Dangerous operation detection system and method based on big data

Abstract

The invention belongs to the technical field of dangerous work detection, and particularly relates to a dangerous work detection system and method based on big data. According to the invention, by combining multi-dimensional data such as operation behaviors, operation environments and operation scenes and integrating behavior characteristics of operators, potential dangerous behaviors can be identified more accurately, false alarms and missing alarms can be reduced, operation risks can be judged quickly through real-time data acquisition, alarms can be given out before the dangerous behaviors occur, response time can be won for the operators, and the operation safety can be improved. The method can reduce the accident rate, introduces the operation environment compensation information and the operation scene compensation information, can dynamically adjust the parameters of the danger detection model according to different operation environments and scenes, is suitable for various complex working conditions, can predict the behavior trend of an operator, can evaluate the potential risk of the operator, and provides a scientific basis for operation safety management.

Description

Technical Field

[0001] The present invention belongs to the technical field of hazardous operation detection, and particularly relates to a big data-based hazardous operation detection system and method. Background Art

[0002] With the continuous development of industrialization and automation, the working environment has become increasingly complex, and the working conditions and behavior patterns of workers have also become diversified. Especially in the field of hazardous operations, such as construction, mining, petrochemical and other industries, workers face relatively high safety risks. Traditional hazardous operation monitoring methods mostly rely on manual inspections or regular checks, which have deficiencies such as slow response speed, narrow coverage, and inability to monitor in real time, resulting in many potential hazards not being detected in time and increasing the risk of accidents.

[0003] In recent years, with the development of technologies such as big data, Internet of Things, and artificial intelligence, data such as the behavior of workers, working environment, and working scenarios can be collected in real time through sensors, monitoring devices, or wearable devices, and analyzed and processed through a big data platform to provide more accurate and real-time hazardous operation monitoring and warning functions. The big data-based hazardous operation detection method can collect multi-dimensional information such as the behavior data, environment data, and scenario data of workers in real time, and combine preset safety standards and hazardous operation characteristics to perform behavior prediction, risk assessment, and hazard warning, thereby effectively reducing the incidence of operation accidents.

[0004] However, existing hazardous operation detection methods only rely on the behavior data of workers, ignoring the impact of the working environment and working scenarios on operation safety, and unable to comprehensively evaluate the risks of workers. Most are based on fixed safety standards and simple algorithm models, and cannot dynamically adapt to changes in different working environments and working scenarios, resulting in unsatisfactory detection effects under complex working conditions. Summary of the Invention

[0005] The purpose of the present invention is to provide a big data-based hazardous operation detection method, which can dynamically adjust the parameters of the hazard detection model, adapt to changes in different working environments and scenarios, improve the accuracy and reliability of hazardous operation detection, and has high practical value.

[0006] The technical solution adopted by the present invention is specifically as follows: A big data-based hazardous operation detection method, comprising: Obtaining the operation behavior data of a worker, determining whether the operation behavior data meets a first preset condition, and if not, determining that the operation behavior of the worker is abnormal and marking it as an abnormal operation behavior; Obtaining the historical operation preference data of the worker, and obtaining the operation behavior information of the worker according to the abnormal operation behavior and the historical operation preference data; Obtain the operation environment data of the operator, and obtain the corresponding operation environment compensation information according to the operation environment data; Obtain the operation scenario data of the operator, and obtain the corresponding operation scenario compensation information according to the operation scenario data; Obtain the predicted behavior feature information according to the operation behavior information, the operation environment compensation information and the operation scenario compensation information; Obtain multi-source dangerous operation data, and obtain corresponding multiple dangerous behavior feature information according to the multi-source dangerous operation data; Obtain the behavior feature similarity information between the predicted behavior feature information and each dangerous behavior feature information, determine whether the behavior feature similarity information meets the second condition, if not, determine that the operator is in the dangerous operation range, and send an alarm message.

[0007] In a preferred solution, the steps of obtaining the operation behavior data of the operator, judging whether the operation behavior data meets the first preset condition, and if not, determining that the operation behavior of the operator is abnormal and marking it as an abnormal operation behavior include: Obtain the operation behavior data of the operator; Obtain multiple operation behavior feature vectors according to the operation behavior data; Obtain the standard operation behavior data, and obtain the corresponding standard operation behavior feature vectors corresponding to each operation behavior feature vector according to the standard operation behavior data; Obtain the abnormal operation value according to the multiple operation behavior feature vectors and the multiple standard operation behavior feature vectors; Obtain the standard operation behavior feature threshold Judge whether the abnormal operation value exceeds the standard operation behavior feature threshold; If the abnormal operation value exceeds the standard operation behavior feature threshold, determine that the operation behavior of the operator is abnormal and mark it as an abnormal operation behavior.

[0008] In a preferred solution, the steps of obtaining the historical operation preference data of the operator and obtaining the operation behavior information of the operator according to the abnormal operation behavior and the historical operation preference data include: Construct a historical period; Obtain the historical operation preference data of the operator within the historical period; Obtain corresponding multiple historical operation preference behavior vectors according to the historical operation preference data; Obtain the abnormal operation behavior vector corresponding to the abnormal operation behavior; Obtain the operation behavior value according to the abnormal operation behavior vector and the multiple historical operation preference behavior vectors, and mark it as the operation behavior information.

[0009] In a preferred solution, the steps of constructing a historical period include: Obtain the time nodes marked as abnormal operation behaviors and mark them as the end time of the historical period; Obtain the corresponding operation behavior characteristics according to the operation behavior data; Obtain a behavior duration table, where the behavior duration table includes multiple operation behavior characteristics and the historical durations corresponding to each operation behavior; Obtain the corresponding historical duration from the duration table according to the operation behavior characteristics; Obtain the start time of the historical period according to the end time of the historical period and the historical duration; Obtain the historical period according to the start time and end time of the historical period.

[0010] In a preferred solution, the steps of obtaining the operation environment data of an operator and obtaining the corresponding operation environment compensation information according to the operation environment data include: Obtain the operation environment data of the operator; Obtain the corresponding multiple operation environment vectors according to the operation environment data; Obtain the corresponding operation environment value according to the multiple operation environment vectors and mark it as the operation environment compensation information.

[0011] In a preferred solution, the steps of obtaining the operation scenario data of an operator and obtaining the corresponding operation scenario compensation information according to the operation scenario data include: Obtain the operation scenario data of the operator; Obtain the corresponding operation scenario characteristics according to the operation scenario data; Obtain a scenario table, where the scenario table includes multiple operation scenario characteristics and the operation scenario values corresponding to each operation scenario characteristic; Obtain the corresponding operation scenario value from the scenario table according to the operation scenario characteristics and mark it as the operation scenario compensation information.

[0012] In a preferred solution, the steps of obtaining the predicted behavior characteristic information according to the operation behavior information, operation environment compensation information, and operation scenario compensation information include: Obtain the corresponding operation behavior value according to the operation behavior information; Obtain the corresponding operation environment value according to the operation environment compensation information; Obtain the corresponding operation scenario value according to the operation scenario compensation information; Obtain the corresponding predicted behavior characteristic value according to the operation behavior value, operation environment compensation value, and operation scenario compensation value; Obtain a characteristic table, where the characteristic table includes multiple predicted behavior characteristic interval values and the predicted behavior characteristic information corresponding to each predicted behavior characteristic interval value.

[0013] In a preferred embodiment, the steps of obtaining the behavioral feature similarity information between the predicted behavioral feature information and each hazardous behavioral feature information, determining whether the behavioral feature similarity information meets the second condition, and if not, determining that the operator is within the hazardous operation range and sending an alarm message include: Obtaining the corresponding predicted behavioral feature vector according to the predicted behavioral feature information; Obtaining the hazardous behavioral feature vector corresponding to each hazardous behavioral feature information according to multiple hazardous behavioral feature information; Obtaining the behavioral feature similarity value between the predicted behavioral feature vector and each hazardous behavioral feature vector, and marking it as the behavioral feature similarity information; Obtaining the behavioral feature similarity threshold; Determining whether each behavioral feature similarity value exceeds the behavioral feature similarity threshold; If there is a behavioral feature similarity value that exceeds the behavioral feature similarity threshold among multiple behavioral feature similarity values, it is determined that the operator is within the hazardous operation range and an alarm message is sent.

[0014] The present invention also provides a big data-based hazardous operation detection system for the above-mentioned big data-based hazardous operation detection method, including: An abnormality judgment module, configured to obtain the operation behavior data of the operator, determine whether the operation behavior data meets the first preset condition, and if not, determine that the operation behavior of the operator is abnormal and mark it as abnormal operation behavior; An operation preference module, configured to obtain the historical operation preference data of the operator, and obtain the operation behavior information of the operator according to the abnormal operation behavior and the historical operation preference data; An operation environment module, configured to obtain the operation environment data of the operator, and obtain the corresponding operation environment compensation information according to the operation environment data; An operation scenario module, configured to obtain the operation scenario data of the operator, and obtain the corresponding operation scenario compensation information according to the operation scenario data; A predicted behavior module, configured to obtain the predicted behavioral feature information according to the operation behavior information, the operation environment compensation information, and the operation scenario compensation information; A hazardous behavior module, configured to obtain multi-source hazardous operation data, and obtain corresponding multiple hazardous behavioral feature information according to the multi-source hazardous operation data; A hazardous judgment module, configured to obtain the behavioral feature similarity information between the predicted behavioral feature information and each hazardous behavioral feature information, determine whether the behavioral feature similarity information meets the second condition, and if not, determine that the operator is within the hazardous operation range and send an alarm message.

[0015] And, a big data-based hazardous operation detection terminal, including: one or more processors; a storage device having one or more programs stored thereon; When one or more programs are executed by one or more processors, the one or more processors implement a dangerous operation detection method based on big data.

[0016] The technical effects achieved by the present invention are: The present invention combines multi-dimensional data such as operating behavior, operating environment and operating scene, and integrates the behavioral characteristics of operators to more accurately identify potential dangerous behaviors and reduce false alarms and missed alarms. Through real-time data collection, it can quickly determine operating risks and issue alarms before dangerous behaviors occur, thereby buying reaction time for operators and reducing the accident rate. The present invention introduces operating environment compensation information and operating scene compensation information, and can dynamically adjust the parameters of the hazard detection model according to different operating environments and scenes. It is applicable to a variety of complex working conditions, can predict the behavioral trends of operators and evaluate their potential risks, and provide a scientific basis for operational safety management. Multi-source dangerous operation data is integrated to build a comprehensive dangerous behavior feature library, which helps to improve the accuracy and reliability of hazard detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of the method provided by the present invention; Figure 2 It is a system module diagram provided by the present invention. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0019] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0020] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure or characteristic that may be included in at least one implementation of the present invention. The phrase "in a preferred embodiment" that appears in different places in this specification does not refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0021] Secondly, the present invention is described in detail in conjunction with schematic diagrams. When describing the embodiments of the present invention in detail, for the convenience of explanation, the schematic diagrams are only examples and should not limit the scope of protection of the present invention.

[0022] Please refer to the attached Figure 1 As shown, a big data-based dangerous operation detection method is provided, including: S1. Obtain the operation behavior data of the operator, and determine whether the operation behavior data meets the first preset condition. If not, determine that the operation behavior of the operator is abnormal and mark it as an abnormal operation behavior; S2. Obtain the historical operation preference data of the operator, and obtain the operation behavior information of the operator according to the abnormal operation behavior and the historical operation preference data; S3. Obtain the operation environment data of the operator, and obtain the corresponding operation environment compensation information according to the operation environment data; S4. Obtain the operation scenario data of the operator, and obtain the corresponding operation scenario compensation information according to the operation scenario data; S5. Obtain the predicted behavior feature information according to the operation behavior information, the operation environment compensation information and the operation scenario compensation information; S6. Obtain multi-source dangerous operation data, and obtain corresponding multiple dangerous behavior feature information according to the multi-source dangerous operation data; S7. Obtain the behavior feature similarity information between the predicted behavior feature information and each dangerous behavior feature information, and determine whether the behavior feature similarity information meets the second condition. If not, determine that the operator is in the dangerous operation range and send an alarm message.

[0023] As in the above steps S1 to S7, the operating behavior data of the operators is collected in real time through sensors, monitoring equipment or wearable devices, and compared with the preset safety standards (the first preset condition). If the operating behavior does not meet the standards, it will be marked as abnormal behavior. By obtaining the historical operating preference data of the operators and combining it with the current abnormal operating behavior, the operating personnel's behavior information can be obtained more accurately. The operating environment data (such as environmental gas composition, temperature, humidity, noise, vibration, etc.) is collected, and the operating environment compensation information is generated based on these data. The operating scene data (such as the current operating type, the use of operating tools, etc.) is further collected to generate the operating scene compensation information to ensure that the hazard detection algorithm can adapt to the characteristics of different operating scenes. The operating behavior information, the operating environment compensation information and the operating scene compensation information are integrated to generate the predicted behavior feature information of the operators, which is used to simulate their possible behavior trajectories and potential risks. Multi-source dangerous operation data is obtained from the big data platform, and the dangerous behavior feature information therein is extracted to construct a dangerous The behavior feature library compares the predicted behavior feature information with each dangerous behavior feature in the dangerous behavior feature library. If the behavior feature similarity exceeds the second preset condition, it is determined that the operator is in a dangerous operation range, and an alarm message is immediately issued to prompt relevant personnel to take protective measures. Combining multi-dimensional data such as operating behavior, operating environment and operating scene, and integrating the behavioral characteristics of the operators, it can more accurately identify potential dangerous behaviors and reduce false alarms and missed alarms. Through real-time data collection, it can quickly determine the operating risks and issue an alarm before the dangerous behavior occurs, so as to gain reaction time for the operators and reduce the accident rate. The introduction of operating environment compensation information and operating scene compensation information can dynamically adjust the parameters of the danger detection model according to different operating environments and scenes. It is suitable for a variety of complex working conditions, can predict the behavioral trends of operators and evaluate their potential risks, and provide a scientific basis for operation safety management. Multi-source dangerous operation data is integrated to build a comprehensive dangerous behavior feature library, which helps to improve the accuracy and reliability of danger detection.

[0024] In a preferred embodiment, the steps of obtaining the operation behavior data of the operator, determining whether the operation behavior data meets the first preset condition, and if not, determining that the operation behavior of the operator is abnormal and marking it as abnormal operation behavior include: S101, obtaining the operation behavior data of the operator; S102, obtaining a plurality of operation behavior feature vectors according to the operation behavior data; S103, acquiring standard operation behavior data, and acquiring a standard operation behavior feature vector corresponding to each operation behavior feature vector according to the standard operation behavior data; S104, obtaining an abnormal operation value according to a plurality of operation behavior feature vectors and a plurality of standard operation behavior feature vectors; S105. Obtain the threshold of standard operation behavior characteristics S106. Determine whether the abnormal operation value exceeds the threshold of standard operation behavior characteristics; If the abnormal operation value exceeds the threshold of standard operation behavior characteristics, it is determined that the operation behavior of the operator is abnormal and marked as an abnormal operation behavior.

[0025] In the above steps S101 to S106, the real-time operation behavior data of the operator is obtained through sensors, monitoring devices or other acquisition tools. These data may include information such as action trajectories, postures, speeds, etc. The obtained operation behavior data is subjected to feature extraction to generate multiple operation behavior feature vectors. These feature vectors can describe the key attributes of the operation behavior, such as displacement, angle change, action rhythm, etc. The corresponding standard operation behavior feature vectors are extracted from the preset standard operation behavior data. These standard feature vectors define the operation behavior patterns that meet the safety requirements and are used for comparison with the actual behavior. The abnormal operation value is calculated based on multiple operation behavior feature vectors and multiple standard operation behavior feature vectors. The calculation formula of the abnormal operation value is , where X represents the abnormal operation value, q represents the numbers of the operation behavior feature vectors and multiple standard operation behavior feature vectors, q = 1, 2, 3... h, represents the qth operation behavior feature vector, represents the ith standard operation behavior feature vector. According to the operation type, environmental conditions and safety requirements, the threshold of standard operation behavior characteristics is set. This threshold is used to distinguish normal behavior from abnormal behavior. Determine whether the abnormal operation value exceeds the threshold of standard operation behavior characteristics. If it exceeds the threshold, it indicates that there is a significant deviation between the current operation behavior and the standard behavior. Determine that this behavior is an abnormal operation behavior and mark it. It can accurately capture the deviation of the operation behavior, improve the accuracy of abnormal behavior recognition, ensure the real-time discovery of abnormal behavior during the operation process, and take corresponding measures in time. The threshold of standard operation behavior characteristics can be dynamically adjusted according to different operation types, environmental conditions and personnel characteristics to adapt to various complex working conditions and expand the application scope.

[0026] In a preferred embodiment, the steps of obtaining the operation behavior information of the operator according to the abnormal operation behavior and the historical operation preference data include: S201. Construct a historical time period; S202. Obtain the historical operation preference data of the operator within the historical time period; S203. Obtain multiple corresponding historical operation preference behavior vectors according to the historical operation preference data; S204. Obtain the abnormal operation behavior vector corresponding to the abnormal operation behavior; S205. Obtain the operation behavior value based on the abnormal operation behavior vector and multiple historical operation preference behavior vectors, and mark it as operation behavior information.

[0027] In the above steps S201 to S205, according to the operation type, time range, and behavior characteristics, construct a historical period as the time window for obtaining historical operation preference data. The historical period can be a continuous time period (such as the past week) or scattered key time points (such as high-intensity operation time periods). Within the constructed historical period, obtain the historical operation preference data of the operation personnel through a database or real-time collection tool. These data reflect the behavior characteristics and habitual patterns of the operation personnel in different situations. Extract the features of the historical operation preference data to generate multiple historical operation preference behavior vectors. Each vector corresponds to a specific behavior characteristic, such as action frequency, operation duration, path selection, etc. According to the currently detected abnormal operation behavior, extract its features and generate an abnormal operation behavior vector. Calculate the operation behavior value through the abnormal operation behavior vector and the historical operation preference behavior vector. The calculation formula for the operation behavior value is , where Z represents the operation behavior value, i represents the number of the historical operation preference behavior vector, i = 1, 2, 3…n, represents the i-th historical operation preference behavior vector, Y represents the abnormal operation behavior vector. According to the calculation result, mark it as operation behavior information, which can comprehensively understand its behavior habits, provide a more reliable reference basis for the determination of abnormal behaviors, can identify the personalized behavior characteristics of the operation personnel, and avoid false alarms caused by behavior differences.

[0028] In a preferred implementation manner, the steps of constructing the historical period include: S2011. Obtain the time node marked as an abnormal operation behavior and mark it as the end time of the historical period; S2012. Obtain the corresponding operation behavior characteristics according to the operation behavior data; S2013. Obtain the behavior duration table, where the behavior duration table includes multiple operation behavior characteristics and the historical duration corresponding to each operation behavior; S2014. Obtain the corresponding historical duration from the duration table according to the operation behavior characteristics; S2015. Obtain the start time of the historical period according to the end time of the historical period and the historical duration; S2016. Obtain the historical period according to the start time and end time of the historical period.

[0029] In the above steps S2011 to S2016, identify the time nodes marked as abnormal operation behaviors, and use this time node as the end time of the historical period. The end time is a key point of the historical period. According to the operation behavior data, extract the operation behavior characteristics related to the abnormal behavior. These characteristics are used to associate the data in the historical duration table. Load the preset behavior duration table. The behavior duration table contains multiple operation behavior characteristics and their corresponding historical durations. The historical duration defines the reference time range of each operation behavior under normal conditions. According to the extracted operation behavior characteristics, find and obtain the corresponding historical duration from the behavior duration table. The historical duration reflects the normal execution time of this operation behavior, providing a basis for determining the historical period. According to the end time of the historical period and the corresponding historical duration, calculate the start time of the historical period forward. The start time defines the behavior reference range before the abnormal behavior occurs. Combine the start time and the end time of the historical period to generate a complete historical period. The historical period covers a period of time before the abnormal behavior occurs, which is used to analyze the change trend of the operation behavior. Dynamically obtaining the historical duration according to different operation behavior characteristics can adapt to the time characteristics of different types of operations and improve the flexibility of historical period construction.

[0030] In a preferred embodiment, the steps of obtaining the operation environment data of the operator and obtaining the corresponding operation environment compensation information according to the operation environment data include: S301. Obtain the operation environment data of the operator; S302. Obtain a corresponding plurality of operation environment vectors according to the operation environment data; S303. Obtain the corresponding operation environment value according to the plurality of operation environment vectors and mark it as the operation environment compensation information.

[0031] In the above steps S301 to S303, through sensors, monitoring devices or other acquisition tools, the data of the operation environment where the operator is located is obtained in real time. The operation environment data includes but is not limited to environmental gas composition, temperature, humidity, noise, light intensity, air quality, etc. Feature extraction is performed on the obtained operation environment data to generate a plurality of operation environment vectors. Each vector corresponds to an environmental characteristic, such as a temperature vector, a humidity vector, a noise vector, etc. According to the plurality of operation environment vectors, calculate the operation environment value. The calculation formula of the operation environment value is , where H represents the operation environment value, j represents the number of the operation environment vector, j = 1, 2, 3... t, represents the jth operation environment vector, which can dynamically reflect the environmental changes on the operation site, ensure the timeliness and accuracy of the compensation information, can adjust the judgment standard of abnormal behavior according to the environmental impact, reduce false alarms and missed alarms caused by environmental changes, and improve the accuracy of dangerous operation detection.

[0032] In a preferred embodiment, the steps of obtaining the operation scenario data of the operator and obtaining the corresponding operation scenario compensation information according to the operation scenario data include: S401. Obtain the operation scenario data of the operator; S402. Obtain the corresponding operation scenario characteristics according to the operation scenario data; S403. Obtain a scenario table, where the scenario table includes multiple operation scenario characteristics and the operation scenario values corresponding to each operation scenario characteristic; S404. Obtain the corresponding operation scenario value from the scenario table according to the operation scenario characteristics, and mark it as the operation scenario compensation information.

[0033] In the above steps S401 to S404, through means such as on-site monitoring, image recognition, and sensing devices, the operation scenario data is collected in real time. The scenario data includes information such as the layout of the operation area, equipment distribution, personnel location, and operation process. According to the collected operation scenario data, the key characteristics of the scenario are identified through a feature extraction algorithm. For example, the equipment type, spatial structure, and operation process complexity in the scenario are extracted. The preset scenario table is loaded. The scenario table records a variety of operation scenario characteristics and their corresponding operation scenario values. According to the extracted operation scenario characteristics, the corresponding operation scenario values are matched in the scenario table, and the operation scenario compensation information is generated. The operation scenario compensation information reflects the impact of the scenario on the operation behavior, can adjust the dangerous operation detection standard for different operation scenarios, effectively reduce misjudgment and missed judgment caused by scenario differences, and can comprehensively consider multi-dimensional scenario characteristics to provide comprehensive compensation support.

[0034] In a preferred embodiment, the steps of obtaining the predicted behavior feature information according to the operation behavior information, operation environment compensation information, and operation scenario compensation information include: S501. Obtain the corresponding operation behavior value according to the operation behavior information; S502. Obtain the corresponding operation environment value according to the operation environment compensation information; S503. Obtain the corresponding operation scenario value according to the operation scenario compensation information; S504. Obtain the corresponding predicted behavior feature value according to the operation behavior value, operation environment compensation value, and operation scenario compensation value; S505. Obtain a feature table, where the feature table includes multiple predicted behavior feature interval values and the predicted behavior feature information corresponding to each predicted behavior feature interval value.

[0035] In the above steps S501 to S505, the operation behavior value is extracted according to the operation behavior information, the operation environment value is extracted according to the operation environment compensation information, the operation scenario value is extracted according to the operation scenario compensation information, and the predicted behavior feature value is calculated by integrating the operation behavior value, the operation environment value and the operation scenario value. The calculation formula of the predicted behavior feature value is W = Z * H * C, where W represents the predicted behavior feature value, Z represents the operation behavior value, H represents the operation environment value, and C represents the operation scenario compensation value. The feature table is loaded. The feature table contains multiple predicted behavior feature interval values and their corresponding predicted behavior feature information. According to the calculated predicted behavior feature value, the corresponding feature interval value in the feature table is matched, and the corresponding predicted behavior feature information is obtained. Analyzing the behavior characteristics of the operator from multiple dimensions improves the accuracy of the predicted behavior features, can dynamically adapt to complex and changeable operation conditions, and ensures the timeliness and reliability of the prediction results.

[0036] In a preferred embodiment, the steps of obtaining the behavior feature similarity information between the predicted behavior feature information and each dangerous behavior feature information, determining whether the behavior feature similarity information meets the second condition, and if not, determining that the operator is in the dangerous operation range and sending an alarm message include: S701. Obtain the corresponding predicted behavior feature vector according to the predicted behavior feature information; S702. Obtain the dangerous behavior feature vector corresponding to each dangerous behavior feature information according to multiple dangerous behavior feature information; S703. Obtain the behavior feature similarity value between the predicted behavior feature vector and each dangerous behavior feature vector, and mark it as the behavior feature similarity information; S704. Obtain the behavior feature similarity threshold; S705. Determine whether each behavior feature similarity value exceeds the feature similarity threshold; If there is a behavior feature similarity value that exceeds the feature similarity threshold among multiple behavior feature similarity values, it is determined that the operator is in the dangerous operation range and an alarm message is sent.

[0037] In the above steps S701 to S705, according to the generated predicted behavior feature information, it is transformed into a standardized predicted behavior feature vector, and each dangerous behavior feature vector is extracted from the dangerous behavior feature information to establish a dangerous behavior feature library for matching. Each vector represents a known dangerous behavior pattern, and the behavior feature similarity value between the predicted behavior feature vector and each dangerous behavior feature vector is calculated. The calculation formula of the feature similarity value is , where Denoted as the feature similarity value, A is denoted as the predicted behavior feature vector, B is denoted as the dangerous behavior feature vector. A behavior feature similarity threshold is set to distinguish safe behaviors from dangerous behaviors. This threshold can be determined through historical data analysis or expert experience. Compare each behavior feature similarity value with the behavior feature similarity threshold. If there is a situation where any of the multiple behavior feature similarity values exceeds the threshold, it is determined that the current operation behavior matches the dangerous behavior feature, indicating that the operator is in the dangerous operation range, and an alarm message is immediately sent to prompt relevant personnel to take necessary safety measures. It can accurately identify whether the operator's behavior is close to known dangerous behavior patterns, avoiding misjudgment or missed judgment, can quickly respond to changes in the operator's behavior to achieve dynamic monitoring, and can cover diverse dangerous behavior patterns to adapt to different operation scenarios.

[0038] Please refer to the appendix Figure 2 As shown, the present invention also provides a big data-based dangerous operation detection system for the above-mentioned big data-based dangerous operation detection method, including: Anomaly judgment module, used to obtain the operation behavior data of the operator, and judge whether the operation behavior data meets the first preset condition. If it does not meet, it is determined that the operation behavior of the operator is abnormal and marked as an abnormal operation behavior; Operation preference module, used to obtain the historical operation preference data of the operator, and obtain the operation behavior information of the operator according to the abnormal operation behavior and the historical operation preference data; Operation environment module, used to obtain the operation environment data of the operator, and obtain the corresponding operation environment compensation information according to the operation environment data; Operation scenario module, used to obtain the operation scenario data of the operator, and obtain the corresponding operation scenario compensation information according to the operation scenario data; Predicted behavior module, used to obtain predicted behavior feature information according to the operation behavior information, operation environment compensation information and operation scenario compensation information; Dangerous behavior module, used to obtain multi-source dangerous operation data, and obtain corresponding multiple dangerous behavior feature information according to the multi-source dangerous operation data; Dangerous judgment module, used to obtain the behavior feature similarity information between the predicted behavior feature information and each dangerous behavior feature information, and judge whether the behavior feature similarity information meets the second condition. If it does not meet, it is determined that the operator is in the dangerous operation range and an alarm message is sent.

[0039] As described above, the anomaly judgment module obtains the operation behavior data of the operator, extracts multiple operation behavior feature vectors, compares them with the standard operation behavior data, calculates the anomaly operation value, and determines whether the anomaly operation value exceeds the standard operation behavior feature threshold. If it exceeds, it is marked as an abnormal operation behavior. The operation preference module extracts the historical operation preference data of the operator, constructs historical time periods, generates multiple historical operation preference behavior vectors based on the historical operation preference data and abnormal operation behavior information, combines them with the abnormal operation behavior vectors, obtains the operation behavior value, and generates operation behavior information. The operation environment module obtains the operation environment data in real time, extracts multiple operation environment vectors, calculates the operation environment compensation information based on the characteristic values of the environment data, and is used to correct the environmental impact of the operation behavior. The operation scenario module collects the operation scenario data, extracts relevant features according to the scenario characteristics, and generates the operation scenario compensation information through scenario table matching. The prediction behavior module integrates the operation behavior information, operation environment compensation information, and operation scenario compensation information to generate the predicted behavior characteristic value, and outputs the predicted behavior characteristic information through characteristic table matching. The dangerous behavior module extracts the dangerous behavior characteristic information from the multi-source dangerous operation data, constructs a dangerous behavior characteristic vector library, which covers a variety of known dangerous behavior patterns and serves as a reference standard for similarity matching. The danger judgment module calculates the behavior characteristic similarity value between the predicted behavior characteristic information and the dangerous behavior characteristic information, compares the similarity value with the behavior characteristic similarity threshold. If it exceeds the threshold, it determines that the operator is in the dangerous operation range and immediately issues an alarm message to remind to take safety measures. By comprehensively combining the behavior characteristics of the operator and multi-source dangerous operation data, accurate dangerous behavior recognition is achieved. It can collect operation behavior, environment, and scenario data in real time, dynamically update the predicted behavior characteristic information, achieve rapid response to changes in the operation environment, and can more accurately evaluate the behavior characteristics of the operator, reducing misjudgment.

[0040] And, a big data-based dangerous operation detection terminal, comprising: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the big data-based dangerous operation detection method.

[0041] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art in this technical field, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention. The structures, devices, and operation methods not specifically described and explained in the present invention, unless otherwise specified and limited, are implemented according to the conventional means in the art.

Claims

1. A dangerous operation detection method based on big data, characterized in that: include: Acquire the operation behavior data of the operator, determine whether the operation behavior data meets the first preset condition, and if not, determine that the operation behavior of the operator is abnormal and mark it as abnormal operation behavior; Obtain the historical operation preference data of the operator, and obtain the operation behavior information of the operator based on the abnormal operation behavior and the historical operation preference data; Obtain the working environment data of the operator, and obtain the corresponding working environment compensation information according to the working environment data; Obtain the operation scene data of the operator, and obtain the corresponding operation scene compensation information according to the operation scene data; Acquire predicted behavior feature information according to operation behavior information, operation environment compensation information and operation scene compensation information; Acquire multi-source dangerous operation data, and obtain corresponding multiple dangerous behavior feature information based on the multi-source dangerous operation data; Obtain the behavior characteristic similarity information between the predicted behavior characteristic information and each dangerous behavior characteristic information, and determine whether the behavior characteristic similarity information meets the second condition. If not, determine that the operator is located in a dangerous operation range and issue an alarm message.

2. The method for detecting dangerous operations based on big data according to claim 1 is characterized in that: The steps of obtaining the operation behavior data of the operator and determining whether the operation behavior data meets the first preset condition, and if not, determining that the operation behavior of the operator is abnormal and marking it as abnormal operation behavior, include: Obtaining the operating behavior data of operators; Acquire multiple operation behavior feature vectors according to the operation behavior data; Acquire standard operation behavior data, and acquire a standard operation behavior feature vector corresponding to each operation behavior feature vector according to the standard operation behavior data; Acquire abnormal operation values ​​according to a plurality of operation behavior feature vectors and a plurality of standard operation behavior feature vectors; Get the threshold value of standard operation behavior characteristics Determine whether the abnormal operation value exceeds the standard operation behavior characteristic threshold; If the abnormal operation value exceeds the standard operation behavior characteristic threshold, the operation behavior of the operator is judged to be abnormal and marked as abnormal operation behavior.

3. The method for detecting dangerous operations based on big data according to claim 1 is characterized in that: The steps of obtaining the historical operation preference data of the operator and obtaining the operation behavior information of the operator according to the abnormal operation behavior and the historical operation preference data include: Constructing historical periods; Obtain historical operation preference data of operators within a historical period; Acquire corresponding multiple historical job preference behavior vectors according to the historical job preference data; Obtain an abnormal operation behavior vector corresponding to the abnormal operation behavior; The operation behavior value is obtained according to the abnormal operation behavior vector and multiple historical operation preference behavior vectors, and marked as operation behavior information.

4. The method for detecting dangerous operations based on big data according to claim 3 is characterized in that: The steps to construct a historical period include: Get the time node marked as abnormal operation behavior and mark it as the end time of the historical period; Obtaining corresponding operation behavior features according to the operation behavior data; Obtaining a behavior duration table, wherein the behavior duration table includes multiple operation behavior features and a historical duration corresponding to each operation behavior; Obtain the corresponding historical duration from the duration table according to the operation behavior characteristics; Get the start time of the historical period based on the end time and historical duration of the historical period; Get the historical period based on the start time and end time of the historical period.

5. The method for detecting dangerous operations based on big data according to claim 1, characterized in that: The steps of obtaining the working environment data of the operator and obtaining the corresponding working environment compensation information according to the working environment data include: Obtain the working environment data of the operators; Acquire a plurality of corresponding operating environment vectors according to the operating environment data; Corresponding operating environment values ​​are obtained according to multiple operating environment vectors and marked as operating environment compensation information.

6. The method for detecting dangerous operations based on big data according to claim 1, characterized in that: The steps of obtaining the operation scene data of the operator and obtaining the corresponding operation scene compensation information according to the operation scene data include: Obtaining the operating scene data of the operators; Obtain corresponding operation scene characteristics according to operation scene data; Obtain a scenario table, wherein the scenario table includes a plurality of operation scenario characteristics and an operation scenario value corresponding to each operation scenario characteristic; According to the characteristics of the operation scene, the corresponding operation scene value is obtained from the scene table and marked as the operation scene compensation information.

7. The method for detecting dangerous operations based on big data according to claim 1, characterized in that: The step of obtaining predicted behavior feature information according to the operation behavior information, the operation environment compensation information and the operation scene compensation information includes: Obtaining corresponding operation behavior values ​​according to the operation behavior information; Obtaining corresponding working environment values ​​according to working environment compensation information; Obtaining corresponding operation scene values ​​according to operation scene compensation information; Obtaining corresponding predicted behavior feature values ​​according to the operation behavior value, the operation environment compensation value, and the operation scene compensation value; A feature table is obtained, wherein the feature table includes a plurality of predicted behavior feature interval values ​​and predicted behavior feature information corresponding to each predicted behavior feature interval value.

8. The method for detecting dangerous operations based on big data according to claim 1, characterized in that: The steps of obtaining behavior feature similarity information between the predicted behavior feature information and each dangerous behavior feature information, determining whether the behavior feature similarity information meets the second condition, and if not, determining that the operator is located in a dangerous operation range and issuing an alarm message include: Acquire a corresponding predicted behavior feature vector according to the predicted behavior feature information; Acquire a dangerous behavior feature vector corresponding to each dangerous behavior feature information according to the plurality of dangerous behavior feature information; Obtaining a behavior feature similarity value between the predicted behavior feature vector and each dangerous behavior feature vector, and marking it as behavior feature similarity information; Obtaining behavioral feature similarity threshold; Determine whether each behavior feature similarity value exceeds a feature similarity threshold; If there is a behavior feature similarity value that exceeds the feature similarity threshold among multiple behavior feature similarity values, it is determined that the operator is in a dangerous working range and an alarm message is issued.

9. A dangerous operation detection system based on big data, applied to the dangerous operation detection method based on big data according to any one of claims 1 to 8, characterized in that: include: The abnormality judgment module is used to obtain the operation behavior data of the operator and judge whether the operation behavior data meets the first preset condition. If not, the operation behavior of the operator is judged to be abnormal and marked as abnormal operation behavior; The operation preference module is used to obtain the historical operation preference data of the operator, and obtain the operation behavior information of the operator based on the abnormal operation behavior and the historical operation preference data; The working environment module is used to obtain the working environment data of the operator and obtain the corresponding working environment compensation information according to the working environment data; The operation scene module is used to obtain the operation scene data of the operator and obtain the corresponding operation scene compensation information according to the operation scene data; A prediction behavior module, used to obtain prediction behavior feature information based on operation behavior information, operation environment compensation information and operation scene compensation information; The dangerous behavior module is used to obtain multi-source dangerous operation data and obtain corresponding multiple dangerous behavior feature information according to the multi-source dangerous operation data; The danger judgment module is used to obtain the behavior characteristic similarity information between the predicted behavior characteristic information and each dangerous behavior characteristic information, and determine whether the behavior characteristic similarity information meets the second condition. If not, it is determined that the operator is in a dangerous operation range and an alarm message is issued.

10. A dangerous operation detection terminal based on big data, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When one or more programs are executed by one or more processors, the one or more processors implement the big data-based dangerous work detection method described in any one of claims 1 to 8.

Images