the WBGT prediction model (multivariate DNN regression model) in parallel with the learning process of the above-mentioned WBGT time series prediction model as follows.

[0068] That is, in step S15, the control unit 1 acquires the WBGT data of the reference point at the learning cycle over the learning period under the control of the reference point WBGT data acquisition processing unit 13. The WBGT data of the reference point is acquired by receiving, via the communication I / F unit 4, the WBGT data transmitted from the reference point WBGT measuring device BM that is closest to the worker's work location.

[0069] Then, in step S16, the control unit 1 trains a WBGT prediction model (multivariate DNN regression model) under the control of the work place WBGT prediction processing unit 14. This learning process is performed using a set of WBGT data of the reference point acquired during the learning period and a set of work place WBGT time series predicted values ​​output from the WBGT time series prediction model, and using the set of work place WBGT data acquired by the work place WBGT time series prediction processing unit 12 as a correct answer label.

[0070] As a result, a WBGT prediction model is created that inputs the WBGT value represented by the WBGT data of the reference point and outputs a WBGT prediction value for a predetermined period from the present to the future at the workplace. The data of this trained WBGT prediction model is stored in the storage unit 32.

[0071] (1-2-3) Learning the heart rate time series prediction model The control unit 1 of the heat stroke countermeasure support server SV executes the learning process of the heart rate time series prediction model in parallel with the learning process of the WBGT time series prediction model and the WBGT prediction model (multivariate DNN regression model) as described below.

[0072] That is, in step S17, the control unit 1 acquires heartbeat data measured by the worker's heartbeat sensor S1 at the learning cycle over the learning period under the control of the heartbeat data acquisition processing unit 18. The heartbeat data is acquired by receiving, via the communication I / F unit 4, the heartbeat data transmitted from the worker terminal UT carried by the worker.

[0073] Then, under the control of the heart rate time series prediction processing unit 19, the control unit 1 trains the heart rate time series prediction model in step S18. This learning process is performed using the set of heart rate data acquired during the learning period. As a result, a heart rate time series prediction model is created that receives the worker's heart rate extracted from the heart rate data as input and outputs a time series prediction value of the worker's heart rate for a predetermined period from the present to the future. The trained data of the heart rate time series prediction model is stored in the storage unit 36.

[0074] (2) Operational Phase (2-1) Preparation FIG. 8 is a flowchart showing an example of the processing procedure and processing contents of the learning processing executed by the control unit 1 of the heat stroke countermeasure support device SV in the operation phase.

[0075] Prior to the start of work, the control unit of the heatstroke prevention support device SV, under the control of the attribute information acquisition processing unit 16, receives the worker's attribute information, for example from an administrator terminal, via the communication I / F unit 4 in step S30, and stores the received attribute information in the data storage unit 3.

[0076] The attribute information includes identification information of the worker (worker ID) and identification information indicating the age and work place (work place ID). Among these, the worker ID and the work place ID are used to select the WBGT time series prediction model, the WBGT prediction model (multivariate DNN regression model), and the heart rate time series prediction model to be used for prediction. The attribute information may also include information indicating gender, suitability for work, etc.

[0077] Prior to starting work, the worker also inputs a clothing equipment level, which indicates the clothing and equipment that the worker will wear and put on, into the worker terminal UT.

[0078] 9a in Fig. 9 shows an example of the relationship between the clothing and equipment worn and attached by the worker and the clothing equipment levels associated with these clothing and each other. For example, when the worker wears clothing equivalent to a thick long-sleeved jacket and long pants, the worker refers to the above correspondence table and inputs the clothing equipment level "IL3".

[0079] In response to this, the control unit 1 of the heat stroke prevention support device SV, under the control of the clothing equipment information acquisition processing unit 17, receives the clothing equipment level input by the worker US from the worker terminal UT via the communication I / F unit 4 in step S31, and stores the received clothing equipment level in the data memory unit 3.

[0080] (2-2) Heatstroke prevention support processing It is assumed that a worker starts a predetermined task at a specified work location, and in response, the manager inputs a support start request at the manager terminal.

[0081] When the control unit 1 of the heat stroke countermeasure support device SV detects the input of the support start request in step S32, it executes the heat stroke countermeasure support process for the worker at preset support time intervals as follows: The support time intervals are set to, for example, one minute, like the learning cycle described above, but if it is expected that the work intensity of the worker and the hot environment will not change frequently, they may be set to any time longer than one minute, for example, 10 minutes or 15 minutes.

[0082] (2-2-1) Determining the level of hot environment First, in step S33, the control unit 1 of the heat stroke countermeasure support device SV receives the reference point WBGT data transmitted from the reference point WBGT measuring device MB via the communication I / F unit 4 under the control of the reference point WBGT data acquisition processing unit 13.

[0083] Next, in step S34, under the control of the workplace WBGT prediction processing unit 14, the control unit 1 inputs the WBGT value of the reference point represented by the received WBGT data of the reference point and the workplace WBGT time series prediction value output from the WBGT time series prediction model at this point into the WBGT prediction model as explanatory variables.

[0084] As a result, the WBGT prediction model outputs a WBGT prediction value for the workplace, which is the WBGT value for the workplace estimated based on the WBGT value for the reference point, and further reflects the WBGT time series prediction value for the workplace. For example, the WBGT prediction model outputs prediction values ​​for the WBGT data for the workplace on the current day, one day later, and two days later.

[0085] Once the WBGT predicted value for the work location is obtained, the control unit 1, under the control of the hot environment level determination processing unit 15, in step S35, refers to the hot environment level determination table stored in the memory unit 33 and determines the hot environment level corresponding to the WBGT predicted value for the work location for the current day, one day later, and two days later.

[0086] For example, assume that the hot environment level determination table is currently defined as shown in 10a of Fig. 10. In this state, if the predicted WBGT value for the workplace one day later is 26°C, the hot environment level determination processing unit 15 refers to the hot environment level determination table and determines the hot environment level to be "EL2".

[0087] (2-2-2) Determining the level of work intensity Next, under the control of the heartbeat data acquisition processing unit 17, the control unit 1 receives the worker's current heartbeat data measured by the heartbeat sensor S1 from the worker terminal UT via the communication I / F unit 4 in step S36.

[0088] Next, in step S37, under the control of the heart rate time series prediction processor 19, the control unit 1 inputs the heart rate extracted from the received heart rate data as an explanatory variable to the heart rate time series prediction model stored in the storage unit 36. As a result, the heart rate time series prediction model outputs heart rate time series prediction values ​​for the current day, one day later, and two days later.

[0089] Next, under the control of the activity intensity level judgment processing unit 20, the control unit 1 first calculates predicted values ​​of the activity intensity (METs) of the worker on the day, one day later, and two days later in step S38. The predicted values ​​of METs are calculated from the time-series predicted values ​​of the heart rate output from the heart rate time-series prediction model and the heart rate at rest. The calculation of the METs value may reflect the maximum heart rate in the past.

[0090] Next, in step S39, the work intensity level determination processing unit 20 determines the work intensity level corresponding to the calculated predicted values ​​of the work intensity (METs) for the current day, one day later, and two days later by referring to the work intensity level determination table stored in the memory unit 35.

[0091] For example, assume that the work intensity level judgment table is currently defined as shown in 9b of Fig. 9. In this state, if the predicted value of the worker's work intensity (METs) one day later is "4 METs or more on average," the work intensity level judgment processor 20 refers to the work intensity level judgment table and judges the work intensity level to be "ML3."

[0092] Next, in step S40, the work intensity level determination processing unit 20 refers to the work intensity x clothing equipment determination table stored in the memory unit 35 to determine a work intensity level that reflects the clothing equipment level obtained before the start of work in the work intensity (METs) level determined for each of the current day, one day later, and two days later.

[0093] For example, assume that the work intensity x clothing determination table is currently defined as shown in 9c of Fig. 9. In this state, if the work intensity level of the worker one day later is determined to be "ML3" and the clothing level is determined to be "IL3", the work intensity level determination processor 20 refers to the work intensity x clothing determination table and determines the work intensity level reflecting the clothing level to be "ML3=4".

[0094] (2-2-3) Determining the risk level of heat stroke Next, under the control of the heatstroke risk level determination processing unit 21, in step S41, the control unit 1 refers to the heatstroke risk level determination table stored in the memory unit 34 to determine the worker's heatstroke risk level on the current day, one day later, and two days later.

[0095] For example, assume that the heatstroke risk level determination table is defined as shown in 10b of Fig. 10. In this state, if the hot environment level is "EL2" (25°C or higher and lower than 28°C) as described above, and the work intensity level reflecting the clothing level is "ML3=4", the work intensity level determination processor 20 refers to the heatstroke risk level determination table and determines that "heatstroke risk level=4". This "heatstroke risk level=4" indicates a "high risk".

[0096] In addition, "heatstroke risk level = 1" indicates "slight risk," "heatstroke risk level = 2" indicates "mild risk," "heatstroke risk level = 3" indicates "medium risk," and "heatstroke risk level = 5" indicates "very high risk."

[0097] Fig. 11 shows another example of a determination using the heatstroke risk level determination table. In this example, even if the hot environment level is as high as EL4 (31°C or higher), the work intensity level reflecting the clothing level is as low as "1," so the heatstroke risk level is determined to be "3" (medium risk). This example prevents overestimation of the heatstroke risk level, making it possible to maintain work efficiency.

[0098] On the other hand, as another example, even if the hot environment level is relatively low at EL3 (less than 31°C), if the work intensity level reflecting the clothing equipment level is high at "5," the heatstroke risk level is determined to be "5" (or higher). According to this example, the underestimation of the heatstroke risk level is prevented, and the occurrence of heatstroke accidents during work can be prevented in advance, thereby maintaining the safety of workers.

[0099] (2-2-4) Output of heat stroke risk level judgment information When the heatstroke risk level is determined, the control unit 1 finally transmits information indicating the heatstroke risk level determination result from the communication I / F unit 4 to a terminal previously set as a notification destination in step S42 under the control of the heatstroke risk level determination information output processing unit 22. The notification destination may be, for example, a worker terminal UT carried by the worker to be supported, or an administrator terminal used by the administrator.

[0100] Information that indicates the result of the heatstroke risk level assessment can be, for example, an alarm that uses a sound or an announcement message that uses voice synthesis technology, or an indication device such as an LED that lights up or blinks, or a combination of these. In this case, it is advisable to change the tone and intermittent cycle of the alarm, the content of the announcement message, the lighting color, and the blinking cycle according to the heatstroke risk level.

[0101] (2-2-5) Termination control In step S43, the control unit 1 monitors whether a support end request has been input. If a support end request has not been input in this state, the control unit 1 returns to step S33 and repeatedly executes a series of support processes in steps S33 to S42 at a preset support time interval.

[0102] On the other hand, for example, if the worker finishes the designated work and the manager inputs a support end request in response to the request at the manager terminal, the control unit 1 of the heat stroke countermeasure support device SV ends the support process and returns to the standby state when the control unit 1 detects the support end request in step S43.

[0103] (effect) As described above, in one embodiment, the worker's heart rate data acquired during work is input into a trained heart rate time series prediction model to predict the worker's heart rate from the present to the future, and the worker's work intensity (METs) is calculated based on this predicted heart rate value, the worker's age, and the resting heart rate acquired before work.Then, the worker's clothing level reported in advance is reflected in the calculated work intensity level to determine a work intensity level reflecting the clothing level.

[0104] In parallel with this, WBGT data representing the heat index at a reference point is obtained, and the obtained WBGT data at the reference point and the time series predicted value of WBGT at the workplace output from the trained WBGT time series prediction model are input into the trained WBGT prediction model to predict the WBGT value from the present to the future at the workplace to be estimated, and the hot environment level of the workplace is calculated based on this WBGT predicted value.

[0105] Then, based on the work intensity level reflecting the clothing level and the hot environment level, the worker's current to future heatstroke risk level is predicted, and the predicted heatstroke risk level is notified to, for example, the worker or a manager.

[0106] Therefore, according to one embodiment, it is possible to predict the risk level of heat stroke for each worker and for each workplace based on both the work intensity level that reflects the workload and attributes of each worker and further reflects the clothing of the worker, and the hot environment level that reflects the WBGT value of each workplace. Therefore, it is possible to predict the risk level of heat stroke with a higher accuracy than when the risk level of heat stroke is determined only by the WBGT or only by the physical condition of the worker.

[0107] In addition, since the WBGT value at a workplace is estimated based on the WBGT data at a reference point, there is no need to install a WBGT measuring device at each workplace, which reduces equipment costs.

[0108] Furthermore, by using a multivariate DNN regression model as a WBGT prediction model, and inputting a time series prediction value based on the WBGT value of the work place measured by the worker's temperature and humidity sensor S2 in addition to the WBGT value of the reference point as an explanatory variable into this multivariate DNN regression model, it is possible to predict WBGT taking into account the variation in WBGT values ​​for each place where the worker actually works. Therefore, by using this WBGT prediction value, it is possible to improve the accuracy of determining the hot environment level, and thereby maintain a high level of accuracy in determining the heatstroke risk level.

[0109] [Other embodiments] (1) In one embodiment, a case has been described in which a risk level of heat stroke from the present to the future is predicted and the result is output. However, it is also possible to make the heat stroke countermeasure support device SV create a plan for preventing heat stroke accidents on the day, a plan for future safety measures for workers, a schedule arrangement plan, etc. based on the prediction result, send them to the manager's terminal, and present them to the manager.

[0110] (2) In the heat stroke countermeasure support device SV, information showing the predicted results of the heat stroke risk level may be stored in the data storage unit 3, and this stored predicted information of the heat stroke risk level may be later analyzed to create analytical information showing tendencies of high-risk work locations, time periods, situations, etc., and transmit the analytical information to the manager's terminal. In this way, the manager can refer to the analytical information to look back on past work environments and situations and utilize it for future safety measures, etc.

[0111] (3) Each processing function and each predictive model provided in the heatstroke countermeasure support device SV may not be concentrated in a single server computer or personal computer, but may be distributed across multiple server computers or personal computers that can transmit data to each other.

[0112] Although the embodiment of the present invention has been described in detail above, the above description is merely an example of the present invention in every respect. It goes without saying that various improvements and modifications can be made without departing from the scope of the present invention. In other words, in implementing the present invention, specific configurations according to the embodiment may be appropriately adopted.

[0113] In short, this invention is not limited to the above-mentioned embodiment as it is, and in the implementation stage, the components can be modified and embodied without departing from the gist of the invention. In addition, various inventions can be formed by appropriately combining multiple components disclosed in the above-mentioned embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, components from different embodiments may be appropriately combined. [Explanation of symbols]

[0114] US…Worker S1…Heart rate sensor S2: Temperature and humidity sensor BM…Reference point WBGT measuring device PS…Oil refinery plant P1~P8: Work area SV: Heat stroke prevention support device NW…Network 1...Control unit 2...Program memory section 3. Data storage section 4. Communication I / F section 5. Bus 11...Workplace WBGT data acquisition processing unit 12...Workplace WBGT time series prediction processing unit 13...Reference point WBGT data acquisition processing unit 14...Workplace WBGT prediction processing unit 15…Hot environment level judgment processing unit 16...Attribute information acquisition processing unit 17...Clothing equipment information acquisition processing unit 18...Heart rate data acquisition processing unit 19...Heart rate time series prediction processing unit 20...Work intensity level determination processing unit 21…Heat stroke risk level determination processing unit 22...Heat stroke risk level determination information output processing unit 31…WBGT time series prediction model data storage unit 32…WBGT prediction model data storage section 33...Hot environment level determination table storage unit 34...Heat stroke risk level determination table storage unit 35...Work intensity level determination table storage unit 36...Heart rate time series prediction model data storage unit

Claims

1. A first processing unit that acquires attribute information related to a physical condition of a subject; A second processing unit that calculates a time series predicted value of biological information reflecting the workload of the subject; a third processing unit that calculates an activity intensity level of the subject based on the time-series predicted value of the biological information and the attribute information; A fourth processing unit that predicts a heat index at a work location of the subject; A fifth processing unit that calculates a hot environment level in the work place based on the predicted value of the heat index; A sixth processing unit that determines a heat stroke risk level of the subject based on the work intensity level and the hot environment level; a seventh processing unit that outputs information representing the heat stroke risk level determination result; A heat stroke prevention support device equipped with the above.

2. The heatstroke prevention support device of claim 1, further comprising an eighth processing unit that calculates a clothing equipment level that indicates the degree to which the clothing and equipment used by the subject affects the work, and reflects the clothing equipment level in the work intensity level.

3. The second processing unit includes: A first machine learning model that receives a feature amount of the subject's heart rate data as an input and outputs a time series prediction value of the heart rate data; a processing unit that acquires heart rate data of the subject at a first time point, and inputs a feature amount of the acquired heart rate data at the first time point into the first machine learning model as an explanatory variable, thereby obtaining a time series prediction value of the heart rate data of the subject at a second time point after the first time point from the first machine learning model; The heat stroke countermeasure support device according to claim 1 .

4. The second processing unit includes: The heatstroke prevention support device of claim 3, further comprising a first learning processing unit that, in a learning phase, acquires a set of heart rate data of the subject while working for a predetermined period prior to the first time point, and creates the first machine learning model trained to output a time series prediction value of the heart rate data by having the first machine learning model learn using features of the acquired set of heart rate data.

5. The fourth processing unit is A second machine learning model that receives a heat index of a reference location as an input and outputs a predicted value of the heat index of the work location; a heat index prediction processing unit that acquires a heat index of the reference point at a first time point, inputs the acquired heat index of the reference point as an explanatory variable into the second machine learning model, and acquires a predicted value of the heat index of the work place at a second time point that is later than the first time point from the second machine learning model; The heat stroke countermeasure support device according to claim 1 .

6. The fourth processing unit is A third machine learning model that receives the heat index of the work place as an input and outputs a time series prediction value of the heat index of the work place, The second machine learning model is configured to receive as input a heat index at the reference point and a time series predicted value of the heat index at the work location output from the third machine learning model, and to output a predicted value of the heat index at the work location; The heat index prediction processing unit acquires a heat index of the reference point at the first time point, inputs the acquired heat index of the reference point and a time-series predicted value of the heat index of the work place output from the third machine learning model into the second machine learning model as explanatory variables, and acquires a predicted value of the heat index of the work place at a second time point after the first time point from the second machine learning model. The heat stroke countermeasure support device according to claim 5.

7. The fourth processing unit is The heatstroke prevention support device of claim 6, further comprising a second learning processing unit that creates the trained second machine learning model, which in the learning phase acquires a set of heat indices at the reference point and a set of heat indices at the work location for a predetermined period of time past the first time point, treats the acquired set of heat indices at the work location as a correct label, and operates to output a predicted value of the heat index at the work location by having the second machine learning model learn using the set of heat indices at the reference point and the time series predicted value of the heat index at the work location output from the third machine learning model.

8. A heat stroke prevention method executed by an information processing device, comprising: A process of acquiring attribute information related to the subject's physical condition; obtaining a time series predicted value of biological information reflecting the workload of the subject; determining an activity intensity level of the subject based on the time-series predicted value of the biological information and the attribute information; predicting a heat index at the subject's workplace; determining a hot environment level in the work place based on the predicted value of the heat index; determining a risk level of heat stroke of the subject based on the work intensity level and the hot environment level; outputting information representing the heat stroke risk level determination result; A heat stroke prevention support method comprising the steps of:

9. 8. A program for causing a processor included in a heat stroke countermeasure support device to execute at least one of processes executed by first to eighth processing units included in the heat stroke countermeasure support device according to claim 1.

Images