CN111822517B - Lithium battery pole piece rolling mill thickness control system based on cloud platform BP neural network - Google Patents

Abstract

The invention relates to a lithium battery pole piece rolling mill thickness control system based on a cloud platform BP neural network, which comprises a remote monitoring layer, a communication module, a field control layer and a field device layer. The remote monitoring layer is mainly used for realizing interconnection of field equipment and a cloud platform server through a communication module, collecting information such as tension, roll gap size and pole piece thickness of a winding and unwinding position in real time, obtaining deviation of pole piece thickness control through information arrangement and BP neural network algorithm analysis, optimizing and modifying a control strategy of the field control layer according to the deviation, and completing remote control and diagnosis of the lithium battery pole piece rolling mill. The invention can predict the actual processing thickness of the lithium battery pole piece in advance and compare the predicted actual processing thickness with the set processing thickness, and sends a modification strategy through a far-layer monitoring layer so as to intelligently modify the main control parameters, thereby realizing the closed-loop feedback control of equipment, greatly improving the processing precision of the pole piece thickness, and solving the problem of low precision of the thickness control system of the traditional lithium battery pole piece rolling mill.

Description

Lithium battery pole piece rolling mill thickness control system based on cloud platform BP neural network

Technical Field

The invention relates to the technical field of lithium battery pole piece rolling mill control systems, in particular to a lithium battery pole piece rolling mill thickness control system based on a cloud platform BP neural network.

Background

In lithium battery pole piece rolling mill control system, a plurality of constitutional units have involved feedback control: the device comprises a roll gap adjusting structure, a pole piece belt deviation rectifying structure at a winding and unwinding part and a tension control structure. The roll gap adjusting structure is used for adjusting the gap between the main roll and the roll to complete the adjustment of the rolling force and the thickness of the pole piece; the winding and unwinding pole piece belt deviation rectifying structure is used for carrying out deviation adjustment on the pole piece belt, and avoiding tower-shaped winding and the like caused by the deviation of the pole piece so as to influence the quality of the pole piece; the winding and unwinding tension control system is used for maintaining the stable tension of the pole pieces in the winding and unwinding process, ensuring the flatness of the rolled pole pieces and avoiding the phenomena of wave and the like caused by uneven extension of the pole pieces. The rolling process of the lithium battery pole piece can enable the matrix foil tape and active substances attached to the surface of the matrix foil tape to be more compact, so that the compaction density of the battery is maintained in a certain range space, the rate capability of the battery is kept, and the thickness precision of the lithium battery pole piece directly determines the chemical performance of the battery. In order to comply with the development trend of current intelligent manufacturing, a novel intelligent control strategy is provided for a control part of a rolling mill, and the reliability and stability of the operation of the rolling mill and the rolling speed and precision of a pole piece are improved, which is a necessary way for promoting the industrialization and scale production of the lithium battery pole piece. The prior art has the following defects:

1. the thickness of the pole piece in the production of the traditional lithium battery pole piece rolling mill is influenced by factors such as winding and unwinding tension change, rolling speed, rolling force, roll gap size and the like, and a large error exists between the actual thickness and the set thickness and is difficult to accurately control.

2. The traditional rolling mill equipment management lags behind, needs skilled workman to adjust each part according to the operation experience, can't carry out intelligent operation and control to the rolling mill.

3. Although the control effect of a high-end hydraulic servo thickness automatic control lithium battery pole piece rolling mill (AGC rolling mill) in the market at present is good, the manufacturing cost is high, and the high-end hydraulic servo thickness automatic control lithium battery pole piece rolling mill cannot be popularized and applied on a large scale.

4. The existing control system only realizes an online monitoring function, does not establish an accurate thickness prediction model, cannot reversely modify the parameters of the pole piece rolling mill, and does not realize closed-loop control between equipment and a cloud platform.

5. The thickness control system belongs to a complex industrial process, related control objects, production technology and process flow are complex, and some working conditions are obtained only by the experience of engineers. The traditional modeling mode has many defects, namely the model is too simple and cannot reflect the characteristics of a control system; if not, the model is too complicated and is difficult to implement; if the difference between the model and the actual model is too large, the established model has no significance; or whether the model is too single and not representative.

Based on the existing defects, it is necessary to design a high-precision and intelligent control system of a lithium battery pole piece rolling mill.

Disclosure of Invention

Aiming at the problems, the invention provides a lithium battery pole piece rolling mill thickness control system based on a cloud platform BP neural network.

In order to achieve the purpose, the technical scheme adopted by the invention is as follows:

a lithium battery pole piece rolling mill thickness control system based on a cloud platform BP neural network is characterized in that the control system comprises a remote monitoring layer, a communication module, a field control layer and a field device layer; the remote monitoring layer is used for realizing interconnection of a field control layer and a cloud platform server through a communication module, and a lithium battery pole piece rolling mill expert system library and a BP neural network pole piece rolling mill thickness prediction model are arranged in the cloud platform server, so that the cloud platform server can perform reverse adjustment on the field control layer of the rolling mill; and the field control layer and the field device layer are communicated with each other through RS 485.

The working process of the communication module is as follows:

firstly, user configuration, work pin initialization, timer setting and registration connection callback function are carried out to determine that the communication module is connected with the cloud platform server;

next, receiving the rolling mill thickness control parameter data through a serial port circuit on the main control system and sending the rolling mill thickness control parameter data to the cloud platform server, skipping to the communication module no matter whether the rolling mill thickness control parameter data are sent to the cloud platform server or not, receiving data fed back from the cloud platform server, feeding back the data to a main control chip on the main control system if the rolling mill thickness control parameter data are fed back from the cloud platform server, and then adjusting a field device layer to complete one-time closed-loop feedback control;

if the communication module does not receive the data sent from the cloud platform server, the step of receiving serial port data is returned, namely the rolling mill thickness control parameter data is received through the serial port circuit on the main control system, and circulation is performed again.

The communication module adopts an ESP8266 wireless module, a wireless module external circuit is carried through a chip serial port on a main control system, corresponding driving program design and development are completed, the main control chip transmits acquired data to the wireless module through the serial port, and a TCP/IP communication protocol based on a Socket interface is adopted for data transmission with the cloud platform server, so that the cloud platform server can carry out reverse control on rolling mill equipment, and the closed-loop control of a pole piece rolling mill production line is realized.

The remote monitoring layer comprises a cloud platform server, the cloud platform server selects a Linux operating system cloud platform server of the Alababa company, a B/S architecture is selected as the system architecture, the cloud platform server needs to be provided with a database and a WEB server for data storage and WEB project mounting, data interaction between the field control layer and the cloud platform server is achieved by adopting a TCP/IP communication protocol, and data interaction between the cloud platform server and the WEB client is achieved by adopting an HTTP communication protocol.

The cloud database is a Redis cloud database, and not only is data sent by a rolling mill master control system stored, but also operation data generated by a WEB client side is recorded. The process of downloading and decompressing the Redis cloud database can be completed through a wget command and a tar command in a Linux system, and then compiling and installing are completed through a make command.

The WEB client is used for remotely logging in the cloud platform to check the operation information of the control system, the users need to be subjected to authority division, different users have different management authorities and access authorities, unregistered strangers cannot access and browse production data related to lithium battery pole pieces, and loss of enterprises caused by misoperation of the system by the strangers is avoided.

The cloud platform server acquires information of control parameters of a field control layer in real time, information gathering is achieved, the deviation between the predicted thickness and the set thickness is calculated through a BP neural network pole piece rolling mill thickness prediction model, and a deviation optimization scheme is given through an expert system library. The expert system library of the lithium battery pole piece rolling mill comprises various fault solutions of the lithium battery pole piece rolling mill and a control parameter optimization scheme of a field control layer of which the deviation between the predicted thickness and the set thickness of the rolling mill is within 0.1mm, wherein the parameter optimization scheme of the field control layer can be provided when the deviation is within 0.1mm, the regulation and control precision of the optimization scheme is within 0.02mm, and the fault solution can be provided when the rolling mill breaks down. When the deviation between the predicted thickness and the set thickness of the rolling mill is more than 0.1mm, the rolling mill is in a fault condition, and the rolling mill needs to be reset manually.

The deviation is calculated as the difference between the predicted thickness given by the prediction model and the set thickness set for the production line at the beginning of production.

The field control layer comprises a main control system and an HMI (human machine interface) man-machine interaction system, the main control system takes STM32F407ZET6 as a main control chip, data transmission is carried out between the RS485 communication serial port and the field device layer, the main control system is responsible for realizing combined control of all device components of the device layer, the main control system receives deviation correction and tensile sensing signals, control signals are output through algorithm analysis, and a driver is used for driving a motor to operate to realize correction of the position and the tension of the pole piece strip. The HMI human-computer interaction system adopts a touch screen of Beijing Kunlun Tongtai, and data interaction is realized between the HMI human-computer interaction system and the main control system by adopting a ModbusRTU communication protocol, so that the real-time observation of control parameters of a field control layer and the fault information of the rolling mill can be realized, and the alarm is given when the rolling mill breaks down.

The field device layer mainly contains the basic electrical components and devices involved in the closed-loop control process. The device mainly comprises a deviation correcting and tension sensor, a servo amplifier, a frequency converter, a driving motor and other basic equipment structures. The equipment is the basis of closed-loop control, a sensor realizes the real-time acquisition of a deviation-rectifying tension signal, a servo driver realizes the control of a motor, and a frequency converter and the motor drive a mechanical structure to operate.

The master control system mainly comprises 5 control parts of unwinding deviation correction, unwinding pole piece tension, roll gap adjustment, winding deviation correction and winding pole piece tension. The deviation rectifying function of the winding and unwinding pole piece belt is to correct the position of the pole piece belt which is deviated in the operation process. During the transmission process of the pole piece belt, the pole piece can be deviated due to factors such as abrasion or deviation of a driving roller, disturbance of a mechanical structure or uneven distribution of the pole piece belt. The tension of the unwinding pole piece and the tension of the winding pole piece are generated due to the speed difference of the unwinding speed, the rolling speed and the winding speed in the pole piece rolling process.

And establishing a large database of the parameters of the lithium battery pole piece rolling mill, wherein the sample data of the BP neural network is selected from the large database of the parameters of the lithium battery pole piece rolling mill, so that the range of the selected sample data is wide and various, extreme conditions are eliminated in numerical value, the selected sample data has representativeness and authenticity, and the accuracy of a subsequent control process is ensured. Loading actual operation data of the lithium battery pole piece rolling mill equipment in the database, and acquiring the actual operation data of the lithium battery pole piece rolling mill equipment from the established large parameter database of the lithium battery pole piece rolling mill as a sample set; the actual operation data comprises the entrance thickness of the pole piece, the exit thickness of the pole piece, the entrance width of the pole piece, the rolling speed, the rolling force, the winding tension of the pole piece, the unwinding tension of the pole piece, the length of a roll gap and the actual measurement thickness of the pole piece.

The BP neural network topological structure is divided into an input layer, an output layer and a hidden layer.

Specifically, the input quantity of the input layer is a variable parameter of a control system of the pole piece rolling mill, and the variable parameter comprises: the thickness of the pole piece inlet, the thickness of the pole piece outlet, the width of the pole piece inlet, the rolling speed, the rolling force, the pole piece unwinding tension, the pole piece winding tension and the actual roll gap of the roll; the final aim is to obtain accurate target thickness of the pole piece, so that the output layer is determined to be the actually measured thickness of the pole piece, and the number of nodes of the opposite output layer is 1; the number of the nodes of the hidden layer is 5.

And an expert system library of the lithium battery pole piece rolling mill is arranged on the server in the cloud platform, and the expert system library comprises various fault solutions of the lithium battery pole piece rolling mill and a parameter optimization scheme when the predicted thickness is inconsistent with the actually measured thickness. The expert system library obtains parameter deviation of pole piece thickness control through sorting data information uploaded by the main control system and analysis based on a BP neural network algorithm, gives an optimization scheme of rolling mill equipment control parameters, judges the fault condition of the rolling mill, and displays the fault condition to a user in real time through a human-computer interaction system. The expert system library can be continuously completed by technicians with built-in database knowledge, and the accuracy of the expert system library is kept.

A construction method of a BP neural network pole piece rolling mill thickness prediction model comprises the following steps:

(1) constructing a large database of a lithium battery pole piece rolling mill, loading actual operation data of various lithium battery pole piece rolling mills in the database, and acquiring the actual operation data of the lithium battery pole piece rolling mill as a sample set;

the actual operation data comprises pole piece inlet thickness, pole piece outlet thickness, pole piece inlet width, rolling speed, rolling force, pole piece rolling tension, pole piece unreeling tension, roll gap length and pole piece actual measurement thickness;

(2) constructing a topological structure of the BP neural network;

(3) determining parameters of the BP neural network in the step (2) by using the sample set in the step (1);

(4) and loading the BP neural network model in a cloud platform server, and controlling the thickness of the pole piece according to the change of parameters of each actual operation data in the operation of the lithium battery pole piece rolling mill to adjust the thickness of the pole piece in real time.

Compared with the prior art, the invention has the beneficial effects that:

1. according to the invention, relevant parameter data can be obtained through the field control layer, the remote monitoring layer and the cloud platform server of the rolling mill production line, the parameters can be optimized, the optimized data is finally utilized to reversely adjust the field equipment layer of the rolling mill production line, and the field equipment layer can also optimize the real-time production data into the cloud platform server through the field control layer and the communication module, so that the closed-loop control of the rolling mill production line is realized.

2. According to the invention, because parameters of the lithium battery pole piece rolling mill are complex and changeable during operation, and improper data can be easily selected by taking artificial measurement data as sample data, so that the subsequent control error is larger, the method adopts a big data method (big database) to select BP neural network sample data, so that the range of the selected sample data is wide and various, extreme conditions are eliminated in numerical value, the sample data is representative and true, and the accuracy of the subsequent control process is ensured.

3. According to the invention, an accurate BP neural network model is established, the thickness of the lithium battery pole piece to be rolled is predicted and compared with the set thickness by utilizing the nonlinear fitting capacity of the BP neural network, and the intelligent adjustment of equipment parameters is realized by a cloud platform feedback signal, so that the problems that the thickness rolling force, the rolling speed, the winding and unwinding tension, the roll gap size and other factors influence in the production of a lithium battery pole piece rolling mill, and a large error exists between the actual thickness and the predicted thickness and is difficult to control are effectively solved.

4. The BP neural network algorithm is placed on a cloud platform server of a remote monitoring layer for calculation, so that the data calculation amount of the master control system is reduced, and the response speed of the master control system is effectively improved.

5. The cloud platform server is loaded with a lithium battery pole piece rolling mill expert system library, can give an optimization scheme of rolling mill equipment control parameters and judge the fault condition of the rolling mill according to data uploaded by a rolling mill master control system, and displays the fault condition to a user in real time through a human-computer interaction system. The expert system can be continuously completed by technicians with built-in database knowledge to maintain the accuracy.

In a word, the control system comprises a hardware framework of an intelligent control system of the rolling mill and the construction of a thickness prediction model of the BP neural network pole piece rolling mill. Because parameters of the lithium battery pole piece rolling mill during operation are complex and changeable, improper data can be easily selected by artificially measured data as sample data, and the subsequent control error is larger, the sample data is selected by using a big data method, so that the selected sample data is wide and various in range, extreme conditions are eliminated by numerical values, the sample data is representative and real, and the accuracy of the subsequent control process is ensured. Meanwhile, information such as tension, roll gap size, pole piece thickness and the like of a winding and unwinding position is collected in real time by utilizing problem analysis capacity of a BP neural network, the information is transmitted to a cloud platform server to be summarized, deviation of pole piece thickness control is obtained through BP neural network algorithm analysis, scheme inquiry is carried out on an expert system library of a lithium battery pole piece rolling mill loaded on the cloud platform server, then field control layer control strategies are optimized and modified, and remote control and diagnosis of the lithium battery pole piece rolling mill are completed. The problem that the actual thickness and the set thickness are difficult to control due to the fact that the rolling mill for the lithium-ion battery pole pieces is influenced by factors such as winding and unwinding tension, rolling force, rolling speed and roll gap size in production is effectively solved, the intelligent degree of a control system is greatly improved, and the adjustment work of workers on equipment is greatly reduced.

Drawings

FIG. 1 is a diagram of a hardware architecture according to the present invention.

Fig. 2 is a schematic structural diagram of a master control system according to the present invention.

Fig. 3 is a flow chart of the operation of the communication module of the present invention.

FIG. 4 is a flow chart of a control method according to the present invention.

FIG. 5 is a schematic diagram of the BP neural network structure according to the present invention.

Detailed Description

The present invention will be further described with reference to specific embodiments.

As shown in fig. 1, the control system architecture includes a remote monitoring layer, a communication module, a field control layer, and a field device layer. The remote monitoring layer realizes interconnection of the field device and the cloud platform server through the communication module, and the field control layer and the field device layer are communicated through an RS485 bus.

The remote monitoring layer comprises a cloud platform server, the cloud platform server selects a Linux operating system cloud platform server of the Alababa company, a B/S architecture is selected as the system architecture, the cloud platform server needs to be provided with a database and a WEB server for data storage and WEB project mounting, data interaction between the field control layer and the cloud platform server is achieved by adopting a TCP/IP communication protocol, and data interaction between the cloud platform server and the WEB client is achieved by adopting an HTTP communication protocol.

The cloud database selects a Redis cloud database, the process that the Redis cloud database downloads and decompresses the installation package can be completed through a wget command and a tar command in a Linux system, and then compiling and installing are completed through a make command.

The WEB client side needs to divide the authority of the users, different users have different management authorities and access authorities, unregistered strangers cannot access and browse production data related to the rapier loom, and loss of enterprises caused by misoperation of the strangers on the system is avoided.

The communication module adopts an ESP8266 wireless module, a wireless module external circuit is carried through a chip serial port on the main control system, the main control chip on the main control system transmits acquired data to the wireless module through the serial port, and a TCP/IP communication protocol based on a Socket interface is adopted for data transmission with the cloud platform server, so that the cloud platform server can carry out reverse control on rolling mill equipment, and the closed-loop control of a pole piece rolling mill production line is realized.

The field control layer comprises a main control system and an HMI (human machine interface) man-machine interaction system, the main control system takes STM32F407ZET6 as a main control chip, data transmission is carried out between the RS485 communication serial port and the field device layer, the main control system is responsible for realizing combined control of all device components of the device layer, the main control system receives deviation correction and tensile sensing signals, control signals are output through algorithm analysis, and a driver is used for driving a motor to operate to realize correction of the position and the tension of the pole piece strip. The HMI human-computer interaction system adopts a touch screen of Beijing Kunlun Tongtai, data interaction is realized between the HMI human-computer interaction system and the master control system by adopting a ModbusRTU communication protocol, a slave drive program of the master control system is completed based on the ModbusRTU communication protocol, real-time observation of control parameters of a field control layer and fault information of a rolling mill can be realized, and an alarm is given when the rolling mill breaks down.

The field device layer mainly contains the basic electrical components and devices involved in the closed-loop control process. The device mainly comprises a deviation correcting and tension sensor, a servo amplifier, a frequency converter, a driving motor and other basic equipment structures. The equipment is the basis of closed-loop control, a sensor realizes the real-time acquisition of a deviation-rectifying tension signal, a servo driver realizes the control of a motor, and a frequency converter and the motor drive a mechanical structure to operate.

As shown in fig. 2, the main control flow of the main control system is as follows: the method comprises the steps that firstly, a pole piece to be rolled is required to be released by an unwinding inflatable shaft, unwinding tension control and unwinding deviation correction control are firstly carried out before the pole piece enters a rolling part, then the pole piece to be rolled enters an upper roller and a lower roller of a rolling mill through a guide roller to be rolled, a control part for controlling the rolling speed through controlling the rotating speed of a motor and a gap adjustment control part for controlling the gap between the upper roller and the lower roller are arranged in the rolling process, the rolled pole piece is detected through winding tension control and winding deviation correction control again, and finally winding is carried out through a winding inflatable shaft.

The deviation rectifying function of the winding and unwinding pole piece belt is to correct the position of the pole piece belt which is deviated in the operation process. During the transmission process of the pole piece belt, the pole piece can be deviated due to factors such as abrasion or deviation of a driving roller, disturbance of a mechanical structure or uneven distribution of the pole piece belt.

The tension of the unwinding pole piece and the tension of the winding pole piece are generated due to the speed difference of the unwinding speed, the rolling speed and the winding speed in the pole piece rolling process. The winding and unwinding operations can be stably performed just because of the existence of tension, so that the numerical value modification of the tension is very necessary through the feedback of the cloud platform server.

As shown in fig. 3, the ESP8266 wireless communication module first performs user configuration, initialization of working pins, setting of timers, and registering a connection callback function to determine that the communication module is connected to the cloud platform server. Next, receiving the rolling mill thickness control parameter data through a serial port circuit on the main control system and sending the rolling mill thickness control parameter data to a cloud platform server, skipping to an ESP8266 wireless communication module to receive data fed back from the cloud platform no matter whether the data are sent to the cloud platform server or not, feeding the data back to a main control chip on the main control system if the data are fed back from the cloud platform, and then adjusting a field device layer to complete one-time closed-loop feedback control; if the ESP8266 wireless communication module does not receive the data sent from the cloud platform, judging whether the sending is finished or not, if not, returning to the step of receiving the serial port data, and circulating again; if the transmission is finished, the procedure is finished. The software part mainly comprises the following four parts:

(1) the user configuration program comprises the following steps: the method mainly comprises the steps of setting the working mode of the wireless communication module, connecting a user name and a password of a router, connecting single or multiple paths, and setting the IP address and the port number of a TCP server; initialization procedure of related pins.

(2) And a pin configuration program: the method mainly comprises the configuration of a serial port pin, a wireless module chip selection CH _ PD pin and a reset RST (GPIO16) pin. The input-output mode, the pin rate, etc. of the relevant pins are mainly configured.

(3) Timer clock procedure: after establishing contact with the cloud platform server, sending information to the cloud platform server at intervals to determine whether the connection is normal; and at certain intervals, the connection with the router needs to be confirmed; and if the abnormal situation occurs, reestablishing connection with the cloud platform server or the router immediately, and ensuring that data transmission continues.

(4) Wireless transmission module service function program: the part of programs are mainly used for restarting the wireless transmission module, transmitting AT instructions, transmitting character strings, receiving character strings and the like.

The ESP8266 wireless module is successfully connected to the network, and the connection with the cloud platform server must be maintained all the time to realize the real-time data transmission. And if the wireless module loses contact with the cloud platform server, the wireless module repeatedly sends a request to the cloud platform server to realize reconnection.

As shown in fig. 4, a specific control method according to an embodiment of the present invention is: firstly, a BP neural network prediction model of the thickness of the lithium battery pole piece is established on a cloud platform server, the production process of the lithium battery pole piece rolling mill is dynamically predicted, the predicted thickness close to the actual production result is obtained, and sample data is selected from a large database of the lithium battery pole piece, which is established in advance. Before the rolling mill is produced, the set thickness of the rolling mill is set, then the predicted thickness is compared with the set thickness of the rolling mill, when the deviation is not larger than 0.1mm, data are sent to an expert system library loaded in a cloud platform server for analysis, the expert system library in the cloud platform server gives a parameter optimization scheme according to the analysis result, and the deviation of the optimization scheme can reach within 0.02 mm. The parameter optimization scheme comprises the adjustment of tension, roll gap size, rolling force and rolling speed, the adjustment is fed back to the master control system through the communication module, and the master control system is used for adjusting the field device layer in real time, so that the production precision of the thickness of the pole piece is improved. And the control parameter information and the fault solution of the field control layer are displayed in real time through an industrial touch screen of the man-machine interaction system.

Before the model is established, firstly establishing a large parameter database of the lithium battery pole piece rolling mill, loading actual operation data of the lithium battery pole piece rolling mill in the database, and acquiring the actual operation data of the lithium battery pole piece rolling mill of the system as a sample set; the actual operation data comprises the entrance thickness of the pole piece, the exit thickness of the pole piece, the entrance width of the pole piece, the rolling speed, the rolling force, the winding tension of the pole piece, the unwinding tension of the pole piece, the length of a roll gap and the actual measurement thickness of the pole piece. The sample data of the BP neural network is used for intensively selecting the data of the rolling mill in steady operation from the database, so that the range of the selected sample data is wide and various, extreme conditions are eliminated in numerical value, the sample data has representativeness and authenticity, and the accuracy of the subsequent control process is ensured.

The sample data firstly uses a data normalization method to map the original value range of the processed input data and output data into the range of [ -1,1], so that the relative error and absolute error of each variable can be kept on the same level number, and the sample data is controlled in a uniform range, so that the singular data can be restrained or eliminated, and the training speed of the network is promoted.

The method comprises the steps that sample data obtained after big data screening and data processing are randomly divided into two parts by a BP neural network lithium battery pole piece thickness prediction model, and three quarters of the sample data are selected as training data from the data in the sample data and used for BP neural network training; the other one quarter is used as test data for predicting the BP neural network, and random selection is mainly used for reducing the possible correlation among sample data and improving the accuracy of a prediction result.

The BP neural network parameters include: learning rate, connection weight, activation function, iteration times and expected error, and BP neural parameters can be further optimized through the existing intelligent algorithm.

Wherein, the learning rate is optimally between 0.01 and 0.9, and the specific value can be determined only by repeatedly training and comparing sample data; the connection weight value selects a random number between (-1,1), so that the output value of the neuron weighted by initial data can be close to zero, and the connection weight value of the neuron can be adjusted in the maximum change of an S-shaped activation function; and selecting a Sigmoid function as a structural function of the neural network by the activation function selection function.

Parameter setting value of BP neural network

And the expert system library on the server in the cloud platform comprises various fault solutions of the lithium battery pole piece rolling mill and optimization schemes of field control layer control parameters when the deviation between the predicted thickness and the set thickness is less than 0.1mm, and the optimization schemes are input into the expert system library after being tested in advance by technicians. The expert system library obtains the deviation of pole piece thickness control through the arrangement of data information uploaded by the main control system and the analysis of a BP neural network algorithm, gives an optimization scheme of field control layer control parameters and judges the fault condition of the rolling mill, and displays the fault condition to a user in real time through a human-computer interaction system. The expert system library can be continuously completed by technicians with built-in database knowledge, and the accuracy of the expert system library is kept.

As shown in fig. 5, the input layer, the output layer and the hidden layer of the BP neural network need to be determined for constructing the BP neural network topology structure of the BP neural network lithium battery pole piece thickness prediction model.

Specifically, the input quantity of the input layer is a variable parameter of a control system of the pole piece rolling mill, and the variable parameter comprises: the thickness of the pole piece inlet, the thickness of the pole piece outlet, the width of the pole piece inlet, the rolling speed, the rolling force, the pole piece unwinding tension, the pole piece winding tension and the actual roll gap of the roll; the final aim is to obtain accurate target thickness of the pole piece, so that the output layer is determined to be the actually measured thickness of the pole piece, and the number of nodes of the opposite output layer is 1; the number of the nodes of the hidden layer is selected to be 5.

The invention is not the best known technology.

The above embodiments are merely illustrative of the technical ideas and features of the present invention, and the purpose thereof is to enable those skilled in the art to understand the contents of the present invention and implement the present invention, and not to limit the protection scope of the present invention. All equivalent changes and modifications made according to the spirit of the present invention should be covered within the protection scope of the present invention.

Claims (7)

1. A lithium battery pole piece rolling mill thickness control system based on a cloud platform BP neural network is characterized in that the control system comprises a remote monitoring layer, a communication module, a field control layer and a field device layer; the remote monitoring layer is used for realizing interconnection of a field control layer and a cloud platform server through a communication module, and a lithium battery pole piece rolling mill expert system library and a BP neural network pole piece rolling mill thickness prediction model are arranged in the cloud platform server, so that the cloud platform server can perform reverse adjustment on the field control layer of the rolling mill; the field control layer and the field equipment layer are communicated with each other;

the working process of the communication module is as follows:

firstly, user configuration, work pin initialization, timer setting and registration connection callback function are carried out to determine that the communication module is connected with the cloud platform server;

next, receiving the rolling mill thickness control parameter data through a serial port circuit on the main control system and sending the rolling mill thickness control parameter data to the cloud platform server, skipping to the communication module no matter whether the rolling mill thickness control parameter data are sent to the cloud platform server or not, receiving data fed back from the cloud platform server, feeding back the data to a main control chip on the main control system if the rolling mill thickness control parameter data are fed back from the cloud platform server, and then adjusting a field device layer to complete one-time closed-loop feedback control;

if the communication module does not receive the data sent from the cloud platform server, returning to the step of receiving the serial port data, namely receiving the rolling mill thickness control parameter data through a serial port circuit on the main control system, and circulating again;

the cloud platform server acquires parameter information of a field control layer in real time, the information is summarized, parameter deviation of pole piece thickness control is obtained through a BP neural network pole piece rolling mill thickness prediction model, and the strategy of the field control layer is optimized and modified according to the deviation, so that the remote control of the lithium battery pole piece rolling mill is completed; the expert system library of the lithium battery pole piece rolling mill comprises a lithium battery pole piece rolling mill fault solution scheme and a parameter optimization scheme of which the deviation between the predicted thickness and the set thickness is within 0.1mm, so that the optimization deviation is kept within 0.02 mm;

the thickness prediction model of the BP neural network pole piece rolling mill is divided into an input layer, an output layer and a hidden layer, wherein the input quantity of the input layer comprises the following components: the thickness of the pole piece inlet, the thickness of the pole piece outlet, the width of the pole piece inlet, the rolling speed, the rolling force, the pole piece unwinding tension, the pole piece winding tension and the actual roll gap of the roll; the output layer is actually measured thickness of the pole piece.

2. The lithium battery pole piece rolling mill thickness control system based on the cloud platform BP neural network as claimed in claim 1, wherein the number of nodes of the relative output layer is 1; the number of the nodes of the hidden layer is 5.

3. The lithium battery pole piece rolling mill thickness control system based on the cloud platform BP neural network of claim 1, characterized in that: establishing a large database of lithium battery pole piece rolling mill parameters, loading actual operation data of lithium battery pole piece rolling mill equipment in the database, and acquiring the actual operation data of the lithium battery pole piece rolling mill equipment as a BP neural network sample set; the actual operation data comprises the entrance thickness of the pole piece, the exit thickness of the pole piece, the entrance width of the pole piece, the rolling speed, the rolling force, the winding tension of the pole piece, the unwinding tension of the pole piece, the length of a roll gap and the actual measurement thickness of the pole piece.

4. The lithium battery pole piece rolling mill thickness control system based on the cloud platform BP neural network of claim 1, characterized in that: the field control layer comprises a main control system and an HMI (human machine interface) man-machine interaction system, the main control system takes an STM32F407ZET6 as a main control chip and performs data transmission with the field equipment layer through an RS485 communication serial port, the HMI man-machine interaction system and the main control system adopt a ModbusRTU communication protocol to realize data interaction, and a slave machine driving program of the main control system is completed based on the ModbusRTU communication protocol; the system is responsible for realizing combined control of all equipment components of a field equipment layer and correcting control parameters by adopting an optimization strategy fed back by a remote monitoring layer; the control parameter information of the field control layer and the fault information of the rolling mill can be observed in real time, and an alarm is given when the rolling mill breaks down.

5. The lithium battery pole piece rolling mill thickness control system based on the cloud platform BP neural network of claim 1, characterized in that: the master control system comprises 5 control parts of unwinding deviation correction, unwinding pole piece tension, roll gap adjustment, winding deviation correction and winding pole piece tension; each control section is capable of closed loop adjustment by the master control system.

6. The lithium battery pole piece rolling mill thickness control system based on the cloud platform BP neural network of claim 1, characterized in that: the communication module adopts an ESP8266 wireless module, information transmission between the Internet and the master control system is achieved through WIFI and a serial port, the cloud platform server can carry out reverse control on the rolling mill equipment, and closed-loop control of the pole piece rolling mill production line is achieved.

7. The lithium battery pole piece rolling mill thickness control system based on the cloud platform BP neural network of claim 1, characterized in that: the construction method of the BP neural network pole piece rolling mill thickness prediction model comprises the following steps:

(1) constructing a large database of a lithium battery pole piece rolling mill, loading actual operation data of various lithium battery pole piece rolling mills in the database, and acquiring the actual operation data of the lithium battery pole piece rolling mill as a sample set;

the actual operation data comprises pole piece inlet thickness, pole piece outlet thickness, pole piece inlet width, rolling speed, rolling force, pole piece rolling tension, pole piece unreeling tension, roll gap length and pole piece actual measurement thickness;

(2) constructing a topological structure of the BP neural network;

(3) determining parameters of the BP neural network in the step (2) by using the sample set in the step (1);

(4) and loading the BP neural network model in a cloud platform server, and controlling the thickness of the pole piece according to the change of parameters of each actual operation data in the operation of the lithium battery pole piece rolling mill to adjust the thickness of the pole piece in real time.

Images