A target recognition and detection system and method based on neural network

Abstract

The present invention discloses a target recognition and detection system and method based on a neural network, and relates to the field of industrial production operation and maintenance technology, including: acquiring monitoring videos in an industrial production process, using monitoring images to train a neural network model for personnel identification and equipment identification; obtaining the personnel's proficiency in operating the equipment based on the personnel's operating time and practice operating time; analyzing the monitoring images of the equipment, obtaining the risk value of the equipment based on the personnel's operating proficiency and operating time, and when the risk value of the equipment is greater than or equal to a threshold, contacting maintenance personnel to repair the equipment; retrieving time data in the maintenance record, and judging whether the equipment needs to be repaired according to the operation time of the equipment by different personnel, providing a reference for equipment maintenance, and improving the effect of equipment maintenance; using neural network technology to identify equipment and personnel, without the need for manual intervention, and being able to adapt to industrial production scenarios.

Description

Technical Field

[0001] The present invention relates to the field of industrial production operation and maintenance technology, and specifically to a target recognition and detection system and method based on a neural network. Background Art

[0002] In the field of industrial production, stable operation of equipment is the basis for ensuring the stability of the production process and the safety of personnel; equipment maintenance is an indispensable part of safe production, and is of great significance for ensuring the normal operation of the production system, improving the service life and operating efficiency of equipment; with the development of machine learning technology, neural networks provide better image recognition for various application scenarios, and play an increasingly important role in industrial production; the operating time of equipment has always been an important factor in determining whether the equipment needs maintenance, and monitoring the operating time of equipment during the production process is of great significance for determining whether the equipment needs maintenance; at present, production equipment usually adopts regular maintenance and overhaul methods. Due to differences in the capabilities of personnel operating the equipment, the damage caused to the equipment by personnel operation is not exactly the same, and regular maintenance has problems such as untimely maintenance and low maintenance efficiency; for this reason, how to improve the maintenance effect of equipment has become an urgent problem to be solved. Summary of the invention

[0003] The purpose of the present invention is to provide a target recognition and detection system and method based on a neural network to solve the problems raised in the above background technology.

[0004] In one aspect of the present invention, a method for target recognition and detection based on a neural network is provided, comprising:

[0005] S10, acquiring a monitoring video of the industrial production process, obtaining a monitoring image of each frame based on the monitoring video, and using the monitoring image to train a neural network model for personnel recognition and equipment recognition;

[0006] S11, based on the monitoring images during the production process, obtaining the operation time of the personnel on the equipment, and based on the operation time of the personnel on the equipment and the practice operation time, obtaining the operation proficiency of the personnel on the equipment;

[0007] S12, monitor the equipment in real time, analyze the monitoring image of the equipment, and obtain the risk value of the equipment based on the operator's operating proficiency and operating time. When the risk value of the equipment is greater than or equal to the threshold, contact the maintenance personnel to repair the equipment.

[0008] In step S10, the method of training the neural network model using the monitoring image further includes the following steps:

[0009] S20, finding a surveillance image in which a person exists from the surveillance image, extracting the person's physical features and identity features from the surveillance image in which the person exists, taking the physical features and identity features as input and the person's ID as output, training a person neural network model for identifying the identity of the person appearing in the surveillance image, wherein the identity features refer to the certificate recording the employee's identity information;

[0010] S21, find a surveillance image about the device from the surveillance image, extract the movement and behavior characteristics of the personnel from the surveillance image, take the movement and behavior characteristics of the personnel as input, and the status of the device as output, train the device neural network model for identifying the device status in the surveillance image, wherein the status of the device includes three situations: unmanned operation, single-person operation, and multi-person operation; when the device is identified as being operated by a single person, determine the time when the single person operates the device, and use the personnel neural network model to determine the identity of the person operating the device; when the device is identified as being operated by multiple people, determine the time when the multiple people operate the device, and use the personnel neural network model to determine the identity of the person operating the device.

[0011] In step S11, obtaining the operator's proficiency in operating the equipment based on the operator's operating time on the equipment further includes the following steps:

[0012] S30, based on the identity information of person P, determine the independent operation authority of person P for each device, where the independent operation authority refers to the authority of person P to operate the device independently without the need for guidance or supervision by other persons except P; based on the independent operation authority of all persons including P for each device, obtain the person with independent operation authority for device B;

[0013] S31, obtaining the operation time of the person P on each device from the single-person operation situation in the historical monitoring image; obtaining the practice operation time of the person P on each device from the multi-person operation situation in the historical monitoring image;

[0014] S32, for device B, find a person who has independent operation authority for device B, and determine whether the person who has independent operation authority for device B has the ability to operate other devices A with lower operation requirements than B. 1 , A 2 , …, A n The operation time on B, n is the number of other devices with lower operation requirements than B, the other devices A 1 , A 2 , …, A n The condition that the operation time of all personnel with independent operation authority for device B is not 0 is met;

[0015] For complex and difficult-to-operate equipment B, before personnel operate equipment B, in addition to sufficient training and practice, they may need to have simple operation equipment A related to equipment B. 1 , A 2 , …, A n Supported by operational experience.

[0016] S33, obtaining the information of person P about other equipment A 1 , A 2 , …, A n Operation time t 1 ,t 2 ,…,t n , calculate the impact of personnel P on other equipment A 1 , A 2 , …, A n Operation score y 1 ,y 1 =a 1 t 1 +a 2 t 2 +…+a n t n ; Get the time y that person P practices operating equipment B 2 , based on y 1 and 2 Get the operator P's proficiency Y for equipment B, Y = b 1 y 1 +b 2 y 2 , where a 1 、a 2 , …, a n 、b 1 and b 2 is the coefficient.

[0017] Coefficient a 1 、a 2 , …, a n 、b 1 and b 2 Obtain it by following these steps:

[0018] S40, obtain the permissions of all personnel who have independent operation rights on device B to operate other devices A 1 , A 2 , …, A n The total operation time t s1 ,t s2 ,…,t sn , based on t s1 ,t s2 ,…,t sn Get a 1 、a 2, …, a n ,

[0019] S41, obtain the total time y of all personnel with independent operation authority for device B practicing operation of device B s2 , then b 1 =1-b 2 , y s1 =a 1 t s1 +a 2 t s2 +…+a n t sn ,y s1 For personnel who have independent operation authority over device B, they can control other devices A as a whole. 1 , A 2 , …, A n Operation score;

[0020] The personnel's independent operation authority for equipment is determined by the management personnel. The personnel with independent operation authority for equipment B are skilled personnel in operating equipment B. Conversely, skilled personnel in operating equipment B do not necessarily have independent operation authority for equipment B. For different personnel, before obtaining independent operation authority for equipment B, they must hold independent operation authority for other equipment A. 1 , A 2 , …, A n The operating experience of different personnel on other equipment 1 , A 2 , …, A n Although there are differences in operating experience, for all personnel in the entire production system who have independent operating authority for device B, all personnel have independent operating authority for other devices A. 1 , A 2 , …, A n The ratio of the total operation time of device B to that of device A can reflect the importance of obtaining the independent operation permission of device B for other devices A. 1 , A 2 , …, A n The proportional relationship between the operation time requirements and the personnel who meet the requirements for obtaining the independent operation authority of equipment B is also a skilled person in operating equipment B; the longer the total operation time of the equipment is, the more frequently the equipment is used, and the smaller the effect of improving the operation authority per unit operation time is, so the coefficient a 1 、a 2 , …, a n Positively correlated with the inverse of the total operation time; coefficient a 1 、a 2 , …, a n The sum of is 1, and will not contribute to the operation score y s1Produce a scaling effect; for other devices A 1 , A 2 , …, A n The operation time is carried out separately n Normalized, the total time y that personnel with independent operation authority for device B practice operation on device B s2 After normalization, y s1 and s2 The same scale; operating device B has two requirements, for other devices A 1 , A 2 , …, A n The sum of the operating experience of equipment B and the experience of practicing operating equipment B, and the practice operating time of equipment B is y s2 The larger the coefficient b is, the more difficult it is to operate device B. It takes more time to become familiar with the operation of device B. It is difficult to master the operation of device B in a short time. 2 With y s2 is positively correlated with the inverse of to reflect the operating difficulty of device B.

[0021] Determine whether person P is a skilled operator of device B: If person P's proficiency Y in operating device B is greater than or equal to the proficiency threshold Y thr , then person P is a skilled person in operating device B, otherwise person P is an unskilled person in operating device B; calculate the operating proficiency of all personnel with independent operating authority for device B to find the minimum value after removing outliers, and use the minimum value after removing outliers as the proficiency threshold Y thr The difficulty of obtaining independent operation authority for device B is higher than becoming a skilled operator of device B, so the minimum value after removing abnormal values ​​is taken.

[0022] In step S12, obtaining the risk value of the equipment based on the operator's operating proficiency and operating time of the equipment includes the following steps:

[0023] S50, obtaining the time T during which the device B is operated by a skilled person 1 , Time for unskilled personnel to practice operation T 2 and the time T for unskilled personnel to operate 3 When equipment B is operated by multiple people, if there are both skilled and unskilled personnel operating equipment B, or only skilled personnel operating equipment B, the time equipment B is operated by skilled personnel increases by T group , if there are only unskilled personnel among the personnel operating equipment B, the time equipment B is operated by unskilled personnel increases by T group , T groupis the time when device B is operated by multiple people; when device B is operated by one person, based on the identification result of the operator, if the operator is a skilled operator, the time when device B is operated by the skilled operator increases; if the operator is an unskilled operator and at time T thr Within 1 year, if the equipment B is operated by one person and then by multiple people with skilled personnel, the time for unskilled personnel to practice operating the equipment B will increase by T. indv1 Otherwise, the time when equipment B is operated by unskilled personnel increases by T indv2 ; T thr is the preset time threshold, T indv1 T is the time when device B is in single-operation state when it changes from single-operation state to multi-operation state. indv2 For device B at time T thr The time when device B is in single-operation state and does not change from single-operation state to multi-operation state;

[0024] When equipment B is operated by multiple people, as long as there is a skilled operator operating equipment B, equipment B will not incur losses due to unfamiliarity and misoperation during operation. When there is no skilled operator operating equipment B, misoperation may occur, causing losses to equipment B. When the equipment is operated by one person, if the operator is a skilled operator, losses due to unfamiliarity and misoperation will not occur. If the operator is an unskilled operator, there are two situations. The first situation is that with the permission given by the skilled operator operating equipment B, the unskilled operator practices operation of equipment B, which will cause losses during the practice operation process, and the risk value brought to the equipment is greater than the situation where the skilled operator operating equipment B is present. In this case, the skilled operator operating equipment B will appear later, and equipment B will be transformed from a single-operation situation to a multi-operation situation. The second situation is that the unskilled operator uses equipment B privately, and within the time threshold, equipment B does not change from a single-operation situation to a multi-operation situation, and the losses brought to the equipment will be greater than the first situation. In both the first and second situations, the time for the unskilled operator to practice operation of equipment B increases.

[0025] S51, calculate the risk value f of device B B , f B =W 1 T 1 +W 2 T 2 +W 3 T 3 , W 1 , W 2 and W 3 is the weight, if the risk value is greater than or equal to the risk threshold f thr, then equipment B needs to be repaired, and after the repair is completed, the time T for equipment B to be operated by skilled personnel 1 , Time for unskilled personnel to practice operation T 2 and the time T for unskilled personnel to operate 3 Reset to zero and start increasing again. If the risk value is less than or equal to the risk threshold f thr , the risk value of device B is monitored.

[0026] Risk threshold f thr , weight W 1 , W 2 and W 3 This is determined by the following steps:

[0027] S60, obtain the maintenance record of equipment B, find the time when equipment B is operated by a skilled person, the time when an unskilled person practices operation, and the time when the unskilled person operates, and record the combination of the time when equipment B is operated by a skilled person, the time when the unskilled person practices operation, and the time when the unskilled person operates as data point C according to the maintenance record 1 , C 2 ,…,C m , m is the number of maintenance records of equipment B; 1 , C 2 ,…,C m Perform clustering operation to obtain cluster clusters;

[0028] S61, in the same cluster, randomly select 3 different data points, and make a difference between each element between them to obtain 3 sets of difference values: and Establish a system of three linear equations for W 1 , W 2 and W 3 Solve Traverse all data points in the same cluster and get W 1 , W 2 and W 3 The average value of ; in all maintenance records of equipment B, not limited to the data points in one cluster, using the time when equipment B is operated by skilled personnel, the time when unskilled personnel practice operation, the time when unskilled personnel operate, the weight W 1 , W 2 and W 3 The average value of the risk value of each maintenance record is calculated according to the risk value calculation formula. After removing the abnormal values ​​from the risk values ​​of all maintenance records, the minimum value is taken as the risk threshold f thr ; After obtaining the risk threshold, use the weight W 1 , W 2 and W 3The average value of and the data of equipment B that have not been repaired are used to determine the number of times the maintenance data is mistakenly judged as requiring maintenance;

[0029] S62, based on the number of clusters, multiple groups of W are obtained 1 , W 2 and W 3 The average value of each group W 1 , W 2 and W 3 A risk threshold f is obtained in the same way as in step S61. thr , find W 1 Maximum and minimum values ​​of the average value, W 2 Maximum and minimum values ​​of the average value and W 3 The maximum and minimum values ​​of the average value, for W 1 , W 2 and W 3 Using cross validation within the maximum and minimum values ​​of the average value, we can find the number of times that equipment B is misjudged as requiring maintenance. 1 , W 2 and W 3 The combination of 1 , W 2 and W 3 The final value of W 1 , W 2 and W 3 After the final value is obtained, the risk threshold f is obtained in the same way as in step S61. thr The final value of .

[0030] The time T that equipment B is operated by skilled personnel 1 , Time for unskilled personnel to practice operation T 2 and the time T for unskilled personnel to operate 3 It also reflects the running time of the equipment. The longer the running time of the equipment, the higher the T 1 , T 2 and T 3 In order to find the problems of device B as much as possible, the risk threshold f thr Set as the minimum value of the risk value of all maintenance records after removing abnormal values; T 1 , T 2 and T 3 Normalization operations are also performed to eliminate the impact of scale. In the same cluster, the similarity between data points is high, although the T 1 , T 2 and T 3There are differences between them, but the results of the differences are the same: Equipment B needs to be repaired. It is considered that the risk values ​​of data points in the same cluster are the same. After taking the data of three data points and making a difference, three equations are obtained to calculate W. 1 , W 2 and W 3 Solve; the data points of different clusters are subtracted to obtain W 1 , W 2 and W 3 The difference will be large, and cross-validation is used to find W 1 , W 2 and W 3 The threshold is set as the minimum value of the risk value of all maintenance records after removing the abnormal value, and the W 1 , W 2 and W 3 The indicator of the combined effect becomes the number of times that the data of equipment B that has not been maintained is mistakenly judged as the number of times that maintenance is required.

[0031] In another aspect of the present invention, a target recognition and detection system based on a neural network is provided, comprising: a monitoring video acquisition module, a monitoring image acquisition module, a neural network recognition module, a data storage module, a risk value analysis module and an equipment maintenance module, wherein the output end of the monitoring video acquisition module is connected to the input end of the monitoring image acquisition module and the input end of the data storage module, so as to acquire the monitoring video in the production process; the output end of the monitoring image acquisition module is connected to the input end of the data storage module, so as to decompose and process the monitoring video generated by the monitoring video acquisition module to obtain the monitoring image of each frame; the output end of the neural network recognition module is connected to the input end of the risk value analysis ... The personnel identity and equipment status in the control image are identified to obtain the time when a single person operates the equipment and the time when multiple people operate the equipment; the output end of the data storage module is connected to the input end of the neural network recognition module and the input end of the risk value analysis module, and is used to store the monitoring video, monitoring image and equipment maintenance records in the production process; the output end of the risk value analysis module is connected to the input end of the equipment maintenance module, and the risk value of the equipment is determined based on the time when the equipment is operated by skilled personnel, the time when unskilled personnel practice operation and the time when unskilled personnel operate, and when the equipment needs to be repaired, the equipment information and maintenance requirement information are sent to the equipment maintenance module; the equipment maintenance module is used to contact maintenance personnel to repair the equipment.

[0032] The neural network recognition module also includes a first feature extraction unit, a personnel neural network unit, a second feature extraction unit, a device neural network unit, a device operation time determination unit and a personnel operation time determination unit; the first feature extraction unit is used to find a monitoring image with personnel in the monitoring image, and extract the physical characteristics and identity characteristics of the personnel from the monitoring image with personnel; the personnel neural network unit takes the physical characteristics and identity characteristics as input, and the personnel ID as output, trains the personnel neural network model, and identifies the identity of the personnel appearing in the monitoring image, and the identity characteristics refer to the certificate recording the employee's identity information; the second feature extraction unit finds the monitoring image about the equipment from the monitoring image, and extracts the action and behavior characteristics of the personnel from the monitoring image; the device neural network unit takes the action and behavior characteristics of the personnel as input, and the status of the equipment as output, trains the device neural network model, and identifies the status of the equipment in the monitoring image, and the status of the equipment includes three situations: unmanned operation, single-person operation and multi-person operation;

[0033] The device operation time determination unit is used to determine the time when the device B is operated by a skilled person, the time when the unskilled person practices the operation, and the time when the unskilled person operates the device B. When the device B is operated by multiple people, if there are both skilled and unskilled persons among the persons operating the device B, or there are only skilled persons among the persons operating the device B, the time when the device B is operated by the skilled person increases by T. group , if there are only unskilled personnel among the personnel operating equipment B, the time equipment B is operated by unskilled personnel increases by T group , T group is the time when device B is operated by multiple people; when device B is operated by one person, based on the identification result of the operator, if the operator is a skilled operator, the time when device B is operated by the skilled operator is increased by T indv ; If the operator is unskilled and at time T thr Within 1 year, if the equipment B is operated by one person and then by multiple people with skilled personnel, the time for unskilled personnel to practice operating the equipment B will increase by T. indv1 Otherwise, the time when equipment B is operated by unskilled personnel increases by T indv2 ; T thr is the preset time threshold, T indv1 T is the time when device B is in single-operation state when it changes from single-operation state to multi-operation state. indv2 and T indv For device B at time T thr The time when device B is in single-operation state and does not change from single-operation state to multi-operation state;

[0034] The personnel operation time determination unit is used to determine the operation time of personnel P on device B. When personnel P is a skilled operator and device B is operated by personnel P, the operation time of personnel P on device B increases. The operation of device B by personnel P includes two situations: the device B is operated by personnel P alone and the device B is operated by multiple persons including personnel P. If device B is operated by multiple persons including personnel P, in addition to the increase in the operation time of personnel P on device B, group Among multiple people operating equipment B, including personnel P, the operating time of equipment B for a skilled person is increased by T group For unskilled personnel operating equipment B, the time for practicing the operation of equipment B is increased by T group If equipment B is operated by person P alone and person P is a skilled operator of equipment B, the operating time of person P on equipment B increases by T indv ; If equipment B is operated by person P alone and person P is an unskilled operator of equipment B, the time for person P to practice operating equipment B should be increased.

[0035] The risk value analysis module also includes an equipment risk value calculation and evaluation unit and a personnel proficiency calculation and evaluation unit. The equipment risk value calculation and evaluation unit determines the risk value of the equipment based on the time the equipment is operated by skilled personnel, the time the equipment is practiced, and the time the equipment is operated by unskilled personnel. The risk threshold of the equipment is determined based on the historical maintenance record of the equipment. The risk value of the equipment and the risk threshold are compared to determine whether the equipment needs maintenance. If so, the equipment information and maintenance requirement information are sent to the equipment maintenance module, and the risk value of the equipment is reset to zero after the maintenance is completed; if not, the risk value of the equipment is continuously monitored. The personnel proficiency calculation and evaluation unit obtains the personnel's proficiency in operating the equipment based on the personnel's operation time and the personnel's practice time.

[0036] The data storage module stores equipment information, information about personnel with independent operation authority for the equipment, and information about skilled personnel operating the equipment, and uses maintenance records to store maintenance equipment information, the time the equipment was operated by skilled personnel before maintenance, the time unskilled personnel practiced operation, and the time the unskilled personnel operated.

[0037] Compared with the prior art, the beneficial effects achieved by the present invention are: determining whether personnel can skillfully operate the target equipment based on the personnel's operation and practice time of the equipment; retrieving the time data in the maintenance record, judging whether the equipment needs maintenance based on the operation time of the equipment by different personnel, providing a reference for equipment maintenance, and improving the effect of equipment maintenance; using neural network technology to identify equipment and personnel, without the need for human intervention, and can adapt to industrial production scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0039] Figure 1 It is a structural schematic diagram of a target recognition and detection system based on a neural network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] In the embodiments of the present invention, please refer to Figure 1 , a structural schematic diagram of a target recognition and detection system based on a neural network is provided, comprising: a monitoring video acquisition module, a monitoring image acquisition module, a neural network recognition module, a data storage module, a risk value analysis module and an equipment maintenance module, wherein the output end of the monitoring video acquisition module is connected to the input end of the monitoring image acquisition module and the input end of the data storage module, and is used to acquire the monitoring video in the production process; the output end of the monitoring image acquisition module is connected to the input end of the data storage module, and is used to decompose and process the monitoring video generated by the monitoring video acquisition module to obtain a monitoring image of each frame; the output end of the neural network recognition module is connected to the input end of the risk value analysis module, and is used to decompose and process the monitoring video generated by the monitoring video acquisition module to obtain a monitoring image of each frame; the output end of the neural network recognition module is connected to the input end of the risk value analysis module, and is used to decompose and process the monitoring image generated by the monitoring image ... The personnel identity and equipment status in the image are identified to obtain the time when a single person operates the equipment and the time when multiple people operate the equipment; the output end of the data storage module is connected to the input end of the neural network recognition module and the input end of the risk value analysis module, and is used to store monitoring videos, monitoring images and equipment maintenance records in the production process; the output end of the risk value analysis module is connected to the input end of the equipment maintenance module, and the risk value of the equipment is determined based on the time when the equipment is operated by skilled personnel, the time when unskilled personnel practice operation and the time when unskilled personnel operate, and when the equipment needs to be repaired, the equipment information and maintenance requirement information are sent to the equipment maintenance module; the equipment maintenance module is used to contact maintenance personnel to repair the equipment.

[0042] The neural network recognition module also includes a first feature extraction unit, a personnel neural network unit, a second feature extraction unit, a device neural network unit, a device operation time determination unit and a personnel operation time determination unit; the first feature extraction unit is used to find a monitoring image with personnel in the monitoring image, and extract the physical characteristics and identity characteristics of the personnel from the monitoring image with personnel; the personnel neural network unit takes the physical characteristics and identity characteristics as input, and the personnel ID as output, trains the personnel neural network model, and identifies the identity of the personnel appearing in the monitoring image, and the identity characteristics refer to the certificate recording the employee's identity information; the second feature extraction unit finds the monitoring image about the equipment from the monitoring image, and extracts the action and behavior characteristics of the personnel from the monitoring image; the device neural network unit takes the action and behavior characteristics of the personnel as input, and the status of the equipment as output, trains the device neural network model, and identifies the status of the equipment in the monitoring image, and the status of the equipment includes three situations: unmanned operation, single-person operation and multi-person operation;

[0043] The device operation time determination unit is used to determine the time when the device B is operated by a skilled person, the time when the unskilled person practices the operation, and the time when the unskilled person operates the device B. When the device B is operated by multiple people, if there are both skilled and unskilled persons among the persons operating the device B, or there are only skilled persons among the persons operating the device B, the time when the device B is operated by the skilled person increases by T. group , if there are only unskilled personnel among the personnel operating equipment B, the time equipment B is operated by unskilled personnel increases by T group , T group is the time when device B is operated by multiple people; when device B is operated by one person, based on the identification result of the operator, if the operator is a skilled operator, the time when device B is operated by the skilled operator is increased by T indv ; If the operator is unskilled and at time T thr Within 1 year, if the equipment B is operated by one person and then by multiple people with skilled personnel, the time for unskilled personnel to practice operating the equipment B will increase by T. indv1 Otherwise, the time when equipment B is operated by unskilled personnel increases by T indv2 ; T thr is the preset time threshold, T indv1 T is the time when device B is in single-operation state when it changes from single-operation state to multi-operation state. indv2 and T indv For device B at time T thr The time when device B is in single-operation state and does not change from single-operation state to multi-operation state;

[0044] The personnel operation time determination unit is used to determine the operation time of personnel P on device B. When personnel P is a skilled operator and device B is operated by personnel P, the operation time of personnel P on device B increases. The operation of device B by personnel P includes two situations: the device B is operated by personnel P alone and the device B is operated by multiple persons including personnel P. If device B is operated by multiple persons including personnel P, in addition to the increase in the operation time of personnel P on device B, group Among multiple people operating equipment B, including personnel P, the operating time of equipment B for a skilled person is increased by T group For unskilled personnel operating equipment B, the time for practicing the operation of equipment B is increased by T group If equipment B is operated by person P alone and person P is a skilled operator of equipment B, the operating time of person P on equipment B increases by T indv ; If equipment B is operated by person P alone and person P is an unskilled operator of equipment B, the time for person P to practice operating equipment B should be increased.

[0045] The risk value analysis module also includes an equipment risk value calculation and evaluation unit and a personnel proficiency calculation and evaluation unit. The equipment risk value calculation and evaluation unit determines the risk value of the equipment based on the time the equipment is operated by skilled personnel, the time the equipment is practiced, and the time the equipment is operated by unskilled personnel. The risk threshold of the equipment is determined based on the historical maintenance record of the equipment. The risk value of the equipment and the risk threshold are compared to determine whether the equipment needs maintenance. If so, the equipment information and maintenance requirement information are sent to the equipment maintenance module, and the risk value of the equipment is reset to zero after the maintenance is completed; if not, the risk value of the equipment is continuously monitored. The personnel proficiency calculation and evaluation unit obtains the personnel's proficiency in operating the equipment based on the personnel's operation time and the personnel's practice time.

[0046] The data storage module stores equipment information, information about personnel with independent operation authority for the equipment, and information about skilled personnel operating the equipment, and uses maintenance records to store maintenance equipment information, the time the equipment was operated by skilled personnel before maintenance, the time unskilled personnel practiced operation, and the time the unskilled personnel operated.

[0047] In an embodiment of the present invention, a target recognition and detection method based on a neural network is provided, comprising:

[0048] S10, acquiring a monitoring video of an industrial production process, obtaining a monitoring image of each frame based on the monitoring video, and using the monitoring image to train a neural network model for personnel recognition and equipment recognition; including steps S20 and S21:

[0049] S20, finding a surveillance image in which a person exists from the surveillance image, extracting the person's physical features and identity features from the surveillance image in which the person exists, taking the physical features and identity features as input and the person's ID as output, training a person neural network model for identifying the identity of the person appearing in the surveillance image, wherein the identity features refer to the certificate recording the employee's identity information;

[0050] S21, find a surveillance image about the device from the surveillance image, extract the movement and behavior characteristics of the personnel from the surveillance image, take the movement and behavior characteristics of the personnel as input, and the status of the device as output, train the device neural network model for identifying the device status in the surveillance image, wherein the status of the device includes three situations: unmanned operation, single-person operation, and multi-person operation; when the device is identified as being operated by a single person, determine the time when the single person operates the device, and use the personnel neural network model to determine the identity of the person operating the device; when the device is identified as being operated by multiple people, determine the time when the multiple people operate the device, and use the personnel neural network model to determine the identity of the person operating the device.

[0051] The time can be determined according to the number of frames. The time of one frame is 1 / FP, where FP is the frame rate. When the frame rate is 30FPS and the number of frames for single-player operation is 6000, the single-player operation lasts for 6000 / 30=200 seconds.

[0052] S11, based on the monitoring images during the production process, obtaining the operation time of the personnel on the equipment, and based on the operation time of the personnel on the equipment and the practice operation time, obtaining the operation proficiency of the personnel on the equipment, including steps SS30 to S33:

[0053] S30, based on the identity information of person P, determine the independent operation authority of person P for each device, where the independent operation authority refers to the authority of person P to operate the device independently without the need for guidance or supervision by other persons except P; based on the independent operation authority of all persons including P for each device, obtain the person with independent operation authority for device B;

[0054] S31, obtaining the operation time of the person P on each device from the single-person operation situation in the historical monitoring image; obtaining the practice operation time of the person P on each device from the multi-person operation situation in the historical monitoring image;

[0055] S32, for device B, find a person who has independent operation authority for device B, and determine whether the person who has independent operation authority for device B has the ability to operate other devices A with lower operation requirements than B. 1 , A 2 , …, A n The operation time on B, n is the number of other devices with lower operation requirements than B, the other devices A1 , A 2 , …, A n The condition that the operation time of all personnel with independent operation authority for device B is not 0 is met;

[0056] For device B, personnel may need to have other device A 1 , A 2 , A 3 The personnel who are granted independent operation rights for device B may have experience in operating device C, which is more complex than device B. This is not necessary to consider. Consider that the personnel who have independent operation rights for device B have experience in operating device A, which is not 0. 1 , A 2 , A 3 , if there is device A 4 , and there is a person with independent operation authority on device B who has access to device A 4 The operation time is 0, indicating that the operation device A 4 Experience is not necessary to operate device B; if device B does not require experience in operating other devices, then the operator's proficiency in device B is only related to the time the operator practices operating device B.

[0057] S33, obtaining the information of person P about other equipment A 1 , A 2 , …, A n Operation time t 1 ,t 2 ,…,t n , calculate the impact of personnel P on other equipment A 1 , A 2 , …, A n Operation score y 1 ,y 1 =a 1 t 1 +a 2 t 2 +…+a n t n ; Get the time y that person P practices operating equipment B 2 , based on y 1 and 2 Get the operator P's proficiency Y for equipment B, Y = b 1 y 1 +b 2 y 2 , where a 1 、a 2 , …, a n 、b 1 and b 2is the coefficient; obtain the information of all personnel who have independent operation authority on device B and other devices A 1 , A 2 , …, A n The total operation time y s1 ,t s2 ,…,t sn , based on t s1 ,t s2 ,…,t sn Get a 1 、a 2 , …, a n , Get the total time y spent practicing operation of device B by all personnel who have independent operation rights on device B s2 , then b 1 =1-b 2 , y s1 =a 1 t s1 +a 2 t s2 +…+a n t sn ,y s1 For personnel who have independent operation authority over device B, they can control other devices A as a whole. 1 , A 2 , …, A n Operation score;

[0058] Three people with independent operation authority over device B can control other devices A. 1 , A 2 , A 3 The operation times are (10, 5, 2), (8, 4, 2) and (6, 3, 1). When the three personnel with independent operation rights for device B practice time 1, 1, 1, they first perform normalization operations, 10, 8 and 6 are normalized, 5, 4 and 3 are normalized, 2, 2 and 1 are normalized, and 1, 1 and 1 are normalized. After normalization is completed, the other devices A are normalized. 1 , A 2 , A 3 The operation time is added together to get t s1 ,t s2 ,t s3 , add the training time for device B to get y s2 According to S33, we can get a 1 、a 2 、a 3 、b 1 and b 2 , the time unit is hour.

[0059] Determine whether person P is a skilled operator of device B: If person P's proficiency Y in operating device B is greater than or equal to the proficiency threshold Y thr , then person P is a skilled person in operating device B, otherwise person P is an unskilled person in operating device B; calculate the operating proficiency of all personnel with independent operating authority for device B to find the minimum value after removing outliers, and use the minimum value after removing outliers as the proficiency threshold Y thr The outliers of operation proficiency can be removed by using methods such as box plot method, mean discrimination method, and isolation forest algorithm.

[0060] S12, monitor the equipment in real time, analyze the monitoring image of the equipment, obtain the risk value of the equipment based on the operator's proficiency in operating the equipment and the operating time, and when the risk value of the equipment is greater than or equal to the threshold, contact the maintenance personnel to repair the equipment, which specifically includes steps S50 and S51:

[0061] S50, obtaining the time T during which the device B is operated by a skilled person 1 , Time for unskilled personnel to practice operation T 2 and the time T for unskilled personnel to operate 3 When equipment B is operated by multiple people, if there are both skilled and unskilled personnel operating equipment B, or only skilled personnel operating equipment B, the time equipment B is operated by skilled personnel increases by T group , if there are only unskilled personnel among the personnel operating equipment B, the time equipment B is operated by unskilled personnel increases by T group , T group is the time when device B is operated by multiple people; when device B is operated by one person, based on the identification result of the operator, if the operator is a skilled operator, the time when device B is operated by the skilled operator increases; if the operator is an unskilled operator and at time T thr Within 1 year, if the equipment B is operated by one person and then by multiple people with skilled personnel, the time for unskilled personnel to practice operating the equipment B will increase by T. indv1 Otherwise, the time when equipment B is operated by unskilled personnel increases by T indv2 ; T thr is the preset time threshold, T indv1 T is the time when device B is in single-operation state when it changes from single-operation state to multi-operation state. indv2 For device B at time T thr The time when device B is in single-operation state and does not change from single-operation state to multi-operation state;

[0062] In the case of single-person operation, the threshold Tthr When the time is set to 20 minutes, the unskilled operator P who operates equipment B at the beginning 1 After operating equipment B for 10 minutes, the skilled operator P 2 Appear with P 1 Operate device B together, then P 1 The time for practicing the operation of equipment B is increased by 10 minutes, and the time for unskilled personnel to practice the operation of equipment B is T 2 Add ten minutes; if P 2 No show within 20 minutes, or P 2 is an unskilled person operating equipment B, then the time T of equipment B being operated by the unskilled person 3 Add device B to P 1 Single-person operation time T indv2 , at this time P 1 The time for practicing operation of device B also increases by T indv2 ; Unskilled person P operating equipment B 2 When it occurs, it is converted into a multi-person operation situation to analyze the time.

[0063] Get the risk threshold f thr , weight W 1 , W 2 and W 3 , comprising steps S60 to S62:

[0064] S60, obtain the maintenance record of equipment B, find the time when equipment B is operated by a skilled person, the time when an unskilled person practices operation, and the time when the unskilled person operates, and record the combination of the time when equipment B is operated by a skilled person, the time when the unskilled person practices operation, and the time when the unskilled person operates as data point C according to the maintenance record 1 , C 2 ,…,C m , m is the number of maintenance records of equipment B; 1 , C 2 ,…,C m Perform clustering operation to obtain cluster clusters;

[0065] S61, in the same cluster, randomly select 3 different data points, and make a difference between each element between them to obtain 3 sets of difference values: and Establish a system of three linear equations for W 1 , W 2 and W 3 Solve Traverse all data points in the same cluster and get W 1 , W 2 and W 3The average value of ; in all maintenance records of equipment B, not limited to the data points in one cluster, using the time when equipment B is operated by skilled personnel, the time when unskilled personnel practice operation, the time when unskilled personnel operate, the weight W 1 , W 2 and W 3 The average value of the risk value of each maintenance record is calculated according to the risk value calculation formula. After removing the abnormal values ​​from the risk values ​​of all maintenance records, the minimum value is taken as the risk threshold f thr ; After obtaining the risk threshold, use the weight W 1 , W 2 and W 3 The average value of and the data of equipment B that have not been repaired are used to determine the number of times the maintenance data is mistakenly judged as requiring maintenance;

[0066] S62, based on the number of clusters, multiple groups of W are obtained 1 , W 2 and W 3 The average value of each group W 1 , W 2 and W 3 A risk threshold f is obtained in the same way as in step S61. thr , find W 1 Maximum and minimum values ​​of the average value, W 2 Maximum and minimum values ​​of the average value and W 3 The maximum and minimum values ​​of the average value, for W 1 , W 2 and W 3 Using cross validation within the maximum and minimum values ​​of the average value, we can find the number of times that equipment B is misjudged as requiring maintenance. 1 , W 2 and W 3 The combination of 1 , W 2 and# 3 The final value of 1 、# 2 and W 3 After the final value is obtained, the risk threshold f is obtained in the same way as in step S61. thr The final value of

[0067] Multiple groups of W 1 , W 2 and W 3 The average values ​​of all have maximum and minimum values, which may not appear simultaneously in a group. 1min ,W 1max ]、[W 2min ,W 2max ]、[W 3min ,W3max ] Set 3 steps and use cross validation to find W 1 , W 2 and W 3 The final value of the step size can be set as the quotient of the interval length and the constant constant, such as W 1 The step size can be W 1min , W 1max W 1 The minimum and maximum values ​​of the average value, W 2min , W 2max W 2 The minimum and maximum values ​​of the average value, W 3min , W 3max W 3 The minimum and maximum values ​​of the average.

[0068] S51, calculate the risk value f of device B B , f B =W 1 T 1 +W 2 T 2 +W 3 T 3 , W 1 , W 2 and W 3 is the weight, if the risk value is greater than or equal to the risk threshold f thr , then equipment B needs to be repaired, and after the repair is completed, the time T for equipment B to be operated by skilled personnel 1 , Time for unskilled personnel to practice operation T 2 and the time T for unskilled personnel to operate 3 Reset to zero and start increasing again. If the risk value is less than or equal to the risk threshold f thr , the risk value of device B is monitored.

[0069] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0070] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A target recognition and detection method based on a neural network, characterized in that: The following steps are involved: S10, acquiring a monitoring video of the industrial production process, obtaining a monitoring image of each frame based on the monitoring video, and using the monitoring image to train a neural network model for personnel recognition and equipment recognition; The use of monitoring images to train the neural network model also The method comprises the following steps S20 and S21: S20, finding a surveillance image in which a person exists from the surveillance image, extracting the person's physical features and identity features from the surveillance image in which the person exists, taking the physical features and identity features as input and the person's ID as output, training a person neural network model for identifying the identity of the person appearing in the surveillance image, wherein the identity features refer to the certificate recording the employee's identity information; S21, finding a surveillance image about the device from the surveillance image, extracting the motion and behavior features of the person from the surveillance image, taking the motion and behavior features of the person as input and the state of the device as output, training a device neural network model for identifying the state of the device in the surveillance image, wherein the state of the device includes three situations: unmanned operation, single-person operation, and multi-person operation; when the device is identified as being operated by a single person, determining the time when the single person operates the device, and using the person neural network model to determine the identity of the person operating the device; when the device is identified as being operated by multiple people, determining the time when the multiple people operate the device, and using the person neural network model to determine the identity of the person operating the device; S11, based on the monitoring images during the production process, obtaining the operation time of the personnel on the equipment, and based on the operation time of the personnel on the equipment and the practice operation time, obtaining the operation proficiency of the personnel on the equipment, including steps S30 to S33: S30, based on the identity information of person P, determine the independent operation authority of person P for each device, where the independent operation authority refers to the authority of person P to operate the device independently without the need for guidance or supervision by other persons except P; based on the independent operation authority of all persons including P for each device, obtain the person with independent operation authority for device B; S31, obtaining the operation time of person P on each device from the single-person operation situation in the historical monitoring image; From the multi-person operation situation in the historical monitoring images, obtain the time for person P to practice operation of each device; S32, for device B, find a person who has independent operation authority for device B, and determine whether the person who has independent operation authority for device B has the ability to operate other devices A1, A2, ..., A3, etc. with lower operation requirements than B. n The operation time on B is n, and the number of other devices with lower operation requirements than B is n, and the other devices A1, A2, ..., A n The condition that the operation time of all personnel with independent operation authority for device B is not 0 is met; S33, obtaining the information of person P on other devices A1, A2, ..., A n The operation time t1, t2, ..., t n , calculate the personnel P's influence on other equipment A1, A2, ..., A n The operation score y1, y1 = a1t1 + a2t2 + ... + a n t n ; Get the time y2 that personnel P practices to operate equipment B, and based on y1 and y2, get the operating proficiency Y of personnel P on equipment B, Y = b1y1 + b2y2, where a1, a2, ..., a n , b1 and b2 are coefficients; if the operator P's operating proficiency Y of equipment B is greater than or equal to the proficiency threshold Y thr , then person P is a skilled person operating equipment B, otherwise person P is an unskilled person operating equipment B; S12, monitor the equipment in real time, analyze the monitoring image of the equipment, obtain the risk value of the equipment based on the operator's proficiency in operating the equipment and the operating time, and when the risk value of the equipment is greater than or equal to the threshold, contact the maintenance personnel to repair the equipment; specifically includes steps S50 and S51: S50, obtaining the time T1 of the skilled operator operating the device B, the time T2 of the unskilled operator practicing the operation, and the time T3 of the unskilled operator operating the device B. When the device B is operated by multiple people, if there are both skilled and unskilled operators among the operators of the device B, or there are only skilled operators among the operators of the device B, the time T1 of the skilled operator operating the device B is increased by group , if there are only unskilled personnel among the personnel operating equipment B, the time equipment B is operated by unskilled personnel increases by T group , T group The time when device B is operated by multiple people; When equipment B is operated by one person, based on the identification result of the operator, if the operator is a skilled operator, the time equipment B is operated by the skilled operator increases; if the operator is an unskilled operator and at time T thr Within 1 year, if the equipment B is operated by one person and then by multiple people with skilled personnel, the time for unskilled personnel to practice operating the equipment B will increase by T. indv1 Otherwise, the time when equipment B is operated by unskilled personnel increases by T indv2 ; T thr is the preset time threshold, T indv1 T is the time when device B is in single-operation state when it changes from single-operation state to multi-operation state. indv2 For device B at time T thr The time when device B is in single-operation state and does not change from single-operation state to multi-operation state; S51, calculate the risk value f of device B B , f B =W1T1+W2T2+W3T3, W1, W2 and W3 are weights. If the risk value is greater than or equal to the risk threshold f thr , then equipment B needs to be repaired. After the repair is completed, the time T1 for skilled personnel to operate equipment B, the time T2 for unskilled personnel to practice operation, and the time T3 for unskilled personnel to operate equipment B are reset to zero and start to increase again. If the risk value is less than or equal to the risk threshold f thr , the risk value of device B is monitored.

2. A target recognition and detection method based on a neural network according to claim 1, characterized in that: The coefficients a1, a2, ..., a n , b1 and b2 are obtained by the following steps: S40, obtaining the permissions of all personnel who have independent operation rights on device B to operate other devices A1, A2, ..., A n The total operation time t s1 ,t s2 ,…,t sn , based on t s1 ,t s2 ,…,t sn Get a1, a2, ..., a n , S41, obtain the total time y of all personnel with independent operation authority for device B practicing operation of device B s2 , then b1=1-b2, y s1 =a1t s1 +a2t s2 +…+a n t sn ,y s1 For personnel who have independent operation authority over device B, they can control other devices A1, A2, ..., A n Operation score; Determine whether person P is a skilled operator of device B: If person P's proficiency Y in operating device B is greater than or equal to the proficiency threshold Y thr , then person P is a skilled person operating equipment B, otherwise person P is an unskilled person operating equipment B; Calculate the proficiency of all personnel with independent operation rights on device B, find the minimum value after removing outliers, and use the minimum value after removing outliers as the proficiency threshold Y thr .

3. A target recognition and detection method based on a neural network according to claim 2, characterized in that: The risk threshold f thr , weights W1, W2 and W3 are determined by the following steps: S60, obtaining the maintenance record of equipment B, finding the time when equipment B is operated by a skilled person, the time when an unskilled person practices operation, and the time when the unskilled person operates, and recording the combination of the time when equipment B is operated by a skilled person, the time when the unskilled person practices operation, and the time when the unskilled person operates as data points C1, C2, ..., C m , m is the number of maintenance records of equipment B; for C1, C2, ..., C m Perform clustering operation to obtain cluster clusters; S61, in the same cluster, randomly select 3 different data points, and make a difference between each element between them to obtain 3 sets of difference values: and Establish a system of three linear equations to solve W1, W2 and W3 Traverse all data points in the same cluster to get the average value of W1, W2 and W3; in all maintenance records of equipment B, not limited to data points in one cluster, use the average value of the time when equipment B is operated by skilled personnel, the time when unskilled personnel practice operation, the time when unskilled personnel operate, the weights W1, W2 and W3 to calculate the risk value of each maintenance record according to the risk value calculation formula, remove the outliers from the risk values ​​of all maintenance records and take the minimum value as the risk threshold f thr ; After obtaining the risk threshold, the average value of weights W1, W2, and W3 and the data of equipment B that has not been repaired are used to determine the number of times the maintenance data is misjudged as requiring maintenance; S62, based on the number of clusters, obtain the average values ​​of multiple groups W1, W2 and W3, and obtain a risk threshold f for each group W1, W2 and W3 in the same way as in step S61 thr , find the maximum and minimum values ​​of the average value of W1, the maximum and minimum values ​​of the average value of W2, and the maximum and minimum values ​​of the average value of W3, use cross-validation method for W1, W2 and W3 within the maximum and minimum values ​​of the average value, find the combination of W1, W2 and W3 for which the data of equipment B that has not been repaired is mistakenly judged as requiring the least number of repairs, and use it as the final value of W1, W2 and W3. After obtaining the final values ​​of W1, W2 and W3, obtain the risk threshold f in the same way as in step S61 thr The final value of .

4. A target recognition and detection system based on a neural network, using a target recognition and detection method based on a neural network as claimed in any one of claims 1 to 3, characterized in that: include: A monitoring video acquisition module, a monitoring image acquisition module, a neural network recognition module, a data storage module, a risk value analysis module and an equipment maintenance module, wherein the output end of the monitoring video acquisition module is connected to the input end of the monitoring image acquisition module and the input end of the data storage module, and is used to acquire the monitoring video in the production process; the output end of the monitoring image acquisition module is connected to the input end of the data storage module, and is used to decompose and process the monitoring video generated by the monitoring video acquisition module to obtain the monitoring image of each frame; the output end of the neural network recognition module is connected to the input end of the risk value analysis module, and is used to identify the identity of the personnel and the equipment status in the monitoring image, and obtain the time when a single person operates the equipment and the time when multiple people operate the equipment; The output end of the data storage module is connected to the input end of the neural network identification module and the input end of the risk value analysis module, and is used to store monitoring videos, monitoring images and equipment maintenance records in the production process; the output end of the risk value analysis module is connected to the input end of the equipment maintenance module, and the risk value of the equipment is determined based on the time when the equipment is operated by skilled personnel, the time when unskilled personnel practice operation and the time when unskilled personnel operate the equipment. When the equipment needs to be repaired, the equipment information and maintenance requirement information are sent to the equipment maintenance module; the equipment maintenance module is used to contact maintenance personnel to repair the equipment.

5. A target recognition and detection system based on a neural network according to claim 4, characterized in that: The neural network recognition module also includes a first feature extraction unit, a personnel neural network unit, a second feature extraction unit, a device neural network unit, a device operation time determination unit and a personnel operation time determination unit; the first feature extraction unit is used to find a monitoring image with a person in the monitoring image, and extract the person's physical features and identity features from the monitoring image with the person; the personnel neural network unit uses the physical features and identity features as input and the person's ID as output, trains a personnel neural network model, and identifies the identity of the person appearing in the monitoring image, and the identity feature refers to a certificate that records the employee's identity information; the second feature extraction unit finds the monitoring image about the device from the monitoring image, and extracts the action and behavior features of the person from the monitoring image; The equipment neural network unit takes the action and behavior characteristics of the personnel as input and the status of the equipment as output, trains the equipment neural network model, and recognizes the equipment status in the monitoring image. The equipment status includes three situations: unmanned operation, single-person operation, and multi-person operation; The device operation time determination unit is used to determine the time when the device B is operated by a skilled person, the time when the unskilled person practices the operation, and the time when the unskilled person operates the device B. When the device B is operated by multiple people, if there are both skilled and unskilled persons among the persons operating the device B, or there are only skilled persons among the persons operating the device B, the time when the device B is operated by the skilled person increases by T. group , if there are only unskilled personnel among the personnel operating equipment B, the time equipment B is operated by unskilled personnel increases by T group , T group The time when device B is operated by multiple people; When equipment B is operated by a single person, based on the identification result of the operator, if the operator is a skilled operator, the time equipment B is operated by the skilled operator increases by T indv ; If the operator is an unskilled person and at time T thr Within 1 year, if the equipment B is operated by one person and then by multiple people with skilled personnel, the time for unskilled personnel to practice operating the equipment B will increase by T. indv1 Otherwise, the time when equipment B is operated by unskilled personnel increases by T indv2 ; T thr is the preset time threshold, T indv1 T is the time when device B is in single-operation state when it changes from single-operation state to multi-operation state. indv2 and T indv For device B at time T thr The time when device B is in single-operation state and does not change from single-operation state to multi-operation state; The personnel operation time determination unit is used to determine the operation time of personnel P on device B. When personnel P is a skilled operator and device B is operated by personnel P, the operation time of personnel P on device B increases. The operation of device B by personnel P includes two situations: the device B is operated by personnel P alone and the device B is operated by multiple persons including personnel P. If device B is operated by multiple persons including personnel P, in addition to the increase in the operation time of personnel P on device B, group Among multiple people operating equipment B, including personnel P, the operating time of equipment B for a skilled person is increased by T group For unskilled personnel operating equipment B, the time for practicing the operation of equipment B is increased by T group If equipment B is operated by person P alone and person P is a skilled operator of equipment B, the operating time of person P on equipment B increases by T indv ; If equipment B is operated by person P alone and person P is an unskilled operator of equipment B, the time for person P to practice operating equipment B should be increased.

6. A target recognition and detection system based on a neural network according to claim 4, characterized in that: The risk value analysis module also includes an equipment risk value calculation and evaluation unit and a personnel proficiency calculation and evaluation unit. The equipment risk value calculation and evaluation unit determines the risk value of the equipment based on the time the equipment is operated by skilled personnel, the time the equipment is practiced, and the time the equipment is operated by unskilled personnel. The risk threshold of the equipment is determined based on the historical maintenance record of the equipment. The risk value of the equipment and the risk threshold are compared to determine whether the equipment needs maintenance. If so, the equipment information and maintenance requirement information are sent to the equipment maintenance module, and the risk value of the equipment is reset to zero after the maintenance is completed; if not, the risk value of the equipment is continuously monitored. The personnel proficiency calculation and evaluation unit obtains the personnel's proficiency in operating the equipment based on the personnel's operation time and the personnel's practice time.

7. A target recognition and detection system based on a neural network according to claim 6, characterized in that: The data storage module stores equipment information, information about personnel with independent operation authority for the equipment, and information about skilled personnel operating the equipment, and uses maintenance records to store maintenance equipment information, the time the equipment was operated by skilled personnel before maintenance, the time unskilled personnel practiced operation, and the time the unskilled personnel operated.

Images