Device and method, computer-readable recording medium, and computer program for ai-based job safety analysis creation automation

Abstract

The present invention relates to a technology for creating a job safety analysis by recommending and generating content to be included in the job safety analysis. According to an aspect of the present invention, a method for AI-based job safety analysis creation automation may comprise: searching for a job context corresponding to a JSA title and JSA job content which are input by a user; generating accident types (hereinafter, recommended accident types) corresponding to the user job context and user job steps; generating a recommended risk factor on the basis of a handled chemical substance input by the user; generating a current countermeasure corresponding to an accident type selected by the user from among the recommended accident types; generating a damage severity and an accident frequency corresponding to the job context and the recommended risk factor; and creating a JSA document on the basis of the user job context, the user job steps, the accident type selected by the user, the recommended risk factor, a recommended current countermeasure, the damage severity, and the accident frequency.

Description

Device and method for automating the preparation of AI-based work risk assessments, computer-readable recording media, and computer programs

[0001] The present invention relates to a technique for creating a work risk assessment by recommending and generating content to be included in the work risk assessment.

[0002] For reference, this application claims priority to Korean Patent Application No. 10-2024-0078249, filed June 17, 2024. The entire contents of this priority application are incorporated herein by reference.

[0003] Job Safety Analysis (JSA) is a method of establishing work procedures to perform the work safely by identifying hazardous risk factors at each stage by dividing a specific task into major steps, and preparing preventive measures to eliminate and reduce risks. It is a method to minimize accidents by having employees recognize the risks of the work performed at the workplace, and it is also legally mandated to prevent serious accidents, making it an important part of safety management.

[0004] In writing such work risk assessments, the current method is to have workers write them manually, so the content may differ depending on the writer, and in many cases, they are not familiar with various cases, making it difficult to write the content.

[0005] The purpose of the present invention is to automate the preparation of a work risk assessment by recommending and generating content to be prepared and providing it to a worker.

[0006] According to one aspect of the present invention, an automated method for creating an AI-based work risk assessment comprises the steps of: searching for a work context including a JSA title and JSA work content corresponding to a JSA title and JSA work content input by a user in a work context database; searching for a JSA document including a work context (user work context) selected by a user among the searched work contexts and a work step and a disaster type corresponding to a work step input by a user (hereinafter, “user work step”), thereby generating a disaster type (hereinafter, “recommended disaster type”) corresponding to the user work context and the user work step; searching for a JSA document including a work step and a disaster type corresponding to the user work context and the user work step in the work step database, and generating a recommended risk factor based on the search result and a handling chemical substance input by the user; searching for a JSA document including a work context and a disaster type corresponding to the user work context and the user work step in the work step database, thereby generating a current action (hereinafter, “recommended current action”) corresponding to a disaster type selected by the user among the recommended disaster types; The method may include: a step of searching a JSA document including a work context, risk factors, and accident history corresponding to a user work context and recommended risk factors in a risk factor database, thereby generating damage intensity and accident frequency corresponding to the user work context and recommended risk factors; and a step of creating a JSA document based on the user work context, the user work step, the disaster type selected by the user, the recommended current measures, the recommended risk factors, damage intensity, and accident frequency.

[0007] In one embodiment, the step of searching for a work context may include searching for a work context including a JSA title and JSA work content corresponding to a JSA title and JSA work content input by a user, based on a vector similarity of the JSA title and JSA work content input by a user, in a work context database in which a plurality of work contexts, each including a JSA title and JSA work content, are stored.

[0008] In one embodiment, the step of generating a recommended disaster type may include: searching for a JSA document including a disaster type and a work step corresponding to a user task context and a user task step based on a vector similarity of the user task context and the user task step from a work step database in which a plurality of JSA documents, each including a work context, a risk factor, and a disaster type, are stored; calculating a cosine similarity of the work step and the work context included in the searched JSA document; extracting a JSA document corresponding to the work context and the user task step from among the searched JSA documents based on the cosine similarity by making the cosine similarity of the work step and the cosine similarity of the work step different from each other; and generating a recommended disaster type based on the disaster type included in the extracted JSA document.

[0009] In one embodiment, the step of generating a recommended risk factor may include: searching for a JSA document including a work step and a disaster type corresponding to a user work context and a user work step from a work step database in which a plurality of JSA documents, each including a work context, a work step, a hazard factor, a handling chemical, and a disaster type, are stored; extracting a JSA document including a disaster type corresponding to a disaster type selected by a user from among recommended disaster types from among the searched JSA documents; and generating a risk factor (hereinafter, referred to as a recommended risk factor) corresponding to a handling chemical input by a user using a language model based on the disaster type, the work step, the hazard factor, and the handling chemical included in the extracted JSA document.

[0010] In one embodiment, the step of generating a current action item may include: searching for a JSA document including a task context and a risk factor corresponding to a user task context and a recommended risk factor based on a vector similarity of the user task context and the recommended risk factor from a task step database in which a plurality of JSA documents, each including a risk factor and a task context, are stored; calculating a cosine similarity of the risk factor and the task context included in the searched JSA document; extracting a JSA document corresponding to the recommended risk factor and the task context from among the searched JSA documents based on the cosine similarity by making the cosine similarity of the risk factor and the cosine similarity of the task context different from each other; and generating a current action item from among the action items corresponding to a recommended disaster type from among the action items included in the extracted JSA documents using a language model.

[0011] In one embodiment, the step of generating a damage intensity may include: a step of searching for a JSA document including a work context and a risk factor corresponding to a user work context and a recommended risk factor from a risk factor database in which a plurality of JSA documents, each of which includes a work context, a risk factor, a current measure, and a damage intensity, are stored; and a step of generating a damage intensity corresponding to the user work context and the recommended risk factor based on the risk intensity of the risk factors included in the searched JSA document.

[0012] In one embodiment, the step of generating an accident frequency may include the step of searching for an accident history corresponding to a user work context from a risk factor database in which a plurality of JSA documents, each of which includes a work context and an accident history, are stored; and the step of generating an accident frequency corresponding to the user work context and the recommended risk factor based on the cause of the searched accident history.

[0013] According to another aspect of the present invention, an AI-based work risk assessment writing automation device includes: a work context search unit that searches for a work context including a JSA title and JSA work content corresponding to a JSA title and JSA work content input by a user in a work context database; a recommended disaster type generation unit that searches for a JSA document including a work context (user work context) selected by a user among the searched work contexts and a work step and a disaster type corresponding to a work step (hereinafter, “user work step”) input by a user in a work step database, thereby generating a disaster type (hereinafter, “recommended disaster type”) corresponding to the user work context and the user work step; a recommended risk factor generation unit that searches for a JSA document including a work step and a disaster type corresponding to the user work context and the user work step in the work step database, and generates a recommended risk factor based on the search result and a handling chemical substance input by the user; a current action generation unit that searches for a JSA document including a work context and a risk factor corresponding to the user work context and the recommended risk factor in the work step database, thereby generating a current action (hereinafter, “recommended current action”) corresponding to a disaster type selected by a user among the recommended disaster types; The method may include a damage intensity and accident frequency generation unit that generates damage intensity and accident frequency corresponding to the user task context and recommended risk factors by searching for a JSA document including a task context, risk factors, and accident history corresponding to the user task context and recommended risk factors in a risk factor database; and a JSA document creation unit that creates a JSA document based on the user task context, the user task stage, the disaster type selected by the user, the recommended current measures, and the damage intensity and accident frequency of the recommended risk factors.

[0014] According to another embodiment of another aspect of the present invention, an AI-based work risk assessment writing automation device comprises: a memory storing at least one command; And a processor, wherein the at least one instruction is executed by the processor, so that the device searches for a work context including a JSA title and JSA work content corresponding to a JSA title and JSA work content input by a user in a work context database, searches for a JSA document including a work context (user work context) selected by the user among the searched work contexts and a work step and a disaster type corresponding to a work step (hereinafter, user work step) input by the user, and generates a disaster type (hereinafter, recommended disaster type) corresponding to the user work context and the user work step, searches for a JSA document including a work step and a disaster type corresponding to the user work context and the user work step in the work step database, and generates a recommended risk factor based on the search result and the handling chemical substance input by the user, searches for a JSA document including a work context and a risk factor corresponding to the user work context and the recommended risk factor in the work step database, and generates a current action (hereinafter, recommended current action) corresponding to a disaster type selected by the user among the recommended disaster types, and searches for a JSA document including a work context and a risk factor corresponding to the user work context and the recommended risk factor in the work step database, By searching for a JSA document including a work context, risk factors and accident history corresponding to the context and the recommended risk factors, damage intensity and accident frequency corresponding to the user work context and the recommended risk factors are generated, and a JSA document is created based on the user work context, the user work step, the disaster type selected by the user, the recommended risk factors, the recommended current measures, the damage intensity and the accident frequency.

[0015] According to one aspect of the present invention, it is possible to create an efficient work risk assessment document by recommending items to be included in a work risk assessment document to be written by a worker.

[0016] Figure 1 is a block diagram of an automated device for writing AI-based work risk assessment according to one embodiment of the present invention.

[0017] Figure 2 is a flowchart of an automated method for creating an AI-based work risk assessment according to one embodiment of the present invention.

[0018] FIG. 3 is a drawing showing an example of a work risk assessment document according to one embodiment of the present invention.

[0019] FIG. 4 is a diagram for explaining the generation of a recommended disaster type according to one embodiment of the present invention.

[0020] FIG. 5 is a diagram for explaining the generation of recommended risk factors according to one embodiment of the present invention.

[0021] FIG. 6 is a diagram for explaining the generation of recommended current actions according to one embodiment of the present invention.

[0022] FIG. 7 is a diagram for explaining the generation of damage intensity and accident frequency according to one embodiment of the present invention.

[0023] Figure 8 is a block diagram of an automated device for writing AI-based work risk assessment according to another embodiment of the present invention.

[0024] The advantages and features of the present invention, and the methods for achieving them, will become clearer with reference to the embodiments described in detail below with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various forms. These embodiments are provided solely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the scope of the present invention is defined solely by the claims.

[0025] In describing embodiments of the present invention, specific descriptions of known functions or configurations will be omitted unless actually necessary. Furthermore, the terms described below are defined based on their functions in the embodiments of the present invention and may vary depending on the intent or custom of the user or operator. Therefore, their definitions should be based on the overall content of this specification.

[0026] The terms ‘…bu’, ‘…gi’, etc. used hereinafter mean a unit that processes at least one function or operation, and this can be implemented by hardware, software, or a combination of hardware and software.

[0027]

[0028] Figure 1 is a block diagram of an automated device for writing AI-based work risk assessment according to one embodiment of the present invention.

[0029] Referring to FIG. 1, the AI-based work risk assessment writing automation device (1000) may include an input unit (1100), a work context search unit (1200), a recommended disaster type creation unit (1300), a recommended risk factor creation unit (1400), a current action creation unit (1500), a damage intensity and accident frequency creation unit (1600), a JSA document creation unit (1700), and an output unit (1800). The input unit (1100), the work context search unit (1200), the recommended disaster type creation unit (1300), the recommended risk factor creation unit (1400), the current action creation unit (1500), the damage intensity and accident frequency creation unit (1600), the JSA document creation unit (1700), and the output unit (1800) are exemplary divisions of the functions of the AI-based work risk assessment writing automation device (1000), but are not limited thereto.

[0030] According to an embodiment, the functions of the input unit (1100), the work context search unit (1200), the recommended disaster type generation unit (1300), the recommended risk factor generation unit (1400), the current action generation unit (1500), the damage intensity and accident frequency generation unit (1600), the JSA document creation unit (1700), and the output unit (1800) can be merged / separated and implemented as a series of computer-executable commands included in at least one program.

[0031] The AI-based work risk assessment writing automation device (1000) comprises an input unit (1100), a work context search unit (1200), a recommended disaster type creation unit (1300), a recommended risk factor creation unit (1400), a current action creation unit (1500), a damage intensity and accident frequency creation unit (1600), a JSA document creation unit (1700), and an output unit (1800), each of which is stored in a memory (8200), and when executed by a processor (8100) to be described later, can perform the functions described below.

[0032] First, the input unit (1100) can receive data required for writing a JSA document from an external device using a wired or wireless communication means.

[0033] Additionally, the input unit (1100) can receive commands required for creating a JSA document from a user by using a means such as a user interface.

[0034] In one embodiment, the input unit (1100) can receive input of a JSA title, JSA content, etc. that can confirm the type of work risk assessment document that the user wants to create.

[0035] In one embodiment, the input unit (1100) may receive a command from the user to select one or more pieces of information recommended or generated by the AI-based work risk assessment writing automation device (1000).

[0036] The work context search unit (1200) can search for a work context based on a JSA title and JSA work content, and for this purpose, can include a work context database in which a plurality of work contexts are stored. The work context search unit (1200) can search for a work context including a JSA title and JSA work content corresponding to a JSA title and JSA work content input by a user, based on the vector similarity of the JSA title and JSA work content input by a user, in the work context database in which a plurality of work contexts including a JSA title and JSA work content are stored.

[0037] In one embodiment, the task context may be a single sentence containing the JSA title and the JSA task content.

[0038] In one embodiment, the work context database may vectorize and store a plurality of work contexts, each including a JSA title and JSA work content, so that they can be searched based on vector similarity of JSA titles and JSA work content input by a user.

[0039] The recommended disaster type generation unit (1300) can search for a JSA document based on a work context and a work step, and for this purpose, can include a work step database in which a plurality of JSA documents including work steps and disaster types are stored. The recommended disaster type generation unit (1300) can search for a JSA document including a work step and a disaster type corresponding to a searched work context (or a work context selected by the user from among the searched work contexts) (user work context) and a work step input by the user (hereinafter, user work step) in the work step database, thereby generating a disaster type (hereinafter, recommended disaster type) corresponding to the user work context and the user work step.

[0040] The recommended risk factor generation unit (1400) can search for JSA documents based on work contexts and work steps, and for this purpose, can include a work step database in which multiple JSA documents including work steps and disaster types are stored. The recommended risk factor generation unit (1400) can search for JSA documents including work steps and disaster types corresponding to the user work context and user work steps in the work step database. The recommended risk factor generation unit (1400) can generate recommended risk factors based on the JSA document search results and the handling chemicals input by the user.

[0041] The current action generation unit (1500) can search for a JSA document based on a work context and risk factors, and for this purpose, can include a work step database in which a plurality of JSA documents including work steps and risk factors are stored. The current action generation unit (1500) can generate a current action (hereinafter, a recommended current action) corresponding to a disaster type selected by a user among recommended disaster types by searching for a JSA document including a work context and risk factors corresponding to a user work context and recommended risk factors in the work step database.

[0042] The damage intensity and accident frequency generation unit (1600) can search for JSA documents based on work contexts and risk factors, and for this purpose, can include a risk factor database in which JSA documents including work contexts, risk factors, and accident histories are stored. The damage intensity and accident frequency generation unit (1600) can generate damage intensity and accident frequency corresponding to user work contexts and recommended risk factors by searching for JSA documents including work contexts, risk factors, and accident histories corresponding to user work contexts and recommended risk factors in the risk factor database.

[0043] The JSA document creation unit (1700) can create a JSA document based on the user work context, user work step, disaster type selected by the user, recommended risk factors, recommended current measures, damage intensity, and accident frequency.

[0044] The output unit (1800) can transmit the written JSA document and the information recommended and generated during the writing process to an external device using wired or wireless communication means.

[0045] In addition, the output unit (1800) can output JSA documents written so that the user can recognize them, information recommended and generated during the writing process, etc. using a display device, etc.

[0046]

[0047] Figure 2 is a flowchart of an automated method for creating an AI-based work risk assessment according to one embodiment of the present invention.

[0048] Hereinafter, the above method is described as an example performed by an AI-based work risk assessment writing automation device (1000) illustrated in FIG. 1.

[0049] Referring to FIG. 2, in step S2100, the AI-based work risk assessment writing automation device (1000) can receive a JSA title and JSA work content from a user. The AI-based work risk assessment writing automation device (1000) can search for a work context including a JSA title and JSA work content corresponding to the JSA title and JSA work content input by the user, based on the vector similarity of the JSA title and JSA work content input by the user, in a work context database in which a plurality of work contexts including the JSA title and JSA work content are stored.

[0050] In one embodiment, the task context may be a single sentence containing the JSA title and the JSA task content.

[0051] In one embodiment, the work context database may vectorize and store a plurality of work contexts, each including a JSA title and JSA work content, so that they can be searched based on vector similarity of JSA titles and JSA work content input by a user.

[0052] Additionally, the AI-based work risk assessment writing automation device (1000) can receive a command from the user to select at least one work context from among the searched work contexts.

[0053] In step S2200, the AI-based work risk assessment writing automation device (1000) can receive work steps from a user. The AI-based work risk assessment writing automation device (1000) can search for a JSA document including a work step and a disaster type corresponding to a searched work context (or a work context selected by the user from among the searched work contexts) (user work context) and a work step input by the user (hereinafter, user work step) in a work step database, thereby generating a disaster type (hereinafter, recommended disaster type) corresponding to the user work context and the user work step.

[0054] Additionally, the AI-based work risk assessment writing automation device (1000) can receive a command from the user to select at least one of the recommended disaster types.

[0055] In step S2300, the AI-based work risk assessment creation automation device (1000) can search for a JSA document containing a work step and disaster type corresponding to the user work context and the user work step in the work step database. The AI-based work risk assessment creation automation device (1000) can receive handling chemicals as input from the user. The AI-based work risk assessment creation automation device (1000) can generate recommended risk factors based on the JSA document search results and the handling chemicals input by the user.

[0056] In step S2400, the AI-based work risk assessment creation automation device (1000) can search for a JSA document including a work context and risk factors corresponding to the user work context and recommended risk factors in the work step database, thereby generating a current action (hereinafter, recommended current action) corresponding to a disaster type selected by the user among the recommended disaster types.

[0057] In step S2500, the AI-based work risk assessment creation automation device (1000) can generate damage intensity and accident frequency corresponding to the user work context and recommended risk factors by searching for a JSA document including work context, risk factors, and accident history corresponding to the user work context and recommended risk factors in the risk factor database.

[0058] In step S2600, the AI-based work risk assessment writing automation device (1000) can write a JSA document based on a user work context, a user work step, a disaster type selected by the user, recommended risk factors, recommended current measures, damage intensity, and accident frequency.

[0059]

[0060] FIG. 3 is a drawing showing an example of a work risk assessment document according to one embodiment of the present invention.

[0061] Referring to FIG. 3, the job risk assessment (JSA) document may include information such as the job name (ooo shipping dock reinforced concrete work), at least one job step, at least one risk factor corresponding to at least one job step, the handled chemical substance (handled object) corresponding to at least one job step, the type of accident corresponding to at least one job step, the current measures corresponding to at least one job step, and the current risk level (accident frequency, damage intensity, risk level) corresponding to at least one job step, and the type and content of this information may vary depending on the type, content, team, etc. of the job.

[0062]

[0063] FIG. 4 is a diagram for explaining the generation of a recommended disaster type according to one embodiment of the present invention.

[0064] Hereinafter, the above method is described as an example performed by an AI-based work risk assessment writing automation device (1000) illustrated in FIG. 1.

[0065] An AI-based work risk assessment writing automation device (1000) can search for a JSA document including a work step and a disaster type corresponding to a user work context and a user work step based on vector similarity of the user work context and the user work step in a work step database in which a plurality of JSA documents including each work context, risk factor, and disaster type are stored.

[0066] The AI-based work risk assessment writing automation device (1000) can calculate the cosine similarity of the work steps and work contexts included in the searched JSA document.

[0067] The AI-based work risk assessment writing automation device (1000) can extract JSA documents (e.g., top 50 JSA documents) corresponding to the work context and user work stage from among JSA documents searched based on cosine similarity by setting different weights for the cosine similarity of the work stage and the cosine similarity of the work stage. For example, the weights for the cosine similarity of the work stage and the cosine similarity of the work stage may be 80:20.

[0068] The AI-based work risk assessment writing automation device (1000) can aggregate the duplication frequency of disaster type items included in the extracted JSA document, extract the top N (e.g., 5) disaster types based on the duplication frequency, and generate recommended disaster types based on the extracted disaster types.

[0069]

[0070] FIG. 5 is a diagram for explaining the generation of recommended risk factors according to one embodiment of the present invention.

[0071] Referring to FIG. 5, the AI-based work risk assessment writing automation device (1000) can search for a JSA document including a work step and a disaster type corresponding to a user work context and a user work step based on the vector similarity of the user work context and the user work step in a work step database in which a plurality of JSA documents including a work context, a work step, a risk factor, and a disaster type are stored, respectively. In this case, the AI-based work risk assessment writing automation device (1000) can set different weights for the vector similarities of the user work context and the user work step, for example, 80:20.

[0072] The AI-based work risk assessment writing automation device (1000) can extract JSA documents that include a disaster type corresponding to a disaster type selected by a user from among recommended disaster types among the searched JSA documents. Specifically, the AI-based work risk assessment writing automation device (1000) can filter the searched JSA documents based on the recommended disaster type and extract JSA documents that include a disaster type with the same content as the recommended disaster type.

[0073] The AI-based work risk assessment writing automation device (1000) can generate risk factors (hereinafter, recommended risk factors) corresponding to handling chemicals input by a user using a language model based on the disaster type, work stage, risk factors, and handling chemicals included in the extracted JSA document.

[0074]

[0075] FIG. 6 is a diagram for explaining the generation of recommended current measures according to one embodiment of the present invention.

[0076] Referring to FIG. 6, the AI-based work risk assessment writing automation device (1000) can search for a JSA document including a work context and risk factor corresponding to a user work context and a recommended risk factor based on the vector similarity of the user work context and the recommended risk factor in a work step database in which a plurality of JSA documents, each including a risk factor and a work context, are stored.

[0077] An AI-based work risk assessment writing automation device (1000) can calculate the cosine similarity of risk factors and work contexts included in a searched JSA document.

[0078] The AI-based work risk assessment writing automation device (1000) can extract JSA documents corresponding to recommended risk factors and work contexts from among JSA documents searched based on cosine similarity by making the weights of the cosine similarity of risk factors and the cosine similarity of work contexts different from each other (for example, 75:25). Specifically, the AI-based work risk assessment writing automation device (1000) can extract JSA documents corresponding to recommended risk factors and work contexts by filtering only JSA documents that include risk factors with the same content as recommended risk factors from among the JSA documents searched based on cosine similarity.

[0079] An AI-based work risk assessment writing automation device (1000) can generate a recommended disaster type from the disaster type included in the extracted JSA document using a language model.

[0080] As a specific example, the AI-based work risk assessment writing automation device (1000) can extract the top 5 JSA documents with weighted similarity based on the cosine similarity (weighted 0.75) between risk factors and the cosine similarity (weighted 0.25) between work contexts.

[0081] When the AI-based work risk assessment writing automation device (1000) simply lists the top K measures included in the extracted JSA document based on similarity, a large number of text selections with overlapping meanings, such as “implementation of two-person work” and “prevention through implementation of two-person work”, may occur. Accordingly, the AI-based work risk assessment writing automation device (1000) utilizes the mutual embedding cosine similarity of the action items included in the extracted JSA document to select three current action items with different meanings (can be generated by considering action items with various meanings), considers all possible cases of selecting three of the selected action items (e.g., if there are 10 current action items, the number of cases in which three of them can be selected is 120 (10! / 3! * (10-3)!)), calculates pair-wise cosine similarities for each of the three selected action items (<action item 1 vs. action item 2>, <action item 1 vs. action item 3>, <action item 2 vs. action item 3>), calculates the calculated cosine similarities and sums them, thereby calculating all 120 combinations and selecting the combination with the lowest value (lowest mutual similarity). The AI-based work risk assessment creation automation device (1000) can additionally create new current action items (up to two) based on the above information by providing a few-shot prompt to a language model (e.g., GPT) the risk factors selected by the user from among the recommended risk tolerances and the three calculated current action items.

[0082]

[0083] FIG. 7 is a diagram for explaining the generation of damage intensity and accident frequency according to one embodiment of the present invention.

[0084] Referring to FIG. 7, the AI-based work risk assessment writing automation device (1000) can search for a JSA document including a work context and risk factors corresponding to a user work context and recommended risk factors in a risk factor database in which a plurality of JSA documents, each including a work context, risk factors, current measures, and damage intensity, are stored.

[0085] An AI-based work risk assessment creation automation device (1000) can generate a damage intensity corresponding to a user work context and recommended risk factors based on the risk intensity of risk factors included in a searched JSA document.

[0086] An AI-based work risk assessment writing automation device (1000) can search for an accident history corresponding to a user work context from a risk factor database in which multiple JSA documents, each including a work context and an accident history, are stored.

[0087] An AI-based work risk assessment creation automation device (1000) can generate an accident frequency corresponding to a user work context and recommended risk factors based on the cause of the searched accident history.

[0088]

[0089] Figure 8 is a block diagram of an automated device for writing AI-based work risk assessment according to another embodiment of the present invention.

[0090] As illustrated in FIG. 8, the AI-based work risk assessment preparation automation device (1000) may include at least one element of a processor (8100), a memory (8200), a storage unit (8300), a user interface input unit (8400), and a user interface output unit (8500), which may communicate with each other via a bus (8600). In addition, the AI-based work risk assessment preparation automation device (1000) may also include a network interface (8700) for connecting to a network. The processor (8100) may be a CPU or a semiconductor device that executes processing instructions stored in the memory (8200) and / or the storage unit (8300). The memory (8200) and the storage unit (8300) may include various types of volatile / non-volatile storage media. For example, the memory may include a ROM (8240) and a RAM (8250).

[0091] The processor (81000) can control the overall operation of the AI-based work risk assessment writing automation device (1000) to perform the present invention.

[0092] The devices described above may be implemented as hardware components, software components, and / or a combination of hardware components and software components. For example, the devices and components described in the embodiments may be implemented using one or more general-purpose computers or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable array (FPA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing instructions and responding. The processing device may execute an operating system (OS) and one or more software applications running on the operating system.

[0093] Additionally, the processing device may access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, the processing device is sometimes described as being used alone; however, those skilled in the art will appreciate that the processing device may include multiple processing elements and / or multiple types of processing elements. For example, the processing device may include multiple processors, or a processor and a controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0094] Software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing device to perform a desired operation or may, independently or collectively, command the processing device. The software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave, for interpretation by the processing device or for providing instructions or data to the processing device. The software may also be distributed over networked computer systems and stored or executed in a distributed manner. The software and data may be stored on one or more computer-readable recording media.

[0095]

[0096] The above description is merely an illustrative illustration of the technical idea of ​​the present invention, and those skilled in the art will appreciate that various modifications and variations can be made without departing from the essential quality of the present invention. Therefore, the embodiments disclosed in this specification are not intended to limit the technical idea of ​​the present invention, but rather to illustrate it, and the scope of the technical idea of ​​the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of the rights of the present invention.

Claims

1. A method performed by an automated device for writing AI-based work risk assessments, A step of retrieving a task context including a JSA title and JSA task content corresponding to a JSA title and JSA task content input by a user from a task context database; A step of searching for a JSA document including a work step and a disaster type corresponding to a work context (user work context) selected by the user among the searched work contexts and a work step input by the user (hereinafter, user work step) in a work step database, thereby generating a disaster type (hereinafter, recommended disaster type) corresponding to the user work context and the user work step; A step of searching for a JSA document including a work step and disaster type corresponding to the user work context and the user work step in the work step database, and generating a recommended risk factor based on the search result and the handling chemical substance input by the user; A step of searching for a JSA document including a work context and risk factors corresponding to the user work context and the recommended risk factors in the work step database, thereby generating a current action (hereinafter, recommended current action) corresponding to a disaster type selected by the user among the recommended disaster types; A step of generating damage intensity and accident frequency corresponding to the user work context and the recommended risk factor by searching for a JSA document including work context, risk factor, and accident history corresponding to the user work context and the recommended risk factor in a risk factor database; and A step of creating a JSA document based on the user task context, the user task step, the disaster type selected by the user, the recommended risk factor, the recommended current action, the damage intensity, and the accident frequency; An automated method for creating AI-based work risk assessments including:

2. In paragraph 1, The steps for retrieving the above working context are: In the above work context database where a plurality of work contexts each including a JSA title and JSA work content are stored, a work context including a JSA title and JSA work content corresponding to the JSA title and JSA work content input by the user is searched based on the vector similarity of the JSA title and JSA work content input by the user. An automated method for creating AI-based work risk assessments including:

3. In paragraph 1, The steps for generating the above recommended disaster types are: A step of searching for a JSA document including a work step and a disaster type corresponding to the user work context and the user work step based on a vector similarity of the user work context and the user work step in the work step database where a plurality of JSA documents including a work context, a risk factor, and a disaster type are stored, respectively; A step of calculating the cosine similarity of the work steps and work contexts included in the searched JSA document; A step of extracting a JSA document corresponding to the task context and the user task step from among the searched JSA documents based on the cosine similarity by making the cosine similarity of the task step and the weight of the cosine similarity of the task step different from each other; and A step of generating the recommended disaster type based on the disaster type included in the extracted JSA document; An automated method for creating AI-based work risk assessments including:

4. In paragraph 1, The steps for generating the above recommended risk factors are: A step of searching for a JSA document including a work step and a disaster type corresponding to the user work context and the user work step in the work step database where a plurality of JSA documents including a work context, a work step, a risk factor, a handling chemical substance, and a disaster type are stored; A step of extracting a JSA document including a disaster type corresponding to a disaster type selected by the user among the recommended disaster types from among the searched JSA documents; and A step of generating a risk factor (hereinafter, recommended risk factor) corresponding to the handling chemical substance input by the user using a language model based on the disaster type, work phase, risk factor, and handling chemical substance included in the extracted JSA document; An automated method for creating AI-based work risk assessments including:

5. In paragraph 1, The steps to create the above current measures are: A step of searching for a JSA document including a task context and risk factor corresponding to the user task context and the recommended risk factor based on a vector similarity of the user task context and the recommended risk factor in a task step database in which a plurality of JSA documents including risk factors and task contexts are stored, respectively; A step of calculating the cosine similarity of the risk factors and work contexts included in the searched JSA documents; A step of extracting a JSA document corresponding to the recommended risk factor and the work context from among the searched JSA documents based on the cosine similarity by making the weights of the cosine similarity of the risk factor and the cosine similarity of the work context different from each other; and A step of generating current measures (hereinafter, recommended current measures) from measures corresponding to the recommended disaster type among the measures included in the extracted JSA document using a language model; An automated method for creating an AI-based work risk assessment, including:

6. In paragraph 1, The steps for generating the above damage intensity are: A step of searching for a JSA document including a work context and risk factor corresponding to the user work context and recommended risk factor in the risk factor database in which a plurality of JSA documents including work context, risk factor, current measures, and damage intensity are stored, and A step of generating a damage intensity corresponding to the user work context and recommended risk factors based on the risk intensity of the risk factors included in the searched JSA document; An automated method for creating an AI-based work risk assessment, including:

7. In paragraph 1, The steps for generating the above accident frequency are: A step of searching for an incident history corresponding to the user's work context in a risk factor database in which a plurality of JSA documents, each including a work context and an incident history, are stored; A step of generating an accident frequency corresponding to the user operation context and recommended risk factors based on the cause of the searched accident history; An automated method for creating an AI-based work risk assessment, including:

8. As an automated device for writing AI-based work risk assessments, a memory storing at least one instruction; and Contains a processor, By executing at least one instruction by the processor, the device, In the work context database, search for a work context that includes a JSA title and JSA work content corresponding to the JSA title and JSA work content entered by the user, In the work step database, by searching for a JSA document including a work step and a disaster type corresponding to a work context (user work context) selected by the user among the searched work contexts and a work step input by the user (hereinafter, user work step), a disaster type (hereinafter, recommended disaster type) corresponding to the user work context and the user work step is generated, In the above work step database, a JSA document including the work step and disaster type corresponding to the user work context and the user work step is searched, and a recommended risk factor is generated based on the search result and the handling chemical substance entered by the user. In the above work step database, by searching for a JSA document including a work context and risk factors corresponding to the user work context and the recommended risk factors, a current action corresponding to a disaster type selected by the user among the recommended disaster types (hereinafter, recommended current action) is generated, By searching for a JSA document including a work context, risk factors, and accident history corresponding to the user work context and the recommended risk factors in the risk factor database, damage intensity and accident frequency corresponding to the user work context and the recommended risk factors are generated. Creating a JSA document based on the user task context, the user task step, the disaster type selected by the user, the recommended risk factor, the recommended current action, the damage intensity, and the accident frequency. An automated device for creating AI-based work risk assessments.

9. In paragraph 8, The above processor, In the above work context database where a plurality of work contexts each including a JSA title and JSA work content are stored, a work context including a JSA title and JSA work content corresponding to the JSA title and JSA work content input by the user is searched based on the vector similarity of the JSA title and JSA work content input by the user. An automated device for writing AI-based work risk assessments, including:

10. In paragraph 8, The processor searches for a JSA document including a work step and a disaster type corresponding to the user work context and the user work step based on a vector similarity of the user work context and the user work step in the work step database in which a plurality of JSA documents including a work context, a risk factor, and a disaster type are stored, and Calculate the cosine similarity of the work steps and work contexts contained in the retrieved JSA documents, By making the cosine similarity of the above work step and the weight of the cosine similarity of the above work step different from each other, the JSA document corresponding to the work context and the user work step is extracted from the searched JSA documents based on the cosine similarity, Generating the above recommended disaster types based on the disaster types included in the extracted JSA documents, An automated device for writing AI-based work risk assessments, including:

11. In paragraph 8, The above processor, In the work step database where a plurality of JSA documents including work contexts, work steps, risk factors, handling chemicals, and disaster types are stored, a JSA document including a work step and disaster type corresponding to the user work context and the user work step is searched, Among the searched JSA documents, a JSA document containing a disaster type corresponding to the disaster type selected by the user among the recommended disaster types is extracted, Using a language model based on the disaster type, work stage, risk factors, and handling chemicals included in the extracted JSA document, a risk factor (hereinafter, recommended risk factor) corresponding to the handling chemicals entered by the user is generated. An automated device for creating AI-based work risk assessments.

12. In paragraph 8, The above processor, Searching for a JSA document including a task context and risk factor corresponding to the user task context and the recommended risk factor based on the vector similarity of the user task context and the recommended risk factor in a task step database in which a plurality of JSA documents including risk factors and task contexts are stored, Calculate the cosine similarity of the risk factors and task contexts included in the retrieved JSA documents. By making the weights of the cosine similarity of the risk factor and the cosine similarity of the work context different from each other, the JSA document corresponding to the recommended risk factor and the work context is extracted from among the searched JSA documents based on the cosine similarity. Using a language model, we generate recommended disaster types from the disaster types included in the extracted JSA documents. An automated device for creating AI-based work risk assessments.

13. In paragraph 8, The above processor, In the risk factor database where multiple JSA documents including each work context, risk factor, current action, and damage intensity are stored, a JSA document including a work context and risk factor corresponding to the user work context and recommended risk factor is searched, Generates a damage intensity corresponding to the user task context and recommended risk factors based on the risk intensity of the risk factors included in the searched JSA document. An automated device for creating AI-based work risk assessments.

14. In paragraph 8, The above processor, In the risk factor database where multiple JSA documents, each containing a work context and an accident history, are stored, the accident history corresponding to the user work context is searched. Based on the cause of the retrieved accident history, generate an accident frequency corresponding to the user operation context and recommended risk factors. An automated device for creating AI-based work risk assessments.

15. A non-transitory computer-readable recording medium storing computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, A step of retrieving a task context including a JSA title and JSA task content corresponding to a JSA title and JSA task content input by a user from a task context database; A step of searching for a JSA document including a work step and a disaster type corresponding to a work context (user work context) selected by the user among the searched work contexts and a work step input by the user (hereinafter, user work step) in a work step database, thereby generating a disaster type (hereinafter, recommended disaster type) corresponding to the user work context and the user work step; A step of searching for a JSA document including a work step and disaster type corresponding to the user work context and the user work step in the work step database, and generating a recommended risk factor based on the search result and the handling chemical substance input by the user; A step of searching for a JSA document including a work context and risk factors corresponding to the user work context and the recommended risk factors in the work step database, thereby generating a current action (hereinafter, recommended current action) corresponding to a disaster type selected by the user among the recommended disaster types; A step of generating damage intensity and accident frequency corresponding to the user work context and the recommended risk factor by searching for a JSA document including work context, risk factor, and accident history corresponding to the user work context and the recommended risk factor in a risk factor database; and A method comprising: causing the processor to perform a step of creating a JSA document based on the user task context, the user task step, the disaster type selected by the user, the recommended risk factor, the recommended current action, the damage intensity, and the accident frequency. Non-transitory computer-readable recording medium.

Images