Deep UV glue curing method and system

Abstract

The application belongs to the technical field of UV glue curing, and discloses a deep-layer UV glue curing method and system. The method preliminarily sets the incident intensity I0 of an auxiliary ultraviolet curing light source below a to-be-processed deep-layer UV glue product according to the obtained dynamic change result of the refractive index of the UV glue; uses nondestructive flaw detection to detect the overall real-time curing state of the to-be-processed deep-layer UV glue product and acquires a real-time curing state image; a control end is compared with a pre-stored target sample, and according to the comparison information, the auxiliary ultraviolet curing light source is instructed to change the incident intensity I2, so that the incident intensity I2 meets the curing condition of the to-be-processed deep-layer UV glue product. The application adds an auxiliary ultraviolet curing light source at the bottom of the product, the auxiliary ultraviolet curing lamp mainly increases the energy conductivity of the glue and provides part of the energy required for the curing of the glue, and finally enables the deep-layer glue to be completely cured.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of UV glue curing, and particularly relates to a deep-layer UV glue curing method and system. BACKGROUND

[0002] The prior art uses one ultraviolet curing light source to irradiate from the top of the product downwards to cure the UV glue of the product. The conventional UV glue curing is carried out by using a mercury lamp, which has the advantages of good curing effect, firmness and fast surface drying, but has the disadvantages of large power consumption, large volume and inability to be turned on and off at any time. People have carried out continuous research and development, and use LED as the light source to cure the UV glue, but need to replace the photosensitizer, which can reduce the energy consumption and has a small volume, but the cost of the used glue is high, the surface cannot be completely dried, and the overall cost is high.

[0003] In the ultraviolet curing nanoimprint technology, the cured glue is a core material used for preparing components, and the evolution of the mechanical and optical properties in the forming process has a decisive influence on the use performance of the components including optical uniformity. Therefore, dynamic quantitative detection of the evolution of the optical properties (mainly the refractive index) of the cured glue during the curing process under irradiation of ultraviolet light of different intensities is of great significance.

[0004] The refractive index of the cured glue changes little before and after curing, and due to the introduction of the ultraviolet light source, the traditional ellipsometry method cannot complete such measurement tasks, and it is necessary to use optical principles to independently measure and design and build the optical system.

[0005] The prior art invention patent "Method for detecting dynamic optical properties of UV glue curing process" (publication number CN103969220A, publication date 20140806) discloses the following steps: preparing a micro waveguide cavity, pouring UV glue into the cavity; joining the waveguide cavity with two glass sheets coated with a thin film of noble metal on one side to form a double-sided metal-coated waveguide structure; building an incident light path, and curing with a UV lamp from one side; using a position-sensitive detector to detect the lateral displacement of the laser when it propagates in the waveguide structure and undergoes total reflection; and through the lateral displacement of the laser when it propagates in the waveguide structure and produces a guided mode, which is greatly enhanced at the position where the lateral displacement appears a sharp peak, the mode number corresponding to the position, the known incident angle, and the mode eigen equation of the waveguide structure, the dynamic change of the dielectric constant of the UV glue is deduced, and then the dynamic change of the refractive index of the UV glue can be calculated.

[0006] Through the above analysis, the problems and defects of the prior art are that the irradiation intensity of the ultraviolet curing light source is not effectively controlled, so that the bottom of the glue with a thick curing thickness cannot be completely cured. SUMMARY

[0007] To overcome the problems in the related art, the embodiment of the present application provides a deep UV glue curing method and system.

[0008] The technical solution is as follows: the deep UV glue curing method comprises:

[0009] S1, according to the dynamic change result of the refractive index of the obtained UV glue, setting the incident intensity I0 of the auxiliary ultraviolet curing light source below the deep UV glue product to be processed;

[0010] S2, using non-destructive testing to detect the overall real-time curing state of the deep UV glue product to be processed, and obtaining a real-time curing state image;

[0011] S3, the control end receives the real-time curing state image, compares it with the pre-stored target sample, and according to the comparison information, controls the auxiliary ultraviolet curing light source to change the incident intensity I2, so that the changed incident intensity I2 meets the conditions of the overall curing of the deep UV glue product to be processed.

[0012] In step S3, the changed incident intensity I2 meets the conditions of the overall curing of the deep UV glue product to be processed, comprising:

[0013] S301, constructing a real-time curing state image model of the deep UV glue product to be processed, and obtaining the current incident intensity I1 of the auxiliary ultraviolet curing light source; based on the constructed real-time curing state image model of the deep UV glue product to be processed, establishing an adjustment strategy model of the auxiliary ultraviolet curing light source;

[0014] S302, using a deep learning network to extract feature information between state nodes of the current incident intensity I1, and combining messages through a clustering method;

[0015] S303, using a controller of the control end to perform scheduling strategy, using an evaluation module of the control end to evaluate the strategy effect of the controller of the control end, and using a deep learning network to update the state of the current incident intensity I1 state node;

[0016] S304, judging whether the scheduling result meets the preset standard, if yes, outputting the scheduling scheme; otherwise, using a strategy optimization model to train parameters of the deep learning network, the controller of the control end and the evaluation module of the control end, and obtaining the changed incident intensity I0 of the auxiliary ultraviolet curing light source.

[0017] In step S301, obtaining the current incident intensity I1 of the auxiliary ultraviolet curing light source comprises:

[0018] (1) constructing a real-time curing state image model of the deep UV glue product to be processed, as follows:

[0019] H = ((U, S) U K)

[0020] H = ((U, S) U K)

[0021] (2) Obtain the current intensity I1 of the auxiliary ultraviolet curing light source, use the current intensity I1 of the auxiliary ultraviolet curing light source to represent the real-time state of the dynamic scheduling of the auxiliary ultraviolet curing light source, and use the current intensity I1 of the auxiliary ultraviolet curing light source as the characteristic information of U. The characteristic information of U includes: the current intensity I1 state node state, the processing time, the number of subsequent auxiliary ultraviolet curing light source intensity adjustment stages, and the start time of the auxiliary ultraviolet curing light source intensity adjustment stage. The current intensity I1 state node state includes [1, 0, 0], [0, 1, 0] and [0, 0, 1], wherein [1, 0, 0], [0, 1, 0] and [0, 0, 1] respectively represent that the auxiliary ultraviolet curing light source intensity adjustment stage has not started, the auxiliary ultraviolet curing light source intensity adjustment stage is being processed, and the auxiliary ultraviolet curing light source intensity adjustment stage has been completed. The processing time is the processing time of the auxiliary ultraviolet curing light source intensity adjustment stage. The number of subsequent auxiliary ultraviolet curing light source intensity adjustment stages is the number of subsequent auxiliary ultraviolet curing light source intensity adjustment stages of the same to-be-processed deep UV glue product auxiliary ultraviolet curing light source intensity adjustment stage U. The start time of the auxiliary ultraviolet curing light source intensity adjustment stage is the start processing time of the auxiliary ultraviolet curing light source intensity adjustment stage U.

[0022] In step S301, an auxiliary ultraviolet curing light source adjustment strategy model is established, including:

[0023] The auxiliary ultraviolet curing light source adjustment strategy tuple is (M, B, O, T, x), wherein M represents an auxiliary ultraviolet curing light source scheduling state, B represents a scheduling action, O represents a state transition probability, T represents a reward obtained by each action, and the scheduling target is to minimize the completion time; and x represents a discount factor, that is, the influence degree of the current action on future rewards.

[0024] The expression of the reward T obtained by each action is:

[0025]

[0026] In the formula, makespan() is the total processing time of the deep UV glue product sequence to be processed, H is the current state, and H' is the next state.

[0027] In step S302, the feature information between the current incident intensity I1 state nodes is extracted by using a deep learning network, and the messages are combined by a clustering method, including:

[0028] (a) embedding the incident intensity I1 state node, extracting the feature information between different incident intensity I1 state nodes, and the expression is:

[0029]

[0030] In the formula, is the embedding information of the tth generation incident intensity I1 state node, is a posterior incident intensity I1 state node update function, and relu() is a vector concatenation function, is a prior incident intensity I1 state node update function, Rp(U) is a prior incident intensity I1 state node set, and Rq(U) is a posterior incident intensity I1 state node set, is a target incident intensity I1 state node update function, is the embedding information of the k-1th generation incident intensity I1 state node;

[0031] (b) when the incident intensity I1 state node of the auxiliary ultraviolet curing light source incident intensity adjustment stage is completed or does not exist, the incident intensity I1 state node embedding is set to a zero vector with the same dimension as H and no longer participates in state updating;

[0032] (c) the parameters of the input layer, the hidden layer and the output layer of the deep learning network are trained by using a multilayer perception.

[0033] In step S303, the current incident intensity I1 state node state is updated by using a deep learning network, including:

[0034] (i) generating a probability distribution of a target control end selecting an operable auxiliary ultraviolet curing light source incident intensity adjustment stage, and the expression is:

[0035]

[0036] wherein, is a probability distribution value, is a real-time curing state image set of the deep-UV adhesive product to be processed under k nodes at time τ, is the kth node state node embedding information of the incident intensity I1, ζ5 is a scheduling node, H τ is a real-time curing state image set of the deep-UV adhesive product to be processed under k nodes at time τ, is a set of auxiliary ultraviolet curing light source incident intensity adjustment stages available at time τ, m l is a differentiable function that maps the incident intensity I1 state node embedding information to the logarithmic value of each incident intensity I1 state node;

[0037] (ii) a random strategy is used to select the auxiliary ultraviolet curing light source incident intensity adjustment stage, and the state value is solved by using the evaluation module of the control end, and the expression is:

[0038]

[0039] wherein, U() is a set function of the auxiliary ultraviolet curing light source incident intensity adjustment stage, is a real-time curing state image set of the deep-UV adhesive product to be processed under k nodes at time τ, m U is a differentiable function, and U is a set of auxiliary ultraviolet curing light source incident intensity adjustment stages, is a real-time curing state image set of the deep-UV adhesive product to be processed under k nodes at time i, ζ6 is a conversion node;

[0040] (iii) an initial state including the number of processing control ends, the deep-UV adhesive product to be processed, the auxiliary ultraviolet curing light source incident intensity adjustment stage and the corresponding processing time is randomly generated, the sample data is scheduled, and the state transition sample is collected.

[0041] In step S304, the parameters of the strategy optimization model, the deep learning network, the controller of the control end and the evaluation module of the control end are trained, including:

[0042] (A) A policy optimization model is used to update the regulator and control unit of the deep learning network and the control unit. The parameters of the evaluation module are Ξ={ζ1,ζ2,ζ3,ζ4,ζ5,ζ6}, where ζ1,ζ2,ζ3,ζ4,ζ5,ζ6 represent the predecessor node, successor node, disjunction node, target node, scheduling node, and transition node, respectively. The parameters are updated if and only if the incident intensity I1 state node embedding information and the scheduling action improves the scheduling effect. The objective function formula is as follows:

[0043]

[0044] In the formula, The objective function value, For the update function, min() is the minimum function, y i (Ξ) represents the scheduling performance value of the evaluation module at time i. Let ψ be the overall dominance function, 1-ψ be the preceding overall dominance function, and 1+ψ be the following overall dominance function.

[0045] (B) Add the value function error and entropy addition terms to the objective function, and the expression is:

[0046]

[0047] In the formula, The value is obtained by adding the value function error and entropy addition terms to the objective function, where a and β are coefficients, U(H τ ;Ξ) represents the set of auxiliary ultraviolet curing light source incident intensity adjustment stages under parameter constraints of the real-time curing status image set and evaluation module of the deep UV adhesive product to be processed at time τ, and H τ τ represents the set of real-time curing status images of the deep UV adhesive products to be processed, which can be selected at time τ. Ξ represents the parameters of the evaluation module. N represents the set of all auxiliary ultraviolet curing light source incident intensity adjustment stages at time τ. τ For the current time τ The entropy value N under the strategy τ Let τ be the entropy at time τ. The entropy addition strategy is used at time τ.

[0048] (C) continuously moving towards The gradient descent direction is updated for Ξ until convergence.

[0049] Another object of the present invention is to provide a deep UV adhesive curing system, the system implementing the deep UV adhesive curing method, the system comprising:

[0050] The initial incident intensity setting module sets the incident intensity I0 of the auxiliary ultraviolet curing light source below the deep UV adhesive product to be treated, based on the dynamic change results of the UV adhesive refractive index.

[0051] The real-time curing status image acquisition module uses non-destructive testing to detect the overall real-time curing status of the deep UV adhesive product to be processed and acquire real-time curing status images.

[0052] The control terminal receives the real-time curing status image and compares it with the pre-stored target sample. Based on the comparison information, it controls the auxiliary ultraviolet curing light source to change the incident intensity I2 so that the changed incident intensity I2 meets the conditions for overall curing of the deep UV adhesive product to be processed.

[0053] Furthermore, the deep UV adhesive curing system is mounted on a computer device, which includes at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor. When the processor executes the computer program, it performs the functions of the deep UV adhesive curing system.

[0054] Furthermore, the deep UV adhesive curing system is mounted on a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it can realize the functions of the deep UV adhesive curing system.

[0055] Combining all the above technical solutions, the beneficial effects of this invention are as follows: This invention solves the problem of the bottom adhesive failing to cure. By adding an auxiliary ultraviolet curing light source to the bottom of the product, the auxiliary ultraviolet curing lamp mainly increases the energy conductivity of the adhesive and provides some of the energy required for adhesive curing, ultimately enabling the deep adhesive to cure completely.

[0056] This invention provides a feature description of the auxiliary UV curing light source scheduling problem, transforming it into a sequential policy problem. Modeling is based on the policy process, making it applicable to auxiliary UV curing light source scheduling problems of different network sizes. It eliminates the need for repeated training and exhibits strong generalization performance. This invention introduces a policy optimization model to train the neural network, which can stably improve parameter performance. This invention is suitable for dynamic scheduling scenarios. If a problem occurs in the incident intensity adjustment stage of a certain auxiliary UV curing light source, the incident intensity I1 state node and its subsequent incident intensity I1 state nodes can be directly set to zero vector incident intensity I1 state nodes without affecting the scheduling of other deep UV adhesive products to be processed.

[0057] This invention improves the generalization ability of the algorithm for scheduling auxiliary UV curing light sources in manufacturing systems, enabling its application to scheduling problems of different network sizes within the same system. By topologicalizing the scheduling problem network, this invention enhances the generalization ability of the scheduling algorithm, effectively handling the scheduling problem of auxiliary UV curing light sources, and exhibits strong versatility and high adaptability. Attached Figure Description

[0058] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure;

[0059] Figure 1 This is a flowchart of the deep UV adhesive curing method provided in the embodiments of the present invention;

[0060] Figure 2 The flowchart of the control terminal provided in this embodiment of the invention receives the real-time curing status image, compares it with the pre-stored target sample, and instructs the auxiliary ultraviolet curing light source to change the incident intensity I2 according to the comparison information.

[0061] Figure 3 This is a schematic diagram of the deep UV adhesive curing system provided in an embodiment of the present invention;

[0062] In the diagram: 1. Initial setting module for incident intensity; 2. Real-time solidification status image acquisition module; 3. Control terminal. Detailed Implementation

[0063] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0064] Based on the existing patent "A Method for Detecting the Dynamic Optical Properties of UV Adhesive Curing Process" (Publication No. CN103969220A, Publication Date 20140806), this invention obtains the dynamic changes in the refractive index of UV adhesive and further improves the technology based on the results, specifically how to control the irradiation intensity of the ultraviolet curing light source to achieve the effect of complete curing of the bottom of a thick adhesive.

[0065] Example 1, as Figure 1 As shown, the deep UV adhesive curing method provided in this embodiment of the invention includes: changing from one ultraviolet curing light source irradiating the product from above to two ultraviolet curing light sources, one at the top and one at the bottom, curing together. Specifically, it includes:

[0066] S1, Based on the dynamic change results of the UV adhesive refractive index, set the incident intensity I0 of the auxiliary ultraviolet curing light source below the deep UV adhesive product to be treated.

[0067] S2, using non-destructive testing to detect the overall real-time curing status of the deep UV adhesive product to be processed, and to obtain real-time curing status images;

[0068] S3, the control terminal receives the real-time curing status image, compares it with the pre-stored target sample, and controls the auxiliary ultraviolet curing light source to change the incident intensity I2 so that the changed incident intensity I2 meets the conditions for overall curing of the deep UV adhesive product to be processed.

[0069] Example 2, as another possible optimized implementation, such as Figure 2 In step S3, the modified incident intensity I2 is made to meet the conditions for overall curing of the deep UV adhesive product to be treated, including:

[0070] S301, Construct a real-time curing state image model of the deep UV adhesive product to be processed, and obtain the current incident intensity I1 of the auxiliary ultraviolet curing light source; Based on the constructed real-time curing state image model of the deep UV adhesive product to be processed, establish an adjustment strategy model for the auxiliary ultraviolet curing light source.

[0071] S302, uses a deep learning network to extract feature information between state nodes of the current incident intensity I1, and combines messages through clustering methods;

[0072] S303 utilizes the controller at the control end to implement scheduling strategies, uses the evaluation module at the control end to evaluate the effectiveness of the controller strategy, and uses a deep learning network to update the state of the current incident intensity I1 state node.

[0073] S304, determine whether the scheduling result meets the preset standard. If it does, output the scheduling scheme; otherwise, use the strategy optimization model to train the parameters of the deep learning network, the controller at the control end, and the evaluation module at the control end to obtain the changed incident intensity I0 of the auxiliary ultraviolet curing light source.

[0074] In step S301, the current incident intensity I1 of the auxiliary ultraviolet curing light source is obtained, including:

[0075] (1) Construct a real-time curing state image of the deep UV adhesive product to be processed, as shown in the following model:

[0076] H = ((U, S) ∪ K)

[0077] Wherein, H is the set of real-time curing status images of the deep UV adhesive product to be processed; U is the set of incident intensity adjustment stages of the auxiliary ultraviolet curing light source, and the incident intensity adjustment stage of the auxiliary ultraviolet curing light source is represented as the incident intensity I1 state node in U; ​​S is the set of incident intensity adjustment boundaries; the incident intensity adjustment boundary set contains incident intensity adjustment boundaries, and each incident intensity adjustment boundary represents the priority constraint between two consecutive incident intensity adjustment stages of the auxiliary ultraviolet curing light source on the same deep UV adhesive product to be processed; K is the set of incident intensity determination boundaries, which contains incident intensity determination boundaries, and each incident intensity determination boundary represents the control end shared constraint between two incident intensity I1 state nodes. When two incident intensity adjustment stages of the auxiliary ultraviolet curing light source can be processed by the same control end, the current incident intensity I1 state node of the corresponding incident intensity adjustment stage of the auxiliary ultraviolet curing light source is connected to the incident intensity determination boundary.

[0078] (2) Obtain the current incident intensity I1 of the auxiliary ultraviolet curing light source. Use the current incident intensity I1 to represent the real-time dynamic scheduling status of the auxiliary ultraviolet curing light source. Use the current incident intensity I1 as the feature information of U. The feature information of U includes: the current incident intensity I1 state node status, processing time, number of subsequent auxiliary ultraviolet curing light source incident intensity adjustment stages, and start time of the auxiliary ultraviolet curing light source incident intensity adjustment stage. The current incident intensity I1 state node status includes [1, 0, 0], [0, 1, 0], and [0, 0, 1], where [1, 0, 0], [0, 1, 0], and [0, 0, 1] are... [1] respectively indicates that the auxiliary ultraviolet curing light source incident intensity adjustment stage has not yet started, the auxiliary ultraviolet curing light source incident intensity adjustment stage is being processed, and the auxiliary ultraviolet curing light source incident intensity adjustment stage has been completed; the processing time is the processing time of the auxiliary ultraviolet curing light source incident intensity adjustment stage; the number of subsequent auxiliary ultraviolet curing light source incident intensity adjustment stages is the number of subsequent auxiliary ultraviolet curing light source incident intensity adjustment stages U of the same deep UV adhesive product to be processed; the start time of the auxiliary ultraviolet curing light source incident intensity adjustment stage is the start processing time of the auxiliary ultraviolet curing light source incident intensity adjustment stage U.

[0079] In step S301, an auxiliary ultraviolet curing light source adjustment strategy model is established, including:

[0080] The auxiliary UV curing light source adjustment strategy tuple is (M, B, O, T, x), where M represents the auxiliary UV curing light source scheduling state, B represents the scheduling action, O represents the state transition probability, T represents the reward obtained for each action, with the scheduling objective being to minimize the completion time; and x represents the discount factor, which is the degree of influence of the current action on the future reward.

[0081] The expression for the reward T obtained for each action is:

[0082]

[0083] In the formula, makespan() represents the total processing time of the deep UV adhesive product sequence to be processed, H represents the current state, and H' represents the next state.

[0084] Furthermore, in step S302, a deep learning network is used to extract feature information between the current incident intensity I1 state nodes, and messages are combined using a clustering method, including:

[0085] (a) Embed the incident intensity I1 state node and extract the feature information between state nodes with different incident intensities I1. The expression is:

[0086]

[0087] In the formula, Embed information for the state node of incident intensity I1 in generation t. The relu() function is the state node update function for the subsequent incident intensity I1, and it is a vector concatenation function. Let Rp(U) be the state node update function for the preceding incident intensity I1, where Rp(U) is the set of state nodes for the preceding incident intensity I1, and Rq(U) is the set of state nodes for the subsequent incident intensity I1. The state node update function is for the target incident intensity I1. Embed information for the state node of incident intensity I1 in the (k-1)th generation;

[0088] (b) When the incident intensity I1 state node is completed or does not exist during the incident intensity adjustment stage of the auxiliary ultraviolet curing light source, the incident intensity I1 state node is embedded as a zero vector with the same dimension as H and no longer participates in the state update.

[0089] (c) Train the parameters of the input layer, hidden layer, and output layer of the deep learning network using a multilayer perceptron.

[0090] In step S303, the state of the current incident intensity I1 state node is updated using a deep learning network, including:

[0091] (i) The probability distribution of the target control terminal for selecting the operable auxiliary ultraviolet curing light source incident intensity adjustment stage is expressed as:

[0092]

[0093] In the formula, These are probability distribution values. This is a set of real-time curing status images of the deep UV adhesive products to be processed, selectable at k nodes at time τ. Let the incident intensity I1 be the state node embedding information for the k-th node, ζ5 be the scheduling node, and H be the state node embedding information. τ This is a set of real-time curing status images of the deep UV adhesive products to be processed, available at time τ. For the set of auxiliary UV curing light source incident intensity adjustment stages available at time τ, m l () is a differentiable function that maps the embedded information of the incident intensity I1 state node to the logarithm of each incident intensity I1 state node.

[0094] (ii) A random strategy is used to select the stage for adjusting the incident intensity of the auxiliary ultraviolet curing light source. The state value is solved using the evaluation module of the control terminal, and the expression is:

[0095]

[0096] In the formula, U() is the set function for adjusting the incident intensity of the auxiliary ultraviolet curing light source. To provide a set of real-time curing status images of the deep UV adhesive product to be processed, selectable at k nodes at time τ, m U () is a differentiable function, and U is the set of stages for adjusting the incident intensity of the auxiliary ultraviolet curing light source. This is a set of real-time curing status images of the deep UV adhesive products to be processed that can be selected at k nodes at time i, and ζ6 is the conversion node.

[0097] (iii) Randomly generate the initial state including the number of processing control terminals, the deep UV adhesive product to be processed, the processing auxiliary ultraviolet curing light source incident intensity adjustment stage and the corresponding processing time, schedule the sample data, and collect state transition samples.

[0098] In step S304, the parameters of the deep learning network, the regulator at the control end, and the evaluation module at the control end are trained using the policy optimization model, including:

[0099] (A) A strategy optimization model is adopted to update the regulator and control terminal of the deep learning network and the control terminal. The parameters of the evaluation module are Ξ={ζ1,ζ2,ζ3,ζ4,ζ5,ζ6}, where ζ1,ζ2,ζ3,ζ4,ζ5,ζ6 represent the predecessor node, the successor node, the disjunction node, the target node, the scheduling node, and the transformation node, respectively.

[0100] The parameters are updated if and only if the incident intensity I1, state node embedding information, and scheduling actions improve the scheduling effect. The objective function formula is as follows:

[0101]

[0102] In the formula, The objective function value, For the update function, min() is the minimum function, y i (Ξ) represents the scheduling performance value of the evaluation module at time i. Let ψ be the overall dominance function, 1-ψ be the preceding overall dominance function, and 1+ψ be the following overall dominance function.

[0103] (B) Add the value function error and entropy addition terms to the objective function, and the expression is:

[0104]

[0105] In the formula, The value is obtained by adding the value function error and entropy addition terms to the objective function, where a and β are coefficients, U(H τ ;Ξ) represents the set of auxiliary ultraviolet curing light source incident intensity adjustment stages under parameter constraints of the real-time curing status image set and evaluation module of the deep UV adhesive product to be processed at time τ, and H τ τ represents the set of real-time curing status images of the deep UV adhesive products to be processed, which can be selected at time τ. Ξ represents the parameters of the evaluation module. N represents the set of all auxiliary ultraviolet curing light source incident intensity adjustment stages at time τ. τ For the current time τ The entropy value N under the strategy τ Let τ be the entropy at time τ. The entropy addition strategy is used at time τ.

[0106] (C) continuously moving towards The gradient descent direction is updated for Ξ until convergence.

[0107] Example 3, as Figure 3 As shown, the present invention provides a deep UV adhesive curing system, which specifically includes:

[0108] The incident intensity preliminary setting module 1 sets the incident intensity I0 of the auxiliary ultraviolet curing light source below the deep UV adhesive product to be treated based on the obtained dynamic change results of the UV adhesive refractive index.

[0109] The real-time curing status image acquisition module 2 uses non-destructive testing to detect the overall real-time curing status of the deep UV adhesive product to be processed and acquire real-time curing status images.

[0110] Control terminal 3 receives the real-time curing status image and compares it with the pre-stored target sample. Based on the comparison information, it controls the auxiliary ultraviolet curing light source to change the incident intensity I2 so that the changed incident intensity I2 meets the conditions for overall curing of the deep UV adhesive product to be processed.

[0111] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0112] The information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of the present invention. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0113] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this invention. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments.

[0114] This invention also provides a computer device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in any of the above method embodiments.

[0115] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps described in the various method embodiments above.

[0116] This invention also provides an information data processing terminal, which, when executed on an electronic device, provides a user input interface to implement the steps described in the above method embodiments. The information data processing terminal is not limited to mobile phones, computers, or switches.

[0117] This invention also provides a server that, when executed on an electronic device, provides a user input interface to implement the steps described in the above method embodiments.

[0118] This invention provides a computer program product that, when run on an electronic device, enables the electronic device to implement the steps described in the various method embodiments above.

[0119] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0120] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention and within the spirit and principles of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for curing deep UV adhesive, characterized in that, The method includes: S1, Based on the obtained dynamic change results of the UV adhesive refractive index, set the incident intensity of the auxiliary ultraviolet curing light source below the deep UV adhesive product to be treated. ; S2, using non-destructive testing to detect the overall real-time curing status of the deep UV adhesive product to be treated, and to obtain real-time curing status images; S3, the control terminal receives the real-time curing status image, compares it with the pre-stored target sample, and controls the auxiliary ultraviolet curing light source to change the incident intensity based on the comparison information. The changed incident intensity The conditions for overall curing of the deep UV adhesive product to be treated must be met; In step S3, the changed incident intensity is... The conditions for achieving overall curing of the deep UV adhesive product to be treated include: S301, Construct a real-time curing state image model of the deep UV adhesive product to be processed, and obtain the current incident intensity of the auxiliary ultraviolet curing light source. Based on the constructed real-time curing state image model of the deep UV adhesive product to be processed, an auxiliary ultraviolet curing light source adjustment strategy model is established. S302, using a deep learning network to extract the current incident intensity The characteristic information between state nodes is used to combine messages through clustering methods; S303 utilizes a regulator at the control end to implement a scheduling strategy, an evaluation module at the control end to evaluate the effectiveness of the regulator strategy, and a deep learning network to update the current incident intensity. State node status; S304, determine whether the scheduling result meets the preset standard. If it does, output the scheduling scheme; otherwise, use the strategy optimization model to train the parameters of the deep learning network, the controller at the control end, and the evaluation module at the control end to obtain the change in incident intensity of the auxiliary ultraviolet curing light source. .

2. The deep UV adhesive curing method according to claim 1, characterized in that, In step S301, the current incident intensity of the auxiliary ultraviolet curing light source is obtained. ,include: (1) Construct a real-time curing state image of the deep UV adhesive product to be processed. The model is as follows: in, This is a collection of real-time curing status images of deep UV adhesive products to be processed. To assist in adjusting the incident intensity of the ultraviolet curing light source, the auxiliary ultraviolet curing light source incident intensity adjustment stage is in The value in is the incident intensity. State nodes; For the incident intensity adjustment boundary set; the incident intensity adjustment boundary set contains incident intensity adjustment boundaries, each incident intensity adjustment boundary representing the priority constraint between two consecutive auxiliary ultraviolet curing light source incident intensity adjustment stages on the same deep UV adhesive product to be treated. A set of boundary conditions is defined for the incident intensity, the set of incident intensity determination boundary conditions containing incident intensity determination boundaries, each incident intensity determination boundary representing two incident intensities. The control terminals between state nodes share constraints. When the incident intensity adjustment stages of two auxiliary UV curing light sources can be handled by the same control terminal, the current incident intensity of the corresponding auxiliary UV curing light source in the incident intensity adjustment stage is... The state node is connected to the boundary determined by the incident intensity; (2) Obtain the current incident intensity of the auxiliary ultraviolet curing light source Using auxiliary ultraviolet light to cure the current incident intensity of the light source This indicates the real-time dynamic scheduling status of the auxiliary ultraviolet curing light source, utilizing the current incident intensity of the auxiliary ultraviolet curing light source. As Feature information; The characteristic information includes: current incident intensity Status node status, processing time, number of subsequent auxiliary UV curing light source incident intensity adjustment stages, start time of auxiliary UV curing light source incident intensity adjustment stage; the current incident intensity The status node states include [1,0,0], [0,1,0], and [0,0,1], where [1,0,0], [0,1,0], and [0,0,1] represent, respectively, that the auxiliary UV curing light source incident intensity adjustment stage has not yet started, the auxiliary UV curing light source incident intensity adjustment stage is in progress, and the auxiliary UV curing light source incident intensity adjustment stage has been completed; the processing time is the processing time of the auxiliary UV curing light source incident intensity adjustment stage; the subsequent number of auxiliary UV curing light source incident intensity adjustment stages is the number of auxiliary UV curing light source incident intensity adjustment stages for the same deep UV adhesive product to be processed. The number of subsequent auxiliary ultraviolet curing light source incident intensity adjustment stages; the start time of the auxiliary ultraviolet curing light source incident intensity adjustment stage is the auxiliary ultraviolet curing light source incident intensity adjustment stage. The start time of processing.

3. The deep UV adhesive curing method according to claim 1, characterized in that, In step S301, an auxiliary ultraviolet curing light source adjustment strategy model is established, including: The auxiliary UV curing light source adjustment strategy tuple is ,in, Indicates the scheduling status of the auxiliary ultraviolet curing light source. Indicates a scheduling action. Represents the state transition probability. This represents the reward obtained for each action, with the scheduling objective being to minimize the completion time. This represents the discount factor, which is the degree to which the current action affects future returns; Rewards for each action The expression is: In the formula, This represents the total processing time for the entire series of deep UV adhesive products to be processed. This is the current state. This is the next state.

4. The deep UV adhesive curing method according to claim 1, characterized in that, In step S302, the current incident intensity is extracted using a deep learning network. The characteristic information between state nodes is combined into messages using clustering methods, including: (a) Embedded incident intensity State nodes, extracting different incident intensities The characteristic information between state nodes is expressed as follows: In the formula, For the first Substitute incident intensity State nodes embed information. Subsequent incident intensity State node update function, It is a function of concatenating vectors. Preceding incident intensity State node update function, Preceding incident intensity State node set Subsequent incident intensity State node set Target incident intensity State node update function, For the first Substitute incident intensity State nodes embed information; (b) During the adjustment phase of the incident intensity of the auxiliary ultraviolet curing light source When a state node is complete or does not exist, the incident intensity will be... The state node embedding is set to a zero vector with the same dimension as H, and it no longer participates in state updates; (c) Train the parameters of the input layer, hidden layer and output layer of the deep learning network using a multilayer perceptron.

5. The deep UV adhesive curing method according to claim 1, characterized in that, In step S303, the current incident intensity is updated using a deep learning network. The state of a state node includes: (i) The probability distribution of the incident intensity adjustment stage of the operable auxiliary ultraviolet curing light source selected at the target control terminal is expressed as follows: In the formula, These are probability distribution values. In order to be in time A set of real-time curing status images of selectable deep UV adhesive products to be processed at each node. For the first Nodal incident intensity State nodes embed information. As a scheduling node, In order to be in A real-time curing status image set of deep UV adhesive products available for processing at any time. In order to be in A set of adjustable incident intensity stages for the auxiliary UV curing light source that can be selected at any time. For a differentiable function, the incident intensity State node embedding information mapped to each incident intensity The logarithmic value of the state node; (ii) A random strategy is used to select the stage for adjusting the incident intensity of the auxiliary ultraviolet curing light source. The state value is solved using the evaluation module of the control end, and the expression is: In the formula, To assist in adjusting the stage set function of the incident intensity of the ultraviolet curing light source, In order to be in time A set of real-time curing status images of selectable deep UV adhesive products to be processed at each node. It is a differentiable function. To assist in adjusting the incident intensity of the ultraviolet curing light source, In order to be in time A set of real-time curing status images of selectable deep UV adhesive products to be processed at each node. For conversion nodes; (iii) Randomly generate the initial state including the number of processing control terminals, the deep UV adhesive product to be processed, the processing auxiliary ultraviolet curing light source incident intensity adjustment stage and the corresponding processing time, schedule the sample data, and collect state transition samples.

6. The deep UV adhesive curing method as described in claim 1, characterized in that, In step S304, the parameters of the deep learning network, the regulator at the control end, and the evaluation module at the control end are trained using the policy optimization model, including: (A) A policy optimization model is used to update the regulator and control unit of the deep learning network and the control unit. The parameters of the evaluation module are: ,in These represent the preceding node, the following node, the disjunction node, the target node, the scheduling node, and the transition node, respectively. If and only if the incident intensity The state node embedding information and scheduling actions are used to update parameters to improve scheduling efficiency. The objective function formula is as follows: In the formula, , For the update function, It is the minimum function. Let i be the scheduling performance value of the evaluation module at time i. The overall advantage function, For the preceding overall advantage function, This is the overall dominance function for the subsequent order; (B) Add the value function error and entropy addition terms to the objective function, and the expression is: In the formula, The value is obtained by adding the value function error and entropy addition terms to the objective function. All are coefficients. In order to be in A set of real-time curing status images of deep UV adhesive products to be processed, available at any time, and a set of auxiliary UV curing light source incident intensity adjustment stages under parameter constraints of the evaluation module. for A real-time curing status image set of deep UV adhesive products available for processing at any time. To evaluate the parameters of the module, for The entire process of adjusting the incident intensity of the auxiliary ultraviolet curing light source is constantly being implemented. For the present time Entropy value under the strategy for Time entropy, for The time-entropy addition strategy; (C) continuously towards Gradient descent direction update Until it converges.

7. A deep UV adhesive curing system, characterized in that, The system implements the deep UV adhesive curing method as described in any one of claims 1-6, and the system comprises: The initial incident intensity setting module (1) sets the incident intensity of the auxiliary ultraviolet curing light source below the deep UV adhesive product to be treated based on the dynamic change results of the obtained UV adhesive refractive index. ; The real-time curing status image acquisition module (2) uses non-destructive testing to detect the overall real-time curing status of the deep UV adhesive product to be processed and acquires real-time curing status images. Control terminal (3): The control terminal receives the real-time curing status image, compares it with the pre-stored target sample, and controls the auxiliary ultraviolet curing light source to change the incident intensity based on the comparison information. The changed incident intensity The conditions for overall curing of the deep UV adhesive product to be treated must be met.

8. The deep UV adhesive curing system according to claim 7, characterized in that, The system is mounted on a computer device, which includes at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor. When the processor executes the computer program, it performs the functions of the method described above.

9. The deep UV adhesive curing system according to claim 7, characterized in that, The system is mounted on a computer-readable storage medium that stores a computer program, which, when executed by a processor, can perform the functions of the above-described method.

Images