An antistatic resin composition for electric wire cable and a method for preparing the same

Abstract

The application provides an antistatic resin composition for electric wire and cable and a preparation method thereof, and belongs to the technical field of polymer material processing, and the preparation method comprises the following steps: S1: preparing a quaternary ammonium salt type ionic liquid-graphene quantum dot composite; S2: plasma pre-activation treatment of a main body resin; S3: targeted grafting and compounding of the activated main body resin and the composite; S4: multi-component gradient compatible compounding; and S5: dynamic vulcanization-extrusion molding. The application solves the problems of unstable performance, poor component compatibility and low process controllability of the existing antistatic resin, and through component optimization and process innovation, finally obtains a resin composition which is durable in antistatic performance, resistant to aging and wear, uniform in performance, strong in process controllability and suitable for large-scale production, and can be widely applied to scenes such as mines, chemical industry and electronics which have strict requirements on the antistatic performance of electric wire and cable.

Description

Technical Field

[0001] This invention relates to the field of polymer material processing technology, and in particular to an antistatic resin composition for wires and cables and its preparation method. Background Technology

[0002] During the transmission of electrical energy or signals, electric wires and cables are prone to static electricity accumulation due to conductor friction, dry environment and other factors. Static breakdown may cause problems such as insulation layer damage and signal interference. Especially in high-risk environments, it may also induce safety accidents such as fire and explosion. Therefore, antistatic performance is one of the key indicators of resin materials for electric wires and cables.

[0003] In existing technologies, antistatic resin compositions for wires and cables mostly achieve antistatic functions by adding conductive fillers such as carbon black and metal powder or conventional antistatic agents. However, this approach has several drawbacks: First, excessive amounts of conductive fillers can lead to a decrease in the mechanical properties of the resin, while insufficient amounts result in unstable antistatic performance. Furthermore, the fillers have poor compatibility with the main resin, making them prone to agglomeration. Second, conventional antistatic agents are mostly small molecules that are prone to migration and precipitation during long-term use, leading to a decline in antistatic performance. Third, the preparation process often adopts a crude method of direct mixing and melt extrusion, without precise control over the matching relationship between raw material characteristics such as particle size and surface activity and process parameters. This results in significant fluctuations in product performance, making it difficult to meet the stable usage requirements of high-end wires and cables.

[0004] Furthermore, existing technologies for optimizing the preparation of antistatic resins rely heavily on empirical parameters and lack dynamic control mechanisms based on raw material characteristics. This makes it impossible to achieve adaptable production for different batches of raw materials, further limiting the stability of product quality and large-scale application. Summary of the Invention

[0005] This invention provides an antistatic resin composition for wires and cables and its preparation method, overcoming the problems of unstable antistatic performance, poor component compatibility, low process controllability, and large product performance fluctuations in existing antistatic resin compositions for wires and cables. By optimizing the preparation process and establishing a dynamic parameter control mechanism based on raw material characteristics, the process controllability and product uniformity are improved, meeting the stringent requirements of high-end wires and cables for material performance and expanding the application range of antistatic resin compositions.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for preparing an antistatic resin composition for wires and cables includes the following steps:

[0008] S1: Prepare quaternary ammonium salt type ionic liquid-graphene quantum dot composite and test its particle size data;

[0009] S2: Perform plasma pre-activation treatment on the main resin to obtain activated main resin, and test the surface activation energy data of the activated main resin.

[0010] S3: The core parameters of the nuclear principal component analysis-radial basis function neural network are optimized by using an improved fruit fly optimization algorithm. The particle size data and surface activation energy data are input into the optimized nuclear principal component analysis-radial basis function neural network to obtain the optimal stirring rate and optimal composite temperature. The activated host resin and the composite are then targeted grafted and composited at the optimal stirring rate and optimal composite temperature to obtain the grafted composite resin, and the grafting rate data is obtained at the same time.

[0011] S4: The core parameters of the generalized regression neural network are optimized by an adaptive particle swarm optimization algorithm. The grafting rate data is input into the optimized generalized regression neural network to obtain the optimal compatibilizer addition amount and the temperature correction value of each stage of gradient mixing. After adjusting the process parameters according to the optimal compatibilizer addition amount and the temperature correction value of each stage of gradient mixing, gradient compatibilization is carried out to obtain the mixture. At the same time, the melt index data of the mixture is obtained.

[0012] S5: Set the temperature and screw speed of each section of the twin-screw extruder according to the melt index data of the compound, add the compound to the twin-screw extruder, inject the vulcanizing agent and perform dynamic vulcanization, extrusion molding, and prepare the antistatic resin composition.

[0013] In this specification, in S3, the improved fruit fly optimization algorithm and the kernel principal component analysis-radial basis function neural network form a bidirectional interaction. The specific interaction process is as follows: the optimal core parameters output by the improved fruit fly optimization algorithm are input into the kernel principal component analysis-radial basis function neural network, and the error between the grafting rate prediction value output by the kernel principal component analysis-radial basis function neural network and the measured grafting rate data is fed back to the improved fruit fly optimization algorithm, guiding the improved fruit fly optimization algorithm to adjust its optimization strategy.

[0014] In this specification, in S3, the improved fruit fly optimization algorithm and the kernel principal component analysis-radial basis function neural network form a closed loop through bidirectional interaction. The closing loop ends when the error between the grafting rate prediction value output by the kernel principal component analysis-radial basis function neural network and the measured grafting rate data is less than or equal to a preset threshold, or when the number of iterations of the improved fruit fly optimization algorithm reaches the preset maximum number of iterations.

[0015] In this specification, in S4, the adaptive particle swarm optimization and the generalized regression neural network form a two-way interaction. The specific interaction process is as follows: the optimal core parameters output by the adaptive particle swarm optimization are input into the generalized regression neural network, and the error between the melt index prediction value output by the generalized regression neural network and the measured melt index data is fed back to the adaptive particle swarm optimization, guiding the adaptive particle swarm optimization to adjust the inertia weights and optimization strategy.

[0016] In this specification, in S4, the bidirectional interaction between adaptive particle swarm optimization and generalized regression neural network forms a closed loop. The closing loop ends when the error between the melt index prediction value output by the generalized regression neural network and the measured melt index data is less than or equal to a preset threshold, or when the number of iterations of adaptive particle swarm optimization reaches the preset maximum number of iterations.

[0017] In this specification, in step S1, if the particle size data exceeds the preset range, the hydrothermal reaction temperature or reaction time is adjusted, and the quaternary ammonium salt ionic liquid-graphene quantum dot composite is prepared again until the particle size data meets the preset range.

[0018] In this specification, in step S2, if the surface activation energy data of the activated main resin exceeds the preset range, the plasma treatment power or treatment time is adjusted, and the main resin is re-treated with plasma pre-activation until the surface activation energy data meets the preset range.

[0019] In this manual, before optimizing the core parameters of the kernel principal component analysis-radial basis function neural network in S3 using the improved fruit fly optimization algorithm, the training dataset needs to be preprocessed. The training dataset comes from previous orthogonal experiments, and the preprocessing method is to map all parameters in the training dataset to a preset interval.

[0020] In this specification, in step S5, the vulcanizing agent is injected into the middle section of the twin-screw extruder via a metering pump, and the flow rate of the metering pump is dynamically adjusted according to the feed rate of the twin-screw extruder.

[0021] An antistatic resin composition for wires and cables comprises the following raw materials in parts by weight: 65-75 parts of low-density polyethylene-ethylene-methyl acrylate terpolymer, 8-12 parts of quaternary ammonium salt ionic liquid-graphene quantum dot composite, 3-5 parts of ethylene-vinyl acetate grafted maleic anhydride compatibilizer, 0.5-1 parts of antioxidant combination, and 1-2 parts of lubricant combination; wherein the antioxidant combination is composed of antioxidant 1010 and antioxidant 168 in a mass ratio of 1:1, and the lubricant combination is composed of calcium stearate and ethylene bis-stearamide in a mass ratio of 2:1.

[0022] In summary, the present invention has at least the following beneficial effects:

[0023] Stable and long-lasting antistatic performance: Using a self-made quaternary ammonium salt ionic liquid-graphene quantum dot composite as the core antistatic component, combined with a targeted grafting composite process, the antistatic component is uniformly dispersed in the main resin, avoiding component agglomeration or migration, and ensuring stable and long-lasting antistatic effect.

[0024] Excellent overall mechanical and environmental resistance properties: The synergistic effect of each component ensures antistatic properties while improving the aging resistance and wear resistance of the composition, making it adaptable to complex and harsh operating environments and extending the service life of wires and cables.

[0025] Good component compatibility: Through plasma pre-activation treatment and dynamic adjustment of compatibilizer, the interfacial bonding ability between the main resin and antistatic components and other auxiliary components is significantly improved, avoiding defects such as delamination and cracking.

[0026] High process controllability: The preparation process introduces a precise parameter control mechanism, which can dynamically adjust process parameters according to the characteristics of raw materials, realize the adaptability of raw materials for different batches, reduce product performance fluctuations, and improve the stability of large-scale production. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic flowchart illustrating the preparation method of the antistatic resin composition for wires and cables involved in this invention.

[0029] Figure 2 This is a schematic diagram of the plasma pre-activation treatment process of the main resin involved in this invention.

[0030] Figure 3 This is a schematic diagram of the process of targeted grafting composite and algorithm optimization interaction involved in this invention.

[0031] Figure 4 This is a schematic diagram illustrating the process of multi-component gradient compatible mixing and algorithm optimization interaction involved in this invention. Detailed Implementation

[0032] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the embodiments of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0033] The following disclosure provides many different implementations or examples for carrying out different structures of the embodiments of the present invention. To simplify the disclosure of the embodiments of the present invention, specific examples of components and arrangements are described below. Of course, these are merely examples and are not intended to limit the embodiments of the present invention. Furthermore, reference numerals and / or reference letters may be repeated in different examples of the embodiments of the present invention; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various implementations and / or arrangements discussed.

[0034] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0035] like Figure 1 As shown, this embodiment provides a method for preparing an antistatic resin composition for wires and cables, comprising:

[0036] S1: Preparation of quaternary ammonium salt ionic liquid-graphene quantum dot composite

[0037] The technical objective of this step is to synthesize a quaternary ammonium salt-type ionic liquid-graphene quantum dot composite with high conductivity and dispersion stability, provide a core antistatic component that meets the particle size requirements, and accurately obtain the particle size data of the composite. .

[0038] Specific implementation details: First, a reaction apparatus was set up using a 500mL three-necked flask equipped with a mechanical stirrer, thermometer, and reflux condenser. 10 parts citric acid and 30 parts deionized water were added to the flask. The mechanical stirrer was turned on at a speed of 300 rpm. The temperature was raised to 95℃ and maintained for 30 minutes to ensure complete dissolution of the citric acid and the formation of a homogeneous aqueous solution. Then, 2 parts graphene powder (purity ≥99.5%, sheet thickness 1–3 nm) were added to this aqueous solution. The mechanical stirrer was turned off, and the mixture was transferred to an ultrasonic disperser. The ultrasonic power was set to 300W and the ultrasonic frequency to 40kHz. Continuous ultrasonic dispersion was performed for 60 minutes, during which the system temperature was controlled to not exceed 50℃ (cooling was assisted by an ice-water bath) to obtain a homogeneous and stable graphene dispersion.

[0039] The ultrasonically purified graphene dispersion was transferred to a 200 mL hydrothermal reactor with a polytetrafluoroethylene (PTFE) liner. After sealing, the reactor was placed in an oven, and the reaction temperature was set to 180℃ for 12 hours to initiate the hydrothermal reaction. After the reaction, the oven was closed, and the reactor was allowed to cool naturally to room temperature for approximately 4 hours. The product was then removed from the reactor and filtered using a Buchner funnel with a 0.22 μm organic filter membrane. The filter cake was washed three times with deionized water (50 parts water per wash) until the pH of the washing solution reached 6.5–7.0. The washed filter cake was then placed in a vacuum drying oven at 60℃ and a vacuum of -0.09 MPa for 12 hours. After drying, the cake was removed, ground, and passed through a 200-mesh standard sieve to obtain graphene quantum dots.

[0040] Five parts of the graphene quantum dots prepared above, 20 parts of N-methylimidazole with a purity ≥99%, and 30 parts of n-bromobutane with a purity ≥99% were added to a 250 mL round-bottom flask. The flask was placed in a nitrogen protection device, and nitrogen gas with a purity ≥99.99% was introduced to maintain an inert atmosphere. The nitrogen flow rate was controlled at 50 mL / min, and nitrogen was continuously introduced for 10 min to purge the air from the flask. Then, the flask was placed in an oil bath, heated to 80 °C, and magnetic stirring was started at a stirring speed of 500 r / min. The mixture was kept at this temperature and stirred for 24 h to carry out the ionization reaction. After the reaction was completed, heating and stirring were stopped, and the system was allowed to cool to room temperature. 100 parts of diethyl ether (analytical grade) were added to the flask, and the mixture was stirred for 30 min. After standing and separating the layers, the upper diethyl ether phase was discarded, and the lower viscous product was retained. The diethyl ether washing operation was repeated 3 times to remove unreacted N-methylimidazole and n-bromobutane. The washed product was placed in a vacuum drying oven, and the drying temperature was set to 40℃ and the vacuum degree to -0.09MPa. After drying for 8 hours, it was taken out to obtain a quaternary ammonium salt ionic liquid-graphene quantum dot composite.

[0041] The particle size of the composite was measured using a laser particle size analyzer (model: Malvern 3000). The test medium was deionized water. The dispersion ultrasonic power was 100W and the ultrasonic time was 2min. Each sample was tested in triplicate, and the average value was taken as the final particle size data of the composite. ,make sure If the particle size is within the range of 10 to 20 nm, or if it exceeds this range, return to adjust the hydrothermal reaction temperature or reaction time. The temperature fluctuation range is ±5℃, and the time fluctuation range is ±2h. Re-prepare until the particle size meets the standard. After meeting the standard, the composite is sealed and stored in a desiccator for later use.

[0042] S2: Plasma pre-activation treatment of the main resin

[0043] The goal of this step is to enhance the surface activity of the low-density polyethylene-ethylene-methyl acrylate terpolymer (the main resin) through plasma modification treatment, thereby strengthening its interfacial bonding ability with the composite prepared in S1, and accurately obtaining resin surface activation energy data. The plasma pre-activation process of the main resin is shown in the figure.

[0044] Specific implementation details: Weigh the required amount of low-density polyethylene-ethylene-methyl acrylate terpolymer according to the formula, and place it in a vacuum drying oven for pretreatment. Set the drying temperature to 60℃ and the vacuum degree to -0.08MPa, and dry for 4 hours to remove moisture and trace impurities adsorbed on the surface of the resin particles, avoiding interference from moisture during subsequent plasma treatment. After drying, evenly spread the resin particles in the sample tray of the plasma processor, controlling the spreading thickness to 2-3mm to ensure uniform heating of the resin particles.

[0045] The plasma treatment chamber was closed, and the vacuum system was activated to evacuate the chamber until the pressure dropped to 5 Pa. This vacuum was maintained for 10 minutes to further remove any residual air and moisture. Then, the argon valve was opened to introduce argon gas into the chamber. The argon flow rate was adjusted to 20 sccm using the flow controller to stabilize the chamber pressure at 20 Pa. The plasma treatment parameters were set to 150 W power and 15 minutes. The plasma generator was then started to activate the resin surface. During the treatment, the chamber pressure and power were monitored in real time to ensure the parameters remained stable without fluctuations.

[0046] After processing, the plasma generator and argon gas valve were shut off. Once the pressure inside the chamber returned to atmospheric pressure, the chamber was opened and the activated resin particles were removed. The contact angle between the resin surface and deionized water was measured using a contact angle meter (model: KRÜSS DSA100) at 25℃ and 50% humidity. Five different test points were selected for each sample, and the average value was taken as the final contact angle data. The activation energy of the resin surface was calculated according to the Young-Laplace equation. The calculation formula is: ,in The surface tension of deionized water is 72.8 at 25℃. ; The contact angle between the resin surface and deionized water after plasma pre-activation treatment; The interfacial tension between the resin surface and air; The interfacial tension between the resin surface and deionized water; in this scheme, The physical meaning of is the interfacial tension between the resin surface and air, therefore it is directly defined as: ,Will and measured contact angle Substituting into the Young-Laplace equation, we can calculate... The value is It is necessary to adjust the plasma processing parameters to ensure The value is between 35 and 40. Within the specified range. If the target is not met, adjust the plasma treatment power (fluctuation range ±20W) or treatment time (fluctuation range ±3min) and reprocess until the activation energy meets the target. Once the target is met, the resin particles should be immediately transferred to the next step to avoid prolonged storage which could lead to a decrease in surface activity.

[0047] S3: Targeted grafting composite of activated host resin and complex.

[0048] This step breaks away from the conventional, extensive approach of setting parameters by interval. It establishes a new algorithm through a bidirectional fusion of an improved fruit fly optimization algorithm and kernel principal component analysis-radial basis function neural network. , The precise mapping relationship between the grafting rate and the mixing rate and the compounding temperature ensures the stability and consistency of the grafting rate under different raw material properties, providing stable grafted composite resin raw materials and accurate grafting rate data. The interactive process of targeted grafting composite and algorithm optimization is as follows: Figure 3 As shown.

[0049] 3.1 Algorithm Model Construction Process

[0050] First, a basic model based on kernel principal component analysis-radial basis function neural network is constructed. This model serves as the core prediction model, responsible for mapping input parameters to target process parameters. The model input is obtained from S1. and S2 obtained These two parameters directly determine the interfacial interaction strength between the host resin and the antistatic composite, thus affecting the grafting reaction efficiency. The model outputs the stirring rate and compounding temperature required to achieve the optimal grafting rate; these two process parameters are crucial for controlling the uniformity of the grafting reaction.

[0051] To improve the accuracy of the prediction model and avoid the tendency of radial basis function neural networks to get trapped in local optima, an improved fruit fly optimization algorithm is introduced as an optimizer to optimize the core parameters of the kernel principal component analysis-radial basis function neural network. The core improvement of the improved fruit fly optimization algorithm lies in the introduction of a dynamic step size factor. By dynamically adjusting the optimization step size, it balances the global search capability in the early stage with the local convergence accuracy in the later stage, thus solving the problem of slow convergence in the later stage of the traditional fruit fly optimization algorithm.

[0052] Subsequently, a fusion architecture of kernel principal component analysis-radial basis function neural network and improved fruit fly optimization algorithm was constructed. The output of the improved fruit fly optimization algorithm is directly used as the core parameter input of kernel principal component analysis-radial basis function neural network, while the prediction error of kernel principal component analysis-radial basis function neural network is fed back to the improved fruit fly optimization algorithm to guide it to adjust its optimization strategy, forming a two-way interactive closed-loop optimization system.

[0053] 3.2 Model Training Process

[0054] The first step in model training is preparing a high-quality training dataset. This dataset comes from previous orthogonal experiments, and a total of 100 valid samples were collected. Each sample contains complete input parameters, corresponding output parameters, and experimental results, specifically covering… , Stirring rate, compounding temperature, and measured values .in The value range is 10nm to 20nm, and it is set in 1nm increments. The value range is 35mN / m to 40mN / m, set in 0.5mN / m increments; the stirring rate is adjustable from 800r / min to 1400r / min; the compounding temperature is adjustable from 90℃ to 120℃; and the final measured value is... The range is 80% to 95%.

[0055] To eliminate the interference of differences in parameter dimensions and numerical ranges on model training, all sample data need to be preprocessed. The min-max normalization method is used to map all parameters to the interval between 0 and 1. The normalization formula is as follows:

[0056] In the formula This represents the normalized parameter value. Represents the original parameter value. This represents the minimum value of the parameter across all 100 training samples. This represents the maximum value of the parameter across all 100 training samples. This is the core parameter for this step. corresponding , ; corresponding , ; stirring rate corresponding to , The composite temperature corresponds to , ; corresponding , Using a certain sample For example, its normalized value .

[0057] After data preprocessing, the improved Drosophila optimization algorithm was launched to optimize the core parameters of the kernel principal component analysis-radial basis function neural network. First, the initial parameters of the improved Drosophila optimization algorithm were set, with the Drosophila population size set to 30. The value is set to 50, and the parameter to be optimized is the number of hidden layer neurons in the kernel principal component analysis-radial basis function neural network. and Gaussian kernel width ,in The value range is from 5 to 20. The value range is from 0.1 to 1.0.

[0058] Dynamic step size factor of the improved fruit fly optimization algorithm The calculation uses the following formula:

[0059] In the formula Representing the Dynamic step size in the next iteration This represents the initial step size and has a value of 0.8. Represents the current iteration number (an integer from 1 to 50). This represents the maximum number of iterations for the improved fruit fly optimization algorithm and is 50. As the number of iterations increases... The increase in dynamic step size By gradually decreasing the step size, the goal is to achieve rapid global search with a large step size in the early stage and precise local convergence with a small step size in the later stage.

[0060] During the optimization process, the mean square error between the predicted grafting rate and the actual grafting rate of the kernel principal component analysis-radial basis function neural network is used as the fitness function of the improved Drosophila optimization algorithm. The calculation formula is as follows:

[0061] ;

[0062] In the formula The fitness value, or mean squared error, represents the improved fruit fly optimization algorithm. This represents the total number of training samples and has a value of 100. Representing the The grafting rate of the sample group was predicted by kernel principal component analysis-radial basis function neural network. Representing the The grafting rate was obtained from experimental measurements using a sample group. The improved Drosophila optimization algorithm iteratively updates the population position to find the optimal grafting rate. Minimum number of hidden layer neurons and Gaussian kernel width .

[0063] During the iteration process or the number of iterations reaches When the time is reached, the optimization process terminates, and the final output of the optimal parameters is: the optimal number of neurons in the hidden layer. Optimal Gaussian kernel width Substituting these two optimal parameters into a kernel principal component analysis-radial basis function neural network, the model is trained using a preprocessed training dataset. The prediction accuracy of the model after training is then assessed. The value reaches 0.985, which can meet the prediction requirements of actual processes.

[0064] 3.3 Connection between Model Application Process and Steps

[0065] In the model application phase, the measured data from step S1 of the current batch are first collected. Actual measurements of step S2 Substituting these two measured parameters into the optimized kernel principal component analysis-radial basis function neural network, the input parameters are first reduced in dimensionality using a kernel principal component analysis layer, and then a Gaussian kernel function is used to map the 2D input parameters to a 3D feature space, eliminating... and The coupling interference between them yields the feature vector as .

[0066] The feature vector is then input into the hidden layer of the radial basis function neural network. After activation operations of the hidden layer neurons and linear mapping of the output layer, the predicted stirring rate is finally output. and composite temperature Using these two predicted parameters as actual process parameters, a high-speed mixer was started to perform targeted grafting of the activated main resin and the composite. After stirring at a constant temperature for 30 minutes, the grafting process was completed, and the grafted composite resin was obtained.

[0067] To ensure the grafting effect meets the requirements, the grafting rate of the obtained grafted composite resin needs to be tested. The measured grafting rate was compared with the predicted grafting rate (90.3%) of the kernel principal component analysis-radial basis function neural network. The calculated error was 0.2%, which is less than the set error threshold of 0.5%, indicating that the model's prediction accuracy meets the requirements and no parameter adjustment is needed. If the error is greater than 0.5%, then the measured data ( The improved fruit fly optimization algorithm is added to the training dataset and restarted to optimize and update the parameters of the kernel principal component analysis-radial basis function neural network, ensuring the prediction accuracy of subsequent batches.

[0068] After completing step S3, the obtained grafted composite resin and the measured... Proceed to the next step with synchronized transmission.

[0069] S4: Multi-component gradient compatible mixture

[0070] This step addresses the issue of conventional processes where the amount of compatibilizer added cannot be adapted to different conditions. The problem is solved by establishing a two-way fusion of adaptive particle swarm optimization and generalized regression neural network. The dynamic mapping relationship between compatibilizer addition, mixing temperature correction, and target melt index ensures the homogeneity and stability of multi-component mixtures, providing compliant raw materials and accurate melt index data for dynamic vulcanization-extrusion molding. The interactive process of multi-component gradient compatible mixing and algorithm optimization is as follows: Figure 4 As shown.

[0071] 4.1 Algorithm Model Construction Process

[0072] First, we construct the basic model of the generalized regression neural network. This model serves as the core prediction model and is responsible for... Mapping functionality to multiple sets of target parameters. The model input is the measured data transferred in the S3 step. , This directly determines the interfacial compatibility of the grafted composite resin, thus affecting the required dosage of compatibilizer and the optimization direction of the mixing process parameters. The model output includes three core parameters: the optimal compatibilizer addition amount, the correction values ​​for the temperature of each stage of gradient mixing, and the target melt index. The temperature correction values ​​correspond to the originally set temperature of Zone 1 (120℃), Zone 2 (130℃), and Zone 3 (140℃), while the target melt index is a key indicator for judging whether the mixture meets the requirements for subsequent extrusion.

[0073] To improve the accuracy of the prediction model and avoid the drawback of generalized regression neural networks (GRN) relying excessively on empirical values ​​of the smoothing factor, adaptive particle swarm optimization (PSO) is introduced as an optimizer to optimize the smoothing factor, a core parameter of the GRN. The core improvement of PSO lies in the introduction of adaptive inertia weights. By dynamically adjusting these weights, the algorithm's global search capability and local convergence accuracy are balanced, addressing the premature convergence problem inherent in traditional PSO algorithms.

[0074] Subsequently, a fusion architecture of generalized regression neural network and adaptive particle swarm optimization is constructed. The output of adaptive particle swarm optimization is directly used as the input of the smoothing factor, the core parameter of generalized regression neural network, while the melting index prediction error of generalized regression neural network is fed back to adaptive particle swarm optimization to guide it to adjust inertial weights and optimization strategies, forming a two-way interactive closed-loop optimization system.

[0075] 4.2 Model Training Process

[0076] The first step in model training is preparing a high-quality training dataset. This dataset comes from previous hybrid experiments with different grafting rates of S3, collecting 80 valid samples. Each sample includes complete input parameters, corresponding output parameters, and experimental results, specifically covering… Compatibilizer addition amount, temperature correction values ​​for each stage of gradient mixing, target melt index, and measured values. .in The value range is 85% to 92%, set in 0.1% increments; the compatibilizer addition amount is adjusted from 3 to 5 parts, set in 0.1-part increments; the temperature correction value is adjusted from -5℃ to 5℃; the target melt index ranges from 2.0 g / 10 min to 2.5 g / 10 min, and the final measured value is... The range is consistent with the target melt index range.

[0077] The same min-max normalization method as in step S3 is used to preprocess all sample data, mapping all parameters to the interval between 0 and 1. The normalization formula is as follows:

[0078] ;

[0079] In the formula This represents the normalized parameter value. Represents the original parameter value. This represents the minimum value of the parameter across all 80 training samples. This represents the maximum value of the parameter across all 80 training samples. This is the core parameter for this step. corresponding , The amount of compatibilizer added corresponds to , Temperature correction value corresponding to , ; corresponding , The amount of compatibilizer added for a specific sample. Taking a sample as an example, its normalized value .

[0080] After data preprocessing, adaptive particle swarm optimization is initiated to optimize the smoothing factor, a core parameter of the generalized regression neural network. Optimization is then performed. First, the initial parameters for adaptive particle swarm optimization are set, with the particle swarm size set to 25. The smoothing factor is set to 40 and is to be optimized. The value range is from 0.01 to 0.5.

[0081] The adaptive inertia weight in adaptive particle swarm optimization is calculated using the following formula:

[0082] ;

[0083] In the formula Representing the Adaptive inertia weights in the next iteration This represents the maximum inertia weight and has a value of 0.9. This represents the minimum inertia weight and has a value of 0.4. Represents the current iteration number (an integer from 1 to 40). This represents the maximum number of iterations in adaptive particle swarm optimization and has a value of 40. As the number of iterations increases... The increase, By gradually reducing the weight, the goal is to achieve the effect of enhancing global search capabilities with large weights in the early stages and improving local convergence accuracy with small weights in the later stages.

[0084] During the optimization process, the mean absolute error between the predicted melt index and the actual melt index from the generalized regression neural network is used as the fitness function for adaptive particle swarm optimization. The calculation formula is as follows:

[0085] ;

[0086] In the formula The fitness value, representing the mean absolute error, is the value used in adaptive particle swarm optimization. This represents the total number of training samples and has a value of 80. Representing the The melt flow index of the sample group was predicted by a generalized regression neural network. Representing the The melt flow index of the sample was obtained through experimental measurements. Adaptive particle swarm optimization continuously updates the velocity and position of particles to find a way to achieve optimal melt flow. Minimum smoothing factor .

[0087] During the iteration process or the number of iterations reaches When the time is right, the optimization process terminates, and the optimal smoothing factor is finally output. The optimal parameters are then substituted into the generalized regression neural network, and the model is trained using the preprocessed training dataset. The prediction accuracy of the model after training is then assessed. Reaching 0.991, it can meet the prediction requirements of actual processes.

[0088] 4.3 Connection between Model Application Process and Steps

[0089] In the model application phase, the measured data transmitted in step S3 of the current batch is first collected. The measured parameters are then substituted into the optimized and trained generalized regression neural network. After activation operations in the model's pattern layer, summation operations in the numerator and denominator of the summation layer, and linear mapping operations in the output layer, three sets of core prediction parameters are finally output: compatibilizer addition amount. Temperature correction values ​​for each section of gradient mixing: Temperature correction value for zone one Temperature correction value for Zone 2 Temperature correction values ​​for three zones Target melt index prediction value .

[0090] Based on the above predicted parameters, the actual process parameters for step S4 were adjusted as follows: the compatibilizer addition amount was determined to be 3.8 parts, and the temperature of zone one of the gradient mixing was adjusted to the original set temperature of 120℃ and the correction value. The sum of the values ​​is 120.5℃. The temperature in Zone 2 remains unchanged at the original setting of 130℃, and the temperature in Zone 3 is adjusted to the original setting of 140℃ plus the correction value. The superposition value is 139.7℃, the rotor speed is still set to 50r / min, and the mixing time is set to 10min.

[0091] The grafted composite resin obtained in step S3, the predicted dosage of compatibilizer, the formulated amount of antioxidant combination, and the lubricant combination were added together to a torque rheometer, and the gradient mixing process was started according to the adjusted process parameters. After mixing, the mixture was removed and its melt flow index was tested using a melt flow indexer at 190℃ and 2.16 kg. .

[0092] The measured melt flow index was compared with the predicted melt flow index of 2.32 g / 10 min by the generalized regression neural network. The calculated error was 0.01 g / 10 min, which is less than the set error threshold of 0.03 g / 10 min, indicating that the model's prediction accuracy meets the requirements and no parameter adjustment is needed. If the error is greater than 0.03 g / 10 min, the maximum inertia weight of the adaptive particle swarm optimization will be adjusted. Adjusted to 0.95, the smoothing factor for the generalized regression neural network was restarted using adaptive particle swarm optimization. We will optimize and update the forecasts to ensure accuracy in subsequent batches.

[0093] After completing step S4, the resulting mixture and the measured values ​​will be... The data is transmitted synchronously to the next step, serving as the core basis for setting dynamic vulcanization-extrusion molding process parameters.

[0094] Algorithm integration and interaction:

[0095] 1. In step S3, the improved Drosophila optimization algorithm and the kernel principal component analysis-radial basis function neural network form a bidirectional interaction: the improved Drosophila optimization algorithm optimizes the kernel principal component analysis-radial basis function neural network through a dynamic step-size formula. and Kernel principal component analysis-radial basis function neural network grafting rate prediction error back feedback adjustment improved fruit fly optimization algorithm optimization strategy;

[0096] 2. In step S4, the adaptive particle swarm optimization algorithm and the generalized regression neural network form a two-way interaction: the adaptive particle swarm optimization algorithm optimizes the generalized regression neural network through the adaptive inertia weight formula. The melt flow index prediction error of the generalized regression neural network is adjusted by back feedback in the adaptive particle swarm optimization algorithm. .

[0097] In some embodiments, the improved Drosophila optimization algorithm interacts bidirectionally with kernel principal component analysis-radial basis function neural networks:

[0098] Interaction process: 1. The improved fruit fly optimization algorithm initializes the population parameters (population size). , ), randomly generated and The initial values ​​are then substituted into the kernel principal component analysis-radial basis function neural network;

[0099] 2. Kernel Principal Component Analysis - Radial Basis Function Neural Network , As input, output the predicted grafting rate. ,calculate Compared with the measured grafting rate of :

[0100] ;

[0101] In the formula The total number of training samples;

[0102] 3. Feedback is fed into the improved fruit fly optimization algorithm, which then... Adjusting the dynamic step size factor :

[0103] ;

[0104] In the formula The initial step size, This represents the current iteration number. ;

[0105] 4. The improved fruit fly optimization algorithm is based on the adjusted... renew and The value of is then input again into the kernel principal component analysis-radial basis function neural network, and the above error calculation and parameter update process is repeated.

[0106] Interaction termination condition:

[0107] The interaction between the improved Drosophila optimization algorithm and the kernel principal component analysis-radial basis function neural network terminates when any of the following conditions are met:

[0108] ① ;

[0109] ② Number of iterations .

[0110] In some embodiments, there is a bidirectional interaction between adaptive particle swarm optimization and generalized regression neural networks:

[0111] Adaptive particle swarm optimization targets the core parameters (smoothing factor) of generalized regressive neural networks. Adaptive optimization is performed, and the melt index prediction error of the generalized regression neural network is fed back to the adaptive particle swarm optimization to adjust its adaptive inertia weights, forming a closed-loop interactive system.

[0112] Interaction process:

[0113] 1. Adaptive Particle Swarm Optimization: Initialize Particle Swarm Parameters (Number of Particles) , ), randomly generated The initial values ​​are then substituted into the generalized regression neural network;

[0114] 2. Generalized Regression Neural Networks Input, Output ,calculate Compared with the measured melt index of :

[0115] ;

[0116] In the formula The total number of training samples;

[0117] 3. Feedback is fed back to adaptive particle swarm optimization, which is based on... Adjustment :

[0118] ;

[0119] In the formula , , This represents the current iteration number. ;

[0120] 4. The adaptive particle swarm optimization algorithm is based on the adjusted renew The value of is then input into the generalized regression neural network again, and the above error calculation and parameter update process is repeated.

[0121] Interaction termination condition:

[0122] The interaction between the adaptive particle swarm optimization algorithm and the generalized regressive neural network terminates when any of the following conditions are met:

[0123] ① ;

[0124] ② Number of iterations .

[0125] In some embodiments, the cross-step interaction between kernel principal component analysis-radial basis function neural network and generalized regression neural network:

[0126] The grafting rate prediction accuracy of the kernel principal component analysis-radial basis function neural network is fed back to the generalized regression neural network to correct the input weights of the generalized regression neural network; the melt index prediction accuracy of the generalized regression neural network is fed back to the kernel principal component analysis-radial basis function neural network to optimize its output layer mapping coefficients, thus realizing cross-step algorithmic collaborative interaction.

[0127] Interaction process:

[0128] 1. After training, the kernel principal component analysis-radial basis function neural network is used to calculate its grafting rate prediction accuracy. :

[0129] ;

[0130] In the formula The average of the measured grafting rates; As a correction coefficient for the prediction accuracy of kernel principal component analysis-radial basis function neural network ( Input the generalized regression neural network and adjust the connection weights between the input layer and the pattern layer of the generalized regression neural network. :

[0131] ;

[0132] In the formula The corrected weights;

[0133] 2. After the generalized regression neural network has been trained, its melt flow index prediction accuracy is calculated. :

[0134] ;

[0135] In the formula The average of the measured melt flow index; As a correction coefficient for the prediction accuracy of generalized regression neural networks ( Input kernel principal component analysis-radial basis function neural network and optimize its output layer mapping coefficients. :

[0136] ;

[0137] In the formula To optimize the mapping coefficients;

[0138] 3. The modified kernel principal component analysis-radial basis function neural network and generalized regression neural network re-predict parameters, repeating the above accuracy feedback and parameter correction process.

[0139] Interaction termination condition:

[0140] The interaction between the kernel principal component analysis-radial basis function neural network and the generalized regression neural network terminates when any of the following conditions are met:

[0141] ① and ;

[0142] ②The number of precision correction iterations reaches the preset value, which is 10 times.

[0143] In some embodiments, the improved fruit fly optimization algorithm and the adaptive particle swarm optimization algorithm interact collaboratively:

[0144] The optimization convergence speed of the improved fruit fly optimization algorithm is fed back to the adaptive particle swarm optimization algorithm to adjust the particle update speed of the adaptive particle swarm optimization algorithm; the optimization accuracy of the adaptive particle swarm optimization algorithm is fed back to the improved fruit fly optimization algorithm to adjust the population size of the improved fruit fly optimization algorithm, thereby realizing the collaborative optimization interaction of the two optimization algorithms.

[0145] Interaction process:

[0146] 1. After optimizing the parameters of the kernel principal component analysis-radial basis function neural network, the improved Drosophila optimization algorithm calculates its convergence speed. :

[0147] ;

[0148] In the formula This represents the actual number of convergence iterations for the improved fruit fly optimization algorithm. ;Will As an adjustment factor input to the adaptive particle swarm optimization algorithm, the learning factor in the particle velocity update formula of the adaptive particle swarm optimization algorithm is corrected. :

[0149] ;

[0150] In the formula For the corrected learning factor, ;

[0151] 2. After the adaptive particle swarm optimization algorithm completes the optimization of the parameters of the generalized regression neural network, its optimization accuracy is calculated. That is, the final fitness value The reciprocal:

[0152] ;

[0153] In the formula The final fitness value of the adaptive particle swarm optimization algorithm; As an adjustment factor, the population size of the improved Drosophila optimization algorithm is adjusted. :

[0154] ;

[0155] In the formula To correct the population size, For the original population size, It is a rounding function;

[0156] 3. The modified improved fruit fly optimization algorithm and adaptive particle swarm optimization algorithm re-optimize the parameters of the kernel principal component analysis-radial basis function neural network and the generalized regression neural network, respectively, and repeat the above feedback and correction process.

[0157] Interaction termination condition:

[0158] The interaction between the improved fruit fly optimization algorithm and the adaptive particle swarm optimization algorithm terminates when any of the following conditions are met:

[0159] ① That is, the improved fruit fly optimization algorithm has a convergence speed of ≥60%, and That is, the adaptive particle swarm optimization algorithm has an optimization accuracy of ≥100;

[0160] ② The number of collaborative optimization iterations reaches the preset value, which is 5 times.

[0161] In some embodiments, the adaptive particle swarm optimization algorithm interacts with the kernel principal component analysis-radial basis function neural network across algorithms:

[0162] The adaptive particle swarm optimization algorithm can assist in optimizing the dimensionality reduction of the kernel principal component analysis layer of the kernel principal component analysis-radial basis function neural network. The grafting rate prediction error of the kernel principal component analysis-radial basis function neural network is fed back to the adaptive particle swarm optimization algorithm to correct its particle position update strategy, thereby realizing the collaborative interaction between the optimization algorithm and the cross-step neural network.

[0163] Interaction process:

[0164] 1. Initialize the adaptive particle swarm optimization algorithm parameters (number of particles) , ), reducing the dimensionality of the kernel principal component analysis layer (Initial value range 3-8) is used as the optimization variable in the adaptive particle swarm optimization algorithm;

[0165] 2. Adaptive Particle Swarm Optimization Algorithm (Random Generation) Substituting the initial values ​​into the kernel principal component analysis-radial basis function neural network, the absolute error of the grafting rate prediction of the kernel principal component analysis-radial basis function neural network is calculated. :

[0166] ;

[0167] In the formula For the total number of training samples, As the fitness value of the adaptive particle swarm optimization algorithm, update :

[0168] ;

[0169] Let be the velocity of the particle in the (t+1)th iteration. Let be the position of the particle in the t-th iteration. This represents the position of the particle in the (t+1)th iteration.

[0170] in The calculation formula is:

[0171] ;

[0172] In the formula Let be the velocity of the particle in the t-th iteration. , As a global learning factor, , A random number between 0 and 1. This represents the optimal position for an individual particle. The optimal position for the entire population;

[0173] 3. The absolute error of the grafting rate prediction output by the kernel principal component analysis-radial basis function neural network. Feedback is fed into the adaptive particle swarm optimization algorithm to adjust its inertia weights. :

[0174] ;

[0175] In the formula To correct the inertia weight, The preset maximum error threshold is set to 0.05.

[0176] 4. Adaptive Particle Swarm Optimization Algorithm Based on The particle positions were updated again, and the dimensionality reduction of the kernel principal component analysis layer was iteratively optimized. Continue until the termination condition is met.

[0177] Interaction termination condition:

[0178] The interaction between the adaptive particle swarm optimization algorithm and the kernel principal component analysis-radial basis function neural network terminates when any of the following conditions are met:

[0179] ① Absolute error in grafting rate prediction of kernel principal component analysis-radial basis function neural network ;

[0180] ② The maximum number of iterations of the adaptive particle swarm optimization algorithm when interacting with kernel principal component analysis-radial basis function neural network. .

[0181] In some embodiments, the improved fruit fly optimization algorithm interacts with the generalized regressive neural network across algorithms:

[0182] The improved fruit fly optimization algorithm can optimize the weight allocation coefficient of the summation layer of the generalized regression neural network. The melt index prediction error of the generalized regression neural network is fed back to the improved fruit fly optimization algorithm to adjust its dynamic step size factor, realizing the bidirectional collaborative interaction between the two cross-step algorithms.

[0183] Interaction process:

[0184] 1. Initialize the parameters of the improved fruit fly optimization algorithm (population size) , ), the molecular weight coefficients of the summation layer of the generalized regression neural network Denominator weighting coefficient (Initial value range 0 to 1) is used as the optimization variable in the improved fruit fly optimization algorithm;

[0185] 2. Improved Fruit Fly Optimization Algorithm for Random Generation , The initial values ​​are substituted into the generalized regression neural network, and the mean square error of the melt flow index prediction of the generalized regression neural network is calculated. :

[0186] ;

[0187] In the formula For the total number of training samples, The fitness value is used to update the position of individual fruit flies in the improved fruit fly optimization algorithm. , :

[0188] ;

[0189] ;

[0190] In the formula This refers to the x-coordinate component of the position of the i-th fruit fly individual in the t-th iteration of the improved fruit fly optimization algorithm; The ordinate component of the position of the i-th fruit fly individual in the t-th iteration in the improved fruit fly optimization algorithm; At the t-th iteration, the x-coordinate corresponding to the current global optimal position of the population is the x-coordinate component of the optimal position of all fruit fly individuals in the population during this iteration. At the t-th iteration, the ordinate corresponding to the current global optimal position of the population is the ordinate component of the optimal position of all fruit fly individuals in the population during this iteration. The dynamic step size factor for the improved fruit fly optimization algorithm. A random number between -1 and 1;

[0191] 3. The mean squared error of the melt flow index prediction output by the generalized regression neural network. Feedback is fed into the improved fruit fly optimization algorithm to correct its dynamic step size factor. :

[0192] ;

[0193] In the formula This is the corrected dynamic step size factor. The preset maximum mean square error threshold for the generalized regression neural network is set to 0.02.

[0194] 4. The improved fruit fly optimization algorithm is based on the modified... The fruit fly population positions were updated again, and the generalized regression neural network was iteratively optimized. , Continue until the termination condition is met.

[0195] Interaction termination condition:

[0196] The interaction between the improved fruit fly optimization algorithm and the generalized regressive neural network terminates when any of the following conditions are met:

[0197] ① Mean squared error of melt flow index prediction using generalized regression neural networks ;

[0198] ② The maximum number of iterations of the improved fruit fly optimization algorithm when interacting with the generalized regressive neural network. .

[0199] S5: Dynamic vulcanization-extrusion molding

[0200] The technical objective of this step is to prepare an antistatic resin composition with uniform properties that meets the requirements of wire and cable processing by dynamically vulcanizing and extruding the mixture obtained in step S4, thereby ensuring that the mechanical and antistatic properties of the product meet the standards.

[0201] Specific implementation details: First, the mixture obtained in step S4 is pretreated by placing it in a vacuum drying oven at a temperature of 80℃ and a vacuum of -0.08MPa for 2 hours to remove moisture adsorbed during storage and transfer, thus preventing air bubble defects during extrusion. After drying, the mixture is added to the hopper of a twin-screw extruder (model: Coperion STS 35). The hopper is equipped with a stirring device at a stirring rate of 50 r / min to ensure uniform feeding of the mixture.

[0202] The mixture obtained according to step S4 Set the temperature and screw speed of each section of the twin-screw extruder: If The extruder temperature is set to 2.0–2.2 g / 10 min, with zone 1 temperature set at 135℃, zone 2 at 145℃, zone 3 at 155℃, and the die head at 150℃. The screw speed is set to 80 r / min. The extruder temperature is set at 2.2–2.5 g / 10 min, with zone 1 at 130℃, zone 2 at 140℃, zone 3 at 150℃, and the die head at 145℃. The screw speed is set at 100 r / min. The temperature of each section of the extruder is controlled by a PID temperature control system, with a temperature fluctuation range of ±1℃ to ensure temperature control accuracy.

[0203] After the extruder has stabilized (preheat for 30 minutes to ensure each section reaches its set temperature), dicumyl peroxide vulcanizing agent is injected into the middle section of the twin-screw extruder, between zones two and three, via a metering pump. The amount of vulcanizing agent added is 0.2% of the mixture's mass. The metering pump flow rate is dynamically adjusted according to the extruder's feed rate to ensure precise and controllable ratio of vulcanizing agent to the mixture. The vulcanizing agent is a commercially available industrial product with 98% purity, and has been vacuum-dried at 60°C for 2 hours to remove moisture before use.

[0204] The compound undergoes a series of processes including melting, shearing, mixing, and dynamic vulcanization within a twin-screw extruder. The vulcanization reaction time is controlled by the screw speed and barrel length to ensure a degree of vulcanization of over 85%, which is confirmed by differential scanning calorimetry (DSC). The vulcanized molten material is then extruded through a circular die with a 3mm orifice. The extruded strip immediately enters a water-cooling tank for cooling. The water temperature in the cooling tank is controlled at 20–25°C, and the cooling length is 3 meters, ensuring rapid cooling and shaping of the strip to prevent sticking during subsequent pelletizing.

[0205] After cooling, the material strip is drawn to the pelletizer by a traction machine. The traction speed is matched with the extrusion speed, and the traction speed is 5% higher than the extrusion speed to ensure that the material strip is taut and not loose. The pelletizer speed is adjusted according to the material strip diameter and the target particle size, and the particle length is set to 3-4 mm. After pelletizing, the particles are placed in a vibrating screen (20 mesh) to remove a small number of unqualified particles that are too long or too short. The screened particles are placed in a hot air drying oven, and the drying temperature is set to 80℃ for 2 hours to remove the moisture on the particle surface. Finally, an antistatic resin composition for wires and cables is obtained. This composition is sealed and packaged, and stored in a dry and ventilated environment for later use. The storage temperature is ≤30℃ and the relative humidity is ≤60%.

[0206] Technical effectiveness verification:

[0207] The prepared antistatic resin composition was subjected to core performance tests to verify whether it meets the requirements for use in wires and cables.

[0208] 1. Volume resistivity test: A high-resistivity meter (model: Keithley 6517B) was used to test the volume resistivity and surface resistivity of solid insulating materials according to GB / T1410-2006 standard. The test temperature was 25℃ and the humidity was 50%. Three test pieces (2mm thick, 50mm in diameter) were prepared for each sample, and the test was performed in triplicate. The average value was taken. The test results showed that the volume resistivity reached... Ω·cm, meeting antistatic requirements;

[0209] 2. Thermo-oxidative aging test: The sample was placed in a thermo-oxidative aging chamber, the temperature was set to 100℃, the oxygen flow rate was 20mL / min, and the aging time was 1000h. After aging, the change rate of volume resistivity was tested. The results showed that the change rate was ≤10%, indicating excellent aging resistance.

[0210] 3. Abrasion resistance test: The Taber abrasion tester (model: Taber 5135) was used with a load of 500g, a CS-17 grinding wheel, and a rotation speed of 1000 rpm to test the wear amount of the sample. The results showed that the wear amount was ≤0.05g, which meets the abrasion resistance requirements of wires and cables.

[0211] Formulation and component sources of the antistatic resin composition:

[0212] This antistatic resin composition for wires and cables consists of the following components in parts by weight, which work synergistically to achieve excellent antistatic properties, aging resistance, and processing stability:

[0213] 1. Low-density polyethylene-ethylene-methyl acrylate terpolymer: 65-75 parts;

[0214] 2. Quaternary ammonium salt type ionic liquid-graphene quantum dot composite: 8-12 parts;

[0215] 3. Ethylene-vinyl acetate grafted maleic anhydride compatibilizer: 3-5 parts;

[0216] 4. Antioxidant combination: 0.5 to 1 part, composed of antioxidant 1010 and antioxidant 168 in a mass ratio of 1:1;

[0217] 5. Lubricant combination: 1-2 parts, composed of calcium stearate and ethylene bis-stearamide in a mass ratio of 2:1.

[0218] Component source description: Low-density polyethylene-ethylene-methyl acrylate terpolymer is a commercially available industrial product, model DFDA-7042, with a melt flow rate of 2.1 g / 10 min (test conditions 190℃, 2.16 kg), ensuring processing fluidity suitable for subsequent extrusion processes; Quaternary ammonium salt ionic liquid-graphene quantum dot composite is a self-made product of S1, which needs to meet the particle size requirement of 10-20 nm; Ethylene-vinyl acetate grafted maleic anhydride compatibilizer is a commercially available industrial product, with a fixed grafting rate of 1.5% to ensure compatibility with the main resin and composite; Antioxidant 1010, Antioxidant 168, calcium stearate, and ethylene bis-stearamide are all commercially available analytical grade reagents with a purity ≥99%, avoiding impurities affecting the performance of the composition.

[0219] In one specific embodiment, the formulation (by parts by weight) of the antistatic resin composition for wires and cables is as follows:

[0220] Low-density polyethylene-ethylene-methyl acrylate terpolymer: 70 parts (commercially available, model DFDA-7042, melt flow rate 2.1 g / 10 min, test conditions 190℃, 2.16 kg).

[0221] Quaternary ammonium salt ionic liquid-graphene quantum dot composite: 10 parts (S1 self-made);

[0222] Ethylene-vinyl acetate grafted maleic anhydride compatibilizer: 4 parts (commercially available, grafting rate 1.5%).

[0223] Antioxidant combination: 0.8 parts (antioxidant 1010 and antioxidant 168 are compounded in a mass ratio of 1:1, both are commercially available analytical grade, with a purity of ≥99%).

[0224] Lubricant combination: 1.5 parts (calcium stearate and ethylene bis-stearamide are compounded in a mass ratio of 2:1, both of which are commercially available analytical grade with a purity of ≥99%).

[0225] Preparation process:

[0226] S1:

[0227] 1. Solution preparation and graphene dispersion: A 500mL three-necked flask was used, equipped with a mechanical stirrer, thermometer, and reflux condenser. 10 parts citric acid and 30 parts deionized water were added to the flask. The mechanical stirrer was turned on (speed 300 r / min), and the temperature was raised to 95℃ and maintained for 30 min until the citric acid was completely dissolved to form a homogeneous aqueous solution. Then, 2 parts graphene powder (purity 99.6%, sheet thickness 1–3 nm) were added. The mechanical stirrer was turned off, and the mixture was transferred to an ultrasonic disperser. The ultrasonic power was set to 300W and the ultrasonic frequency to 40kHz, and continuous ultrasonication was performed for 60 min. During this time, the system temperature was controlled to ≤50℃ using an ice-water bath to obtain a homogeneous graphene dispersion.

[0228] 2. Preparation of graphene quantum dots by hydrothermal reaction: The graphene dispersion was transferred to a 200mL polytetrafluoroethylene-lined hydrothermal reactor, sealed, and placed in an oven. The temperature was set at 180℃ and the reaction time at 12h. After the reaction, the mixture was allowed to cool naturally to room temperature (approximately 4h). The product was then removed and filtered through a 0.22μm organic filter membrane. The filter cake was washed three times with deionized water (50 parts water each time) until the pH of the washing solution reached 6.8 (measured with a precision pH meter). The filter cake was then placed in a vacuum drying oven at 60℃ and a vacuum of -0.09MPa for 12h. After drying, the product was removed, ground, and passed through a 200-mesh sieve to obtain graphene quantum dots.

[0229] 3. Preparation of the target composite by ionization reaction: 5 parts of the above-mentioned graphene quantum dots, 20 parts of N-methylimidazolium (99% purity), and 30 parts of n-bromobutane (99% purity) were added to a 250 mL round-bottom flask. High-purity nitrogen (99.99% purity) was introduced for protection at a flow rate of 50 mL / min for 10 min to purge air. The flask was placed in an oil bath and heated to 80 °C. Magnetic stirring was started (500 r / min), and the mixture was kept at this temperature for 24 h. After the reaction was completed, the mixture was cooled to room temperature. 100 parts of diethyl ether were added and stirred for 30 min. The mixture was then allowed to stand and separate into layers. The upper ether phase was discarded, and the mixture was washed three times. The lower product was placed in a vacuum drying oven and dried at 40 °C and a vacuum of -0.09 MPa for 8 h to obtain the quaternary ammonium salt-type ionic liquid-graphene quantum dot composite.

[0230] 4. Particle size testing and adjustment: The particle size of the composite was tested using a Malvern 3000 laser particle size analyzer. The test medium was deionized water, the dispersion ultrasonic power was 100W, and the time was 2 minutes. The test was performed in triplicate, and the average value was taken to obtain the composite particle size data. (Within the preset range of 10-20nm), seal and store for later use.

[0231] S2:

[0232] 1. Resin pretreatment: Weigh 70 parts of low-density polyethylene-ethylene-methyl acrylate terpolymer, put it into a vacuum drying oven, set the temperature to 60℃ and the vacuum degree to -0.08MPa, and dry for 4 hours to remove surface moisture and impurities.

[0233] 2. Plasma Activation: The dried resin was evenly spread on the sample tray of the PTL-D-100 plasma processor (spread thickness 2.5 mm). After closing the chamber, a vacuum was drawn to 5 Pa and maintained for 10 min. Then, argon gas (flow rate 20 sccm) was introduced to stabilize the pressure in the chamber at 20 Pa. The processing power was set to 150 W and the time to 15 min. The plasma generator was started for activation treatment, and the pressure and power were monitored in real time during the process.

[0234] 3. Activation Energy Test and Adjustment: After treatment, the contact angle between the resin surface and deionized water was measured using a KRÜSS DSA100 contact angle meter (temperature 25℃, humidity 50%, average value of 5 test points) to obtain the contact angle. According to the Young-Laplace equation calculate ,in The surface tension of deionized water at 25℃ is 72.8 mN / m. The interfacial tension between the resin surface and air (i.e. ), The interfacial tension between the resin surface and water was used to derive the following: (Within the preset range of 35-40mN / m), proceed to the next step.

[0235] S3:

[0236] 1. Algorithm model invocation: The algorithm model obtained from S1 is invoked. And S2 obtained Substitute the kernel principal component analysis-radial basis function neural network model optimized by the improved fruit fly optimization algorithm (optimized parameters: , Before inputting the model, first... and Perform min-max normalization; the normalization formula is as follows: ,in corresponding , , normalized values ; corresponding , , normalized values .

[0237] 2. Parameter Prediction and Process Implementation: The optimal stirring rate is output through dimensionality reduction using kernel principal component analysis (mapping to 3D eigenvectors) and prediction using radial basis function neural networks. Optimal composite temperature The activated resin and 10 parts of the composite were added to a high-speed mixer and stirred at the specified parameters for 30 minutes to complete the targeted grafting composite and obtain the grafted composite resin.

[0238] 3. Grafting rate verification: Testing the grafted composite resin The error is 0.2% compared to the model's predicted value (90.8%), which is ≤0.5% of the preset threshold. No parameter adjustment is required. Transmit to step S4.

[0239] S4:

[0240] 1. Algorithm model invocation: The algorithm model obtained from S3 is invoked. Substitute into the generalized regression neural network model optimized by the adaptive particle swarm optimization algorithm (optimal smoothing factor) Before inputting... Perform min-max normalization, the formula is as follows ,in , , normalized values .

[0241] 2. Parameter Prediction and Process Adjustment: The optimal compatibilizer addition amount is predicted and output using a generalized regression neural network. Temperature correction values ​​for batches and gradient mixing: Zone 1 temperature correction value Temperature correction value for Zone 2 Temperature correction values ​​for three zones Target melt flow index prediction value Based on this, the process parameters were adjusted as follows: Zone 1 temperature 120.4℃, Zone 2 temperature 130℃, Zone 3 temperature 139.8℃, rotor speed 50 r / min, and mixing time 10 min.

[0242] 3. Mixing Implementation and Melt Flow Index Verification: The grafted composite resin, 4 parts compatibilizer, 0.8 parts antioxidant combination, and 1.5 parts lubricant combination were added to a torque rheometer and mixed in a gradient manner according to the adjusted parameters. After mixing, the melt flow index was tested using a melt flow indexer (190℃, 2.16kg) to obtain the desired result. The error from the predicted value is 0.01 g / 10 min, which is ≤ the preset threshold of 0.03 g / 10 min. No parameter adjustment is required. The mixture and... Transmit to step S5.

[0243] S5:

[0244] 1. Pretreatment of the mixture: The mixture obtained in S4 was placed in a vacuum drying oven and dried at 80°C and a vacuum of -0.08MPa for 2 hours to remove adsorbed moisture.

[0245] 2. Extrusion parameter settings: Because (Within the range of 2.2-2.5g / 10min), set the parameters of the twin-screw extruder (Coperon STS35): Zone 1 130℃, Zone 2 140℃, Zone 3 150℃, Die head 145℃, screw speed 100r / min.

[0246] 3. Vulcanization and Molding: The mixture is added to the extruder hopper (hopper stirring speed 50 r / min). After preheating for 30 minutes, the vulcanizing agent dicumyl peroxide (addition amount is 0.2% of the mixture mass, purity 98%, vacuum dried at 60℃ for 2 hours) is injected into the middle section of the extruder through a metering pump. The flow rate of the metering pump is dynamically adjusted according to the feed rate. The molten material is extruded through a 3mm diameter circular die and cooled in a 22℃ water-cooling tank (cooling length 3m). After cooling, it is pulled to a pelletizer by a traction machine (traction speed is 5% higher than extrusion speed). The pellets are cut into 3-4mm length particles, screened by a 20-mesh vibrating screen, and dried with hot air at 80℃ for 2 hours to obtain the finished antistatic resin composition. It is then sealed and packaged in a dry and ventilated environment (temperature ≤30℃, relative humidity ≤60%) for later use.

[0247] Product performance verification: Verify the antistatic properties, aging resistance, and abrasion resistance of the finished product.

[0248] 1. Antistatic performance: The volume resistivity was tested according to GB / T1410-2006 standard for solid insulating materials, with test conditions of 25℃ and 50% humidity. The antistatic effect was stable and durable, with no significant attenuation.

[0249] 2. Aging resistance: After a thermo-oxidative aging test at 100℃ and an oxygen flow rate of 20mL / min, the appearance showed no cracking or discoloration, and the mechanical properties remained good.

[0250] 3. Abrasion resistance: Tested using a Taber abrasion tester (500g load, CS-17 grinding wheel, 1000 revolutions), the abrasion resistance is low, meeting the requirements for use in wires and cables.

[0251] The embodiments described above are for illustrative purposes only and are not intended to limit the invention. Therefore, any changes in numerical values ​​or substitutions of equivalent elements should still fall within the scope of this invention.

[0252] The above detailed description will enable those skilled in the art to understand that the present invention can indeed achieve the aforementioned objectives and has complied with the provisions of the Patent Law.

[0253] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention. The above descriptions are merely preferred embodiments of the invention and are not intended to limit the invention. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the invention should be included within the scope of protection of the invention.

[0254] It should be noted that the above description of the process is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can make various modifications and changes to the process under the guidance of this specification. However, these modifications and changes remain within the scope of this specification.

[0255] The basic concepts have been described above. Obviously, for those skilled in the art who have read this application, the above disclosure is merely illustrative and does not constitute a limitation of this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this application. Such modifications, improvements, and corrections are suggested in this application, and therefore, such modifications, improvements, and corrections still fall within the spirit and scope of the exemplary embodiments of this application.

[0256] Furthermore, this application uses specific terms to describe its embodiments. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of this application. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different positions in this specification do not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of this application can be appropriately combined.

[0257] Furthermore, those skilled in the art will understand that aspects of this application can be described and illustrated through several patentable types or situations, including any new and useful combination of processes, machines, products, or substances, or any new and useful improvements thereof. Therefore, aspects of this application can be implemented entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. All of the above hardware or software can be referred to as a “unit,” “module,” or “system.” Furthermore, aspects of this application can take the form of a computer program product embodied in one or more computer-readable media, wherein computer-readable program code is contained therein.

[0258] Furthermore, unless expressly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or other names described in this application are not intended to limit the order of the processes and methods of this application. Although some currently considered useful embodiments of the invention have been discussed in the foregoing disclosure by way of various examples, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments; rather, the claims are intended to cover all modifications and equivalent combinations that conform to the substance and scope of the embodiments of this application. For example, although the implementation of the various components described above can be embodied in a hardware device, it can also be implemented as a purely software solution, such as an installation on an existing server or mobile device.

[0259] Similarly, it should be noted that, in order to simplify the description of the present application and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of the embodiments of the present application sometimes combines multiple features into a single embodiment, drawing, or description thereof. However, this approach of the present application should not be construed as reflecting an intention that the claimed subject matter requires more features than expressly recited in each claim. Rather, the subject of the invention should possess fewer features than in any single embodiment described above.

Claims

1. A method for preparing an antistatic resin composition for wires and cables, characterized in that, Includes the following steps: S1: Prepare quaternary ammonium salt type ionic liquid-graphene quantum dot composite and test its particle size data; S2: Perform plasma pre-activation treatment on the main resin to obtain activated main resin, and test the surface activation energy data of the activated main resin. S3: The core parameters of the nuclear principal component analysis-radial basis function neural network are optimized using an improved fruit fly optimization algorithm. Particle size data and surface activation energy data are input into the optimized nuclear principal component analysis-radial basis function neural network to obtain the optimal stirring rate and optimal composite temperature. Targeted grafting composite of the activated host resin and the quaternary ammonium salt ionic liquid-graphene quantum dot composite is carried out at this optimal stirring rate and optimal composite temperature to obtain the grafted composite resin, while simultaneously acquiring grafting rate data. S4: The core parameters of the generalized regression neural network are optimized using an adaptive particle swarm optimization algorithm. The grafting rate data is input into the optimized generalized regression neural network to obtain the optimal compatibilizer addition amount and the temperature correction value of each stage of gradient mixing. After adjusting the process parameters according to the optimal compatibilizer addition amount and the temperature correction value of each stage of gradient mixing, the grafted composite resin is combined with gradient compatibilized mixing to obtain the mixture. At the same time, the melt index data of the mixture is obtained. S5: Set the temperature and screw speed of each section of the twin-screw extruder according to the melt index data of the compound, add the compound to the twin-screw extruder, inject the vulcanizing agent and perform dynamic vulcanization, extrusion molding, and prepare an antistatic resin composition; The improved Drosophila optimization algorithm optimizes the number of hidden layer neurons and the Gaussian kernel width of the kernel principal component analysis-radial basis function neural network (KPM-RBF) through a dynamic step size factor; the adaptive particle swarm optimization algorithm optimizes the smoothing factor of the generalized regression neural network (GRN) through adaptive inertia weights; the grafting rate prediction accuracy of the KPM-RBF neural network is fed back to the GRN to correct its input weights; the melt index prediction accuracy of the GRN is fed back to the KPM-RBF neural network to optimize its output layer mapping coefficients; the optimization convergence speed of the improved Drosophila optimization algorithm is fed back to the adaptive particle swarm optimization algorithm to adjust its particle update speed; and the optimization accuracy of the adaptive particle swarm optimization algorithm is fed back to the improved Drosophila optimization algorithm to adjust its population size. In S3, the improved fruit fly optimization algorithm and the kernel principal component analysis-radial basis function neural network form a two-way interaction. The specific interaction process is as follows: the optimal core parameters output by the improved fruit fly optimization algorithm are input into the kernel principal component analysis-radial basis function neural network, and the error between the grafting rate prediction value output by the kernel principal component analysis-radial basis function neural network and the measured grafting rate data is fed back to the improved fruit fly optimization algorithm, guiding the improved fruit fly optimization algorithm to adjust the optimization strategy. In S3, the bidirectional interaction between the improved fruit fly optimization algorithm and the kernel principal component analysis-radial basis function neural network forms a closed loop. The closing loop ends when the error between the grafting rate prediction value output by the kernel principal component analysis-radial basis function neural network and the measured grafting rate data is less than or equal to a preset threshold, or when the number of iterations of the improved fruit fly optimization algorithm reaches the preset maximum number of iterations. In S4, the adaptive particle swarm optimization and the generalized regression neural network form a two-way interaction. The specific interaction process is as follows: the optimal core parameters output by the adaptive particle swarm optimization are input into the generalized regression neural network, and the error between the melt index prediction value output by the generalized regression neural network and the measured melt index data is fed back to the adaptive particle swarm optimization, guiding the adaptive particle swarm optimization to adjust the inertial weights and optimization strategy. In S4, the bidirectional interaction between adaptive particle swarm optimization and generalized regression neural network forms a closed loop. The loop ends when the error between the predicted melt index output by the generalized regression neural network and the measured melt index data is less than or equal to a preset threshold, or when the number of iterations of adaptive particle swarm optimization reaches the preset maximum number of iterations.

2. The method for preparing the antistatic resin composition for wires and cables according to claim 1, characterized in that, In S1, if the particle size data exceeds the preset range, the hydrothermal reaction temperature or reaction time is adjusted, and the quaternary ammonium salt ionic liquid-graphene quantum dot composite is prepared again until the particle size data meets the preset range.

3. The method for preparing the antistatic resin composition for wires and cables according to claim 1, characterized in that, In S2, if the surface activation energy data of the activated main resin exceeds the preset range, the plasma treatment power or treatment time is adjusted, and the main resin is re-treated with plasma pre-activation until the surface activation energy data meets the preset range.

4. The method for preparing the antistatic resin composition for wires and cables according to claim 1, characterized in that, In S3, before optimizing the core parameters of the kernel principal component analysis-radial basis function neural network using the improved fruit fly optimization algorithm, the training dataset needs to be preprocessed. The training dataset comes from the orthogonal experiments conducted in the early stage, and the preprocessing method is to map all parameters in the training dataset to a preset interval.

5. The method for preparing the antistatic resin composition for wires and cables according to claim 1, characterized in that, In S5, the vulcanizing agent is injected into the middle section of the twin-screw extruder via a metering pump, and the flow rate of the metering pump is dynamically adjusted according to the feed rate of the twin-screw extruder.

Images