A process monitoring method and system for UV label paper production

Abstract

This application relates to the technical field of UV label paper production process monitoring technology, and more particularly to a process monitoring method and system for UV label paper production. The method comprises: reading preset production process type information stored in an RFID chip embedded in a target UV label paper; the preset type information includes at least one of the following: line width error of the RFID antenna embedded in the target UV label paper, spacing error of the RFID antenna, silver paste printing thickness, contact resistance at the connection point between the RFID chip and the RFID antenna, ambient temperature, and ambient humidity; and inputting the read preset type information into a trained target neural network model to obtain a read failure rate for the RFID chip embedded in the target UV label paper. This invention can improve the efficiency and accuracy of process monitoring for UV label paper production.

Description

Technical Field

[0001] The present invention relates to the technical field of UV label paper production process monitoring, and in particular to a process monitoring method and system for UV label paper production. Background Art

[0002] During the production of UV label paper, which embeds the RFID chip and antenna into flexible substrates like PET, production parameters directly impact the RFID chip's readability. Traditional UV label production monitoring methods rely primarily on manual spot checks or independent monitoring of a single parameter to determine RFID chip readability, making it difficult to ensure efficient and accurate monitoring. Improving the efficiency and accuracy of process monitoring during UV label production is a pressing issue. Summary of the Invention

[0003] The present invention aims to provide a process monitoring method and system for UV label paper production, so as to improve the efficiency and accuracy of process monitoring of UV label paper production.

[0004] According to the present invention, a process monitoring method for UV label paper production is provided, comprising the following steps:

[0005] S100, reading preset type information of the production process stored in the RFID chip embedded on the target UV label paper; the preset type information includes at least one of the following information: line width error of the RFID antenna embedded on the target UV label paper, spacing error of the RFID antenna, silver paste printing thickness, contact resistance of the connection point between the RFID chip and the RFID antenna, ambient temperature and ambient humidity.

[0006] S200: Input the read preset type information into the trained target neural network model to obtain the reading failure rate of the RFID chip embedded in the target UV label paper. The acquisition process of the trained target neural network model includes:

[0007] S210, read the preset type information of the production process stored in the RFID chip embedded in each historical UV label paper in the historical UV label paper set; the historical UV label paper is the UV label paper produced by the target UV label paper production equipment within the historical time period, and the target UV label paper production equipment is the equipment that produces the target UV label paper.

[0008] S220: Perform fuzzy processing on each preset type information of each historical UV label paper read to obtain the fuzzy processed preset type information.

[0009] S230: Obtain the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper according to the preset type information after fuzzification processing and the preset fuzzy rule library.

[0010] S240, judging whether each historical UV label paper is a positive sample or a negative sample according to the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper, and retraining the trained initial neural network model according to the judgment result to obtain a trained target neural network model.

[0011] Compared with the prior art, the present invention has at least the following beneficial effects:

[0012] The present invention infers the RFID chip's read failure rate by reading preset information stored in the RFID chip embedded in UV label paper, including line width error, spacing error, silver paste thickness, contact resistance, and environmental parameters (temperature and humidity). This replaces the traditional manual process of determining the RFID chip's read failure rate, improving the efficiency of determining the RFID chip's read failure rate. Furthermore, the present invention uses a fuzzy algorithm to first delineate positive and negative samples during the acquisition of the model used to infer the RFID chip's read failure rate. This allows the trained initial neural network model (trained based on production data from other UV label paper production equipment, not the target UV label paper production equipment) to be retrained based on the judgment results of the positive and negative samples. This improves the adaptability of the trained target neural network model to the target UV label paper production equipment (i.e., the equipment that produces the target UV label paper and the historical UV label paper) and increases the accuracy of the RFID chip read failure rate output by the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0014] Figure 1 This is a flow chart of a process monitoring method for UV label paper production provided in Example 1 of the present invention;

[0015] Figure 2 This is a flowchart of the process of acquiring a trained target neural network model provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0017] Example 1:

[0018] According to this embodiment, Figure 1 As shown, a process monitoring method for UV label paper production is provided, comprising the following steps:

[0019] S100, reading preset type information of the production process stored in the RFID chip embedded on the target UV label paper; the preset type information includes at least one of the following information: line width error of the RFID antenna embedded on the target UV label paper, spacing error of the RFID antenna, silver paste printing thickness, contact resistance of the connection point between the RFID chip and the RFID antenna, ambient temperature and ambient humidity.

[0020] As an optional specific implementation method, an AOI visual inspection system (accuracy ±1μm) is used to collect antenna line width error and spacing error; a laser thickness gauge (accuracy ±1μm) is used to scan the silver paste printing thickness; an automatic flying probe tester (accuracy ±1mΩ) is used to detect contact resistance; and a temperature and humidity sensor (accuracy ±1℃ / ±2%RH) is used to monitor the ambient temperature and humidity.

[0021] As an optional specific implementation, the production parameters stored in the RFID chip are read by an RFID reader; optionally, the storage format of the preset type information is binary encoding (each preset type information occupies 16 bits, and the accuracy is retained to two decimal places).

[0022] S200: Input the read preset type information into the trained target neural network model to obtain the reading failure rate of the RFID chip embedded in the target UV label paper.

[0023] In this embodiment, Figure 2 As shown, the process of acquiring the trained target neural network model includes:

[0024] S210, read the preset type information of the production process stored in the RFID chip embedded in each historical UV label paper in the historical UV label paper set; the historical UV label paper is the UV label paper produced by the target UV label paper production equipment within the historical time period, and the target UV label paper production equipment is the equipment that produces the target UV label paper.

[0025] In this embodiment, the historical UV label paper and the target UV label paper were produced by the same UV label paper production equipment. The historical time period is the time before the target UV label paper production equipment produced the target UV label paper. Optionally, the historical time period is a set of historical times that are less than or equal to a preset time period since the target UV label paper production equipment produced the target UV label paper, where the preset time period is an empirical value.

[0026] S220: Perform fuzzy processing on each preset type information of each historical UV label paper read to obtain the fuzzy processed preset type information.

[0027] Those skilled in the art will appreciate that any fuzzification process in the prior art falls within the scope of protection of the present invention. As an optional embodiment, the fuzzified information obtained by fuzzifying any preset type of information corresponds to three levels: low, medium, and high. As another optional embodiment, the fuzzified information obtained by fuzzifying any preset type of information corresponds to five levels: low, lower, medium, higher, and high.

[0028] S230: Obtain the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper according to the preset type information after fuzzification processing and the preset fuzzy rule library.

[0029] In this embodiment, the preset fuzzy rule base is a pre-established fuzzy rule base; as an optional specific implementation, the preset type information includes the line width error of the RFID antenna embedded in the target UV label paper, the spacing error of the RFID antenna, the silver paste printing thickness, and the contact resistance at the connection point between the RFID chip and the RFID antenna. Then, the form of each rule is: If the line width error of the RFID antenna is A, the spacing error of the RFID antenna is B, the silver paste printing thickness is C, and the contact resistance at the connection point between the RFID chip and the RFID antenna is D, then the RFID chip read failure rate is F. Where A is any fuzzy line width error, B is any fuzzy spacing error, C is any fuzzy silver paste printing thickness, D is any fuzzy contact resistance, and F is any fuzzy read failure rate.

[0030] As a specific embodiment, the blurred line width error is a low line width error, a medium line width error, or a high line width error; the blurred line width error is a low spacing error, a medium spacing error, or a high spacing error; the blurred silver paste printing thickness is a low silver paste printing thickness, a medium silver paste printing thickness, or a high silver paste printing thickness; the blurred contact resistance is a low contact resistance, a medium contact resistance, or a high contact resistance. As another optional specific embodiment, the blurred line width error is a low line width error, a relatively low line width error, a medium line width error, a relatively high line width error, or a high line width error; the blurred line width error is a low spacing error, a relatively low spacing error, a medium spacing error, a relatively high spacing error, or a high spacing error; the blurred silver paste printing thickness is a low silver paste printing thickness, a relatively low silver paste printing thickness, a medium silver paste printing thickness, a relatively high silver paste printing thickness, or a high silver paste printing thickness; the blurred contact resistance is a low contact resistance, a relatively low contact resistance, a medium contact resistance, a relatively high contact resistance, or a high contact resistance.

[0031] As an optional specific embodiment, the fuzzy read failure rate is a low read failure rate, a medium read failure rate, or a high read failure rate. As another optional specific embodiment, the fuzzy read failure rate is a low read failure rate, a relatively low read failure rate, a medium read failure rate, a relatively high read failure rate, or a high read failure rate.

[0032] S240, judging whether each historical UV label paper is a positive sample or a negative sample according to the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper, and retraining the trained initial neural network model according to the judgment result to obtain a trained target neural network model.

[0033] In this embodiment, the trained initial neural network model is trained based on the production data of other UV label paper production equipment, and is not trained using the production data of the target UV label paper production equipment. Therefore, the trained initial neural network model may not be suitable for the target UV label paper production equipment, that is, the trained initial neural network model may have a low accuracy in inferring the reading failure rate of the RFID chip based on the preset type information of the production process stored in the RFID chip embedded on the UV label paper produced by the target UV label paper production equipment. Therefore, this embodiment also retrains the trained initial neural network model based on the data corresponding to the historical UV label paper produced by the target UV label paper production equipment to improve the adaptability of the trained target neural network model to the target UV label paper production equipment, and improve the accuracy of the reading failure rate of the RFID chip (embedded on the UV label paper produced by the target UV label paper production equipment) output by the trained target neural network model.

[0034] As an optional specific implementation, judging whether each historical UV label paper is a positive sample or a negative sample based on the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper includes: if the fuzzy reading failure rate of the RFID chip corresponding to a certain historical UV label paper matches the corresponding true reading failure rate, then the historical UV label paper is judged to be a positive sample; otherwise, the historical UV label paper is judged to be a negative sample. Optionally, the fuzzy reading failure rate of the RFID chip corresponding to the historical UV label paper matches the corresponding true reading failure rate means that the true reading failure rate of the RFID chip corresponding to the historical UV label paper falls within the reading failure rate range corresponding to the fuzzy reading failure rate. It should be understood that any fuzzy reading failure rate corresponds to a reading failure rate range, and the reading failure rate range corresponding to any fuzzy reading failure rate is preset.

[0035] As an optional embodiment, the target neural network model includes an input layer, a hidden layer, and an output layer. As an optional embodiment, the preset type information includes the line width error of the RFID antenna embedded in the target UV label paper, the spacing error of the RFID antenna, the silver paste printing thickness, and the contact resistance of the connection point between the RFID chip and the RFID antenna; correspondingly, the number of nodes corresponding to the input layer is 4, the hidden layer includes a first hidden layer and a second hidden layer, the number of neurons in the first hidden layer is 32, and the activation function is ReLU; the number of neurons in the second hidden layer is 16, and the activation function is ReLU; the number of nodes in the output layer is 1, and the activation function is Sigmoid.

[0036] As a preferred embodiment, S240 includes:

[0037] S241, if the judgment result is that a certain historical UV label paper is a positive sample, the preset type information corresponding to the historical UV label paper and the corresponding actual reading failure rate constitute a sample data and append it to the positive sample library; otherwise, the preset type information corresponding to the historical UV label paper and the corresponding actual reading failure rate constitute a sample data and append it to the negative sample library.

[0038] S242, selecting samples from the positive sample library and the negative sample library according to a preset positive-negative sample ratio to perform a first round of retraining on the trained initial neural network model; the preset positive-negative sample ratio is the ratio of positive samples and negative samples used in the training process of obtaining the trained initial neural network model.

[0039] In this embodiment, the ratio of positive and negative samples for retraining the trained initial neural network model is determined based on the ratio of positive samples and negative samples used in the training process of the trained initial neural network model. This is beneficial to maintaining the feature distribution pattern learned by the trained initial neural network model, effectively inheriting the implicit bias established by the trained initial neural network model under a specific sample distribution, and continuing the trained initial neural network model's ability to identify specific types of faults. It can achieve accurate modeling of the target UV label paper production equipment while retaining general knowledge.

[0040] S243, if the difference between the accuracy of the trained initial neural network model after the first round of retraining and the accuracy before the first round of retraining is less than or equal to a preset accuracy threshold, the neural network model obtained after the first round of retraining of the trained initial neural network model is determined as the trained target neural network model; otherwise, samples are selected from the positive sample library and the negative sample library according to the preset positive and negative sample ratio to perform a second round of retraining on the initial neural network model.

[0041] Optionally, the preset accuracy threshold is an empirical value.

[0042] In this embodiment, if the difference between the accuracy rate of the trained initial neural network model after the second round of retraining and the accuracy rate before the second round of retraining is less than or equal to a preset accuracy threshold, the neural network model obtained after the second round of retraining of the trained initial neural network model is determined as the trained target neural network model; otherwise, samples are selected from the positive sample library and the negative sample library according to the preset positive-negative sample ratio to perform a third round of retraining on the trained initial neural network model. This process is repeated until the difference between the accuracy rate of the trained initial neural network model after retraining and the accuracy rate before retraining is less than or equal to the preset accuracy threshold, and the neural network model obtained after retraining the trained initial neural network model is determined as the trained target neural network model.

[0043] This embodiment infers the RFID chip's read failure rate by reading preset information such as line width error, spacing error, silver paste thickness, contact resistance, and environmental parameters (temperature and humidity) stored in the RFID chip embedded in the UV label paper. This replaces the traditional manual process of determining the RFID chip's read failure rate, improving the efficiency of determining the RFID chip's read failure rate. Furthermore, this embodiment uses a fuzzy algorithm to first delineate positive and negative samples during the process of acquiring the model used to infer the RFID chip's read failure rate. This allows the trained initial neural network model (trained based on production data from other UV label paper production equipment, not the target UV label paper production equipment) to be retrained based on the judgment results of the positive and negative samples. This improves the adaptability of the trained target neural network model to the target UV label paper production equipment (i.e., the equipment that produces the target UV label paper and the historical UV label paper) and increases the accuracy of the RFID chip read failure rate output by the model.

[0044] Example 2:

[0045] This embodiment provides a process monitoring system for UV label paper production, including: a processor and a storage medium, wherein the storage medium stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded and executed by the processor to implement the following steps:

[0046] S100, reading preset type information of the production process stored in the RFID chip embedded on the target UV label paper; the preset type information includes at least one of the following information: line width error of the RFID antenna embedded on the target UV label paper, spacing error of the RFID antenna, silver paste printing thickness, contact resistance of the connection point between the RFID chip and the RFID antenna, ambient temperature and ambient humidity.

[0047] S200: Input the read preset type information into the trained target neural network model to obtain the reading failure rate of the RFID chip embedded in the target UV label paper. The acquisition process of the trained target neural network model includes:

[0048] S210, read the preset type information of the production process stored in the RFID chip embedded in each historical UV label paper in the historical UV label paper set; the historical UV label paper is the UV label paper produced by the target UV label paper production equipment within the historical time period, and the target UV label paper production equipment is the equipment that produces the target UV label paper.

[0049] S220: Perform fuzzy processing on each preset type information of each historical UV label paper read to obtain the fuzzy processed preset type information.

[0050] S230: Obtain the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper according to the preset type information after fuzzification processing and the preset fuzzy rule library.

[0051] S240, judging whether each historical UV label paper is a positive sample or a negative sample according to the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper, and retraining the trained initial neural network model according to the judgment result to obtain a trained target neural network model.

[0052] Although some specific embodiments of the present invention have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A process monitoring method for UV label paper production, characterized in that: The following steps are involved: S100, reading preset type information of the production process stored in the RFID chip embedded in the target UV label paper; the preset type information includes at least one of the following information: line width error of the RFID antenna embedded in the target UV label paper, spacing error of the RFID antenna, silver paste printing thickness, contact resistance of the connection point between the RFID chip and the RFID antenna, ambient temperature, and ambient humidity; S200, inputting the read preset type information into the trained target neural network model to obtain a reading failure rate of the RFID chip embedded in the target UV label paper; The process of obtaining the trained target neural network model includes: S210, reading preset type information of the production process stored in the RFID chip embedded in each historical UV label paper in the historical UV label paper set; the historical UV label paper is the UV label paper produced by the target UV label paper production equipment within the historical time period, and the target UV label paper production equipment is the equipment that produces the target UV label paper; S220, performing fuzzy processing on each preset type information of each historical UV label paper read to obtain the fuzzy processed preset type information; S230, obtaining the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper according to the preset type information after fuzzification processing and the preset fuzzy rule library; S240, judging whether each historical UV label paper is a positive sample or a negative sample according to the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper, and retraining the trained initial neural network model according to the judgment result to obtain a trained target neural network model.

2. The process monitoring method for UV label paper production according to claim 1, characterized in that: The S240 includes: S241, if the judgment result is that a certain historical UV label paper is a positive sample, the preset type information corresponding to the historical UV label paper and the corresponding actual reading failure rate constitute a sample data and append it to the positive sample library; otherwise, the preset type information corresponding to the historical UV label paper and the corresponding actual reading failure rate constitute a sample data and append it to the negative sample library; S242, selecting samples from the positive sample library and the negative sample library according to a preset positive-negative sample ratio to perform a first round of retraining on the trained initial neural network model; the preset positive-negative sample ratio is the ratio of positive samples to negative samples used in the training process of obtaining the trained initial neural network model; S243, if the difference between the accuracy of the trained initial neural network model after the first round of retraining and the accuracy before the first round of retraining is less than or equal to a preset accuracy threshold, the neural network model obtained after the first round of retraining of the trained initial neural network model is determined as the trained target neural network model; otherwise, samples are selected from the positive sample library and the negative sample library according to the preset positive and negative sample ratio to perform a second round of retraining on the initial neural network model.

3. The process monitoring method for UV label paper production according to claim 1, characterized in that: Judging whether each historical UV label paper is a positive sample or a negative sample according to the fuzzy reading failure rate of the RFID chip corresponding to each historical UV label paper includes: if the fuzzy reading failure rate of the RFID chip corresponding to a certain historical UV label paper matches the corresponding true reading failure rate, then the historical UV label paper is judged to be a positive sample; otherwise, the historical UV label paper is judged to be a negative sample.

4. The process monitoring method for UV label paper production according to claim 1, characterized in that: The target neural network model includes an input layer, a hidden layer and an output layer.

5. A process monitoring system for UV label paper production, characterized in that: The system includes: a processor and a storage medium, wherein the storage medium stores at least one program, and the at least one program is loaded and executed by the processor to implement the process monitoring method for UV label paper production according to any one of claims 1 to 4.

Images