Automatic monitoring terminal, personnel working time estimation method and estimation device

Abstract

The present application discloses an Internet of Things automatic monitoring terminal for equipment operators, an estimation method and an estimation device for the working time of equipment operators, and belongs to the field of intelligent manufacturing and Internet of Things technology. The Internet of Things automatic monitoring terminal includes: a plug and a terminal body; the plug is connected to the terminal body; a switch module and an RFID card module are arranged on the plug; the terminal body is provided with a WiFi module, a Bluetooth module, a clock module, a watchdog module, a circuit status indicator module, a button module, a buzzer module, an external interface module, a debugging interface module, an RFID reading module, a three-phase voltage and current detection module, a main control chip module and a power conversion module. The three-phase voltage and current detection module can be used to read the equipment operation data such as the current, voltage, and power value of the monitored equipment, evaluate the working status of the equipment operator, and according to the punch-in information of the equipment operator collected by the RFID reading module, the working time of the equipment operator when on duty can be counted and output.

Description

Technical Field

[0001] The present application belongs to the technical field of intelligent manufacturing and the Internet of Things, and specifically relates to an Internet of Things automatic monitoring terminal for equipment operators, a method for estimating the working time of personnel, and an estimation device. Background Art

[0002] With the advent of the Internet of Things era, the industrial application field has begun to integrate various technologies and set off a new wave of industrial revolution, namely the Industrial 4.0 revolution. The purpose of the Industrial 4.0 revolution is to use IT technology to enhance the energy of traditional manufacturing industries, transform them into smart factories with fully automated production, and realize intelligent production, personalized customization and mass production. The core of the Industrial Internet of Things is to use communication technologies such as local networks or the Internet to connect sensors, controllers, personnel and objects in new ways, forming a network that connects people and objects, objects and objects, and realizes information and intelligence, thereby maximizing factory efficiency. At present, scholars focus on studying the application of various artificial intelligence algorithms in the Industrial Internet of Things. For example, Kan C et al. have made various improvements to the Industrial Internet of Things through machine learning and deep learning, which has improved the production efficiency and customer satisfaction of the factory. NhuVH et al. used deep learning algorithms for data analysis of identification, classification and decision-making in the Industrial Internet of Things, thereby providing efficient and accurate resource scheduling. However, existing research on the Industrial Internet of Things focuses on optimizing product production, but has not improved traditional infrastructure and cannot effectively supervise employee work efficiency.

[0003] However, in many traditional labor-intensive industries such as sewing machines, hosiery machines, and printing machines, there are a large number of equipment operators. At this time, the management of equipment operators still has the following problems: there are many equipment operators, and many of them still use the traditional clock-in and clock-out system to count the attendance of equipment operators, which has few functions. The superior administrators cannot accurately know the actual work situation of the equipment operators and effectively assign work to them, and cannot manage and supervise them accurately to individuals. This will lead to some lazy equipment operators who are lazy and take advantage of the situation, and their work efficiency is low, but they get the same salary as other personnel. Some equipment operators work actively and have a large workload, but their salary performance is difficult to reflect. The managers cannot assign other work in the first place, which also reduces their work efficiency. Therefore, for a large number of equipment operators, managers cannot accurately manage individuals, cannot accurately evaluate the work efficiency of equipment operators, and it is difficult to achieve a fair and just performance reward and punishment system, thereby reducing the work enthusiasm of equipment operators and the work efficiency of the company. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide an Internet of Things automatic supervision terminal for equipment operators, a method for estimating the working hours of equipment operators, and an evaluation device, which can solve the technical problem that for a large number of equipment operators, managers are unable to accurately manage individuals, cannot accurately evaluate the work efficiency of equipment operators, and find it difficult to implement a fair and just performance reward and punishment system, thereby reducing the work enthusiasm of equipment operators and the work efficiency of the company.

[0005] In order to solve the above technical problems, this application is implemented as follows:

[0006] In a first aspect, an embodiment of the present application provides an Internet of Things automatic supervision terminal for equipment operators, characterized in that it includes a plug and a terminal body;

[0007] The plug is connected to the terminal body;

[0008] The plug is provided with a switch module and an RFID card module;

[0009] The switch module is used to start the power supply;

[0010] The RFID card module is used to provide device ID;

[0011] The terminal body is equipped with a WiFi module, a Bluetooth module, a clock module, a watchdog module, a circuit status indicator module, a button module, a buzzer module, an external interface module, a debugging interface module, an RFID reading module, a three-phase voltage and current detection module, a main control chip module and a power conversion module;

[0012] WiFi module is used to realize wireless LAN communication;

[0013] The Bluetooth module is used to communicate with external devices;

[0014] The clock module is used to provide system clock;

[0015] The watchdog module is used to feed the dog at a scheduled time by program;

[0016] The circuit status indicator module is used to display different colors according to the circuit status;

[0017] The button module is used for reset and switch functions;

[0018] The buzzer module is used for alarm;

[0019] The external interface module is used to provide 5V 2A current to the outside and communicate;

[0020] The debugging interface module is used for program upgrade and debugging;

[0021] The RFID reading module is used to read the RFID tag of the equipment operator;

[0022] The three-phase voltage and current detection module is used to collect the real-time current, voltage and power of the equipment;

[0023] The main control chip module is used to control the equipment and process data;

[0024] The power conversion module is used to provide power to the device.

[0025] In a second aspect, an embodiment of the present application provides a method for estimating the working time of an equipment operator, characterized in that the method is applied to the Internet of Things automatic supervision terminal of the first aspect, and the estimation method includes:

[0026] S101: Install the IoT automatic monitoring terminal on the monitored equipment, read the equipment operation data M such as the current, voltage, and power value of the monitored equipment through the three-phase voltage and current detection module, and record the working status of the current equipment operator. When the equipment operator is in the working state, the working status value is recorded as 1, otherwise it is 0, and the 4×g value of the working time evaluation is obtained. s dimensional data set, where g s Indicates the number of elements in the data set, x n Represents the nth element in the data set;

[0027] S102: Obtain training set and test set;

[0028] S103: Training input weight matrix W A And the output weight matrix W B ;

[0029] S104: According to the equipment operation data M, input weight matrix W A And the output weight matrix W B , calculate the work evaluation result matrix R of the current data through formula 1, and calculate the work evaluation result r of the current data through formula 2 l ;

[0030] R=W B ·g(W A M T )Formula 1

[0031] Among them, g() represents the activation function;

[0032]

[0033] Among them, r bl It represents the probability that the lth data is judged as work status b during work evaluation.

[0034] S105: According to the work evaluation results l, evaluate the working status of the equipment operator, and count and output the working time of the equipment operator when he / she is on duty based on the punch-in information of the equipment operator collected by the RFID reading module.

[0035] In a third aspect, an embodiment of the present application provides a device for estimating the working time of an equipment operator, characterized in that it is applied to an Internet of Things automatic supervision terminal according to claims 1 to 4, and the estimation device includes:

[0036] The detection module is used to install the IoT automatic monitoring terminal on the monitored equipment, read the equipment operation data M such as the current, voltage, and power value of the monitored equipment through the three-phase voltage and current detection module, and record the working status of the current equipment operator. When the equipment operator is in the working state, the working state value is recorded as 1, otherwise it is 0, and the 4×g value of the working time evaluation is obtained. s dimensional data set, where g s Indicates the number of elements in the data set, x n Represents the nth element in the data set;

[0037] Acquisition module, used to obtain training sets and test sets;

[0038] Training module, used to train the input weight matrix W A And the output weight matrix W B ;

[0039] Evaluation module, used to evaluate the device operation data M and input weight matrix W A And the output weight matrix W B , calculate the work evaluation result matrix R of the current data through formula 1, and calculate the work evaluation result r of the current data through formula 2 l ;

[0040] R=W B ·g(W A M T )Formula 1

[0041] Among them, g() represents the activation function;

[0042]

[0043] Among them, r bl It represents the probability that the lth data is judged as work status b during work evaluation.

[0044] Statistics module for evaluating results based on work l , evaluate the working status of the equipment operator, and count and output the working time of the equipment operator when he / she is on duty based on the punch-in information of the equipment operator collected by the RFID reading module.

[0045] In the embodiment of the present application, the IoT automatic monitoring terminal is installed on the monitored equipment, and the equipment operation data such as the current, voltage, and power value of the monitored equipment are read through the three-phase voltage and current detection module to evaluate the working status of the equipment operator, and the working time of the equipment operator when on duty is counted and output according to the punch-in information of the equipment operator collected by the RFID reading module. Accurately manage and evaluate the working time and work efficiency of each equipment operator, and then make a fair and just performance reward and punishment system, which improves the work enthusiasm of the equipment operator and the work efficiency of the company. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is one of the appearance structure diagrams of an Internet of Things automatic supervision terminal provided in an embodiment of the present application;

[0047] Figure 2 This is one of the appearance structure diagrams of an Internet of Things automatic supervision terminal provided in an embodiment of the present application;

[0048] Figure 3 It is a hardware structure diagram provided by an embodiment of the present application;

[0049] Figure 4 It is a flow chart of a method for estimating the working time of an equipment operator provided in an embodiment of the present application;

[0050] Figure 5 It is a structural schematic diagram of a device for estimating the working time of an equipment operator provided in an embodiment of the present application.

[0051] 1-plug, 101-switch module, 102-RFID card module;

[0052] 2-terminal body, 201-WiFi module, 202-Bluetooth module, 203-clock module, 204-watchdog module, 205-circuit status indicator module, 206-button module, 207-buzzer module, 208-external interface module, 209-debugging interface module, 210-RFID reading module, 211-three-phase voltage and current detection module, 212-main control chip module, 213-power conversion module;

[0053] 3-socket;

[0054] 4- Cloud server;

[0055] 5-User’s mobile phone.

[0056] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0058] The following, in conjunction with the accompanying drawings, describes in detail the method for estimating the safety management level of a large commercial complex provided by the embodiment of the present application through specific embodiments and their application scenarios.

[0059] Embodiment 1

[0060] Reference Figure 1 and Figure 2 , shows the appearance structure diagram of an Internet of Things automatic supervision terminal provided in an embodiment of the present application.

[0061] The present application provides an Internet of Things automatic monitoring terminal for equipment operators, which is characterized by including a plug 1 and a terminal body 2.

[0062] Plug 1 is connected to the terminal body 2, and plug 1 is an effective way for the terminal body 2 to establish a connection with the outside world. Optionally, through the connection between plug 1 and an external socket, the terminal body 2 can establish a communication connection with the external socket; optionally, through the connection between plug 1 and an external socket, the external socket can supply power to the terminal body 2.

[0063] The plug 1 is provided with a switch module 101 and an RFID card module 102 .

[0064] The switch module 101 is used to start the power switch module. The power supply can be turned on or off through the switch module 101.

[0065] Optionally, the switch module 101 specifically includes 1 switch.

[0066] The RFID card module 102 is used to provide a device ID. Different IoT automatic monitoring terminals have different device IDs, which can further facilitate management.

[0067] Optionally, the number of the RFID card module 102 is one.

[0068] The terminal body 2 is provided with a WiFi module 201, a Bluetooth module 202, a clock module 203, a watchdog module 204, a circuit status indicator light module 205, a button module 206, a buzzer module 207, an external interface module 208, a debugging interface module 209, an RFID reading module 210, a three-phase voltage and current detection module 211, a main control chip module 212 and a power conversion module 213.

[0069] The WiFi module 201 is used to implement wireless local area network communication.

[0070] Optionally, the WiFi module 201 uses an ESP8266 chip, which supports wireless 80211b / g / n standards and three working modes: STA / AP / STA+AP, has a built-in CTP / IP protocol line, can have multiple TCPClient connections, can receive rich SocketAT commands, has a UART / GPIO data communication interface, supports remote firmware upgrades (OTA), has a built-in 32-bit MCU, can also serve as an application processor, has ultra-low energy consumption, and is suitable for battery-powered applications.

[0071] The Bluetooth module 202 is used to communicate with external devices.

[0072] Optionally, the Bluetooth module 202 uses the ST17H26 chip, which is a low-power chip with an internal space of 16K. It also supports an external EEPROM chip anti-lost device and is widely used in smart lights / group-controlled lights, smart locks, and other smart home IoT products.

[0073] The order information can be received through the WiFi module 201 and the Bluetooth module 202. After the equipment operator completes the assigned order, he can receive the next order through the terminal, and the time management and work arrangement are accurate to the individual. Since different tasks are assigned to individuals, each person has a clear division of labor, which reduces the laziness of employees. At the same time, the working hours and work completion of the equipment operator can be estimated, and the corresponding reward and punishment system can be implemented to improve the income and treatment of efficient personnel and improve the company's work efficiency. The clock module 203 is used to provide the system clock.

[0074] Optionally, the clock module 203 uses a DS1302 chip, which contains a real-time clock / calendar and 31 bytes of static RAM, and communicates with the single-chip microcomputer through a simple serial interface. The real-time clock / calendar circuit provides information on seconds, minutes, hours, days, weeks, months, and years, and the number of days per month and leap years can be automatically adjusted. The clock operation can be determined by the AM / PM indication to use a 24-hour or 12-hour format. The DS1302 can simply communicate with the single-chip microcomputer in a synchronous serial manner, using only three port lines: RST reset, I / O data line, and SCLK serial clock. The read / write data of the clock / RAM is communicated in a character group mode of one byte or up to 31 bytes. The DS1302 has very low power consumption when working, and the power is less than 1mW when maintaining data and clock information.

[0075] The watchdog module 204 is used to feed the watchdog at a scheduled time.

[0076] Optionally, the system program controls the watchdog module 204 to feed the watchdog regularly. If the watchdog is not fed within 3 seconds, the program will automatically restart to protect the program from an infinite loop.

[0077] Optionally, the watchdog module 204 uses a DS1232 chip, which has a power indicator light and an input and output indicator light, supports TTL communication, four-digit digital tube display programming, is controlled by two-digit key input programming, a single input supports 0-5V analog quantity and relay control output, an external crystal oscillator 11.0592M, the output has photoelectric isolation, anti-reverse connection protection, overcurrent protection, and continuous current protection, and has a watchdog function to prevent the system from crashing.

[0078] The circuit status indicator light module 205 is used to display different colors according to the circuit status.

[0079] Optionally, the number of circuit status indicator light modules 205 is 5.

[0080] Optionally, the circuit status indicator light module 205 has an independent light board with 5 LED patch lights in color TBD, which serve as an indication of the box status and function, and change with the key switch function / start / stop. The on and off of the light can be controlled by the main chip GPIO and the brightness of the light is sufficient, so that the staff can clearly distinguish the brightness of the light under indoor lighting conditions.

[0081] The button module 206 is used for resetting and switching functions.

[0082] Optionally, the case module 206 includes 2 buttons, one of which is a pinhole reset button, which can be reset by a short press and restored to factory settings by a long press, and the other is a function button for easy touch.

[0083] The buzzer module 207 is used for alarming.

[0084] Optionally, the ringing sound may be controlled through a GPIO interface of the main control chip module 212 .

[0085] Optionally, the buzzer module 207 uses a buzzer 0905 chip, which is an active electromagnetic direct-insert ultra-small and thin chip with an operating voltage of 3V / 5V / 12V, a size of 9mm in diameter and 5.5mm in height, and is environmentally friendly, high-temperature resistant and wave-solderable.

[0086] The external interface module 208 is used to provide 5V 2A current to the outside and perform communication.

[0087] Optionally, the external interface module 208 has a USB interface, can provide 5V2A current to the outside, and can communicate, meet the USB2.0 standard, have OTG function, and has an RS232 serial port, which can communicate with the device serial port; it has a TTL interface and can support 5V TTL communication.

[0088] The debugging interface module 209 is used for program upgrading and debugging.

[0089] Optionally, the debugging interface module 209 is easy to plug and unplug without disassembling the housing, and there is no risk of electric shock in the high-voltage area during plugging and unplugging.

[0090] The RFID reading module 210 is used to read the RFID tag of the equipment operator.

[0091] The RFID reading module 210 can be used for daily attendance clocking. The RFID reading will not misread the RFID tags of adjacent workstations, and can accurately read the RFID tags on fixed sockets, thereby starting and shutting down the estimated working time of the terminal operator.

[0092] Optionally, the RFID reader module uses the MFRC522 chip, which can accurately read the RFID tag of the equipment operator, thereby recording the operator's attendance and opening and closing the estimation of working time. The chip uses advanced modulation and demodulation concepts to integrate all types of passive contactless communication methods and protocols at 13.56MHz. Supports 14443A compatible transponder signals. The digital part processes ISO014443A frames and error detection. In addition, it also supports the fast CRYPTO1 encryption algorithm for verification of MIFARE series products. MFRC522 supports MIFARE series higher-speed contactless communication, with a two-way data transmission rate of up to 424kbit / s. It uses SPi mode for communication with the host, which is conducive to reducing wiring, reducing the size of PCB boards and reducing costs.

[0093] The three-phase voltage and current detection module 211 is used to collect the real-time current, voltage and power of the equipment.

[0094] Optionally, the three-phase voltage and current detection module 211 adopts a JBLGK-350 module, which is an industrial-grade voltage and current module that can collect AC three-phase voltage, current, active power, reactive power, power factor, accumulated power, frequency, phase sequence, temperature and other electrical parameters. The three-phase phase voltage is 0-370V, the collection accuracy is 0.5%, the three-phase line voltage is 0-640V, the collection frequency is 25Hz, the power consumption is 24V / 1W, and it has the characteristics of built-in watchdog protection, wide temperature operating range, isolation lightning protection and surge protection, and insulated terminal anti-vibration.

[0095] The main control chip module 212 is used to control the device and perform data processing.

[0096] Optionally, the main control chip module 212 adopts the RK3308 chip, which is designed for special audio and IoT applications. The CPU adopts a quad-core ARM Cortex-A35 (64-bit architecture, high performance, low power consumption), integrates a high-performance CODEC (8-channel ADC + 2-channel DAC), directly supports a maximum of 8-channel analog MIC array + acquisition, does not require an external ADC, integrates mainstream digital audio input and output interfaces, and develops a hardware voice detection module (VAD) for low-power applications, significantly reducing the system BOM cost.

[0097] The power conversion module 213 is used to provide power to the device.

[0098] Optionally, the power conversion module 213 uses the LM2596 chip, which is a 3A current output step-down switch type integrated voltage regulator chip produced by Texas Instruments (TI), which contains a fixed frequency oscillator (150KHZ) and a reference voltage regulator (1.23v), and has a complete protection circuit, current limit, thermal shutdown circuit, etc. Using this device, only a few peripheral devices are needed to form a high-efficiency voltage regulator circuit.

[0099] Reference Figure 3 , showing a hardware structure diagram provided in an embodiment of the present application.

[0100] Further, in the embodiment of the present application, the main control chip module 212 is respectively connected to the WiFi module 201, the Bluetooth module 202, the clock module 203, the watchdog module 204, the circuit status indicator module 205, the button module 206, the buzzer module 207, the external interface module 208, the debugging interface module 209, the RFID reading module 210 and the three-phase voltage and current detection module 211;

[0101] The power conversion module 213 is respectively connected to the WiFi module 201, the Bluetooth module 202, the clock module 203, the watchdog module 204, the circuit status indicator module 205, the button module 206, the buzzer module 207, the external interface module 208, the debugging interface module 209, the RFID reading module 210, the three-phase voltage and current detection module 211 and the main control chip module 212;

[0102] The switch module 101 is connected to the power conversion module 213 to realize the opening and closing of the power conversion module 213 .

[0103] Further, in the embodiment of the present application, the IoT automatic supervision terminal also includes a smart socket 3, a cloud server 4 and a user mobile phone 5;

[0104] The RFID card module 102 is connected to the smart socket 3, and the RFID card module 102 provides the device ID to the smart socket 3, and the smart socket 3 is used to read the RFID card in the plug 1 and identify the terminal serial number;

[0105] Specifically, the plug 1 and the smart socket 3 are adapted to each other. When the plug 1 is inserted into the smart socket 3, the RFID card module 102 is electrically connected to the smart socket 3. The smart socket 3 can be provided with an RFID card reader, and the RFID card reader collects the RFID tag on the plug 1 to accurately determine each terminal.

[0106] The mutual adaptation of the intelligent socket 3 and the plug 1 to read the RFID tag to determine each terminal has strong practicality and autonomy.

[0107] The smart socket 3 is connected to the cloud server 4 for communication, and the cloud server 4 is used to receive data from the smart socket 3. Further, in the embodiment of the present application, the RFID card module 102 is connected to the smart socket 3, and the RFID card module 102 provides the device ID to the smart socket 3;

[0108] The WiFi module 201 is connected to the cloud server 4 for communication, and the cloud server 4 is used to receive data from the terminal body 2 .

[0109] The Bluetooth module 202 is connected to the user's mobile phone 5 for communication, and the user's mobile phone 5 is used to receive data from the terminal body 2 .

[0110] In the embodiment of the present application, the IoT automatic monitoring terminal is installed on the monitored equipment, and the equipment operation data such as the current, voltage, and power value of the monitored equipment are read through the three-phase voltage and current detection module to evaluate the working status of the equipment operator, and the working time of the equipment operator when on duty is counted and output according to the punch-in information of the equipment operator collected by the RFID reading module. Accurately manage and evaluate the working time and work efficiency of each equipment operator, and then make a fair and just performance reward and punishment system, which improves the work enthusiasm of the equipment operator and the work efficiency of the company.

[0111] Embodiment 2

[0112] Reference Figure 4 , shows a flow chart of a method for estimating the working time of an equipment operator provided in an embodiment of the present application.

[0113] The method for estimating the working hours of equipment operators provided in the embodiment of the present application is applied to the Internet of Things automatic supervision terminal of the first embodiment.

[0114] Estimation methods include:

[0115] S101: Install the IoT automatic monitoring terminal on the monitored equipment, read the equipment operation data M such as the current, voltage, and power value of the monitored equipment through the three-phase voltage and current detection module 211, and record the working status of the current equipment operator. When the equipment operator is in the working state, the working status value is recorded as 1, otherwise it is 0, and the 4×g value of the working time evaluation is obtained. s dimensional data set.

[0116] Among them, g s Indicates the number of elements in the data set, x n Represents the nth element in the dataset.

[0117] S102: Obtain a training set and a test set.

[0118] Optionally, S102 may be completed by sub-steps S1021 to S1024.

[0119] S1021: Initialize training parameters and set thresholds The threshold is 20000. is 5000, parameter a is 0.2, and the threshold of the number of elements in each category in the training set is 200, the number of training iterations is 1, etc. S1022: Calculate the number of elements g in the training set using Formula 3 x , calculate the number of hidden layers Y by formula 4, and calculate the maximum number of elements F in the two categories of working and not working in the training set by formula 5 max ;

[0120]

[0121] Among them, g s Indicates the number of elements in the data set, min() indicates the minimum value function;

[0122]

[0123]

[0124] S1023: Calculate the initial number of elements F of the two categories of working and non-working by using formula 6 b ;

[0125] F b =max(F max ,(g b g x / g s )), b=1 or 0 Formula 6

[0126] Among them, g b Indicates the number of elements in the data set that belong to category b. When b = 1, it means that the operator is working, and b = 0 means that the operator is not working. Let b = 0, c = 1, l = 1, and the lth element of category b is d bl ;

[0127] S1024: If c is less than or equal to F b , then d bl =x c , x c Indicates the cth element in the training set, c = c + 1, l = l + 1, repeat S1024, otherwise if c is greater than F b , and less than or equal to g b , then there are redundant elements, and we can find all d by using formula 7 bl ,l=1,...,F b An element in and element x c The distance is the largest, and the element number l is obtained c , update the element by formula 8,

[0128] l c = argmax(|d bl -x c |), l=1,...,F b Formula 7

[0129] Among them, argmax() represents the element number corresponding to the maximum value, || represents the distance between two elements,

[0130] in, Represents the updated element;

[0131] c=c+1, if c is greater than g b and b=0, then b=1, c=1, and repeat S1024. Otherwise, set all d bl , The training set is formed, and the elements in the data set other than the elements in the training set form the test set. S103: Training input weight matrix W A And the output weight matrix W B .

[0132] Optionally, S103 may be completed by sub-steps S1031 to S1034.

[0133] S1031: Based on the training set and the test set, obtain the training matrix H and the test matrix S with the sequence pattern, and randomly initialize the Y×4 dimension matrix W with the input weights in the interval [-1 / 4, 1 / 4] A , let the 2×Y degree matrix with output weights be W B ;

[0134] S1032: Calculate the hidden layer parameter U by formula 9, and calculate the output O of the hidden layer neuron by formula 10;

[0135] U=W A X T Formula 9

[0136] Among them, X T is the transpose of the matrix X;

[0137] O=g(U) Formula 10

[0138] Where g() represents the activation function, which can be expressed as g(u)=(1+e -u ) -1 .

[0139] S1033: Calculate the output weight matrix W using formula 11 B , calculate the classification evaluation result Q of the training set through formula 12, and obtain the classification evaluation result q of each element in the training set through formula 13 l ;

[0140]

[0141] Among them, T represents the association label matrix between the hidden layer and the output layer, is the pseudo-inverse matrix of O;

[0142] Q=W B O Formula 12

[0143]

[0144] Among them, q l Indicates the length is F b The classification results of the training set elements, q bl It represents the probability that the lth data in the training set is judged as working state b. According to the classification evaluation results of the training set, the classification evaluation accuracy of the training set is calculated;

[0145] If the classification evaluation accuracy is greater than the threshold K1, proceed to S1034, otherwise the Adam gradient descent algorithm is used to update the weight clustering W A , repeat S1032;

[0146] S1034: Calculate the classification evaluation result matrix Z of the test set by using formula 14, and obtain the classification evaluation result of each element in the test set by using formula 15;

[0147] Z=W B g(W A S T )Formula 14

[0148]

[0149] Among them, z bl It represents the probability that the lth data in the test set is judged to be in working state b;

[0150] According to the classification evaluation results of the test set, calculate the classification evaluation accuracy of the test set;

[0151] If the classification evaluation accuracy of the test set is greater than the threshold K2, the training ends and the trained input weight matrix W is obtained. A And the output weight matrix W B Otherwise, randomly initialize the Y×4 dimension matrix W with input weights in the interval [-1 / 4,1 / 4] A , the number of training iterations is increased by 1; if the number of training iterations is greater than the threshold K3, it means that the training has not converged, the algorithm ends, and an alarm message of "training has not converged" is output, otherwise S1032 is repeated.

[0152] S104: According to the equipment operation data M, input weight matrix W A And the output weight matrix W B , calculate the work evaluation result matrix R of the current data through formula 1, and calculate the work evaluation result r of the current data through formula 2 l ;

[0153] R=W B g(W A M T )Formula 1

[0154] Among them, g() represents the activation function;

[0155]

[0156] Among them, r bl It represents the probability that the lth data is judged as work status b during work evaluation.

[0157] S105: According to the work evaluation results l , evaluate the working status of the equipment operator, and count and output the working time of the equipment operator when he / she is on duty based on the punch-in information of the equipment operator collected by the RFID reading module 210.

[0158] In the embodiment of the present application, the working time evaluation method of equipment operators based on fast extreme learning first collects data sets through the Internet of Things automatic supervision terminal, and according to the size of the data set, obtains a low-scale and highly reliable training set by selecting the number of hidden layers, the number of output layers, the number of training sets, and the elements of the training set, thereby reducing the calculation, storage and pseudo-inversion of large-scale matrices and effectively improving the speed of training. According to the screened training set, an extreme learning machine is used to realize the training and working time estimation of the model, thereby reducing the training error, and having good generalization ability, it can be applied to the working time estimation of operators of different types of equipment.

[0159] In the embodiments of the present application, the problem of estimating the working hours of operators of different types of equipment is effectively solved, and the system has the characteristics of easy installation, flexible expansion, convenient management, and interconnection, thereby improving the accuracy of working time evaluation of equipment operators, reducing the interference of different equipment on the evaluation, and making time management and work arrangements accurate to individual employees.

[0160] In the embodiment of the present application, the IoT automatic monitoring terminal is installed on the monitored equipment, and the equipment operation data such as the current, voltage, and power value of the monitored equipment are read through the three-phase voltage and current detection module to evaluate the working status of the equipment operator, and the working time of the equipment operator when on duty is counted and output according to the punch-in information of the equipment operator collected by the RFID reading module. Accurately manage and evaluate the working time and work efficiency of each equipment operator, and then formulate a fair and just performance reward and punishment system, which improves the work enthusiasm of equipment operators and the work efficiency of the company.

[0161] Embodiment 3

[0162] Reference Figure 5 , shows a schematic structural diagram of a device 50 for estimating the working time of an equipment operator provided in an embodiment of the present application.

[0163] The device 50 for estimating the working time of an equipment operator provided in the embodiment of the present application is applied to the Internet of Things automatic supervision terminal of the first embodiment.

[0164] The estimation device 50 comprises:

[0165] The detection module 501 is used to install the IoT automatic monitoring terminal on the monitored equipment, read the equipment operation data M such as the current, voltage, and power value of the monitored equipment through the three-phase voltage and current detection module 211, and record the working status of the current equipment operator. When the equipment operator is in the working state, the working state value is recorded as 1, otherwise it is 0, and the 4×g value of the working time evaluation is obtained. s dimensional data set, where g s Indicates the number of elements in the data set, x n Represents the nth element in the data set;

[0166] An acquisition module 502 is used to acquire a training set and a test set;

[0167] Training module 503, used to train the input weight matrix W A And the output weight matrix W B ;

[0168] Evaluation module 504, used to evaluate the device operation data M, input weight matrix W A And the output weight matrix W B , calculate the work evaluation result matrix R of the current data through formula 1, and calculate the work evaluation result r of the current data through formula 2 l ;

[0169] R=W B g(W A M T )Formula 1

[0170] Among them, g() represents the activation function;

[0171]

[0172] Among them, r bl It represents the probability that the lth data is judged as work status b during work evaluation.

[0173] Statistics module 505, used to evaluate the work results l , evaluate the working status of the equipment operator, and count and output the working time of the equipment operator when he / she is on duty based on the punch-in information of the equipment operator collected by the RFID reading module 210.

[0174] Optionally, the acquisition module 502 is specifically used for:

[0175] Initialize training parameters and set thresholds The threshold is 20000. is 5000, parameter a is 0.2, and the threshold of the number of elements in each category in the training set is 200, the number of training iterations is 1, etc.

[0176] The number of elements g in the training set is calculated by formula 3 x , calculate the number of hidden layers Y by formula 4, and

[0177] Formula 5 calculates the maximum number of elements F in the two categories of working and not working in the training set max ;

[0178]

[0179] Among them, g s Indicates the number of elements in the data set, min() indicates the minimum value function;

[0180]

[0181]

[0182] The initial number of elements F in the two categories of working and non-working is calculated by formula 6 b ;

[0183] F b =max(F max ,(g b g x / g s )), b=1 or 0 Formula 6

[0184] Among them, g b Indicates the number of elements in the data set that belong to category b. When b = 1, it means that the operator is working, and b = 0 means that the operator is not working. Let b = 0, c = 1, l = 1, and the lth element of category b is d bl ;

[0185] If c is less than or equal to F b , then d bl =x c , x c Indicates the cth element in the training set, c = c + 1, l = l + 1, re-update the element, otherwise if c is greater than F b , and less than or equal to g b , then there are redundant elements, and we can find all d by using formula 7 bl ,l=1,...,F b An element in and element x c The distance is the largest, and the element number l is obtainedc , update the element by formula 8,

[0186] l c = argmax(|d bl -x c |), l=1,...,F b Formula 7

[0187] Among them, argmax() represents the element number corresponding to the maximum value, || represents the distance between two elements,

[0188] in, Represents the updated element;

[0189] c=c+1, if c is greater than gb and b=0, then b=1, c=1, and repeat S1024, otherwise, all d bl , The training set is formed, and the elements in the data set other than the elements in the training set form the test set. Optionally, the training module 503 is specifically used to:

[0190] According to the training set and test set, obtain the training matrix H and test matrix S with sequential patterns, and randomly initialize the Y×4 dimensional matrix W with input weights in the interval [-1 / 4,1 / 4] A , let the 2×Y degree matrix with output weights be W B ;

[0191] The hidden layer parameter U is calculated by formula 9, and the output O of the hidden layer neuron is calculated by formula 10;

[0192] U=W A X T Formula 9

[0193] Among them, X T is the transpose of the matrix X;

[0194] O=g(U) Formula 10

[0195] Where g() represents the activation function, which can be expressed as g(u)=(1+e -u ) -1 .

[0196] The output weight matrix W is calculated by formula 11 B , calculate the classification evaluation result Q of the training set through formula 12, and obtain the classification evaluation result q of each element in the training set through formula 13 l ;

[0197]

[0198] Among them, T represents the association label matrix between the hidden layer and the output layer, is the pseudo-inverse matrix of O;

[0199] Q=W B O Formula 12

[0200]

[0201] Among them, q l Indicates the length is F b The classification results of the training set elements, q bl It represents the probability that the lth data in the training set is judged as working state b. According to the classification evaluation results of the training set, the classification evaluation accuracy of the training set is calculated;

[0202] If the classification evaluation accuracy is greater than the threshold K1, the classification evaluation result of the test set is calculated, otherwise the Adam gradient descent algorithm is used to update the weight clustering W A , recalculate the hidden layer parameters U;

[0203] The classification evaluation result matrix Z of the test set is calculated by formula 14, and the classification evaluation result of each element in the test set is obtained by formula 15;

[0204] Z=W B ·g(W A S T )Formula 14

[0205]

[0206] Among them, z bl It represents the probability that the lth data in the test set is judged to be in working state b;

[0207] According to the classification evaluation results of the test set, calculate the classification evaluation accuracy of the test set;

[0208] If the classification evaluation accuracy of the test set is greater than the threshold K2, the training ends and the trained input weight matrix W is obtained. A And the output weight matrix W B Otherwise, randomly initialize the Y×4 dimension matrix W with input weights in the interval [-1 / 4,1 / 4] A , the number of training iterations is increased by 1; if the number of training iterations is greater than the threshold K3, it means that the training has not converged, the algorithm ends, and an alarm message of "training has not converged" is output, otherwise the hidden layer parameters are recalculated.

[0209] The estimation device 50 provided in the embodiment of the present application can implement each process implemented in the above method embodiment, and will not be described again here to avoid repetition.

[0210] In the embodiment of the present application, the working time evaluation method of equipment operators based on fast extreme learning first collects data sets through the Internet of Things automatic supervision terminal, and according to the size of the data set, obtains a low-scale and highly reliable training set by selecting the number of hidden layers, the number of output layers, the number of training sets, and the elements of the training set, thereby reducing the calculation, storage and pseudo-inversion of large-scale matrices and effectively improving the speed of training. According to the screened training set, an extreme learning machine is used to realize the training and working time estimation of the model, thereby reducing the training error, and having good generalization ability, it can be applied to the working time estimation of operators of different types of equipment.

[0211] In the embodiments of the present application, the problem of estimating the working hours of operators of different types of equipment is effectively solved, and the system has the characteristics of easy installation, flexible expansion, convenient management, and interconnection, thereby improving the accuracy of working time evaluation of equipment operators, reducing the interference of different equipment on the evaluation, and making time management and work arrangements accurate to individual employees.

[0212] In the embodiment of the present application, the IoT automatic monitoring terminal is installed on the monitored equipment, and the equipment operation data such as the current, voltage, and power value of the monitored equipment are read through the three-phase voltage and current detection module to evaluate the working status of the equipment operator, and the working time of the equipment operator when on duty is counted and output according to the punch-in information of the equipment operator collected by the RFID reading module. Accurately manage and evaluate the working time and work efficiency of each equipment operator, and then make a fair and just performance reward and punishment system, which improves the work enthusiasm of the equipment operator and the work efficiency of the company.

[0213] The virtual device in the embodiment of the present application may be a device, or a component, an integrated circuit, or a chip in a terminal.

[0214] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A method for estimating the working time of equipment operators, characterized in that: Applied to the Internet of Things automatic supervision terminal, the Internet of Things automatic supervision terminal includes a plug and a terminal body; the plug is connected to the terminal body; the plug is provided with a switch module and an RFID card module; the switch module is used to start the power switch module; the RFID card module is used to provide the device ID; the terminal body is provided with a WiFi module, a Bluetooth module, a clock module, a watchdog module, a circuit status indicator module, a button module, a buzzer module, an external interface module, a debugging interface module, an RFID reading module, a three-phase voltage and current detection module, a main control chip module and a power conversion module; the WiFi module is used to realize wireless LAN communication; the Bluetooth module is used to realize communication with external devices; the clock module is used to provide the system clock; the watchdog module is used for program timing feeding; the circuit status indicator module is used to display different colors according to the circuit status; the button module is used for reset and switching functions; the buzzer module is used for alarm; the external interface module is used to provide 5V2A current to the outside and communicate; the debugging interface module is used for program upgrade and debugging; the RFID reading module is used to read the RFID tag of the equipment operator; the three-phase voltage and current detection module is used to collect the real-time current, voltage and power of the equipment; The main control chip module is used to control the device and perform data processing; The power conversion module is used to provide power to the device; The estimation method includes: S101: Install the IoT automatic monitoring terminal on the monitored device, read the device operation data M such as current, voltage, power value of the monitored device through the three-phase voltage and current detection module, and record the working status of the current device operator. When the device operator is in working state, the working status value is recorded as 1, otherwise it is 0, and the 4×g value of working time evaluation is obtained. s dimensional data set, where g s Indicates the number of elements in the data set, x n Represents the nth element in the data set; S102: Obtain training set and test set; S103: Training input weight matrix W A And the output weight matrix W B ; S104: Based on the device operation data M and the input weight matrix W A and the output weight matrix W B , calculate the work evaluation result matrix R of the current data through formula 1, and calculate the work evaluation result r of the current data through formula 2 l ; R=W B g(W A M T ) Formula 1 Among them, g() represents the activation function; Among them, r bl It represents the probability that the lth data is judged as work status b during work evaluation; S105: Based on the work evaluation result l , evaluate the working status of the equipment operator, and count and output the working time of the equipment operator when he is on duty based on the punch-in information of the equipment operator collected by the RFID reading module.

2. The estimation method according to claim 1, characterized in that The step S102: obtaining a training set and a test set specifically includes: S1021: Initialize training parameters and set thresholds The threshold is 20000. is 5000, parameter a is 0.2, and the threshold of the number of elements in each category in the training set is 200, the number of training iterations is 1, etc.; S1022: Calculate the number of elements g in the training set using Formula 3 x , the number of hidden layers Y is calculated by formula 4, and the maximum number of elements F of the two categories of working and not working in the training set is calculated by formula 5 max ; Among them, g s Indicates the number of elements in the data set, min() indicates the minimum value function; S1023: Calculate the initial number of elements F of the two categories of working and non-working by using formula 6 b ; F b =max(F max ,(g b g x / g s )),b=1or0 Formula 6 Among them, g b represents the number of elements belonging to category b in the data set; when b=1, it means the operator is working, and b=0 means the operator is not working. Let b=0, c=1, l=1, the lth element of category b is d bl ; S1024: If c is less than or equal to F b , then d bl =x c , x c Indicates the cth element in the training set, c = c + 1, l = l + 1, repeat S1024, otherwise if c is greater than F b , and less than or equal to g b , then there are redundant elements, and we can find all d by using formula 7 bl ,l=1,...,F b An element in and element x c The distance is the largest, and the element number l is obtained c , update the element by formula 8, l c = argmax(|d bl -x c |), l=1,...,F b Formula 7 Among them, argmax() represents the element number corresponding to the maximum value, || represents the distance between two elements, in, Represents the updated element; c=c+1, if c is greater than g b and b=0, then b=1, c=1, and repeat S1024. Otherwise, set all d bl , The training set is composed of the elements in the data set except the elements in the training set, and the test set is composed of the elements in the data set.

3. The estimation method according to claim 1, characterized in that S103: Training input weight matrix W A And the output weight matrix W B , specifically including: S1031: According to the training set and the test set, obtain a training matrix H and a test matrix S with a sequence pattern, and randomly initialize a Y×4 dimensional matrix W with input weights in the interval [-1 / 4, 1 / 4] A , let the 2×Y degree matrix with output weights be W B ; S1032: Calculate the hidden layer parameter U by formula 9, and calculate the output O of the hidden layer neuron by formula 10; U=W A X T Formula 9 Among them, X T is the transpose of the matrix X; O=g(U) Formula 10 Where g() represents the activation function, which can be expressed as g(u)=(1+e -u ) -1 ; S1033: Calculate the output weight matrix W by formula 11 B , calculate the classification evaluation result Q of the training set through formula 12, and obtain the classification evaluation result q of each element in the training set through formula 13 l ; Among them, T represents the association label matrix between the hidden layer and the output layer, is the pseudo-inverse matrix of O; Q=W B O Formula 12 Among them, q l Indicates the length is F b The classification results of the training set elements, q bl represents the probability that the lth data in the training set is judged to be in working state b; according to the classification evaluation results of the training set, the classification evaluation accuracy of the training set is calculated; If the classification evaluation accuracy is greater than the threshold K1, proceed to S1034, otherwise the Adam gradient descent algorithm is used to update the weight clustering W A , repeat S1032; S1034: Calculate the classification evaluation result matrix Z of the test set by using formula 14, and obtain the classification evaluation result of each element in the test set by using formula 15; Z=W B g(W A S T ) Formula 14 Among them, z bl It represents the probability that the lth data in the test set is judged to be in working state b; According to the classification evaluation results of the test set, calculate the classification evaluation accuracy of the test set; If the classification evaluation accuracy of the test set is greater than the threshold K2, the training ends and the trained input weight matrix W is obtained. A and the output weight matrix W B Otherwise, randomly initialize the Y×4 dimension matrix W with input weights in the interval [-1 / 4,1 / 4] A , the number of training iterations is increased by 1; if the number of training iterations is greater than the threshold K3, it means that the training has not converged, the algorithm ends, and an alarm message of "training has not converged" is output, otherwise S1032 is repeated.

4. A device for estimating the working time of equipment operators, characterized in that: Applied to the Internet of Things automatic supervision terminal, the Internet of Things automatic supervision terminal includes a plug and a terminal body; the plug is connected to the terminal body; the plug is provided with a switch module and an RFID card module; the switch module is used to start the power switch module; the RFID card module is used to provide the device ID; the terminal body is provided with a WiFi module, a Bluetooth module, a clock module, a watchdog module, a circuit status indicator module, a button module, a buzzer module, an external interface module, a debugging interface module, an RFID reading module, a three-phase voltage and current detection module, a main control chip module and a power conversion module; the WiFi module is used to realize wireless LAN communication; the Bluetooth module is used to realize communication with external devices; the clock module is used to provide the system clock; the watchdog module is used for program timing feeding; the circuit status indicator module is used to display different colors according to the circuit status; the button module is used for reset and switching functions; the buzzer module is used for alarm; the external interface module is used to provide 5V2A current to the outside and communicate; the debugging interface module is used for program upgrade and debugging; the RFID reading module is used to read the RFID tag of the equipment operator; the three-phase voltage and current detection module is used to collect the real-time current, voltage and power of the equipment; The main control chip module is used to control the device and perform data processing; The power conversion module is used to provide power to the device; The estimation device comprises: The detection module is used to install the IoT automatic monitoring terminal on the monitored device, read the current, voltage, power value and other equipment operation data M of the monitored device through the three-phase voltage and current detection module, and record the working status of the current equipment operator. When the equipment operator is in the working state, the working status value is recorded as 1, otherwise it is 0, and the 4×g value of the working time evaluation is obtained. s dimensional data set, where g s Indicates the number of elements in the data set, x n Represents the nth element in the data set; Acquisition module, used to obtain training sets and test sets; Training module, used to train the input weight matrix W A And the output weight matrix W B ; An evaluation module is used to evaluate the device operation data M and the input weight matrix W. A and the output weight matrix W B , calculate the work evaluation result matrix R of the current data through formula 1, and calculate the work evaluation result r of the current data through formula 2 l ; R=W B g(W A M T ) Formula 1 Among them, g() represents the activation function; Among them, r bl It represents the probability that the lth data is judged as work status b during work evaluation; Statistics module for evaluating the results of the work l , evaluate the working status of the equipment operator, and count and output the working time of the equipment operator when he is on duty based on the punch-in information of the equipment operator collected by the RFID reading module.

5. The estimation device according to claim 4, characterized in that The acquisition module is specifically used for: Initialize training parameters and set thresholds The threshold is 20000. is 5000, parameter a is 0.2, and the threshold of the number of elements in each category in the training set is 200, the number of training iterations is 1, etc.; The number of elements g in the training set is calculated by formula 3 x , the number of hidden layers Y is calculated by formula 4, and the maximum number of elements F of the two categories of working and not working in the training set is calculated by formula 5 max ; Among them, g s Indicates the number of elements in the data set, min() indicates the minimum value function; The initial number of elements in the two categories of working and not working is calculated by formula 6; F b =max(F max ,(g b g x / g s )),b=1or0 Formula 6 Among them, g b represents the number of elements belonging to category b in the data set; when b=1, it means the operator is working, and b=0 means the operator is not working. Let b=0, c=1, l=1, the lth element of category b is d bl ; If c is less than or equal to F b , then d bl =x c , x c Indicates the cth element in the training set, c = c + 1, l = l + 1, re-update the element, otherwise if c is greater than F b , and less than or equal to g b , then there are redundant elements, and we can find all d by using formula 7 bl ,l=1,...,F b An element in and element x c The distance is the largest, and the element number l is obtained c , update the element by formula 8, l c = argmax(|d bl -x c |), l=1,...,F b Formula 7 Among them, argmax() represents the element number corresponding to the maximum value, || represents the distance between two elements, in, Represents the updated element; c=c+1, if c is greater than g b and b=0, then b=1, c=1, and repeat S1024. Otherwise, set all d bl , The training set is composed of the elements in the data set except the elements in the training set, and the test set is composed of the elements in the data set.

6. The estimation device according to claim 4, characterized in that The training module is specifically used for: According to the training set and the test set, a training matrix H and a test matrix S with a sequence pattern are obtained, and a Y×4 dimensional matrix W with input weights is randomly initialized in the interval [-1 / 4, 1 / 4] A , let the 2×Y degree matrix with output weights be W B ; The hidden layer parameter U is calculated by formula 9, and the output O of the hidden layer neuron is calculated by formula 10; U=W A X T Formula 9 Among them, X T is the transpose of the matrix X; O=g(U) Formula 10 Where g() represents the activation function, which can be expressed as g(u)=(1+e -u ) -1 ; The output weight matrix W is calculated by formula 11 B , calculate the classification evaluation result Q of the training set through formula 12, and obtain the classification evaluation result q of each element in the training set through formula 13 l ; Among them, T represents the association label matrix between the hidden layer and the output layer, is the pseudo-inverse matrix of O; Q=W B O Formula 12 Among them, q l Indicates the length is F b The classification results of the training set elements, q bl represents the probability that the lth data in the training set is judged to be in working state b; according to the classification evaluation results of the training set, the classification evaluation accuracy of the training set is calculated; If the classification evaluation accuracy is greater than the threshold K1, the classification evaluation result of the test set is calculated, otherwise the Adam gradient descent algorithm is used to update the weight clustering W A , recalculate the hidden layer parameters U; The classification evaluation result matrix Z of the test set is calculated by formula 14, and the classification evaluation result of each element in the test set is obtained by formula 15; Z=W B g(W A S T ) Formula 14 Among them, z bl It represents the probability that the lth data in the test set is judged to be in working state b; According to the classification evaluation results of the test set, calculate the classification evaluation accuracy of the test set; If the classification evaluation accuracy of the test set is greater than the threshold K2, the training ends and the trained input weight matrix W is obtained. A and the output weight matrix W B Otherwise, randomly initialize the Y×4 dimension matrix W with input weights in the interval [-1 / 4,1 / 4] A , the number of training iterations is increased by 1; if the number of training iterations is greater than the threshold K3, it means that the training has not converged, the algorithm ends, and an alarm message of "training has not converged" is output, otherwise the hidden layer parameters are recalculated.

Images