Personnel safety risk assessment method, system and equipment in electric power operation and medium

Abstract

The invention relates to the technical field of electric power safety, and discloses a personnel safety risk assessment method, system and device in electric power operation and a medium. Removing operation sample data in which environment data abnormity and health index value abnormity of other non-current environment types exist at the same time to obtain an environment sample of the current environment type; training by adopting the environment sample of each environment type to generate an environment clamping model corresponding to the environment type; processing the target job sample data of the target work order type by adopting each environment clamping model, and carrying out weighted calculation on the processing result to obtain a target health index comprehensive prediction value; and according to the target health index comprehensive prediction value and the health index value in the target operation sample data, evaluating the health risk of the operator. The scheme can effectively reduce the potential safety hazard of operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power safety, and in particular to a personnel safety risk assessment method, system, device and medium in power operation. BACKGROUND

[0002] With the rapid development of the power industry at the present stage and the increasing complexity of the actual environment of power operation, the working environment of the front-line personnel in the power system is also becoming more and more severe. The current severe working environment threatens the occupational health of the front-line personnel in the power system. The power grid is currently working hard to eliminate these threats.

[0003] There are different types of operations in the front-line work of the power system. When the operating personnel perform different types of operations in different environments, different safety risk problems will occur. The occupational health management system for the front-line personnel in the power industry is not perfect at present, and it is difficult to effectively obtain the changes of the physical health index values of the operating personnel in different types of operations. There is no specific management method for the changes of the physical health index values of the operating personnel in different types of operations. This results in the inability to guarantee the physical health of the front-line workers, so that the operating personnel are prone to safety hazards during the execution of power operation. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a personnel safety risk assessment method, system, device and medium in power operation, which can solve the problem that the occupational health management system for the front-line personnel in the power industry is not perfect, and it is difficult to effectively obtain the changes of the physical health index values of the operating personnel in different types of operations and lack of specific management method.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a personnel safety risk assessment method in power operation, comprising: Obtaining a plurality of operation sample data of a target work order type that has been completed, each operation sample data comprising operation progress data, environment data and health index value of operating personnel; For each environment type, removing the operation sample data in the plurality of operation sample data that simultaneously has other non-current environment type environment data anomaly and health index value anomaly, and taking each operation sample data retained as an environment sample corresponding to the current environment type; For each environment type, using the corresponding environment sample to construct and train to obtain an environment holding model of the current environment type; The input of the environment holding model is the job progress data in the environment sample and the environment data of the current environment type to which the job progress data belongs, and the output is a predicted value of a sample health index; For target job sample data of the target work order type, each of the environment holding models of the target work order type is used for processing to obtain a plurality of target health index predicted values, and the plurality of target health index predicted values are weighted and calculated based on the obtained target weight ratio to obtain a target health index comprehensive predicted value; According to the target health index comprehensive predicted value and the health index value in the target job sample data, the health risk of the job personnel corresponding to the target job sample data is evaluated.

[0007] As a preferred scheme of the personnel safety risk assessment method in the power operation, wherein: for each environment type, the job sample data in which the environment data of other non-current environment types and the health index value are abnormal at the same time is removed from the plurality of job sample data, and each of the remaining job sample data is taken as an environment sample corresponding to the current environment type, including: For each environment type, it is determined whether there is job sample data in which the environment data of other non-current environment types is in a preset first abnormal interval in the plurality of job sample data; For the plurality of job sample data, the job sample data in which the environment data of the other non-current environment types is in the first abnormal interval and the health index value is in a preset second abnormal interval at the same time is removed, and each of the remaining job sample data is taken as an environment sample corresponding to the current environment type.

[0008] As a preferred scheme of the personnel safety risk assessment method in the power operation, wherein: for each environment type, the job sample data in which the environment data of other non-current environment types and the health index value are abnormal at the same time is removed from the plurality of job sample data, and each of the remaining job sample data is taken as an environment sample corresponding to the current environment type, including: For each environment type, the plurality of job sample data is divided into a plurality of data groups according to a plurality of environment value intervals corresponding to the current environment type, and based on the job sample data in which the health index value is abnormal in each of the data groups, a data space range determined by the environment data of other non-current environment types in the job sample data is circled; For the plurality of data space ranges of the current environment type, the smallest one is selected as a first abnormal space range; For the plurality of job sample data, job sample data in which the environment data of the other non-current environment type is in the first abnormal space range and the health indicator value is in a preset second abnormal interval is removed, and each remaining job sample data is used as an environment sample corresponding to the current environment type.

[0009] As a preferred scheme of the method for assessing personnel safety risks in power operation, for each environment type, a corresponding environment sample is used to build and train an environment holding model of the current environment type, including: For the environment sample corresponding to each environment type, job progress data in the environment sample and environment data of the current environment type are input into a pre-built first neural network model for processing, and a first sample health indicator prediction value is output. The health indicator value in the environment sample is used as a first sample health indicator measured value, a first loss function is built based on the first sample health indicator measured value and the first sample health indicator prediction value, and the first neural network model is trained based on the first loss function to obtain an environment holding model of the current environment type. The environment type includes any of the following types independently or a combination of at least two types: time period, geographical condition, weather condition, noise, and light.

[0010] As a preferred scheme of the method for assessing personnel safety risks in power operation, the method further includes: Based on the job sample data and the sample health indicator prediction values output by the environment holding models, a matching model corresponding to the target work order type is built and trained, the input of the matching model is the environment data in the job sample data, and the output is a sample weight matching used for weighted calculation of each sample health indicator prediction value output by each environment holding model. The environment data in the target job sample data is input into the matching model for processing to obtain the target weight matching.

[0011] This preferred scheme can obtain a reasonable sample weight matching according to the job sample data and the output of the environment holding model by building and training the matching model. This enables accurate determination of the target weight matching according to the environment data when the target job sample data is processed.

[0012] As a preferred scheme of the power operation personnel safety risk assessment method, wherein: the plurality of operation sample data and the sample health index prediction value output by each of the environment holding models are used to build and train a matching model corresponding to the target work order type, including: The environment data in the operation sample data is input into a pre-built second neural network model for processing, and a sample weight matching used for weighted calculation of each sample health index prediction value output by each of the environment holding models is output. The operation progress data and the environment data in the operation sample data are input into each of the corresponding environment holding models for processing, and a plurality of second sample health index prediction values are output. The plurality of second sample health index prediction values are weighted calculated using the sample weight matching, and a sample health index comprehensive prediction value is obtained. The health index value in the operation sample data is taken as a sample health index comprehensive measured value, a second loss function is built based on the sample health index comprehensive measured value and the sample health index comprehensive prediction value, and the second neural network model is trained based on the second loss function, and a matching model corresponding to the target work order type is obtained.

[0013] As a preferred scheme of the power operation personnel safety risk assessment method, wherein: according to the target health index comprehensive prediction value and the health index value in the target operation sample data, the health risk of the operation personnel corresponding to the target operation sample data is evaluated, including: When the health index comprehensive prediction value and the health index comprehensive measured value are both abnormal, it is determined that the operation personnel is in a high health risk; When only one of the health index comprehensive prediction value and the health index comprehensive measured value is abnormal, it is determined that the operation personnel is in a medium health risk; When the health index comprehensive prediction value and the health index comprehensive measured value are both not abnormal, it is determined that the operation personnel is in a low health risk or no health risk.

[0014] In a second aspect, the present application provides a power operation personnel safety risk assessment system, comprising: A data acquisition module is configured to acquire a plurality of operation sample data of a target work order type, each operation sample data comprising operation progress data, environment data and health index value of an operation personnel. a preprocessing module configured to remove, for each environment type, job sample data in which environment data anomalies and health indicator value anomalies of other non-current environment types coexist, and to retain each job sample data as an environment sample corresponding to the current environment type; a model construction module configured to, for each environment type, construct and train an environment holding model of the current environment type using the corresponding environment sample; the environment holding model takes as input job progress data in the environment sample and environment data of the current environment type, and outputs a sample health indicator prediction value; a prediction module configured to, for target job sample data of a target work order type, process the target job sample data using each environment holding model of the target work order type to obtain multiple target health indicator prediction values, and perform weighted calculation on the multiple target health indicator prediction values based on a target weight ratio to obtain a target health indicator comprehensive prediction value; an evaluation module configured to evaluate a health risk of a job worker corresponding to the target job sample data based on the target health indicator comprehensive prediction value and a health indicator value in the target job sample data.

[0015] In a third aspect, the present application provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method described above when executing the computer program.

[0016] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the method described above.

[0017] Compared with the prior art, the present application has the beneficial effect that the present application proposes a method for evaluating personnel safety risks in electric power operations, which removes, for each environment type, job sample data in which environment data anomalies and health indicator value anomalies of other non-current environment types coexist, to obtain environment samples of the current environment type, trains and generates environment holding models of each environment type using the environment samples of the environment type, processes target job sample data of a target work order type using each environment holding model, and performs weighted calculation on the processing results to obtain a target health indicator comprehensive prediction value, and evaluates a health risk of a job worker based on the target health indicator comprehensive prediction value and a health indicator value in the target job sample data. The present application can effectively evaluate the health risks of power grid workers to reduce operation safety hazards. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only represent some of the embodiments of the present application, and all other drawings obtained by those of ordinary skill in the art without creative effort based on these drawings should also fall within the protection scope of the present application.

[0019] Figure 1 The method flow chart of the personnel safety risk assessment method in power operation provided by an embodiment of the present application.

[0020] Figure 2 The flow chart of the method of acquiring environment samples of the personnel safety risk assessment method in power operation provided by an embodiment of the present application.

[0021] Figure 3 The flow chart of the method of acquiring environment samples of the personnel safety risk assessment method in power operation provided by an embodiment of the present application.

[0022] Figure 4 The flow chart of the training method of the environment holding model of the personnel safety risk assessment method in power operation provided by an embodiment of the present application.

[0023] Figure 5 The flow chart of the training method of the proportioning model of the personnel safety risk assessment method in power operation provided by an embodiment of the present application.

[0024] Figure 6 The internal structure diagram of the electronic device of the personnel safety risk assessment method in power operation provided by an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings in the specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the protection scope of the present application.

[0026] Embodiment 1, refer to Figures 1-4 The first embodiment of the present application provides a personnel safety risk assessment method in power operation, which comprises: As Figure 1 shown, the personnel safety risk assessment method in power operation comprises the following steps.

[0027] Step 110: Obtain a plurality of job sample data of the target work order type, each of which includes job progress data, environmental data, and health index values of the job personnel.

[0028] The work order type can be classified according to multiple dimensions such as work type (power transmission, power transformation, power distribution, etc.), job content, technical elements contained, and responsibility skills. In order to accurately classify the work order type, the present embodiment classifies the work order according to the above-mentioned dimensions to obtain the work order type. The target work order type can be any work order type of the existing work order type. For any work order of any work order type, by sampling the relevant data at each cycle time point of the job progress development, the job sample data at each cycle time point for the work order can be obtained. Each job sample data can include job progress data, environmental data, and health index values of the job personnel.

[0029] Job progress data: data related to the description of job execution progress, including but not limited to job progress duration, work order completion progress and effect, etc.

[0030] Environmental data: description data of the environment in which the job personnel is located, such as time period (daytime, nighttime, etc.), geography (surrounding facility profile, spatial size and / or degree of enclosure of the work environment, altitude, etc.), weather conditions (wind, rain, temperature, humidity, seasonal changes, etc.), noise, light, etc.

[0031] Health index values are health index value information reflecting the health of the job personnel collected by wearing devices, image acquisition devices, and other facilities during the job process, such as health information of visceral function, mental state, and basic vital signs. Those skilled in the art can set health index value items according to actual device capabilities and needs, which are not listed here.

[0032] In an optional embodiment, for a plurality of work orders of the target work order type, a plurality of job sample data collected during the execution of each work order can be collected to obtain a plurality of job sample data of the target work order type. In the present embodiment, the plurality of job sample data of the target work order type obtained above are not classified and distinguished according to the specific work order, but are directly subordinate to the target work order type without distinction, which can better reflect the job sample distribution of a certain target work order type as a whole.

[0033] Step 120: For each environment type, remove the job sample data in which the environmental data and health index values of other non-current environment types are abnormal from the plurality of job sample data, and retain each job sample data as an environment sample corresponding to the current environment type.

[0034] The environment type can include, but is not limited to, any one of the following types independently or a combination of at least two types: time period, geographical condition, weather condition, noise, and light.

[0035] In this embodiment, in order to better learn the influence of the environment data of each environment type on the health index value from the plurality of job sample data of the target work order type, when preparing the sample data, for any current environment type, job sample data that is mainly influenced by the environment data of the environment type and affects the health index value is selected as much as possible, and such job sample data is used as the environment sample corresponding to the current environment type.

[0036] In an optional embodiment, in the process of preparing the environment sample corresponding to each environment type, the reverse thinking can be considered, that is, for a certain current environment type, job sample data that simultaneously has environment data anomaly and health index value anomaly of other non-current environment types is first selected from the plurality of job sample data of the target work order type, and then the selected job sample data is removed from all the job sample data of the target work order type, and each job sample data that is left is used as the environment sample corresponding to the current environment type.

[0037] It should be noted that if there are other non-current environment type environment data abnormalities and health indicator value abnormalities in a job sample data, it can be considered that the other non-current environment type environment data has a boosting effect on the abnormality of the health indicator value, and does not meet the requirement that the current environment type environment data alone has a boosting effect on the health indicator value, and therefore cannot be used as the environment sample corresponding to the current environment type; if there are other non-current environment type environment data abnormalities / abnormalities and health indicator value abnormalities in a job sample data, it is considered that the other non-current environment type environment data does not have a boosting effect on the abnormality of the health indicator value regardless of its state, and the environment data of the current environment type still has an independent boosting effect on the health indicator value, so it can be used as the environment sample corresponding to the current environment type; if there are other non-current environment type environment data abnormalities and health indicator value abnormalities in a job sample data, it is considered that the other non-current environment type environment data is not the cause of the health indicator value abnormality, and does not have a boosting effect on the abnormality of the health indicator value. The cause of the health indicator value abnormality is most likely due to the boosting effect of the environment data of the current environment type on the state of the health indicator value, and even this boosting effect is the independent boosting of the environment data of the current environment type, or at least the boosting effect of the combination of the other non-current environment type environment data with the same weight probability. Based on the probability, the embodiment also retains such job sample data as the environment sample corresponding to the current environment type.

[0038] In summary, for each environment type, the embodiment only removes a job sample data when there are other non-current environment type environment data abnormalities and health indicator value abnormalities in the job sample data, and retains the remaining job sample data as the environment sample corresponding to the current environment type. After this step, the above-mentioned multiple job sample data originally belonging to the target work order type are respectively partially removed according to each environment type, and each job sample data retained after the removal becomes an environment sample corresponding to the current environment type. These environment samples can be used to learn the boosting effect of the environment data of each environment type on the health indicator value.

[0039] Step 130: For each environment type, an environment boosting model of the current environment type is constructed and trained using the corresponding environment sample. The input of the environment boosting model is the job progress data in the environment sample and the environment data of the current environment type, and the output is the sample health indicator prediction value.

[0040] In an optional implementation, to learn the impact of work orders of a target work order type on the health indicator values ​​of operators during work order execution, under the influence of environmental data of each environmental type individually, this example constructs and trains an environmental support model for each environmental type faced under the target work order type, using corresponding environmental samples. The input to each environmental support model is the work progress data from the corresponding environmental sample and the environmental data of the current environmental type; the output is the predicted health indicator value. To distinguish it from other health indicator values, this embodiment refers to the predicted health indicator value obtained based on the training samples during model training as the "sample health indicator prediction value."

[0041] Step 140: For the target operation sample data of the target work order type, process it using the environmental support model of the target work order type to obtain multiple target health indicator prediction values, and calculate the multiple target health indicator prediction values ​​by weighting based on the obtained target weight ratio to obtain the comprehensive prediction value of the target health indicator.

[0042] For the target work order type, based on the aforementioned steps, we can learn the influence of environmental data of each environmental type on the health index values ​​of workers during the work process, i.e., the environmental enhancement model. In the actual prediction process, by combining the predicted values ​​of multiple health indicators obtained under various environmental types in a certain ratio, we can predict the comprehensive health index value of workers.

[0043] In an optional implementation, for a given target work sample data of a target work order type, the target work sample data can be the currently collected work sample data corresponding to a target work order that is in operation, or the work sample data collected at a certain historical moment corresponding to a target work order that has been completed. The target work sample data is processed using the environmental support models of the target work order type that have been trained as described above, and multiple predicted health indicator values ​​are obtained. In order to distinguish them from other predicted health indicator values, this embodiment refers to the predicted health indicator value obtained based on the given target sample during the model application process as the "target health indicator prediction value". Based on the pre-acquired target weight ratio for the target work order type, the multiple target health indicator prediction values ​​obtained above are weighted and calculated to obtain the comprehensive prediction value of the target health indicator.

[0044] In this embodiment, to comprehensively process the predicted health indicators obtained under various environmental types for the target work order type according to a certain ratio, a weighting ratio is introduced. This weighting ratio refers to the proportional relationship between multiple weight values ​​used in the weighted calculation of the predicted health indicators under various environmental types. Since the weighting ratio in this step is used for the target work sample data, it is also called the "target weighting ratio." The determination of the weighting ratio can be based on empirical values ​​summarized from historical work sample data, or it can be based on model prediction values ​​obtained after training a neural network model. This embodiment does not limit the method of obtaining the weighting ratio.

[0045] Step 150: Assess the health risks of workers corresponding to the target work sample data based on the comprehensive predicted values ​​of the target health indicators and the health indicator values ​​in the target work sample data.

[0046] In one optional implementation, during the execution of a work order of a target work order type, operators can periodically acquire target work sample data. This allows for the prediction, based on historical experience, of the health indicator values ​​that operators are likely to possess, considering various environmental types within these target work sample data. This indirectly suggests the presence of health risks among operators. Simultaneously, the health indicator values ​​included in the target work sample data represent actual, detected health indicators possessed by operators, allowing for direct inference of whether they have health risks. However, both direct and indirect inference have limitations, as detailed below.

[0047] Direct speculation highlights the randomness of health risks. To give an extreme example, in a relatively suitable working environment, a worker's health indicators may fluctuate randomly due to unforeseen circumstances. However, if this abnormality disappears quickly and does not cause any substantial harm to the worker's health or the work process, then this actual abnormality in health indicators can be characterized as a random abnormality. The worker does not have any substantial health risk, or the probability of having a health risk is relatively low.

[0048] Indirect speculation highlights the inevitability of health risks. To give an extreme example, in a relatively unsuitable working environment, the health indicators of workers should theoretically be highly likely to be abnormal. However, if the actual measured health indicators are not abnormal, it does not mean that the workers' current health status is necessarily fine, or that the probability of abnormal health indicators appearing in the short term is still relatively high. The health risks of workers and the potential risks they bring to the work process cannot be ignored. Therefore, such possible abnormal health indicators can be regarded as an inevitable abnormality that is highly likely to occur, and workers still have a high probability of health risks.

[0049] It should be noted that, based on the above analysis, this embodiment considers both the randomness and inevitability of health risks. When assessing the health risks of workers, it will evaluate the health risks of the workers corresponding to the target work sample data based on the comprehensive predicted value of the target health indicators obtained from the target work sample data acquired in real time and the actual health indicator values ​​obtained from the target work sample data (also known as the "comprehensive measured value of the target health indicators"). This will allow for a more comprehensive and complete assessment of the probability of health risks to workers.

[0050] In one optional implementation, the specific methods for assessing the health risk of workers corresponding to the target work sample data based on the comprehensive predicted value of the target health indicators and the health indicator values ​​in the target work sample data include, but are not limited to, the following steps.

[0051] When both the comprehensive predicted value and the comprehensive measured value of health indicators are abnormal, the worker is determined to be at a high health risk. When only one of the comprehensive predicted value of health indicators and the comprehensive measured value of health indicators is abnormal, the worker is determined to be at a medium health risk. When both the comprehensive predicted value and the comprehensive measured value of health indicators are not abnormal, the worker is determined to be at a low level of health risk or no health risk.

[0052] It should be noted that when both the comprehensive predicted value and the comprehensive measured value of health indicators are abnormal, regardless of historical inevitability or the randomness of the current state, it indicates that the abnormal health indicator values ​​of the workers truly reflect their physical health status, and that the workers are indeed in a relatively high health risk state. Therefore, this state is defined as the workers being in a high-risk health condition. However, when only one of the comprehensive predicted value or the comprehensive measured value of health indicators is abnormal, it indicates that the workers' physical health status, whether based on historical inevitability or the randomness of the current state, cannot fully and accurately determine whether the abnormal health indicator value is due to... The health indicators can accurately reflect the physical health status of workers. Conservatively speaking, workers are mostly in a state of moderate health risk, so this state is defined as medium health risk. When both the comprehensive predicted value and the comprehensive measured value of health indicators are not abnormal, it indicates that the workers' physical health status, whether from historical inevitability or current randomness, can accurately show that the workers' health indicator values ​​are not abnormal and can accurately reflect the workers' physical health status. Workers are mostly in a state of no or relatively low health risk, so this state is defined as low health risk or no health risk.

[0053] The method for assessing personnel safety risks in power operations provided in this implementation acquires multiple work sample data of the completed target work order type. Each work sample data includes work progress data, environmental data, and health indicator values ​​of the workers. For each environmental type, work sample data that simultaneously exhibits abnormal environmental data and abnormal health indicator values ​​in other non-current environmental types is removed, and each retained work sample data is used as the environmental sample corresponding to the current environmental type. For each environmental type, an environmental enhancement model for the current environmental type is constructed and trained using the corresponding environmental sample. The input of the environmental enhancement model is the work progress data and environmental data in the environmental sample, and the output is the predicted value of the sample health indicator. For the target work sample data of the target work order type, the environmental enhancement models of the target work order type are used for processing to obtain multiple predicted values ​​of target health indicators. Based on the obtained target weight ratio, the multiple predicted values ​​of target health indicators are weighted and calculated to obtain a comprehensive predicted value of target health indicators. Based on the comprehensive predicted value of target health indicators and the health indicator values ​​in the target work sample data, the health risk of the workers corresponding to the target work sample data is assessed. This solution employs an environmental enhancement model, which can more clearly and accurately learn the impact of different environmental types on the health indicators of workers under specific work order types. By weighting the predicted health indicators output by each environmental enhancement model, a comprehensive predicted health indicator value is obtained. This comprehensive predicted health indicator value is then combined with the corresponding measured values ​​to obtain a more accurate and comprehensive health risk assessment result for workers. This effectively enables a comprehensive assessment of the health risks of workers during the work process, allowing for early risk warnings and reducing safety hazards during electrical work.

[0054] Based on the above embodiments, this embodiment further optimizes the method for assessing personnel safety risks in power operations. During step 120, two specific execution methods are provided, namely two methods for obtaining environmental samples.

[0055] like Figure 2 As shown in the figure, the method for obtaining environmental samples provided in this embodiment includes the following steps.

[0056] Step 210: For each environment type, determine whether there are other environmental data that are not in the current environment type and are in the preset first abnormal interval among multiple job sample data.

[0057] For each environment type, to determine whether environmental data of other non-current environment types in the job sample data are abnormal, an abnormality interval for judging whether environmental data of each environment type is abnormal can be pre-set based on empirical values, and the abnormality intervals corresponding to the environmental data of other non-current environment types are collectively referred to as the "first abnormality interval". It can be understood that the first abnormality interval covers the abnormality intervals corresponding to the environmental data of each environment type in the other non-current environment types. In this embodiment, it is defined that for any job sample data, if at least one environment type's environmental data exists in the corresponding abnormality interval among the environmental data of other non-current environment types, then the environmental data of other non-current environment types in the job sample data are considered to be collectively in the first abnormality interval.

[0058] In an optional implementation, for each environment type, it is determined whether there is at least one type of environment data in the environmental data of other non-current environment types in the job sample data that is in the corresponding abnormal interval. If so, it is determined that the environmental data of other non-current environment types in the job sample data is in the first abnormal interval.

[0059] Step 220: For multiple job sample data, remove job sample data that simultaneously contains environmental data of other non-current environment types in the first abnormal interval and health indicator values ​​in the preset second abnormal interval, and use each retained job sample data as the environmental sample corresponding to the current environment type.

[0060] To determine whether health indicator values ​​in the work sample data are abnormal, an abnormality range for each type of health indicator can be pre-defined based on empirical values. This range, encompassing all health indicator types, is referred to as the "second abnormality range." Essentially, this second abnormality range covers the abnormality ranges for all health indicator types. If at least one health indicator type in the work sample data falls within its corresponding abnormality range, then all health indicator types in the work sample data are considered to fall within the second abnormality range.

[0061] In an optional implementation, for the aforementioned multiple job sample data, it can be determined whether at least one type of health indicator value in each health indicator type of each job sample data falls within the corresponding abnormal interval. If so, all health indicator values ​​in that job sample data are determined to be in the second abnormal interval, and simply denoted as "health indicator value is in the second abnormal interval". For the aforementioned multiple job sample data, under the current environment type, job sample data that simultaneously contains environmental data from other non-current environment types in the first abnormal interval and health indicator values ​​in the second abnormal interval are removed, and each retained job sample data is used as the environment sample corresponding to the current environment type.

[0062] Figure 2 In the illustrated embodiment, for the current environment type, by using a pre-set first abnormality interval for judging whether other environmental data that are not of the current environment type are abnormal, and a second abnormality interval for judging whether health indicator values ​​are abnormal, the abnormality of these two types of data can be quickly determined, thereby quickly obtaining the environmental sample corresponding to the current environment type from the above multiple work sample data.

[0063] like Figure 3 As shown in the figure, another method for obtaining environmental samples provided in this embodiment includes the following steps.

[0064] Step 310: For each environment type, divide the multiple work sample data into multiple data groups according to the multiple environmental value ranges corresponding to the current environment type. Based on the work sample data containing abnormal health indicator values ​​in each data group, delineate the data space range determined by other environmental data of non-current environment types in the work sample data.

[0065] Among them, multiple adjacent environmental value intervals can be pre-set for each environmental type, and the environmental value of each environmental type in the above multiple operation sample data can fall into an environmental value interval under the corresponding environmental type.

[0066] In an optional implementation, for each environment type, the aforementioned multiple job sample data can be divided according to the environmental value range to which their environmental values ​​belong under the current environment type. Job sample data corresponding to environmental values ​​belonging to the same environmental value range are grouped into the same data group, thus forming multiple data groups. Each data group corresponds to an environmental value range under the current environment type. Within any data group under the current environment type, it can be assumed that the environmental values ​​of each job sample data in this data group are similar under the current environment type, and the influence of these environmental values ​​on the health indicator values ​​in the current data group is also similar. If some health indicator values ​​are abnormal and some are not abnormal in the current data group, it can be inferred that the cause of the abnormal health indicator values ​​is likely due to the influence of abnormal environmental data from other non-current environment types, and the spatial distribution of abnormal environmental data from other non-current environment types is consistent with the spatial distribution of abnormal health indicator values. Therefore, based on the distribution of abnormal health indicator values ​​in the work sample data within each data group, a data space range defined by the distribution of other environmental data of non-current environment types in these work sample data can be delineated. It can be assumed that within this data space range, since the ratio of the corresponding work sample data with abnormal health indicator values ​​to the total work sample data within this data space range reaches a preset ratio, it can be assumed that the proportion of abnormal data in other environmental data of non-current environment types within this data space range also reaches a preset ratio. In other words, this data space range is considered to be a concentrated distribution area where abnormal environmental data of other non-current environment types occur.

[0067] Step 320: For the multiple data space ranges of the current environment type, select the smallest space range as the first abnormal space range.

[0068] In one optional implementation, for multiple data space ranges of the current environment type, if the data space range is small, it indicates that the environmental value interval has a relatively small impact on the abnormal health indicator values ​​in the data group, and the abnormal health indicator values ​​are mainly caused by the influence of environmental data from other non-current environment types within the data space range. However, as the data space ranges corresponding to different environmental value intervals increase, this phenomenon indicates that the influence of the environmental value interval on the abnormal health indicator values ​​in the data group increases, leading to an increase in the number of abnormal health indicator values. Since these newly added abnormal health indicator values ​​are not influenced by environmental data from other non-current environment types, the portion of the data space range determined by the newly added abnormal health indicator values ​​can also be considered as not being the portion of environmental data from other non-current environment types that is abnormal. Furthermore, this conclusion becomes increasingly apparent as the data space range increases. Therefore, in this embodiment, for multiple data space ranges corresponding to the current environment type, the smallest data space range is selected as the space range for determining abnormal environmental data from other non-current environment types, also denoted as the "first abnormal space range".

[0069] Step 330: For multiple job sample data, remove job sample data that simultaneously contains environmental data of other non-current environment types within the first abnormal space range and health indicator values ​​within the preset second abnormal range, and use each retained job sample data as the environmental sample corresponding to the current environment type.

[0070] For details regarding the second abnormal interval, please refer to the aforementioned content, which will not be repeated here.

[0071] In an optional implementation, for the aforementioned multiple job sample data, it can be determined whether at least one type of health indicator value in each health indicator type of each job sample data falls within the corresponding abnormal interval. If so, all health indicator values ​​in that job sample data are determined to be within the second abnormal interval, and simply denoted as "health indicator value is within the second abnormal interval". For the aforementioned multiple job sample data, under the current environment type, job sample data that simultaneously contains environmental data of other non-current environment types within the first abnormal space range and health indicator values ​​within the second abnormal interval are removed, and each retained job sample data is used as the environment sample corresponding to the current environment type.

[0072] Figure 3In the illustrated embodiment, for the current environment type, a first abnormal space range for judging whether other environmental data that are not of the current environment type are abnormal is determined by data statistics, and a second abnormal interval for judging whether health indicator values ​​are abnormal. This allows for the rapid determination of abnormalities in these two types of data, thereby quickly obtaining the environmental sample corresponding to the current environment type from the aforementioned multiple work sample data.

[0073] Based on the above embodiments, this embodiment further optimizes the method for assessing personnel safety risks in power operations. During step 130, a training process for an environmentally supported model is presented.

[0074] like Figure 4 As shown in the figure, the training method of the environment-supported model provided in this embodiment includes the following steps.

[0075] Step 410: For each environmental sample corresponding to each environmental type, input the operation progress data and the environmental data of the current environmental type in the environmental sample into the pre-built first neural network model for processing, and output the predicted value of the first sample health index.

[0076] Regarding the network structure of the first neural network model, this embodiment does not limit it and may include, but is not limited to, any one of the following: fully connected neural network (FCN), convolutional neural network (CNN), and residual network (ResNet).

[0077] In an optional implementation, under the target work order type, for each environmental sample corresponding to each environmental type, the work progress data and the environmental data of the current environmental type in the environmental sample can be extracted and input into the pre-built first neural network model for processing, outputting a predicted health indicator value. To distinguish it from the sample health indicator prediction values ​​obtained by other models during training, this embodiment refers to the sample health indicator prediction value obtained by the first neural network model based on the environmental samples during training as the "first sample health indicator prediction value." This first sample health indicator prediction value describes the impact of the work progress data in the current environmental sample on the predicted health indicator value of the workers, under the influence of the environmental data of the current environmental type.

[0078] Step 420: Use the health indicator values ​​in the environmental samples as the measured values ​​of the first sample health indicators, construct a first loss function based on the measured values ​​and predicted values ​​of the first sample health indicators, and train the first neural network model based on the first loss function to obtain the environmental support model of the current environmental type.

[0079] In an optional implementation, during the training process, the health indicator values ​​in the environmental samples can be used as the measured values ​​of the first sample health indicators, i.e., the label values; a first loss function is constructed based on the difference between the measured values ​​of the first sample health indicators and the corresponding predicted values ​​of the first sample health indicators; and the first neural network model is trained based on the first loss function, and the network parameters of the first neural network model are adjusted, so that after the training process is completed, an environment-supported model of the current environment type is obtained.

[0080] The form of the first loss function is not limited and may include, but is not limited to, the squared loss function, the absolute loss function, etc.

[0081] In some embodiments, the first loss function may include: Where Loss1 is the loss value of the first loss function, and xi is the predicted value of the health indicator of the i-th first sample. Let be the measured value of the health indicator of the i-th first sample, and n be the number of samples.

[0082] In this embodiment, a pre-constructed first neural network model is trained based on environmental samples corresponding to each environment type. During the training process, a first loss function is constructed based on the measured value of the first sample health index in the environmental samples and the predicted value of the first sample health index output by the model to constrain the training results, thereby obtaining the environment-enhanced model of the current environment type. This provides a convenient and fast training method for the environment-enhanced model.

[0083] Example 2, refer to Figure 5 Based on the above embodiments, this embodiment further optimizes the method for assessing personnel safety risks in power operations. The execution of step 140 involves "target weight allocation," and this embodiment specifically provides a method for obtaining the target weight allocation: first, a weight allocation model for predicting the weight allocation is obtained through neural network model training; then, the target weight allocation is obtained based on the weight allocation model.

[0084] Regarding the matching model, it can be based on the work sample data and the predicted values ​​of sample health indicators output by each environmental support model. The matching model corresponding to the target work order type can be constructed and trained. The input of the matching model is the environmental data in the work sample data, and the output is the sample weight matching used to calculate the weighted ratio of each sample health indicator predicted value output by each environmental support model.

[0085] In an optional implementation, to learn the proportion of the impact of environmental data from various environmental types on the health indicator values ​​of workers during work order execution, this example uses the aforementioned multiple work sample data under the target work order type, as well as the sample health indicator prediction values ​​output by each environmental support model under the target work order type, as training samples to construct and train the matching model corresponding to the target work order type. The input of this matching model is the environmental data in the aforementioned work sample data, and the output is the weight ratio used to calculate the weighted health indicator prediction values ​​of each sample output by each environmental support model. To distinguish it from other weight ratios, this embodiment refers to the weight ratio obtained based on the training samples during model training as the "sample weight ratio".

[0086] like Figure 5 As shown, regarding the proportioning model, the training method for the proportioning model provided in this embodiment includes the following steps.

[0087] Step 510: Input the environmental data in the work sample data into the pre-built second neural network model for processing, and output the sample weight ratio for weighted calculation of the predicted health index values ​​of each sample output by each environmental support model.

[0088] Regarding the network structure of the second neural network model, this embodiment does not limit it and may include, but is not limited to, any one of the following: fully connected neural network (FCN), convolutional neural network (CNN), and residual network (ResNet).

[0089] In an optional implementation, under the target work order type, using the aforementioned work sample data as training samples, environmental data can be extracted from the work sample data and input into a pre-built second neural network model for processing. The output is a weighted ratio used to weight the predicted health index values ​​of each sample output by the model for each environment. To distinguish it from the target weight ratio obtained during model application, this embodiment refers to the weight ratio obtained by the second neural network model based on the work sample data during training as the "sample weight ratio." This sample weight ratio describes the proportion of each environment type's environmental data in the impact on the health index values ​​of the workers under the current target work order type, when the work progress data in the work sample data is jointly influenced by environmental data from each environment type. It is understood that this proportion is not fixed but changes with the environmental data of each environment type in each work sample data. By training the model, this changing pattern can be learned, thereby enabling flexible setting of the weight ratio.

[0090] Step 520: Input the work progress data and environmental data from the work sample data into the corresponding environmental support models for processing, and output multiple predicted values ​​of the second sample health indicators.

[0091] In an optional implementation, under the target work order type, the aforementioned work sample data can be used as training samples. The work progress data and environmental data for each environmental type within the work sample data are extracted and input into the corresponding trained environmental support models for processing, outputting multiple predicted health indicator values. To distinguish these values ​​from the predicted health indicator values ​​obtained by the second neural network model based on the work sample data during training, this embodiment refers to the predicted health indicator values ​​as "second sample health indicator prediction values." These second sample health indicator prediction values ​​describe the predicted impact of the work progress data in the work sample data on the health indicator values ​​of the workers under the influence of environmental data for any environmental type.

[0092] Step 530: The predicted values ​​of multiple second-sample health indicators are weighted by the sample weight ratio to obtain the comprehensive predicted value of the sample health indicators.

[0093] Specifically, based on the same work sample data, multiple predicted health index values ​​for second samples can be obtained through trained environmental support models, and sample weight ratios can be obtained through the training of a second neural network model. Then, according to the correspondence of environmental types, the predicted health index values ​​for second samples corresponding to the same environmental type are multiplied by the corresponding weight values ​​in the sample weight ratios, and then all are added together to achieve a weighted calculation, resulting in a comprehensive predicted health index value. To distinguish it from the comprehensive predicted health index value of the target health index during model application, this embodiment refers to the comprehensive predicted health index value during model training as the "comprehensive predicted health index value of the sample".

[0094] Step 540: Use the health index values ​​in the work sample data as the comprehensive measured values ​​of the sample health indexes. Construct a second loss function based on the comprehensive measured values ​​and the comprehensive predicted values ​​of the sample health indexes. Train the second neural network model based on the second loss function to obtain the matching model corresponding to the target work order type.

[0095] In an optional implementation, during the training process, the health index values ​​in the work sample data can be used as the comprehensive measured values ​​of the sample health index, i.e., the label values; a second loss function is constructed based on the difference between the comprehensive measured values ​​of the sample health index and the corresponding comprehensive predicted values ​​of the sample health index; and the second neural network model is trained based on the second loss function, and the network parameters of the second neural network model are adjusted, so that after the training process is completed, the matching model of the target work order type is obtained.

[0096] The form of the second loss function is not limited and may include, but is not limited to, the squared loss function, the absolute loss function, etc.

[0097] In some embodiments, the second loss function may include: Where Loss2 is the loss value of the second loss function, and yj is the comprehensive predicted value of the health index of the j-th sample. Let be the comprehensive measured value of the health indicators of the j-th sample, and m be the number of samples.

[0098] In this embodiment, under the target work order type, a pre-constructed second neural network model is trained based on multiple work sample data and the predicted values ​​of sample health indicators output by each environmental support model obtained after the multiple work sample data have been trained. During the training process, a second loss function is constructed based on the comprehensive measured values ​​of sample health indicators in the work sample data and the comprehensive predicted values ​​of sample health indicators output by the model to constrain the training results, thereby obtaining the matching model of the target work order type, providing a convenient and fast training method for the matching model.

[0099] After training and obtaining the matching model, the environmental data from the target task sample data can be input into the matching model for processing to obtain the target weight matching.

[0100] In this embodiment, the environmental data in the current target operation sample data is processed by a pre-trained matching model to obtain the target weight matching. During the application of the model, the target weight matching can change with the changes in the environmental data of each environmental type in the given target operation sample data, thereby achieving the effect of accurately and flexibly setting the target weight matching.

[0101] Example 3, referring to Figure 6 This embodiment also provides a personnel safety risk assessment system for power operations, including: The data acquisition module is used to acquire multiple work sample data of the target work order type that have been completed. Each work sample data includes work progress data, environmental data and health indicator values ​​of the workers. The preprocessing module is used to remove job sample data that contains both abnormal environmental data and abnormal health indicator values ​​from multiple job sample data for each environment type, and to use each retained job sample data as the environment sample corresponding to the current environment type. The model building module is used to build and train an environment-enhancing model for each environment type using corresponding environment samples. The input to the environmental support model is the work progress data and environmental data of the current environmental type in the environmental sample, and the output is the predicted value of the health index of the sample. The prediction module is used to process the target operation sample data of the target work order type using the environmental support models of the target work order type to obtain multiple target health indicator prediction values. Based on the obtained target weight ratio, the multiple target health indicator prediction values ​​are weighted and calculated to obtain the comprehensive prediction value of the target health indicator. The assessment module is used to assess the health risks of workers corresponding to the target work sample data based on the comprehensive predicted values ​​of the target health indicators and the health indicator values ​​in the target work sample data.

[0102] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0103] This embodiment also provides an electronic device, which can be a terminal, and its internal structure diagram can be as follows: Figure 6 As shown, the electronic device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for assessing personnel safety risks in power operations. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the device's casing, or an external keyboard, touchpad, or mouse.

[0104] This embodiment also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps: Acquire multiple work sample data of the target work order type that have been completed. Each work sample data includes work progress data, environmental data, and health indicator values ​​of the workers. For each environment type, remove job sample data that simultaneously contains abnormal environmental data and abnormal health indicator values ​​from multiple job sample data, and use each retained job sample data as the environment sample corresponding to the current environment type; For each environment type, corresponding environment samples are used to construct and train an environment enhancement model for the current environment type. The input to the environmental support model is the work progress data and environmental data of the current environmental type in the environmental sample, and the output is the predicted value of the health index of the sample. For the target operation sample data of the target work order type, the environmental support models of the target work order type are used for processing to obtain multiple target health indicator prediction values. Based on the obtained target weight ratio, the multiple target health indicator prediction values ​​are weighted and calculated to obtain the comprehensive target health indicator prediction value. Based on the comprehensive predicted values ​​of target health indicators and the health indicator values ​​in the target work sample data, the health risks of the workers corresponding to the target work sample data are assessed.

[0105] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0106] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0107] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for assessing personnel safety risks in electrical operations, characterized in that, include: Acquire multiple work sample data of the target work order type that have been completed. Each work sample data includes work progress data, environmental data, and health indicator values ​​of the workers. For each environment type, remove job sample data that simultaneously contains abnormal environmental data and abnormal health indicator values ​​from the multiple job sample data, and use each retained job sample data as the environment sample corresponding to the current environment type; For each environment type, corresponding environment samples are used to construct and train an environment enhancement model for the current environment type. The input to the environmental support model is the work progress data and environmental data of the current environmental type in the environmental sample, and the output is the predicted value of the sample health index. For the target operation sample data of the target work order type, the environmental support model of each target work order type is used for processing to obtain multiple target health indicator prediction values. Based on the obtained target weight ratio, the multiple target health indicator prediction values ​​are weighted and calculated to obtain the comprehensive target health indicator prediction value. Based on the comprehensive predicted value of the target health indicators and the health indicator values ​​in the target work sample data, the health risk of the workers corresponding to the target work sample data is assessed.

2. The method for assessing personnel safety risks in power operations as described in claim 1, characterized in that, For each environment type, the process involves removing job sample data that simultaneously exhibits abnormal environmental data and abnormal health indicator values ​​in other non-current environment types from the multiple job sample data sets, and using each retained job sample data set as the environment sample corresponding to the current environment type, including: For each environment type, determine whether there are other environmental data that are not in the current environment type and are in a preset first abnormal range among the multiple job sample data; For the multiple job sample data, the job sample data that simultaneously contains environmental data of other non-current environment types in the first abnormal interval and the health indicator value in the preset second abnormal interval is removed, and each retained job sample data is used as the environmental sample corresponding to the current environment type.

3. The method for assessing personnel safety risks in power operations as described in claim 2, characterized in that, For each environment type, the process involves removing job sample data that simultaneously exhibits abnormal environmental data and abnormal health indicator values ​​in other non-current environment types from the multiple job sample data sets, and using each retained job sample data set as the environment sample corresponding to the current environment type, including: For each environment type, the multiple job sample data are divided into multiple data groups according to the multiple environmental value ranges corresponding to the current environment type. Based on the job sample data containing abnormal health indicator values ​​in each data group, the data space range determined by other environmental data of non-current environment type in the job sample data is delineated. For the multiple data space ranges of the current environment type, the one with the smallest space range is selected as the first abnormal space range; For the multiple job sample data, job sample data that simultaneously contains environmental data of other non-current environment types within the first abnormal space range and the health indicator value within the preset second abnormal interval are removed, and each retained job sample data is used as the environmental sample corresponding to the current environment type.

4. The method for assessing personnel safety risks in power operations as described in claim 3, characterized in that, For each environment type, using corresponding environment samples, an environment enhancement model for the current environment type is constructed and trained, including: For each type of environment sample, the operation progress data and the environmental data of the current environment type in the environmental sample are input into the pre-built first neural network model for processing, and the predicted value of the health index of the first sample is output. The health index values ​​in the environmental samples are used as the measured values ​​of the first sample health index. A first loss function is constructed based on the measured values ​​and predicted values ​​of the first sample health index. The first neural network model is trained based on the first loss function to obtain the environmental support model of the current environmental type. The environmental type includes any one of the following types, either as an independent type or as a combination of at least two: time period, geographical conditions, weather conditions, noise, and light.

5. The method for assessing personnel safety risks in power operations as described in claim 4, characterized in that, The method further includes: Based on the work sample data and the predicted values ​​of sample health indicators output by each of the environmental support models, a matching model corresponding to the target work order type is constructed and trained. The input of the matching model is the environmental data in the work sample data, and the output is the sample weight ratio used to perform weighted calculation on the predicted values ​​of each sample health indicator output by each of the environmental support models. The environmental data from the target operation sample data is input into the matching model for processing to obtain the target weighting ratio.

6. The method for assessing personnel safety risks in power operations as described in claim 5, characterized in that, The process of constructing and training a matching model corresponding to the target work order type based on multiple work sample data and the predicted values ​​of sample health indicators output by each of the environmental support models includes: The environmental data in the operation sample data is input into the pre-built second neural network model for processing, and the output is a sample weight ratio used to calculate the weighted health index prediction values ​​of each sample output by each of the environmental support models. The operation progress data and environmental data in the operation sample data are input into the corresponding environmental support models for processing, and multiple second sample health indicator prediction values ​​are output. The predicted values ​​of the multiple second sample health indicators are weighted using the sample weight ratio to obtain the comprehensive predicted value of the sample health indicators. The health indicator values ​​in the work sample data are used as the comprehensive measured values ​​of the sample health indicators. A second loss function is constructed based on the comprehensive measured values ​​and the comprehensive predicted values ​​of the sample health indicators. The second neural network model is then trained based on the second loss function to obtain the matching model corresponding to the target work order type.

7. The method for assessing personnel safety risks in power operations as described in claim 6, characterized in that, Based on the comprehensive predicted value of the target health indicators and the health indicator values ​​in the target work sample data, an assessment of the health risk of the workers corresponding to the target work sample data is performed, including: When both the comprehensive predicted value of the health indicator and the comprehensive measured value of the health indicator are abnormal, the worker is determined to be at a high health risk. When only one of the comprehensive predicted value of the health indicator and the comprehensive measured value of the health indicator is abnormal, the worker is determined to be at a medium health risk. When both the comprehensive predicted value and the comprehensive measured value of the health indicators are not abnormal, the worker is determined to be at a low level of health risk or have no health risk.

8. A personnel safety risk assessment system for power operations, using the method described in any one of claims 1 to 7, characterized in that, include: The data acquisition module is used to acquire multiple work sample data of the target work order type that have been completed. Each work sample data includes work progress data, environmental data and health indicator values ​​of the workers. The preprocessing module is used to remove, for each environment type, job sample data that simultaneously contains abnormal environmental data and abnormal health indicator values ​​of other non-current environment types from the multiple job sample data, and to use each retained job sample data as the environment sample corresponding to the current environment type; The model building module is used to build and train an environment-enhancing model for each environment type using corresponding environment samples. The input to the environmental support model is the work progress data and environmental data of the current environmental type in the environmental sample, and the output is the predicted value of the sample health index. The prediction module is used to process the target operation sample data of the target work order type using the environmental support model of the target work order type to obtain multiple target health indicator prediction values, and to perform weighted calculation on the multiple target health indicator prediction values ​​based on the obtained target weight ratio to obtain the comprehensive prediction value of the target health indicator. The assessment module is used to assess the health risks of the workers corresponding to the target work sample data based on the comprehensive predicted value of the target health indicator and the health indicator values ​​in the target work sample data.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the personnel safety risk assessment method in power operations as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the personnel safety risk assessment method in power operations according to any one of claims 1 to 7.

Images