Preparation process of optical cable coating material

Abstract

The invention provides an optical cable coating material preparation process which comprises the following steps: dissolving a dispersant polyvinyl alcohol (PVA) in deionized water to form a pretreatment solution; under the protection of inert gas, monomer vinyl chloride (VCM) is added into the pretreatment solution, and a stable dispersion system is formed; adding an initiator potassium persulfate (KPS) to promote a polymerization reaction; the invention relates to the technical field of synthesis of high polymer materials, in particular to a PVC (polyvinyl chloride) composite stabilizer, which effectively inhibits thermal decomposition of PVC in the processing process, prolongs the processing temperature range, reduces generation of HCL (hydrogen chloride) and remarkably improves thermal stability by adding a calcium-zinc composite stabilizer and a composite stabilizer of nano silicon dioxide after the polymerization reaction is completed. The prepared cable sheath polyvinyl chloride is more suitable for cable application in a high-temperature environment, and by adding the nano silicon dioxide, not only is the mechanical strength of the PVC improved, but also the impact resistance and the flexibility of the PVC are improved, and large-scale production is easy.

Description

Technical Field

[0001] The invention relates to the technical field of polymer material synthesis, in particular to a preparation process of optical cable coating material. Background Art

[0002] Optical cables are manufactured to meet optical, mechanical or environmental performance specifications. They are communication cable components that use one or more optical fibers placed in a sheath as transmission media and can be used individually or in groups. Optical cables are mainly composed of optical fibers, plastic protective sleeves and plastic sheaths. The cable sheath refers to the plastic sheath, which is generally polyvinyl chloride, a polymer formed by the polymerization of vinyl chloride monomers in the presence of peroxides, azo compounds and other initiators or by free radical polymerization under the action of light and heat. Vinyl chloride homopolymers and vinyl chloride copolymers are collectively referred to as vinyl chloride resins.

[0003] Polyvinyl chloride (PVC) is a widely used synthetic polymer material. Due to its excellent mechanical properties, electrical insulation properties, chemical corrosion resistance and good processing properties, it is widely used in construction, packaging, wires and cables, daily necessities and other fields. However, there are some problems in the processing and use of PVC, such as poor thermal stability, narrow processing temperature range, easy decomposition to produce HCL gas, leading to equipment corrosion and environmental pollution.

[0004] In order to solve the above problems, the performance of PVC is usually improved by adding stabilizers, plasticizers, impact modifiers, etc. Among them, heat stabilizers are the key components to improve the thermal stability of PVC. Common heat stabilizers include lead salts, calcium-zinc composite stabilizers, organic tin stabilizers, etc., but lead salts are gradually being eliminated due to environmental problems, and organic tin stabilizers are expensive. Although calcium-zinc composite stabilizers are environmentally friendly, their thermal stability is limited. Although the addition of plasticizers and impact modifiers can improve the flexibility and impact strength of PVC, it will also reduce its mechanical strength and heat resistance.

[0005] Therefore, developing a PVC preparation method that can effectively improve the thermal stability of PVC without affecting or even improving its other properties has important practical application value. Summary of the invention

[0006] In order to overcome the existing problems, the embodiment of the present application provides a process for preparing an optical cable sheath material. By adding a composite stabilizer, the thermal decomposition of PVC during processing is effectively inhibited, the processing temperature range is extended, the generation of HCL is reduced, and the thermal stability is significantly improved, so that the prepared cable sheath polyvinyl chloride is more suitable for cable applications in high temperature environments. In addition, by adding nano-silicon dioxide, not only the mechanical strength of PVC is improved, but also its impact resistance and flexibility are improved. The operation steps are simple, the requirements for production equipment are not high, it is easy to mass produce, and the production cost is controlled.

[0007] The technical solution adopted by the embodiment of the present application to solve the technical problem is:

[0008] A process for preparing an optical cable coating material, the process comprising the following steps:

[0009] Step 1, dissolving dispersant polyvinyl alcohol (PVA) in deionized water, wherein the ratio of dispersant polyvinyl alcohol (PVA) to deionized water is 1:10-1:50, to form a pretreatment solution;

[0010] Step 2: Under the protection of inert gas, vinyl chloride monomer (VCM) is added to the pretreatment solution to form a stable dispersion system.

[0011] Among them, the monomer vinyl chloride (VCM) is mixed with the dispersant polyvinyl alcohol (PVA) and deionized water under the protection of inert gas to form a stable dispersion system. During the pretreatment process, the temperature is controlled at 5-15°C to increase the solubility of the dispersant and ensure the uniform distribution of the subsequent initiator;

[0012] Step 3: adding initiator potassium persulfate (KPS) to promote the polymerization reaction, the amount of initiator potassium persulfate (KPS) is 0.01%-1% of the mass of the monomer vinyl chloride (VCM), and the polymerization reaction is carried out at a temperature range of 40° C.-80° C.,

[0013] After the initiation stage, a new composite stabilizer is added. The stabilizer is composed of a calcium-zinc composite stabilizer and nano-silicon dioxide in a mass ratio of 3:1. The composite stabilizer can significantly improve the thermal stability of polyvinyl chloride. At the same time, the addition of nano-silicon dioxide can enhance the mechanical properties and weather resistance of the material.

[0014] Step 4: After the polymerization reaction is completed, a calcium zinc composite stabilizer and a nano-silicon dioxide composite stabilizer are added, wherein the amount of the calcium zinc composite stabilizer and the nano-silicon dioxide composite stabilizer is 0.5%-5% of the total mass.

[0015] The polymerization product is filtered, washed and dried to obtain high-stability polyvinyl chloride particles. In this stage, deionized water is used for washing to remove unreacted monomers and initiator residues to ensure the purity and stability of the final product.

[0016] Preferably, the prepared polyvinyl chloride (PVC) is used for insulation of wires and cables.

[0017] Preferably, OMA is used to optimize the Swin Transformer model to optimize the formula ratio and process parameters in the preparation process of the optical cable coating material. The algorithm output optimizes the ratio of polyvinyl alcohol (PVA) to deionized water, the polymerization reaction temperature, the amount of initiator (KPS), the amount of calcium zinc composite stabilizer and the amount of nano-silicon dioxide, thereby improving product performance and production efficiency. The specific steps are as follows:

[0018] Step 1: Data collection and preprocessing

[0019] In order to accurately obtain process parameters and performance indicators, various sensors and hardware devices can be used. The following are the data collection sensors and hardware devices used:

[0020] Temperature sensor: Thermal resistor temperature sensor, used to monitor the polymerization reaction temperature and ensure that the reaction is carried out within the set temperature range;

[0021] Pressure sensor: used to measure the pressure inside the reactor to ensure the safety and stability of the reaction conditions;

[0022] Flow meter: used to measure the flow of different components to ensure that each component is added according to the set ratio;

[0023] Mass flow controller: used to accurately control the flow of gas to ensure that the reaction is carried out in an inert atmosphere;

[0024] pH sensor: used to monitor the pH value of the reaction system to ensure the stability of the system and the smooth progress of the reaction;

[0025] Viscometer: Rotational viscometer, used to measure the viscosity of the reaction process and the final product, to evaluate the progress of the reaction and product performance;

[0026] Mechanical properties testing equipment: universal testing machine, used to test the mechanical strength, impact resistance and flexibility of the prepared PVC samples, test items: tensile strength, compression strength, impact strength;

[0027] Thermal analysis equipment: Thermogravimetric analyzer (TGA), used to test the thermal stability of the prepared PVC samples and evaluate their performance in high temperature environments;

[0028] Data acquisition system: used to integrate and record data from various sensors and test equipment for subsequent analysis and model training. Modules: temperature module, pressure module, flow module, pH module, viscosity module;

[0029] Control system: Distributed control system (DCS) is used to monitor and adjust process parameters in real time to ensure the stability and optimization of the preparation process;

[0030] Data preprocessing:

[0031] Data cleaning: processing missing values; deleting samples with missing values ​​and filling missing values;

[0032] Handling outliers: Outliers will affect the training and prediction of the model and need to be processed using the Z-score method;

[0033] Data standardization: In order for the model to better learn the relationship between features, the data needs to be standardized;

[0034] Dataset division: In order to train, optimize, and evaluate the model, the dataset needs to be divided into training set, validation set, and test set;

[0035] Through the above sensors and hardware equipment, various parameters in the preparation process of optical cable coating materials can be accurately monitored and controlled to ensure the consistency, accuracy and completeness of data, providing a solid data foundation for subsequent intelligent optimization;

[0036] Step 2: Swin Transformer model establishment

[0037] Input embedding: converting input features into a fixed-length vector representation,

[0038] X = [x1, x2, ..., x n ]

[0039] Where X is the input feature vector, x n is the nth feature, n is the number of samples;

[0040] Patch partitioning: Divide the input feature vector into patches of fixed size, each patch contains several features.

[0041] Xpatch=[X[i:i+P-1]|i=1,P,2P,...,(n / P-1)P+1]

[0042] Among them, Xpatch represents the set of patch feature vectors after division, X is the input feature vector, P is the size of the patch, i represents the starting position index of the patch, and n represents the length of the input feature vector;

[0043] Linear embedding: Linear embedding is performed on each patch to map it into a high-dimensional space.

[0044] E patch =W emb X patch

[0045] Among them, W emb is the weight matrix of the linear embedding, E patch represents the patch feature vector after linear embedding, X patch represents the input patch feature vector;

[0046] Hierarchical structure: Swin Transformer extracts features through multiple hierarchical modules (Layer), each of which contains two stages: moving window self-attention (W-MSA) and cross-window self-attention (SW-MSA);

[0047] Moving Window Self-Attention (W-MSA): Multi-head self-attention mechanism is applied within each window to capture local features.

[0048]

[0049] Among them, Attention represents the output of the self-attention mechanism, softmax represents the probability distribution function, Q, K, and V are query, key, and value matrices respectively, and d k is the dimension of the key, QK T is the dot product of the query matrix Q and the transposed matrix of the key matrix K;

[0050] The formula of the multi-head self-attention mechanism is:

[0051] MultiHead(Q,K,V)=Concat(head1,head2,...,head h )W O

[0052]

[0053] Among them, MultiHead represents the output vector of the multi-head self-attention mechanism, Q, K, V are query, key and value matrices respectively, h represents the number of attention heads, head h represents the output of the hth attention head, Concat represents concatenating the outputs of multiple attention heads, Attention represents self-attention calculation, and head i represents the output of the i-th attention head, is the projection matrix, W O is the output matrix;

[0054] Cross-window self-attention (SW-MSA): By offsetting the window, the global features are captured across the window. Let the offset be s and the window size be M. The offset window self-attention formula is:

[0055] X shifted =Shift(X,s)

[0056] Where X is the input feature vector, X shifted represents the feature vector after the offset processing, s is the offset, and Shift represents the window offset operation;

[0057] After applying W-MSA, the feature is then offset back to its original position:

[0058] X Sw-MSA =InverseShift(X shifted , s)

[0059] Among them, X Sw-MSA represents the feature vector after being processed by the cross-window multi-head self-attention mechanism (SW-MSA), InverseShift represents the function of the inverse shift operation, and X shifted represents the feature vector after offset processing, s is the offset;

[0060] Layer Normalization and MLP: In each layered module, Layer Normalization and MLP are applied to further process the features.

[0061] X norm =LayerNorm(X)

[0062] X MLP =MLP(X norm )

[0063] MLP(X)=σ(XW1+b1)W2+b2

[0064] Among them, X norm is the feature vector after layer normalization, LayerNorm represents the function of layer normalization operation, X represents the input feature vector, X MLP represents the output vector after MLP processing, MLP represents multilayer perceptron, MLP(X) represents the output vector of MLP, σ represents the activation function, W1 and W2 are the weight matrices of the first and second layers, b1 and b2 are the bias vectors of the first and second layers;

[0065] Data enhancement and regularization: During the training process, data enhancement and regularization techniques are used to improve the robustness and stability of the model.

[0066] X aug=DataAugmentation(X)

[0067] X reg = Dropout(X aug )

[0068] Among them, X aug is the feature vector after data enhancement processing, DataAugmentatio is the function of data enhancement operation, X represents the original feature vector, X reg is the feature vector after Dropout regularization, and Dropout is the function of Dropout regularization operation;

[0069] Output layer: Through the last layer of MLP, predict the PVC performance indicators.

[0070] Y = MLP(X final )

[0071] In fact, Y is the predicted PVC performance index, MLP stands for multi-layer perceptron, and X final It is the feature representation after all layered modules and regularization processing;

[0072] Step 3: Owl Migration Algorithm (0MA) Optimization Model

[0073] Initialize the population: Assume that the population size is N, and each individual x i Represents a possible solution, including formulation ratios and process parameters,

[0074] x i =[x i1 , x i2 , ..., x id ]

[0075] Among them, x i is the position vector or solution vector of the ith individual, d is the dimension (number of parameters), x id is the dth parameter of the ith individual;

[0076] Fitness function: Fitness function (f(x i )) is used to evaluate the quality of each individual solution. In this application, the mean square error (MSE) can be used as the fitness function.

[0077]

[0078] Among them, f(x i ) is the value of the fitness function, y k is the true value, is the model prediction value, n is the number of samples;

[0079] Hunting and foraging behavior: During hunting and foraging, the owl updates its position according to the fitness value. The position update formula is:

[0080]

[0081] in, is the position of the ith individual in the t+1th generation, is the position of the ith individual in the tth generation, α is the step size factor, rand is a random number, x best is the current optimal position;

[0082] Migration behavior: During the migration process, the owl migrates according to the fitness value to explore new solution space. The migration update formula is:

[0083]

[0084] in, is the position of the ith individual in the t+1th generation, is the position of the ith individual in the tth generation, β is the migration step length, and randn is a normally distributed random number;

[0085] Termination condition: When the maximum number of iterations T is reached or the fitness function value meets a certain threshold ∈, the algorithm stops and outputs the optimal solution.

[0086] ift≥Torf(x best )≤∈,then stop

[0087] Among them, t is the current iteration number, T is the maximum iteration number, f(x best ) is the fitness value of the optimal solution, ∈ is the threshold of the fitness function value, or is the logical OR operator, if is the conditional statement, then stop is the termination operation;

[0088] In the preparation process of optical cable coating materials, the Swin Transformer model is combined with the Owl Migration Algorithm (OMA) to optimize the optimal formula ratio and process parameter combination x best As shown below:

[0089] x best =[x1,x2,x3,x4,x5]

[0090] Wherein, x1 is the ratio of polyvinyl alcohol (PVA) to deionized water, x2 is the polymerization temperature, x3 is the amount of initiator (KPS), x4 is the amount of calcium zinc composite stabilizer, and x5 is the amount of nano-silicon dioxide;

[0091] The OMA optimization algorithm is used to initialize the population, evaluate hunting, foraging, and migration behaviors through the fitness function, and continuously update the individual positions until the optimal formula ratio and process parameter combination is found.

[0092] The advantages of the embodiments of the present application are:

[0093] In the present invention, the addition of the composite stabilizer effectively inhibits the thermal decomposition of PVC during the processing, prolongs the processing temperature range, reduces the generation of HCL, and significantly improves the thermal stability, so that the prepared cable sheath polyvinyl chloride is more suitable for cable applications in high temperature environments. In addition, the addition of nano silicon dioxide not only improves the mechanical strength of PVC, but also improves its impact resistance and flexibility. The operation steps are simple, the requirements on production equipment are not high, large-scale production is easy, and the production cost is controlled.

[0094] The present invention proposes a preparation process for optical cable coating materials based on the Swin Transformer model and the Owl Migration Algorithm (OMA). By intelligently optimizing the formula ratio and process parameters, the performance and production efficiency of the optical cable coating materials are greatly improved. The Patch Partition and multi-head self-attention mechanism are adopted to achieve efficient processing of input features and capture of global features, ensuring that the product has excellent thermal stability, mechanical strength, impact resistance and flexibility in high temperature environments. At the same time, combined with data enhancement and regularization technology, the robustness and generalization ability of the model are significantly improved, and the uncertainty and quality fluctuations in the production process are reduced. This process not only improves the overall performance of the optical cable, but also realizes the intelligence and efficiency of the preparation process. It has important market competitiveness and application prospects, and has greatly promoted the technological progress and innovative development of the optical cable industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] Figure 1 The figure is a schematic diagram of the process for preparing the optical cable coating material of the present invention.

[0096] Figure 2 It is a schematic diagram of the intelligent algorithm optimization process of the optical cable coating material preparation process of the present invention. DETAILED DESCRIPTION

[0097] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, for the convenience of description below, the referenced "upper", "lower", "left", and "right" are the same as the upper, lower, left, and right directions of the drawings themselves. The "first", "second", etc. below are distinguished for the description and have no other special meanings.

[0098] The embodiment of the present application solves the problems in the prior art by providing a process for preparing an optical cable sheath material. By adding a composite stabilizer, the thermal decomposition of PVC during processing is effectively suppressed, the processing temperature range is extended, the generation of HCL is reduced, and the thermal stability is significantly improved, so that the prepared cable sheath polyvinyl chloride is more suitable for cable applications in high temperature environments. In addition, by adding nano-silicon dioxide, not only the mechanical strength of PVC is improved, but also its impact resistance and flexibility are improved. The operation steps are simple, the requirements for production equipment are not high, large-scale production is easy, and the production cost is controlled.

[0099] like Figure 1-2 The technical solution in the embodiment of the present application is to solve the above problems, and the overall idea is as follows:

[0100] Example 1

[0101] This embodiment provides a process for preparing a cable coating material. Figure 1 As shown, the optical cable coating material preparation process includes the following steps:

[0102] Step 1: dissolving the dispersant polyvinyl alcohol (PVA) in deionized water, the ratio of the dispersant polyvinyl alcohol (PVA) to deionized water being 1:10-1:50, to form a pretreatment solution, adding 500 g of deionized water to a polymerization reactor at room temperature, followed by adding 50 g of the dispersant polyvinyl alcohol (PVA), and stirring at 5-15° C. for 30 min until the dispersant polyvinyl alcohol (PVA) is completely dissolved;

[0103] Step 2: Under the protection of inert gas, vinyl chloride monomer (VCM) is added to the pretreatment solution to form a stable dispersion system.

[0104] Wherein, under the protection of inert gas, 1000g of monomer vinyl chloride (VCM) was slowly added into the reactor, and stirring was continued to form a uniform dispersion system;

[0105] Step 3, adding initiator potassium persulfate (KPS) to promote the polymerization reaction, the amount of initiator potassium persulfate (KPS) is 0.01%-1% of the mass of the monomer vinyl chloride (VCM), and the polymerization reaction is carried out at a temperature range of 40° C.-80° C.;

[0106] Among them, after the initiation stage, a new composite stabilizer is added, which is a composite of calcium zinc composite stabilizer and nano silicon dioxide, with a mass ratio of 3:1. The composite stabilizer can significantly improve the thermal stability of polyvinyl chloride, and at the same time, the mechanical properties and weather resistance of the material are enhanced by the addition of nano silicon dioxide. At 40°C, 0.5g of potassium persulfate (KPS) is added to the pretreated system as an initiator to start the polymerization reaction. The reaction lasts for 3 hours at 50°C, and then slowly rises to 60°C for 1 hour to ensure that the polymerization reaction is fully carried out;

[0107] Step 4: After the polymerization reaction is completed, a composite stabilizer of calcium zinc composite stabilizer and nano silicon dioxide is added. The amount of the composite stabilizer of calcium zinc composite stabilizer and nano silicon dioxide is 0.5%-5% of the total mass. In this stage, deionized water is used for washing to remove unreacted monomers and initiator residues to ensure the purity and stability of the final product. After the polymerization reaction is completed, 15g of calcium zinc composite stabilizer and 5g of nano silicon dioxide are added to the reactor and mixed evenly to form a new composite stabilizer. In this step, the addition of nano silicon dioxide helps to improve the mechanical properties and weather resistance of the material.

[0108] The polymerization product is filtered, washed and dried to obtain high-stability polyvinyl chloride particles, the polymerization product is filtered through a filter, and then washed with deionized water to remove unreacted monomers and initiator residues, and the washed polyvinyl chloride particles are dried in a drying device to finally obtain polyvinyl chloride particles with high thermal stability.

[0109] The prepared polyvinyl chloride (PVC) is used for insulation of wires and cables.

[0110] Example 1

[0111] This embodiment provides a process for preparing a cable coating material. Figure 1 As shown, the optical cable coating material preparation process includes the following steps:

[0112] Step 1: dissolving the dispersant polyvinyl alcohol (PVA) in deionized water, the ratio of the dispersant polyvinyl alcohol (PVA) to deionized water being 1:10-1:50, to form a pretreatment solution, adding 500 g of deionized water to a polymerization reactor at room temperature, followed by adding 50 g of the dispersant polyvinyl alcohol (PVA), and stirring at 5-15° C. for 30 min until the dispersant polyvinyl alcohol (PVA) is completely dissolved;

[0113] Step 2: under the protection of inert gas, adding vinyl chloride monomer (VCM) into the pretreatment solution to form a stable dispersion system;

[0114] Wherein, the monomer vinyl chloride (VCM) is mixed with the dispersant polyvinyl alcohol (PVA) and deionized water under the protection of inert gas to form a stable dispersion system. During the pretreatment process, the temperature is controlled at 5-15° C. to improve the solubility of the dispersant and ensure the uniform distribution of the subsequent initiator. Under the protection of inert gas, 1000 g of the monomer vinyl chloride (VCM) is slowly added to the reactor, and stirring is continued to form a uniform dispersion system;

[0115] Step 3, adding initiator potassium persulfate (KPS) to promote the polymerization reaction, the amount of initiator potassium persulfate (KPS) is 0.01%-1% of the mass of the monomer vinyl chloride (VCM), and the polymerization reaction is carried out in a temperature range of 40° C.-80° C., the amount of initiator potassium persulfate (KPS) is adjusted to 0.7 g in the polymerization reactor, the polymerization reaction is carried out at 45° C. for 3.5 hours, and then at 55° C. for 1.5 hours;

[0116] Among them, after the initiation stage, a new composite stabilizer is added. The stabilizer is composed of a calcium-zinc composite stabilizer and nano-silicon dioxide in a mass ratio of 3:1. The composite stabilizer can significantly improve the thermal stability of polyvinyl chloride. At the same time, the addition of nano-silicon dioxide can enhance the mechanical properties and weather resistance of the material.

[0117] Step 4: After the polymerization reaction is completed, a composite stabilizer of calcium zinc composite stabilizer and nano silicon dioxide is added, and the amount of the composite stabilizer of calcium zinc composite stabilizer and nano silicon dioxide is 0.5%-5% of the total mass. After the polymerization reaction is completed, 20g of calcium zinc composite stabilizer and 7g of nano silicon dioxide are added to form a composite stabilizer to further enhance the thermal stability and mechanical properties of polyvinyl chloride;

[0118] Among them, the polymerization reaction product is filtered, washed and dried to obtain high-stability polyvinyl chloride particles. In this stage, deionized water is used for washing to remove unreacted monomers and initiator residues to ensure the purity and stability of the final product. The polymerization product is filtered through a filter and then washed with deionized water to remove unreacted monomers and initiator residues. The washed polyvinyl chloride particles are dried in a drying device to finally obtain polyvinyl chloride particles with high thermal stability.

[0119] The prepared polyvinyl chloride (PVC) is used for insulation of wires and cables.

[0120] Different operating parameters and stabilizer addition ratios of the present invention are demonstrated through Example 1 and Example 2, which can be adjusted according to actual production needs. Through the description of these two embodiments, it can be seen that the method of the present invention has a significant effect in improving the thermal stability and mechanical properties of polyvinyl chloride, while maintaining the simplicity of operation and the controllability of cost.

[0121] like Figure 2 , a process for preparing optical cable coating materials. OMA is used to optimize the Swin Transformer model to optimize the formula ratio and process parameters in the process for preparing optical cable coating materials. The algorithm output optimizes the ratio of polyvinyl alcohol (PVA) to deionized water, polymerization reaction temperature, the amount of initiator (KPS), the amount of calcium zinc composite stabilizer and the amount of nano-silica, thereby improving product performance and production efficiency. The specific steps are as follows:

[0122] Step 1: Data collection and preprocessing

[0123] In order to accurately obtain process parameters and performance indicators, various sensors and hardware devices can be used. The following are the data collection sensors and hardware devices used:

[0124] Temperature sensor: Thermal resistor temperature sensor, used to monitor the polymerization reaction temperature and ensure that the reaction is carried out within the set temperature range;

[0125] Pressure sensor: used to measure the pressure inside the reactor to ensure the safety and stability of the reaction conditions;

[0126] Flow meter: used to measure the flow of different components to ensure that each component is added according to the set ratio:

[0127] Mass flow controller: used to accurately control the flow of gas to ensure that the reaction is carried out in an inert atmosphere;

[0128] pH sensor: used to monitor the pH value of the reaction system to ensure the stability of the system and the smooth progress of the reaction;

[0129] Viscometer: Rotational viscometer, used to measure the viscosity of the reaction process and the final product, to evaluate the progress of the reaction and product performance;

[0130] Mechanical properties testing equipment: universal testing machine, used to test the mechanical strength, impact resistance and flexibility of the prepared PVC samples, test items: tensile strength, compression strength, impact strength;

[0131] Thermal analysis equipment: Thermogravimetric analyzer (TGA), used to test the thermal stability of the prepared PVC samples and evaluate their performance in high temperature environments;

[0132] Data acquisition system: used to integrate and record data from various sensors and test equipment for subsequent analysis and model training. Modules: temperature module, pressure module, flow module, pH module, viscosity module;

[0133] Control system: Distributed control system (DCS) is used to monitor and adjust process parameters in real time to ensure the stability and optimization of the preparation process;

[0134] Data preprocessing:

[0135] Data cleaning: processing missing values; deleting samples with missing values ​​and filling missing values;

[0136] Handling outliers: Outliers will affect the training and prediction of the model and need to be processed using the Z-score method;

[0137] Data standardization: In order for the model to better learn the relationship between features, the data needs to be standardized;

[0138] Dataset division: In order to train, optimize and evaluate the model, the dataset needs to be divided into training set, validation set and test set, with a division ratio of 70% training set, 15% validation set and 15% test set;

[0139] Through the above sensors and hardware equipment, various parameters in the preparation process of optical cable coating materials can be accurately monitored and controlled to ensure the consistency, accuracy and completeness of data, providing a solid data foundation for subsequent intelligent optimization;

[0140] Step 2: Swin Transformer model establishment

[0141] Input embedding: converting input features into a fixed-length vector representation,

[0142] X=[0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1.0]

[0143] Where X is the input feature vector;

[0144] Patch partitioning: Divide the input feature vector into patches of fixed size, each patch contains several features.

[0145] X patch =[X[i:i+P-1]|i=1,P,2P,...,(n / P-1)P+1]

[0146] Among them, X patch represents the set of patch feature vectors after division, X is the input feature vector, P is the size of the patch, set P = 2, i represents the starting position index of the patch, and n represents the length of the input feature vector;

[0147] X patch =[[0.1, 0.2], [0.3, 0.4], [0.5, 0.6], [0.7, 0.8], [0.9, 1.0]]

[0148] Linear embedding: Linear embedding is performed on each patch to map it into a high-dimensional space.

[0149] E patch =W emb X patch

[0150] Among them, W emb is the weight matrix of the linear embedding, E patch represents the patch feature vector after linear embedding, X patch represents the input patch feature vector;

[0151]

[0152] Hierarchical structure: Swin Transformer extracts features through multiple hierarchical modules (Layer), each of which contains two stages: moving window self-attention (W-MSA) and cross-window self-attention (SW-MSA);

[0153] Moving Window Self-Attention (W-MSA): Multi-head self-attention mechanism is applied within each window to capture local features.

[0154]

[0155] Among them, Attention represents the output of the self-attention mechanism, softmax represents the probability distribution function, Q, K, and V are query, key, and value matrices respectively, and d k is the dimension of the key, QK T is the dot product of the query matrix Q and the transposed matrix of the key matrix K;

[0156]

[0157]

[0158] The formula of the multi-head self-attention mechanism is:

[0159] MultiHead(Q,K,V)=Concat(head1,head2,...,head h )W O

[0160]

[0161] Among them, MultiHead represents the output vector of the multi-head self-attention mechanism, Q, K, V are query, key and value matrices respectively, h represents the number of attention heads, head h represents the output of the hth attention head, Concat represents concatenating the outputs of multiple attention heads, Attention represents self-attention calculation, and head i represents the output of the i-th attention head, is the projection matrix, W O is the output matrix;

[0162] Cross-window self-attention (SW-MSA): By offsetting the window, the global features are captured across the window. Let the offset be s and the window size be M. The offset window self-attention formula is:

[0163] X shifted =Shift(X,s)

[0164] Where X is the input feature vector, X shifted represents the feature vector after the offset processing, s is the offset, and Shift represents the window offset operation;

[0165] After applying W-MSA, the feature is then offset back to its original position:

[0166] X SW-MSA =InverseShift(X shifted , s)

[0167] Among them, X SW-MSA represents the feature vector after being processed by the cross-window multi-head self-attention mechanism (SW-MSA), InverseShift represents the function of the inverse shift operation, and X shifted represents the feature vector after offset processing, s is the offset;

[0168] Layer Normalization and MLP: In each layered module, Layer Normalization and MLP are applied to further process the features.

[0169] X norm =LayerNorm(X)

[0170] X MLP =MLP(X norm )

[0171] MLP(X)=σ(XW1+b1)W2+b2

[0172] Among them, X normis the feature vector after layer normalization, LayerNorm represents the function of layer normalization operation, X represents the input feature vector, X MLP represents the output vector after MLP processing, MLP represents multilayer perceptron, MLP(X) represents the output vector of MLP, σ represents the activation function, W1 and W2 are the weight matrices of the first and second layers, b1 and b2 are the bias vectors of the first and second layers;

[0173] Data enhancement and regularization: During the training process, data enhancement and regularization techniques are used to improve the robustness and stability of the model.

[0174] X aug =DataAugmentation(X)

[0175] X reg = Dropout(X aug )

[0176] Among them, X aug is the feature vector after data enhancement processing, DataAugmentatio is the function of data enhancement operation, X represents the original feature vector, X reg is the feature vector after Dropout regularization, and Dropout is the function of Dropout regularization operation;

[0177] Output layer: Through the last layer of MLP, predict the PVC performance indicators.

[0178] Y = MLP(X final )

[0179] In fact, Y is the predicted PVC performance index, MLP stands for multi-layer perceptron, and X final It is the feature representation after all layered modules and regularization processing;

[0180]

[0181] Step 3: Owl Migration Algorithm (OMA) Optimization Model

[0182] Initialize the population: Assume that the population size is N = 3, and each individual x i Representing a possible solution, including formulation ratios and process parameters, the initial population is shown below:

[0183] x1 = [0.02, 60, 0.01, 0.5, 1.0]

[0184] x2=[0.05,65,0.02,1.0,1.5]

[0185] x3=[0.1,70,0.03,1.5,2.0]

[0186] Among them, x i is the position vector or solution vector of the i-th individual;

[0187] Fitness function: Fitness function (f(x i )) is used to evaluate the quality of each individual solution. In this application, the mean square error (MSE) can be used as the fitness function.

[0188]

[0189] Among them, f(x i ) is the value of the fitness function, y k is the true value, is the model prediction value, n is the number of samples;

[0190] Hunting and foraging behavior: During the hunting and foraging process, the owl updates its position according to the fitness value. The position update formula is:

[0191]

[0192] in, is the position of the ith individual in the t+1th generation, is the position of the ith individual in the tth generation, α is the step size factor, rand is a random number, x best is the current optimal position, α=0.1, rand=0.5, x best =[0.05,65,0.02,1.0,1.5];

[0193] Update X1 position:

[0194]

[0195] Update X2 position:

[0196]

[0197] Update X3 position:

[0198]

[0199] Migration behavior: During the migration process, the owl migrates according to the fitness value to explore new solution space. The migration update formula is:

[0200]

[0201] in, is the position of the ith individual in the t+1th generation, is the position of the ith individual in the tth generation, β is the migration step length, randn is a normally distributed random number, β = 0.1, randn = 0.02;

[0202] Update X1 position:

[0203]

[0204] Update X2 position:

[0205]

[0206] Update X3 position:

[0207]

[0208] Termination condition: When the maximum number of iterations T is reached or the fitness function value meets a certain threshold ∈, the algorithm stops and outputs the optimal solution.

[0209] ift≥Torf(x best )≤e,then stop

[0210] Among them, t is the current iteration number, T is the maximum iteration number, f(x best ) is the fitness value of the optimal solution, ∈ is the threshold of the fitness function value, or is the logical OR operator, if is the conditional statement, then stop is the termination operation;

[0211] The fitness function value calculated according to the mean square error is ∈=0.00025, at which point the operation is terminated and the number of iterations is 3.

[0212] In the preparation process of optical cable coating materials, the Swin Transformer model is combined with the Owl Migration Algorithm (OMA) to optimize the optimal formula ratio and process parameter combination x best As shown below:

[0213] x best =[x1,x2,x3,x4,x5]

[0214] Wherein, x1 is the ratio of polyvinyl alcohol (PVA) to deionized water, x2 is the polymerization temperature, x3 is the amount of initiator (KPS), x4 is the amount of calcium zinc composite stabilizer, and x5 is the amount of nano-silicon dioxide;

[0215] Using the OMA optimization algorithm, the population is initialized, and the hunting, foraging and migration behaviors are evaluated through the fitness function, and the individual positions are continuously updated until the optimal formula ratio and process parameter combination are found;

[0216] After 3 iterations, the optimal formula ratio and process parameter combination are found as follows:

[0217] x best =[0.056,65.006,0.026,1.006,1.506]

[0218] According to the results after iterative optimization, the optimal formula ratio and process parameter combination are: the ratio of polyvinyl alcohol (PVA) to deionized water is 0.056 (1:18), the polymerization reaction temperature is 65.006°C, the amount of initiator (KPS) is 0.026% (relative to the mass percentage of monomer vinyl chloride), the amount of calcium zinc composite stabilizer is 1.006% (relative to the total mass percentage), and the amount of nano-silica is 1.506% (relative to the total mass percentage).

[0219] Finally, it should be noted that: Obviously, the above embodiments are only examples for clearly explaining the present invention, and are not intended to limit the implementation methods. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from this are still within the scope of protection of the present invention.

Claims

1. A process for preparing an optical cable coating material, characterized in that: The optical cable coating material preparation process comprises the following steps: Step 1, dissolving a dispersant polyvinyl alcohol (PVA) in deionized water to form a pretreatment solution; Step 2: under the protection of inert gas, adding vinyl chloride monomer (VCM) into the pretreatment solution to form a stable dispersion system; Step 3, adding initiator potassium persulfate (KPS) to promote polymerization reaction; Step 4: After the polymerization reaction is completed, a composite stabilizer of calcium zinc and nano silicon dioxide is added.

2. A process for preparing an optical cable coating material according to claim 1, characterized in that: The ratio of the dispersant polyvinyl alcohol (PVA) to deionized water in the step 1 is 1:10-1:

50.

3. A process for preparing an optical cable coating material according to claim 1, characterized in that: In the step 3, the amount of the initiator potassium persulfate (KPS) used is 0.01%-1% of the mass of the monomer vinyl chloride (VCM).

4. A process for preparing an optical cable coating material according to claim 1, characterized in that: The polymerization reaction in step three is carried out at a temperature range of 40°C to 80°C.

5. A process for preparing an optical cable coating material according to claim 1, characterized in that: In the step 4, the dosage of the composite stabilizer of calcium zinc and nano silicon dioxide is 0.5%-5% of the total mass.

6. A process for preparing an optical cable coating material according to claim 1, characterized in that: The prepared polyvinyl chloride (PVC) is applied to the insulation of wires and cables.

7. A process for preparing an optical cable coating material according to claim 1, characterized in that: It also includes using OMA to optimize the Swin Transformer model to optimize the formula ratio and process parameters in the preparation process of optical cable coating materials. The algorithm output optimizes the ratio of polyvinyl alcohol (PVA) to deionized water, polymerization reaction temperature, the amount of initiator (KPS), the amount of calcium zinc composite stabilizer and the amount of nano-silica, thereby improving product performance and production efficiency. The specific steps are as follows: Step A: Data collection and preprocessing In order to accurately obtain process parameters and performance indicators, various sensors and hardware devices are used. The following are the data collection sensors and hardware devices used: Temperature sensor: Thermal resistor temperature sensor, used to monitor the polymerization reaction temperature and ensure that the reaction is carried out within the set temperature range; Pressure sensor: used to measure the pressure inside the reactor to ensure the safety and stability of the reaction conditions; Flow meter: used to measure the flow of different components to ensure that each component is added according to the set ratio; Mass flow controller: used to accurately control the flow of gas to ensure that the reaction is carried out in an inert atmosphere; pH sensor: used to monitor the pH value of the reaction system to ensure the stability of the system and the smooth progress of the reaction; Viscometer: Rotational viscometer, used to measure the viscosity of the reaction process and the final product, to evaluate the progress of the reaction and product performance; Mechanical properties testing equipment: universal testing machine, used to test the mechanical strength, impact resistance and flexibility of the prepared PVC samples, test items: tensile strength, compression strength, impact strength; Thermal analysis equipment: Thermogravimetric analyzer TGA, used to test the thermal stability of the prepared PVC samples and evaluate their performance in high temperature environments; Data acquisition system: used to integrate and record data from various sensors and test equipment for subsequent analysis and model training. Modules: temperature module, pressure module, flow module, pH module, viscosity module; Control system: Distributed control system DCS, used to monitor and adjust process parameters in real time to ensure the stability and optimization of the preparation process; Data preprocessing: Data cleaning: processing missing values; deleting samples with missing values ​​and filling missing values; Handling outliers: Outliers will affect the training and prediction of the model and need to be processed using the Z-score method; Data standardization: In order for the model to better learn the relationship between features, the data needs to be standardized; Dataset division: In order to train, optimize, and evaluate the model, the dataset needs to be divided into training set, validation set, and test set; Step B: Swin Transformer model building Input embedding: converting input features into a fixed-length vector representation, X=[x1,x2,...,x n ] Where X is the input feature vector, x n is the nth feature, n is the number of samples; Patch partitioning: Divide the input feature vector into patches of fixed size, each patch contains several features. X patch =[X[i:i+P-1]|i=1,P,2P,...,(n / P-1)P+1] Among them, X patch represents the set of patch feature vectors after division, X is the input feature vector, P is the size of the patch, i represents the starting position index of the patch, and n represents the length of the input feature vector; Linear embedding: Linear embedding is performed on each patch to map it into a high-dimensional space. HAVE BEEN patch =W emb X patch Among them, W emb is the weight matrix of the linear embedding, E patch represents the patch feature vector after linear embedding, X patch represents the input patch feature vector; Hierarchical structure: Swin Transformer extracts features through multiple hierarchical modules, each of which contains two stages: moving window self-attention W-MSA and cross-window self-attention SW-MSA; Moving Window Self-Attention W-MSA: A multi-head self-attention mechanism is applied within each window to capture local features. Among them, Attention represents the output of the self-attention mechanism, softmax represents the probability distribution function, Q, K, and V are query, key, and value matrices respectively, and d k is the dimension of the key, QK T is the dot product of the query matrix Q and the transposed matrix of the key matrix K; The formula of the multi-head self-attention mechanism is: MultiHead(Q,K,V)=Concat(head1,head2,...,head h )W O Among them, MultiHead represents the output vector of the multi-head self-attention mechanism, Q, K, V are query, key and value matrices respectively, h represents the number of attention heads, head h represents the output of the hth attention head, Concat represents concatenating the outputs of multiple attention heads, Attention represents self-attention calculation, and head i represents the output of the i-th attention head, is the projection matrix, W O is the output matrix; Cross-window self-attention SW-MSA: By offsetting the window, the global features are captured across the window. Let the offset be s and the window size be M. The offset window self-attention formula is: X shifted =Shift(X,s) Where X is the input feature vector, X shifted represents the feature vector after the offset processing, s is the offset, and Shift represents the window offset operation; After applying W-MSA, the feature is then offset back to its original position: X SW-MSA =InverseShift(X shifted ,s) Among them, X SW-MSA represents the feature vector after being processed by the cross-window multi-head self-attention mechanism SW-MSA, InverseShift represents the function of the inverse shift operation, X shifted represents the feature vector after offset processing, s is the offset; Layer Normalization and MLP: In each layered module, Layer Normalization and MLP are applied to further process the features. X norm =LayerNorm(X) X MLP =MLP(X norm ) MLP(X)=σ(XW1+b1)W2+b2 Among them, X norm is the feature vector after layer normalization, LayerNorm represents the function of layer normalization operation, X represents the input feature vector, X MLP represents the output vector after MLP processing, MLP represents multilayer perceptron, MLP(X) represents the output vector of MLP, σ represents the activation function, W1 and W2 are the weight matrices of the first and second layers, b1 and b2 are the bias vectors of the first and second layers; Data enhancement and regularization: During the training process, data enhancement and regularization techniques are used to improve the robustness and stability of the model. X aug =DataAugmentation(X) X reg =Dropout(X aug ) Among them, X aug is the feature vector after data augmentation processing, DataAugmentation is the function of data augmentation operation, X represents the original feature vector, X reg is the feature vector after Dropout regularization, and Dropout is the function of Dropout regularization operation; Output layer: Through the last layer of MLP, predict the PVC performance indicators. Y=MLP(X final ) In fact, Y is the predicted PVC performance index, MLP stands for multi-layer perceptron, and X final It is the feature representation after all layered modules and regularization processing; Step C: Owl Migration Algorithm OMA Optimization Model Initialize the population: Assume that the population size is N, and each individual x i Represents a possible solution, including formulation ratios and process parameters, x i =[x i1 ,x i2 ,...,x id ] Among them, x i is the position vector or solution vector of the ith individual, d is the dimension, x id is the dth parameter of the ith individual; Fitness function: Fitness function (f(x i )) is used to evaluate the quality of each individual solution. In this application, the mean square error (MSE) can be used as the fitness function. Among them, f(x i ) is the value of the fitness function, y k is the true value, is the model prediction value, n is the number of samples; Hunting and foraging behavior: During the hunting and foraging process, the owl updates its position according to the fitness value. The position update formula is: in, is the position of the i-th individual in the t+1th generation, is the position of the ith individual in the tth generation, α is the step size factor, rand is a random number, x best is the current optimal position; Migration behavior: During the migration process, the owl migrates according to the fitness value to explore new solution space. The migration update formula is: in, is the position of the i-th individual in the t+1th generation, is the position of the ith individual in the tth generation, β is the migration step length, and randn is a normally distributed random number; Termination condition: When the maximum number of iterations T is reached or the fitness function value meets a certain threshold ∈, the algorithm stops and outputs the optimal solution. if t≥Torf(x best )≤∈,then stop Among them, t is the current number of iterations, T is the maximum number of iterations, f(x best ) is the fitness value of the optimal solution, ∈ is the threshold of the fitness function value, or is the logical OR operator, if is the conditional statement, then stop is the termination operation; In the preparation process of optical cable coating materials, the Swin Transformer model is combined with the Owl Migration Algorithm (OMA) to optimize the optimal formula ratio and process parameter combination x best As shown below: x best =[x1,x2,x3,x4,x5] Among them, x1 is the ratio of polyvinyl alcohol (PVA) to deionized water, x2 is the polymerization temperature, x3 is the amount of initiator (KPS), x4 is the amount of calcium zinc composite stabilizer, and x5 is the amount of nano-silicon dioxide; The OMA optimization algorithm is used to initialize the population, evaluate hunting, foraging, and migration behaviors through the fitness function, and continuously update the individual positions until the optimal formula ratio and process parameter combination is found.

Images