Engineering construction safety training method and system based on big data

Abstract

The invention relates to the field of safety practical training, in particular to an engineering construction safety practical training method and system based on big data, and the method comprises the steps: determining a preset analysis condition corresponding to a target person according to a frame selection duration characterization value and a frame selection duration fluctuation value when the answering frequency of the target person is equal to the optimization response frequency; under a first preset analysis condition, determining a compensation mode as point location array compensation or frame parameter compensation according to a point location aggregation similarity value and a point location aggregation representation value corresponding to the test question point location distribution diagram of the target person; when frame parameter compensation is carried out, a frame parameter compensation strategy is determined according to the number proportion of the first anchoring frames and the second anchoring frames, and compensation is carried out based on frame neighborhood parameters or frame areas; under a second preset analysis condition, increasing and adjusting the number of optimization response times; the test questions are dynamically adjusted to improve the safety training efficiency.

Description

Technical Field

[0001] The present invention relates to the field of safety training, and in particular to a method and system for engineering construction safety training based on big data. Background Art

[0002] In the construction industry, safe production is a core guarantee for its development, and safety training is a crucial means of enhancing workers' safety awareness and operational standards. With the development of big data and artificial intelligence technologies, unsafe behavior identification training based on on-site images has become a key area of ​​safety training. This approach analyzes unsafe behaviors (such as not wearing a hard hat or operating machinery inappropriately) in construction site images to train trainees on safety regulations. However, existing safety training typically generates questions based on static question banks or fixed scenarios, failing to dynamically adjust question difficulty or content based on trainees' actual training performance, resulting in inefficient training.

[0003] Chinese Patent Publication No. CN113807990A discloses a construction site safety training method, system, device, and storage medium. The method includes: obtaining a question bank for safety training, current construction site information, and information about construction workers to be trained, generating a personal tag for the construction worker to be trained, matching the personal tag with the question tags in the question bank, and generating a safety training answer sheet that matches the current construction worker to be trained. In the above technical solution, if the number of incorrect answers the construction worker to be trained received the previous day is less than or equal to a preset threshold, all incorrect answers from the previous day are inserted into the current day's safety training answer sheet. If the number of incorrect answers the construction worker to be trained received the previous day is greater than the preset threshold, the number of incorrect answers within the preset threshold is inserted into the current safety training answer sheet. For incorrect answers exceeding the preset threshold, a predetermined number of incorrect answers are selected each day and inserted into the current day's safety training answer sheet, where the predetermined number is less than or equal to the preset threshold. This technical solution can only optimize and add historical incorrect answers to the new answer sheet, and its single optimization method is difficult to meet the training needs of actual scenarios. Summary of the Invention

[0004] To this end, the present invention provides a method and system for engineering construction safety training based on big data, which is used to overcome the problem in the prior art that the method can only optimize the test paper based on repeated training of wrong questions, the optimization method is single, and the test content cannot be dynamically adjusted, resulting in low training efficiency.

[0005] To achieve the above objectives, the present invention provides a method for engineering construction safety training based on big data, comprising:

[0006] When the number of times the target person answers questions is equal to the number of optimized responses, the preset analysis conditions corresponding to the target person are determined based on the box selection time representation value and the box selection time fluctuation value;

[0007] Under the first preset analysis condition, the compensation method is determined to be point array compensation or frame parameter compensation according to the point cluster similarity value and point cluster representation value corresponding to the test point distribution map of the target person;

[0008] When compensating for frame parameters, determining a frame parameter compensation strategy based on frame neighborhood parameters or frame area according to the ratio of the number of the first anchor frame and the second anchor frame;

[0009] Under the second preset analysis condition, an increase adjustment is made to the optimized response times.

[0010] Furthermore, under the first preset analysis condition, a test point distribution map corresponding to each touch delay record is obtained;

[0011] Obtain the point clustering similarity value and point clustering representation value corresponding to the test point distribution map;

[0012] If the point aggregation similarity value is greater than the preset point aggregation similarity value and the point aggregation characterization value is less than or equal to the preset point aggregation characterization value, the compensation method is point array compensation;

[0013] If the point cluster similarity is less than or equal to the preset point cluster similarity or the point cluster representation value is greater than the preset point cluster representation value, the compensation method is frame parameter compensation.

[0014] Furthermore, when the compensation method is point array compensation, the cloud storage test questions are extracted in descending order of the point array deviation, and the required number of cloud storage test questions are compensated;

[0015] The point array deviation is determined based on the array area difference value and the array distance compensation value.

[0016] Furthermore, when the compensation method is frame parameter compensation, the frame parameter compensation strategy is determined according to the ratio of the number of the first anchor frame and the second anchor frame;

[0017] If the quantity ratio is less than the preset quantity ratio, the frame parameter compensation strategy is to compensate based on the frame neighborhood parameters;

[0018] If the quantity ratio is greater than or equal to the preset quantity ratio, the frame parameter compensation strategy is to compensate based on the frame area.

[0019] Furthermore, the anchor box category corresponding to each anchor box is determined according to the frame selection deviation value of the anchor box, and the anchor box category includes a first anchor box whose frame selection deviation value is greater than the allowable frame selection deviation value and a second anchor box whose frame selection deviation value is less than or equal to the allowable frame selection deviation value.

[0020] Furthermore, in the process of extracting cloud-stored test questions, the corresponding storage module is determined according to the optimal update times of the storage module and the engineering diversity.

[0021] Furthermore, under the second preset analysis condition, the number of optimized responses is increased and adjusted according to the difference represented by the frame selection time;

[0022] The increase in the number of optimization responses is positively correlated with the difference in the selection time.

[0023] Furthermore, the preset analysis conditions are determined by the frame selection duration characterization value and the frame selection duration fluctuation value;

[0024] The first preset analysis condition is that the frame selection duration representation value is less than or equal to the preset frame selection duration representation value and the frame selection duration fluctuation value is greater than the preset frame selection duration fluctuation value;

[0025] The second preset analysis condition is that the frame selection duration characterization value is greater than the preset frame selection duration characterization value or the frame selection duration fluctuation value is less than or equal to the preset frame selection duration fluctuation value.

[0026] In addition, the present invention also provides a system for applying the engineering construction safety training method based on big data, which is characterized by comprising:

[0027] A conditional analysis unit, configured to determine a preset analysis condition corresponding to the target person based on the box selection duration representation value and the box selection duration fluctuation value when the target person's number of answers is equal to the optimized number of responses;

[0028] an optimization compensation unit connected to the condition analysis unit, configured to determine, under a first preset analysis condition, based on the point cluster similarity value and the point cluster characterization value corresponding to each touch delay record of the target person, whether the compensation method is point array compensation or frame parameter compensation, and, when compensating for the frame parameters, determine, based on the ratio of the number of the first anchor frame to the number of the second anchor frame, whether the frame parameter compensation strategy is compensation based on the frame neighborhood parameter or compensation based on the frame area;

[0029] a category analysis unit connected to the optimization and compensation unit, configured to determine an anchor box category corresponding to the anchor box according to the box selection deviation value of the anchor box when the optimization and compensation unit performs box parameter compensation;

[0030] a simplified processing unit connected to the condition analysis unit, configured to increase and adjust the number of optimized responses according to the difference in the box selection duration under a second preset analysis condition;

[0031] The cloud data storage unit is connected to the condition analysis unit, the optimization compensation unit, the category analysis unit and the simplified processing unit. The cloud data storage unit includes several storage modules for cloud storage of test questions.

[0032] Compared with the prior art, the beneficial effect of the present invention lies in that the preset analysis conditions corresponding to the target person are determined according to the frame selection time characterization value and the frame selection time fluctuation value in the technical solution of the present invention, and the target person's answering time is reflected by the frame selection time characterization value and the frame selection time fluctuation value, and the compensation method is determined according to the point aggregation similarity value and the point aggregation characterization value corresponding to the target person's test point distribution map according to different preset analysis conditions, or the number of optimized responses is increased and adjusted to make the optimization analysis more accurate, avoiding the problem that the question optimization efficiency is poor and it is difficult to meet the training needs due to a single optimization method.

[0033] Furthermore, under the first preset analysis condition in the technical solution of the present invention, the compensation method is determined to be point array compensation or frame parameter compensation based on the point cluster similarity value and the point cluster characterization value corresponding to the test point distribution map. Compared with the existing technology that only optimizes questions based on historical wrong questions, the present invention analyzes the question image itself and selects different compensation methods accordingly, so that the compensation method is more in line with the actual application scenario, thereby improving the question optimization efficiency and further improving the subsequent training effect.

[0034] Furthermore, in the technical solution of the present invention, the corresponding storage module in the process of extracting cloud-stored test questions is determined according to the preferred update times of the storage module and the engineering diversity, so that the selection of test questions is more in line with the needs of actual scenarios, avoiding the problem of poor test question selection effect caused by manual test question selection or random test question selection. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the engineering construction safety training method based on big data of the present invention;

[0036] Figure 2 A flow chart for determining compensation methods for the present invention;

[0037] Figure 3 A flow chart for determining a frame parameter compensation strategy for the present invention;

[0038] Figure 4 This is a unit connection diagram of the engineering construction safety training system based on big data of the present invention. DETAILED DESCRIPTION

[0039] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0040] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0041] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0042] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0043] See also Figures 1 to 3 As shown, the present invention provides a construction safety training method based on big data, comprising:

[0044] When the number of times the target person answers questions is equal to the number of optimized responses, the preset analysis conditions corresponding to the target person are determined based on the box selection time representation value and the box selection time fluctuation value;

[0045] Under the first preset analysis condition, the compensation method is determined to be point array compensation or frame parameter compensation according to the point cluster similarity value and point cluster representation value corresponding to the test point distribution map of the target person;

[0046] When compensating for frame parameters, determining a frame parameter compensation strategy based on frame neighborhood parameters or frame area according to the ratio of the number of the first anchor frame and the second anchor frame;

[0047] Under the second preset analysis condition, an increase adjustment is made to the optimized response times.

[0048] The present invention is applied to safety training in engineering construction. The target personnel are the personnel who are conducting the training. The target personnel obtain the test paper image through a display device. When the target personnel answers the test question image, the area in the test question image where dangerous behavior is believed to exist is framed on the display device, and the framed area is recorded as an anchor frame. The cloud-stored test questions are the construction site images used in the safety training uploaded by each engineering unit, and each cloud-stored test question stores the shape of the preset area, the position of the area center point of the preset area, and the area of ​​the preset area. The preset area is the area where dangerous behavior exists. The preset area and the corresponding shape of the preset area, the position of the area center point of the preset area, and the area of ​​the preset area are all set by the user, and the user can carry out a large amount of construction in advance. The collection of on-site images and the screening of construction site images with dangerous behaviors, and the corresponding setting of preset areas are contents that are easy for technical personnel in this field to understand and will not be elaborated here; there are preset areas for both test image and cloud-stored test questions, and the test point distribution map is the test image after each preset area is marked (boxed and marked), and the number of times the target personnel answer the questions is the number of times the target personnel conduct practical training on the test image after the determination of the most recent preset analysis conditions is completed. Processing one test image is recorded as one answer number. Among them, when the engineering unit uploads the construction site image, the construction site image is preferentially allocated to the storage module with the least cloud-stored test questions. If the number of storage modules with the least cloud-stored test questions is greater than 1, one of the storage modules with the least cloud-stored test questions is randomly selected for uploading.

[0049] The present invention applies historical records, which include at least one historical process of point aggregation similarity value, point aggregation characterization value, array area difference value, array distance compensation value, frame selection deviation value, frame selection duration characterization value and frame selection duration fluctuation value, and the historical records also correspond to qualified marks, which record whether the historical records meet user needs. Among them, determining whether the historical records meet user needs based on self-set indicators of training effects (such as the accuracy of answering questions) is content that people in this field have mastered and will not be elaborated on.

[0050] Specifically, under the first preset analysis condition, a test point distribution map corresponding to each touch delay record is obtained;

[0051] Obtain the point clustering similarity value and point clustering representation value corresponding to the test point distribution map;

[0052] If the point aggregation similarity value is greater than the preset point aggregation similarity value and the point aggregation characterization value is less than or equal to the preset point aggregation characterization value, the compensation method is point array compensation;

[0053] If the point cluster similarity is less than or equal to the preset point cluster similarity or the point cluster representation value is greater than the preset point cluster representation value, the compensation method is frame parameter compensation.

[0054] The touch delay record is the answer record in which the lower-order selection duration representation value is less than the selection duration representation value in the answer record of the target person's most recent optimized response times. A single answer record includes a test question image processed by the target person, the selection duration corresponding to the test question image, the test question point distribution map, and the selection deviation value of each anchor frame;

[0055] The method for confirming the point aggregation characterization value is to detect the aggregation range area of ​​the test point distribution map corresponding to each touch delay record, and record the average value of the aggregation range area as the point aggregation characterization value. For a single test point distribution map, its aggregation range is the minimum rectangle that can include all preset areas in the test point distribution map; the point aggregation similarity value is the absolute value of the difference between the maximum and minimum values ​​of the aggregation range area corresponding to each test point distribution map.

[0056] The values ​​of the preset point aggregation similarity value and the preset point aggregation characterization value can be determined by the user according to the actual application scenario. It can be understood that the point aggregation similarity value is greater than the preset point aggregation similarity value and the point aggregation characterization value is less than or equal to the preset point aggregation characterization value, which reflects that the preset areas corresponding to the test question images answered by the target person recently are densely distributed and the distribution areas of the preset areas between the test question images are highly similar. Therefore, the compensation method is point array compensation to improve the distribution diversity of the preset areas of subsequent test question images, thereby improving the training effect and avoiding the problem of poor training effect caused by the convergence of questions. Therefore, the greater the user's demand for the distribution diversity of the preset area, the greater the value of the preset point cluster similarity value, and the smaller the value of the preset point cluster characterization value. A value selection method is provided to extract the point cluster similarity value and the point cluster characterization value corresponding to the historical records that meet the user's needs, remove the outliers in the point cluster similarity value and the point cluster characterization value respectively, and record the average values ​​corresponding to the point cluster similarity value and the point cluster characterization value after removing the outliers as the preset point cluster similarity value and the preset point cluster characterization value respectively. Among them, the method of removing outliers includes but is not limited to the 3σ criterion method or the IQR method.

[0057] Specifically, when the compensation method is point array compensation, the cloud storage test questions are extracted in descending order of the point array deviation, and the required number of cloud storage test questions are compensated;

[0058] The point array deviation is determined based on the array area difference value and the array distance compensation value.

[0059] For a single cloud-stored test question, the corresponding point array deviation = array area difference value / preset array area difference value + array distance compensation value / preset array distance compensation value. The array area difference value = the aggregation range area of ​​the cloud-stored test question - the point aggregation representation value. The array distance compensation value is confirmed by obtaining the minimum distance between the preset areas of the cloud-stored test question, and recording the minimum value of the minimum distance as the array distance compensation value.

[0060] The values ​​of the preset array area difference value and the preset array distance compensation value can be set by the user according to the actual application scenario. It can be understood that the higher the user's requirements for the compensation effect of the cloud-stored test questions, the larger the values ​​of the preset array area difference value and the preset array distance compensation value. A value selection method is provided to extract the array area difference value and the array distance compensation value corresponding to the historical records that meet the user's needs, remove the outliers in the array area difference value and the array distance compensation value respectively, and record the corresponding average values ​​of the array area difference value and the array distance compensation value after removing the outliers as the preset array area difference value and the preset array distance compensation value.

[0061] The value of the required quantity is set by the user. It can be understood that the greater the user's demand for cloud storage test question compensation, the greater the required quantity and the richer the number of subsequent test question images. A specific implementation value is provided, and the required quantity = 10.

[0062] Specifically, when the compensation method is frame parameter compensation, the frame parameter compensation strategy is determined according to the ratio of the number of the first anchor frame and the second anchor frame;

[0063] If the quantity ratio is less than the preset quantity ratio, the frame parameter compensation strategy is to compensate based on the frame neighborhood parameters;

[0064] If the quantity ratio is greater than or equal to the preset quantity ratio, the frame parameter compensation strategy is to compensate based on the frame area.

[0065] Quantity ratio = number of first anchor boxes in all touch delay records / number of second anchor boxes. As a special case, if the number of first anchor boxes = 0, there is no need to analyze the quantity ratio, and the fixed selection box parameter compensation strategy is to compensate based on the box neighborhood parameters. If the number of second anchor boxes = 0, there is no need to analyze the quantity ratio, and the box parameter compensation strategy is to compensate based on the box area. It can be understood that there is no situation where the number of first anchor boxes and second anchor boxes is 0 at the same time.

[0066] The value of the preset quantity ratio can be set by the user according to actual conditions. It can be understood that the quantity ratio reflects the accuracy of the anchor frame of the target person. The greater the user's training demand for the accuracy of the anchor frame of the target person, the smaller the preset quantity ratio. Provided is a value, the preset quantity ratio = 20%.

[0067] Specifically, the anchor box category corresponding to each anchor box is determined according to the frame selection deviation value of the anchor box. The anchor box category includes a first anchor box whose frame selection deviation value is greater than the allowable frame selection deviation value and a second anchor box whose frame selection deviation value is less than or equal to the allowable frame selection deviation value.

[0068] The frame selection deviation value of the anchor frame, that is, obtaining the preset area where the anchor frame overlaps, calculating the area of ​​the preset area where the anchor frame is not overlapping, and recording it as the frame selection deviation value. If an anchor frame does not have an overlapping preset area, then the anchor frame does not have a frame selection deviation value, and the anchor frame is an invalid anchor frame. It can be understood that the overlapping preset area can be partial or complete, and there is no need to limit it.

[0069] The value of the allowed box selection deviation value is set by the user. It can be understood that the greater the user's acceptance of the box selection deviation value, the larger the value of the allowed box selection deviation value is. A method for setting the value of the allowed box selection deviation value is provided to extract the box selection deviation value corresponding to the historical records that meet user needs, and record the average value of the box selection deviation value after removing the outliers as the allowed box selection deviation value.

[0070] Specifically, the frame neighborhood parameter=the neighboring object distribution value, where the neighboring object distribution value is the total number of buildings within the relevant range of the preset area.

[0071] For a single preset area, the corresponding correlation range is a circular area. The area center point of the correlation range coincides with the area center point of the preset area, and the entire preset area is within the correlation range. The area of ​​the correlation range is smaller than the area of ​​the test image where the preset area is located. The specific area of ​​the correlation range is set by the user. The greater the user's demand for the analysis range of the frame neighborhood parameters of the preset area, the larger the value of the correlation range. A value is provided, and the area of ​​the correlation range is equal to twice the area of ​​the smallest circular area that can include the preset area.

[0072] When compensation is performed based on the frame neighborhood parameter, the cloud storage test questions are extracted in the required number of cloud storage test questions in descending order of the frame neighborhood parameter;

[0073] When compensation is performed based on the frame area, the required number of cloud-stored test questions are extracted from the cloud-stored test questions in descending order of the frame area for compensation; the frame area corresponding to the cloud-stored test question is the average value of the area of ​​the preset area corresponding to the cloud-stored test question.

[0074] The extracted cloud-stored test questions for compensation are recorded as compensation test questions and added to the test question images for the next target personnel training.

[0075] Specifically, in the process of extracting cloud-stored test questions, the corresponding storage module is determined according to the optimal update times of the storage module and the project diversity.

[0076] When extracting cloud-stored test questions, cloud-stored test questions are only extracted from a single storage module. The preferred update times of each storage module are calculated separately, and the storage modules corresponding to the top three preferred update times are selected in descending order. The storage module with the largest engineering diversity among the three storage modules is selected as the storage module for extracting cloud-stored test questions.

[0077] Among them, the preferred update number is the number of data uploads by each engineering unit corresponding to the storage module. The number of data uploads is the number of times the construction site images are uploaded. There is no limit on the number of construction site images uploaded in a single data upload; the engineering diversity is the total number of engineering units that have uploaded data corresponding to the storage module.

[0078] Specifically, under the second preset analysis condition, the number of optimized responses is increased and adjusted according to the difference represented by the frame selection time;

[0079] The increase in the number of optimization responses is positively correlated with the difference in the selection time.

[0080] The difference in the representation of the frame selection duration = the representation value of the frame selection duration corresponding to the target person - the preset representation value of the frame selection duration.

[0081] Increase the number of optimized responses after adjustment = the number of optimized responses before adjustment + the increase in the number of optimized responses, the increase in the number of optimized responses = k × the difference in box selection time representation, k is the conversion coefficient, the number of optimized responses before adjustment and the value of k are set by the user. It can be understood that the larger the difference in box selection time representation, the worse the answer effect of the target person. Therefore, the larger the value of the optimized response number is, so as to avoid the problem of large data computing resource consumption caused by frequent analysis of test question optimization. In addition, the smaller the number of optimized responses, the greater the frequency of test question optimization. Therefore, the greater the user's acceptance of data computing resource consumption, the smaller the value of k, the larger the number of optimized responses, and the increase in the number of optimized responses is an integer rounded up, and a value is provided, k = 0.4, the number of optimized responses before adjustment = 10.

[0082] Specifically, the preset analysis condition is determined by the frame selection duration characterization value and the frame selection duration fluctuation value;

[0083] The first preset analysis condition is that the frame selection duration representation value is less than or equal to the preset frame selection duration representation value and the frame selection duration fluctuation value is greater than the preset frame selection duration fluctuation value;

[0084] The second preset analysis condition is that the frame selection duration characterization value is greater than the preset frame selection duration characterization value or the frame selection duration fluctuation value is less than or equal to the preset frame selection duration fluctuation value.

[0085] The frame selection duration representation value is the average value of the lower-level frame selection duration representation values ​​of the answer records of the target person's most recent optimized response times. For a single answer record, the corresponding lower-level frame selection duration representation value is confirmed by detecting the interval duration between adjacent anchor frames in the construction time sequence corresponding to the answer record. The interval time is the duration between the construction moments corresponding to two adjacent anchor frames in the construction time sequence, and the average value of the interval duration is recorded as the lower-level frame selection duration representation value of the answer record.

[0086] The fluctuation value of the frame selection duration is recorded as S, and the calculation method of the frame selection duration fluctuation value is:

[0087]

[0088] Among them, Lu is the lower-order box selection time representation value corresponding to the u-th answer record in the answer record of the target person's most recent optimized response times, L0 is the box selection time representation value, P = optimized response times, u = 1, 2, 3, ..., P, and the order of values ​​of u corresponding to the answer record of the target person's most recent optimized response times does not need to be specifically limited, and has no effect on the calculation results.

[0089] The values ​​of the preset frame selection time representation value and the preset frame selection time fluctuation value are set by the user. It can be understood that the higher the user's requirements for the target person's answering fluency, the smaller the preset frame selection time representation value and the smaller the preset frame selection time fluctuation value. A value selection method is provided to extract the frame selection time representation value and the frame selection time fluctuation value corresponding to the historical records that meet the user's needs, and remove the abnormal values ​​in the frame selection time representation value and the frame selection time fluctuation value respectively, and record the corresponding average values ​​of the frame selection time representation value and the frame selection time fluctuation value after removing the abnormal values ​​as the preset frame selection time representation value and the preset frame selection time fluctuation value respectively.

[0090] See also Figure 4 As shown in FIG, which is a unit connection diagram of the engineering construction safety training system based on big data of the present invention, the present invention also provides a system applying the engineering construction safety training method based on big data, characterized in that it includes:

[0091] A conditional analysis unit, configured to determine a preset analysis condition corresponding to the target person based on the box selection duration representation value and the box selection duration fluctuation value when the target person's number of answers is equal to the optimized number of responses;

[0092] an optimization compensation unit connected to the condition analysis unit, configured to determine, under a first preset analysis condition, based on the point cluster similarity value and the point cluster characterization value corresponding to each touch delay record of the target person, whether the compensation method is point array compensation or frame parameter compensation, and, when compensating for the frame parameters, determine, based on the ratio of the number of the first anchor frame to the number of the second anchor frame, whether the frame parameter compensation strategy is compensation based on the frame neighborhood parameter or compensation based on the frame area;

[0093] a category analysis unit connected to the optimization and compensation unit, configured to determine an anchor box category corresponding to the anchor box according to the box selection deviation value of the anchor box when the optimization and compensation unit performs box parameter compensation;

[0094] a simplified processing unit connected to the condition analysis unit, configured to increase and adjust the number of optimized responses according to the difference in the box selection duration under a second preset analysis condition;

[0095] The cloud data storage unit is connected to the condition analysis unit, the optimization compensation unit, the category analysis unit and the simplified processing unit. The cloud data storage unit includes several storage modules for cloud storage of test questions.

[0096] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

[0097] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for engineering construction safety training based on big data, characterized in that: include: When the number of times the target person answers questions is equal to the number of optimized responses, the preset analysis conditions corresponding to the target person are determined based on the box selection time representation value and the box selection time fluctuation value; Under the first preset analysis condition, the compensation method is determined to be point array compensation or frame parameter compensation according to the point cluster similarity value and point cluster representation value corresponding to the test point distribution map of the target person; When compensating for frame parameters, determining a frame parameter compensation strategy based on frame neighborhood parameters or frame area according to the ratio of the number of the first anchor frame and the second anchor frame; Under the second preset analysis condition, an increase adjustment is made to the optimized response times.

2. The engineering construction safety training method based on big data according to claim 1 is characterized in that: Under the first preset analysis condition, obtain the test point distribution map corresponding to each touch delay record; Obtain the point clustering similarity value and point clustering representation value corresponding to the test point distribution map; If the point aggregation similarity value is greater than the preset point aggregation similarity value and the point aggregation characterization value is less than or equal to the preset point aggregation characterization value, the compensation method is point array compensation; If the point cluster similarity is less than or equal to the preset point cluster similarity or the point cluster representation value is greater than the preset point cluster representation value, the compensation method is frame parameter compensation.

3. The engineering construction safety training method based on big data according to claim 2 is characterized in that: When the compensation method is point array compensation, the cloud storage test questions are extracted in descending order of the point array deviation, and the required number of cloud storage test questions are compensated; The point array deviation is determined based on the array area difference value and the array distance compensation value.

4. The engineering construction safety training method based on big data according to claim 2 is characterized in that: When the compensation method is frame parameter compensation, the frame parameter compensation strategy is determined according to the ratio of the number of the first anchor frame and the second anchor frame; If the quantity ratio is less than the preset quantity ratio, the frame parameter compensation strategy is to compensate based on the frame neighborhood parameters; If the quantity ratio is greater than or equal to the preset quantity ratio, the frame parameter compensation strategy is to compensate based on the frame area.

5. The engineering construction safety training method based on big data according to claim 4 is characterized in that: The anchor box category corresponding to each anchor box is determined according to the selection deviation value of the anchor box. The anchor box category includes a first anchor box whose selection deviation value is greater than the allowable selection deviation value and a second anchor box whose selection deviation value is less than or equal to the allowable selection deviation value.

6. The engineering construction safety training method based on big data according to claim 5 is characterized in that: During the process of extracting the cloud-stored test questions, the storage modules corresponding to the extracted cloud-stored test questions are determined based on the optimal update times and engineering diversity of each storage module.

7. The engineering construction safety training method based on big data according to claim 1 is characterized in that: Under the second preset analysis condition, the number of optimized responses is increased and adjusted according to the difference represented by the frame selection time; The increase in the number of optimization responses is positively correlated with the difference in the selection time.

8. The engineering construction safety training method based on big data according to claim 2 or 7 is characterized in that: The preset analysis conditions are determined by the frame selection duration characterization value and the frame selection duration fluctuation value; The first preset analysis condition is that the frame selection duration representation value is less than or equal to the preset frame selection duration representation value and the frame selection duration fluctuation value is greater than the preset frame selection duration fluctuation value; The second preset analysis condition is that the frame selection duration characterization value is greater than the preset frame selection duration characterization value or the frame selection duration fluctuation value is less than or equal to the preset frame selection duration fluctuation value.

9. A system using the engineering construction safety training method based on big data according to any one of claims 1 to 8, characterized in that: include: A conditional analysis unit, configured to determine a preset analysis condition corresponding to the target person based on the box selection duration representation value and the box selection duration fluctuation value when the target person's number of answers is equal to the optimized number of responses; an optimization compensation unit connected to the condition analysis unit, configured to determine, under a first preset analysis condition, based on the point cluster similarity value and the point cluster characterization value corresponding to each touch delay record of the target person, whether the compensation method is point array compensation or frame parameter compensation, and, when compensating for the frame parameters, determine, based on the ratio of the number of the first anchor frame to the number of the second anchor frame, whether the frame parameter compensation strategy is compensation based on the frame neighborhood parameter or compensation based on the frame area; a category analysis unit connected to the optimization and compensation unit, configured to determine an anchor box category corresponding to the anchor box according to the box selection deviation value of the anchor box when the optimization and compensation unit performs box parameter compensation; a simplified processing unit connected to the condition analysis unit, configured to increase and adjust the number of optimized responses according to the difference in the box selection duration under a second preset analysis condition; The cloud data storage unit is connected to the condition analysis unit, the optimization compensation unit, the category analysis unit and the simplified processing unit. The cloud data storage unit includes several storage modules for cloud storage of test questions.

Images