CN-121999280-A - Paperboard state monitoring method and system for wooden supporting plate of papermaking wet pulp packaging machine

CN121999280ACN 121999280 ACN121999280 ACN 121999280ACN-121999280-A

Abstract

The invention provides a method and a system for monitoring the state of a paper board of a wooden supporting plate of a papermaking wet pulp packaging machine, and belongs to the technical field of automatic monitoring. The method comprises the steps of obtaining an original image of a field network camera through an RTSP protocol, carrying out standardized pretreatment, collecting pretreatment images with paper/paper-free states under different illuminations, constructing a training set and training a convolutional neural network model, obtaining and preprocessing the images in real time during production operation, loading a CNN model to carry out state inference, outputting paper or paper-free labels, and finally sending a control signal to a programmable controller through a TCP protocol according to the labels, and driving switching value output to link a packaging machine system. The invention adopts non-contact visual recognition, overcomes the problems of dependence on the physical structure of the supporting plate and misjudgment of indirect judgment logic in the traditional method, realizes stable and accurate state monitoring in a complex illumination environment, and remarkably improves the automation level and the operation reliability of the production line.

Inventors

XU FENG
LI JIANWEN
ZHANG JU
LI NA
CAO YANJUN
Zang Zijia
WANG MENG
YAN SHIYONG

Assignees

山东太阳纸业股份有限公司
南宁太阳纸业有限公司
兖州天章纸业有限公司

Dates

Publication Date: 20260508
Application Date: 20260108

Claims (10)

1. A method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine, comprising: The method comprises the steps of obtaining an original image of a supporting plate by utilizing a network camera on site of a papermaking wet pulp packaging production line through an Rtsp streaming media protocol, carrying out standardized pretreatment on the original image, and generating a standardized pretreatment image; Acquiring a plurality of standardized preprocessed images when the supporting plate is in a paper state and a paper-free state under different ambient lighting conditions, constructing a training image set, labeling a paperboard state label, training a convolutional neural network model by using the labeled training image set, and generating a trained CNN model; When the papermaking wet pulp packaging production line runs, a real-time picture of a supporting plate is obtained through a network camera and is processed to generate a real-time standardized pretreatment image; and driving the switching value output by the programmable controller according to the received signal to link the packaging machine system.
2. The method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine according to claim 1, wherein the step of obtaining an original image of the pallet by using a webcam on site of a papermaking wet pulp packaging production line through an Rtsp streaming media protocol, performing standardized preprocessing on the original image, and generating a standardized preprocessed image comprises the steps of: Logging in a designated network camera and capturing a single frame picture of a supporting plate by utilizing an integrated preset SDK function library through an Rtsp streaming media protocol, and storing the single frame picture as an original BMP image; Reading the original BMP image, separating RGB channels, respectively carrying out histogram equalization treatment on each channel, combining the treated channels to generate a combined image, wherein the histogram equalization treatment comprises the steps of calculating a pixel value cumulative distribution function and mapping according to a calculation result; And converting the combined images into a preset format and specification to generate a standardized preprocessing image.
3. The method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine according to claim 2, wherein the steps of obtaining a plurality of standardized pretreatment images of the pallet in a paper state and a paper-free state under different environmental illumination conditions, constructing a training image set, labeling a cardboard state label, training a convolutional neural network model by using the labeled training image set, and generating a trained CNN model comprise: In a preset period, controlling a network camera to respectively acquire standardized pretreatment images of the supporting plate in two states of paper and paper absence in different periods to form an original training image set; labeling havepaper labels or nohavepaper labels for each image in the original training image set by using an auxiliary labeling tool, generating a labeling image set, performing image enhancement operation on the labeling image set, and outputting a training image set; The method comprises the steps of obtaining a prediction result through forward propagation by a convolutional neural network by utilizing a training image set, calculating cross entropy loss between the prediction result and a real label, adopting a self-adaptive moment estimation optimizer, iteratively updating convolutional kernel weight and bias term parameters in the network through a backward propagation algorithm to minimize a loss function, and solidifying and storing the optimized parameters and a network structure after reaching preset iteration times or performance convergence to generate a CNN model file.
4. The method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine according to claim 3, wherein when the papermaking wet pulp packaging production line is running, a real-time picture of the pallet is acquired and processed by a network camera to generate a real-time standardized pretreatment image, a CNN model is loaded to infer the real-time standardized pretreatment image, and a cardboard state label is output, and the method comprises the following steps: When the papermaking wet pulp packaging production line runs, a real-time picture of a supporting plate is obtained by utilizing a network camera on the site of the papermaking wet pulp packaging production line through an Rtsp streaming media protocol, and standardized pretreatment is carried out on the real-time picture to generate a real-time standardized pretreatment image; Calling a forward computing engine of the CNN model, sequentially executing convolution, activation and pooling operation sequences which are solidified and stored in the CNN model, and outputting a classification probability distribution vector containing probability values of all preset classes through a full connection layer and an output layer; extracting a first probability value P1 corresponding to havepaper categories and a second probability value P0 corresponding to nohavepaper categories from the classification probability distribution vector, comparing the magnitudes of P1 and P0, determining the label corresponding to the label with the larger probability value as a final paperboard state label and outputting the final paperboard state label.
5. The method for monitoring the state of a wooden pallet of a wet pulp packaging machine for paper manufacture according to claim 4, wherein the step of writing corresponding control signals to a programmable controller according to the state label of the paper manufacture through a TCP protocol, and driving a switching value output through the programmable controller according to the received signals to link the packaging machine system comprises the steps of: In a programmable controller, a state mark register address for representing paper-on and paper-off states is predefined and mapped to a physical output point; according to the industrial communication protocol format supported by the programmable controller, assembling a writing instruction message containing the status flag register address and the logic value, and transmitting the writing instruction message through a TCP/IP network; After the programmable controller successfully receives and executes the writing instruction, the state of the internal register is updated, and the corresponding switching value output point position action is driven according to the state, so that the linkage with the main control system of the packaging machine is realized.
6. The method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine according to claim 5, wherein the programmable controller adopts an S7-200Smart series PLC, the industrial communication protocol is an ISO-on-TCP & S7 protocol, and the write instruction message comprises a TPKT head, a COTP layer, an S7 communication head and a parameter data block containing a target address and data.
7. The method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine according to claim 4, wherein when the papermaking wet pulp packaging production line is running, a real-time picture of the pallet is acquired and processed by a network camera to generate a real-time standardized pretreatment image, a CNN model is loaded to infer the real-time standardized pretreatment image, and a cardboard state label is output, and the method further comprises: When the larger one of the first probability value P1 and the second probability value P0 is lower than a preset confidence threshold, judging that the confidence coefficient is insufficient at the present time, automatically adjusting at least one image acquisition parameter of the network camera, re-acquiring and processing a real-time picture of the supporting plate after adjustment, generating a real-time standardized preprocessing image, loading a CNN model for inference, and outputting a paperboard state label.
8. A system for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine, characterized in that the system adopts the method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine according to any one of claims 1 to 7; The system comprises: the image acquisition processing module is used for acquiring an original image of the supporting plate by utilizing a network camera on the site of a papermaking wet pulp packaging production line through an Rtsp streaming media protocol, carrying out standardized pretreatment on the original image, and generating a standardized pretreatment image; The model training module is used for acquiring a plurality of standardized pretreatment images when the supporting plate is in a paper state and a paper-free state under different environmental illumination conditions, constructing a training image set, labeling a paperboard state label, training a convolutional neural network model by using the labeled training image set, and generating a trained CNN model; the real-time deducing module is used for obtaining and processing a real-time picture of the supporting plate through the network camera to generate a real-time standardized pretreatment image when the papermaking wet pulp packaging production line runs; and the linkage control module is used for writing corresponding control signals into the programmable controller through a TCP protocol according to the paperboard state label, and driving the switching value output through the programmable controller according to the received signals so as to link the packaging machine system.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor, when executing the program, carries out the steps of the method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine as claimed in any one of claims 1 to 7.
10. A storage medium having stored thereon a computer program, which when executed by a processor, carries out the steps of the method for monitoring the cardboard state of a wooden pallet of a papermaking wet pulp packaging machine according to any one of claims 1 to 7.

Description

Paperboard state monitoring method and system for wooden supporting plate of papermaking wet pulp packaging machine Technical Field The invention belongs to the technical field of automatic monitoring, and particularly relates to a method and a system for monitoring the state of a paperboard of a wooden supporting plate of a papermaking wet pulp packaging machine. Background In the automatic flow of the papermaking wet pulp packaging production line, the accurate judgment of whether a packaging paper board (wet pulp package) exists on a wooden supporting plate is a key link for realizing continuous and stable operation of equipment. The judgment result directly influences the control logic of subsequent paper taking, packaging, automatic switching of empty/full supporting plates, system start-stop and the like. Traditional monitoring methods rely mainly on physical sensors or indirect inference based on device motion feedback, and have significant limitations in practical applications. One common approach relies on machining a specific central circular through hole in the pallet and detecting the occlusion of the hole with a laser or photo sensor. The holes are shielded when the paper board exists, the state of the sensor changes, and the holes are exposed when no paper exists, so that the state is recovered. However, this method is applied on the premise that the pallet must be prefabricated with this precision hole. In practice, a large number of wooden pallets have not been designed for strength, cost or standardization considerations, resulting in a lack of versatility in this solution, which cannot be implemented in the context of non-hole pallets. Another widely adopted strategy is indirect logic judgment based on the number of "paper take failures". The system typically determines that the pallet is "out of paper" after a number of consecutive paper take actions have failed. The fundamental drawback of this approach is the causal non-uniqueness of its logic. Paper removal failure may result from a variety of factors such as insufficient vacuum suction, abnormal surface humidity of the board, or mechanical positioning errors. Therefore, even under the normal condition that the supporting plate has paper, the paper taking failure can be caused by other faults for multiple times, so that the system is misjudged as 'out of paper', and false alarm is generated or harmful follow-up actions are executed. In summary, the prior art either has insufficient versatility due to the dependence on a specific pallet physical structure, or has poor reliability due to the adoption of an indirect judgment logic, and thus cannot realize direct, reliable and universal on-line monitoring of the existence state of the wooden pallet board. The technical bottleneck directly causes that the packaging machine can not intelligently identify the empty/full supporting plates to finish automatic switching, and can not accurately trigger automatic stop when the paper is actually in shortage, thereby severely restricting the automatic and intelligent level lifting of the production line. Disclosure of Invention Aiming at the problems, the invention aims to provide the paperboard state monitoring method and system for the wooden supporting plate of the papermaking wet pulp packaging machine, which realize the direct, accurate and real-time judgment of the existence state of the wooden supporting plate paperboard and adapt to complex working conditions through the non-contact visual identification based on the convolutional neural network, effectively solve the unreliable problem caused by the fact that the traditional method depends on a specific physical structure and indirect logic judgment, and further remarkably improve the accuracy and the operation continuity of the automatic control of the packaging machine production line. The invention aims to achieve the aim, and the aim is achieved by the following technical scheme: In a first aspect, an embodiment of the present application provides a method for monitoring a cardboard state of a wooden pallet of a papermaking wet pulp packaging machine, including: The method comprises the steps of obtaining an original image of a supporting plate by utilizing a network camera on site of a papermaking wet pulp packaging production line through an Rtsp streaming media protocol, carrying out standardized pretreatment on the original image, and generating a standardized pretreatment image; Acquiring a plurality of standardized preprocessed images when the supporting plate is in a paper state and a paper-free state under different ambient lighting conditions, constructing a training image set, labeling a paperboard state label, training a convolutional neural network model by using the labeled training image set, and generating a trained CNN model; When the papermaking wet pulp packaging production line runs, a real-time picture of a supporting plate is obtained through a network camera and is