Method and system for precise mixing and proportioning of cable jacket masterbatch

Abstract

This invention relates to the field of cable technology, specifically to a method and system for precise mixing and proportioning of cable sheath masterbatch, comprising the following steps: S1, model establishment: establishing a predictive model based on historical production data; S2, data acquisition; S3, proportioning solution; S4, real-time control; S5, model update. This invention establishes a coupled predictive model across performance indicators by explicitly introducing interaction terms into the model, respectively characterizing the regulating effect of flame retardant masterbatch particle size on flame retardant efficiency and the cross-component transfer effect of flame retardant masterbatch moisture content on surface quality. This solves the problem that traditional independent modeling cannot handle the challenges of multi-performance coupling, and improves the prediction accuracy of sheath performance under fluctuating incoming material characteristics.

Description

Technical Field

[0001] This invention relates to the field of sheath material proportioning technology, and in particular to a method and system for precise mixing and proportioning of cable sheath masterbatch. Background Technology

[0002] Key performance indicators of cable sheaths, such as flame retardancy oxygen index and surface roughness, are highly dependent on the precise mass ratio between functional additives like flame retardant masterbatch and color masterbatch and the matrix resin. In actual production, due to fluctuations in upstream batch processes, the effective particle size and moisture content of the masterbatch often exhibit significant batch-to-batch differences. If a fixed ratio is used for production, these fluctuations will directly lead to drift in the performance of the finished product, or even batch-to-batch non-compliance.

[0003] In existing technologies, the common approach to dealing with fluctuations in incoming materials is to establish a predictive model and calculate the adjustment amount of the proportion based on the characteristics of the incoming materials. However, traditional modeling methods usually model the oxygen index and surface roughness independently, and the prediction function of each performance index only includes its own component parameters, ignoring the coupling effect between multiple components. For cable sheaths, changes in the proportion of flame retardant masterbatch and moisture content not only affect the flame retardant performance, but also indirectly affect the surface quality through changes in the dispersion state. This cross-component transmission effect cannot be captured by independent modeling, resulting in insufficient prediction accuracy of the model when the characteristics of incoming materials fluctuate. Summary of the Invention

[0004] This invention aims to solve at least one of the technical problems existing in the prior art. To this end, this invention proposes a method and system for precise mixing and proportioning of cable sheath masterbatch, as detailed below: On one hand, the present invention provides a method for precise mixing and proportioning of cable sheath masterbatch, comprising the following steps: S1, Model Establishment: A predictive model is established based on historical production data. The predictive model is as follows: in, Oxygen index, For surface roughness, This refers to the percentage of flame retardant masterbatch by mass. This refers to the percentage of color masterbatch by mass. For the effective particle size of flame retardant masterbatch, The moisture content of the flame retardant masterbatch, For extruder specific energy consumption, , and , , , , , , These are the regression coefficients; S2, Data Acquisition: Obtain the current production batch and and collect data in real time. ; S3, Proportioning solution: based on the current batch , and target oxygen index Target surface roughness The initial proportion of flame retardant masterbatch that minimizes performance deviation is determined using the prediction model. masterbatch ratio And produce accordingly; among them, for materials that are not yet predictable before feeding... The historical steady-state mean is used as the default value in the solution; S4, Real-time Control: During the production process, based on actual measurements Replace the default value used in step S3, and re-solve using the prediction model to obtain the corrected value. and And the feeding ratio is adjusted; S5, Model Update: Real-time comparison of the prediction model output Compared with actual measurement When the deviation between the two exceeds a threshold, the regression coefficients of the prediction model are incrementally updated based on new sample data, wherein the new sample data includes data from the period in question. and corresponding actual measurements .

[0005] As a further technical solution of the present invention, in step S1, the prediction model is obtained by partial least squares regression training. The training samples are historical batches accumulated from the same production line under steady-state conditions, and each sample contains the batch's... and corresponding actual measurements .

[0006] As a further technical solution of the present invention, in step S3, the initial solution that minimizes the performance deviation is obtained. and Specifically: Based on the proportion of flame retardant masterbatch masterbatch ratio Let the variable be to be determined, provided that the following conditions are met simultaneously: and The sum shall not exceed 50% to ensure that the proportion of base material is not less than 50%; Between 25% and 42%; Between 3% and 8%; Find the set that minimizes the weighted sum of the following two items. and : First item: Predicting the oxygen index With target oxygen index The square of the difference, multiplied by the weight ; Second item: When predicting surface roughness Greater than the target value At that time, take and The square of the difference, multiplied by the weight ;when Less than or equal to When that happens, the value of this item is zero.

[0007] in, and Substitute the prediction model from step S1 into the current batch as well as The default value is calculated. and The preset target value, and These are the preset weighting coefficients.

[0008] As a further technical solution of the present invention, in step S3, the... The default value is the number of recent identical or similar formulations under steady-state production conditions. Statistical mean.

[0009] As a further technical solution of the present invention, in step S4... and The correction is made to transition smoothly in a ramp manner.

[0010] As a further technical solution of the present invention, in step S5, the threshold is: continuous One production batch The absolute value of the prediction deviation is greater than or equal to 0.5; Or, continuous One production batch The absolute value of the prediction deviation is greater than or equal to 0.3 μm; in .

[0011] As a further technical solution of the present invention, in step S5, the incremental update adopts a recursive partial least squares algorithm with a forgetting factor.

[0012] On the other hand, the present invention also provides a precise mixing and proportioning system for cable sheath masterbatch, comprising: The model building module is used to build a predictive model based on historical production data. The predictive model is as follows: in, Oxygen index, For surface roughness, This refers to the percentage of flame retardant masterbatch by mass. This refers to the percentage of color masterbatch by mass. For the effective particle size of flame retardant masterbatch, The moisture content of the flame retardant masterbatch, For extruder specific energy consumption, , and , , , , , , These are the regression coefficients; The data acquisition module is used to obtain the current production batch's... and and collect data in real time. ; The proportioning solution module is used to calculate the proportions based on the current batch. , and target oxygen index Target surface roughness The initial proportion of flame retardant masterbatch that minimizes performance deviation is determined using the prediction model. masterbatch ratio And produce accordingly; among them, for materials that are not yet predictable before feeding... The historical steady-state mean is used as the default value in the solution; The real-time control module is used to monitor the actual production process. Replace the default values ​​used in the proportioning solution module, and re-solve using the prediction model to obtain the corrected values. and And the feeding ratio is adjusted; The model update module is used to compare the output of the prediction model in real time. Compared with actual measurement When the deviation between the two exceeds a threshold, the regression coefficients of the prediction model are incrementally updated based on new sample data, wherein the new sample data includes data from the period in question. and corresponding actual measurements .

[0013] The beneficial effects of this invention are as follows: In this invention, by Explicitly introduced in the model Interactive items, in Explicitly introduced in the model Interactive terms were used to establish a coupled prediction model across performance indicators, which characterized the regulating effect of flame retardant masterbatch particle size on flame retardant efficiency and the cross-component transfer effect of flame retardant masterbatch moisture content on surface quality. This solved the problem that traditional independent modeling could not handle the problem of multi-performance coupling and improved the prediction accuracy of sheath performance under the condition of fluctuating incoming material characteristics.

[0014] In this invention, by obtaining the current production batch and And collect data in real time. Before feeding, the initial oxygen index and target surface roughness are used to solve for the initial value that minimizes performance deviation. and During the production process, actual measurements are used. By replacing the default values, resolving the problem, and correcting the feeding ratio, precise phased control of the proportioning parameters was achieved, solving the problem caused by the feeding stage. The unknown factors caused deviations in the initial mixing ratio.

[0015] In this invention, the output of the prediction model is compared in real time. Compared with actual measurement When the deviation exceeds the threshold, it is based on the inclusion and actual measurement The new sample data incrementally updates the regression coefficients, enabling the model to adapt to slow time-varying factors such as equipment aging and raw material changes. This solves the problem of performance degradation and the need for manual maintenance during long-term model service, ensuring the continuous effectiveness of the proportioning decision throughout the entire life cycle. Detailed Implementation

[0016] In the following description, certain specific details are set forth in order to provide a thorough understanding of various embodiments. However, those skilled in the art will understand that the invention can be practiced without these details. In other instances, well-known structures have not been shown or described in detail to avoid unnecessarily obscuring the description of the embodiments. Unless the context otherwise requires, throughout the specification and appended claims, the word "comprising" should be interpreted in an open-ended, inclusive sense, i.e., as "including but not limited to".

[0017] This invention provides a method for precise mixing and proportioning of cable sheath masterbatch, comprising the following steps: S1, Model Establishment: A predictive model is established based on historical production data. The predictive model is as follows: in, Oxygen index, For surface roughness, This refers to the percentage of flame retardant masterbatch by mass. This refers to the percentage of color masterbatch by mass. For the effective particle size of flame retardant masterbatch, The moisture content of the flame retardant masterbatch, For extruder specific energy consumption, , and , , , , , , These are the regression coefficients; S2, Data Acquisition: Obtain the current production batch and and collect data in real time. ; S3, Proportioning solution: based on the current batch , and target oxygen index Target surface roughness The initial proportion of flame retardant masterbatch that minimizes performance deviation is determined using the prediction model. masterbatch ratio Based on this, materials are added for production; S4, Real-time Control: During the production process, based on actual measurements Replace the default value used in step S3, and re-solve using the prediction model to obtain the corrected value. and And the feeding ratio is adjusted; S5, Model Update: Real-time comparison of the prediction model output Compared with actual measurement When the deviation between the two exceeds a threshold, the regression coefficients of the prediction model are incrementally updated based on new sample data, wherein the new sample data includes data from the period in question. and corresponding actual measurements .

[0018] In this invention, by Explicitly introduced in the model Interactive items, in Explicitly introduced in the model Interactive terms were used to establish a coupled prediction model across performance indicators, which characterized the regulating effect of flame retardant masterbatch particle size on flame retardant efficiency and the cross-component transfer effect of flame retardant masterbatch moisture content on surface quality. This solved the problem that traditional independent modeling or simplified mechanism formulas could not handle the coupling of multiple performance indicators.

[0019] In this invention, by continuously comparing the model's predicted values ​​with the measured values ​​and updating the regression coefficients based on new sample increments when the deviation exceeds the limit, the model achieves self-adaptation to slow time-varying factors such as equipment aging and raw material changes, thus solving the problem of performance degradation and the need for manual maintenance during long-term model service.

[0020] In step S1, the prediction model described in step S1 is trained using partial least squares regression. The training samples are historical batches accumulated from the same production line under steady-state conditions, and each sample contains data from that batch. and corresponding actual measurements .

[0021] Before being used for training, the historical production data must undergo steady-state screening to remove data from the start-up and shutdown phases, material change transition phases, and periods of abnormal production fluctuations. Only samples corresponding to the stable operation of key parameters such as extruder temperature, linear speed, and output within the preset normal range are retained to ensure that the training data can truly reflect the mapping relationship between variables under normal operating conditions.

[0022] In step S3, the initial solution that minimizes the performance deviation is obtained. and Specifically: Based on the proportion of flame retardant masterbatch masterbatch ratio Let the variable be to be determined, provided that the following conditions are met simultaneously: and The sum shall not exceed 50% to ensure that the proportion of base material is not less than 50%; Between 25% and 42%; Between 3% and 8%; Find the set that minimizes the weighted sum of the following two items. and : First item: Predicting the oxygen index With target oxygen index The square of the difference, multiplied by the weight ; Second item: When predicting surface roughness Greater than the target value At that time, take and The square of the difference, multiplied by the weight ;when Less than or equal to When that happens, the value of this item is zero.

[0023] in, and Substitute the prediction model from step S1 into the current batch as well as The default value is calculated. and The preset target value, and These are the preset weighting coefficients.

[0024] The above solution process essentially transforms the target sheath performance into a reverse constraint on the proportioning parameters. Within the process-feasible proportioning space, a set of proportions is found through weighted optimization. and This results in the prediction model output being... and The overall deviation from the target value is minimized; where, when one of the two has already met the target requirements, the optimization target only focuses on the deviation of the other, avoiding over-optimization of the already met indicators.

[0025] In step S3, for materials that are not yet measurable before feeding... The historical steady-state mean is used as the default value in the solution. The default value is the number of recent identical or similar formulations under steady-state production conditions. Statistical mean.

[0026] The default value is taken from the same or similar formulations under steady-state production conditions recently. The statistical mean is based on the historical values ​​of this formulation under the same extruder, similar output, and linear speed conditions. The arithmetic mean obtained from the data calculation uses recent data rather than all historical data in order to reflect the specific energy consumption level of the equipment under the current condition and avoid mismatch between early data and current actual operating conditions due to long-term wear and tear of the equipment.

[0027] In step S4, for and The correction is made to transition smoothly in a ramp manner.

[0028] Smooth transition via ramp refers to the process where, after correction... or When there is a difference between the current actual value and the current value, the setting does not jump directly to the new setting value. Instead, the feed screw speed is controlled to gradually adjust at a preset rate of change until the new target value is reached. This avoids the impact of the step change in the ratio on the uniformity of the melt in the extruder and the surface quality of the sheath. When the difference between the correction amount and the current value does not exceed the preset dead zone, the adjustment is not triggered.

[0029] In step S5, the threshold is: continuous One production batch The absolute value of the prediction deviation is greater than or equal to 0.5; Or, continuous One production batch The absolute value of the prediction deviation is greater than or equal to 0.3 μm; in .

[0030] Among the above thresholds, Deviation limit value 0.5 and The deviation limit of 0.3μm is determined based on the performance acceptance standards and process experience of the sheath product; A deviation exceeding 0.5 indicates a potential substantial shift in the flame retardant rating. A deviation exceeding 0.3 μm means that the surface quality may have deteriorated to the point of being visible to the naked eye. The requirement that an update is only triggered after N consecutive batches exceed the limit is to distinguish between systematic model degradation and occasional quality fluctuations, and to avoid unnecessary model updates caused by individual abnormal samples.

[0031] In step S5, the incremental update employs a recursive partial least squares algorithm with a forgetting factor.

[0032] The recursive partial least squares algorithm with a forgetting factor assigns higher weights to new samples and lower weights to old samples when updating model coefficients. This allows the model to quickly adapt to recent changes in process conditions and raw material characteristics while preserving stable patterns in historical data. This incremental update method only needs to store the current model coefficients and new sample data, without having to access the entire historical database each time, thus meeting the requirements of real-time calculation and long-term operation in industrial settings.

[0033] The precise mixing and proportioning system for cable sheath masterbatch provided by the present invention is described below. The precise mixing and proportioning system for cable sheath masterbatch described below can be referred to in correspondence with the precise mixing and proportioning method for cable sheath masterbatch described above.

[0034] A cable sheath masterbatch precision mixing and proportioning system, comprising: The model building module is used to build a predictive model based on historical production data. The predictive model is as follows: in, Oxygen index, For surface roughness, This refers to the percentage of flame retardant masterbatch by mass. This refers to the percentage of color masterbatch by mass. For the effective particle size of flame retardant masterbatch, The moisture content of the flame retardant masterbatch, For extruder specific energy consumption, , and , , , , , , These are the regression coefficients; The data acquisition module is used to obtain the current production batch's... and and collect data in real time. ; The proportioning solution module is used to calculate the proportions based on the current batch. , and target oxygen index Target surface roughness The initial proportion of flame retardant masterbatch that minimizes performance deviation is determined using the prediction model. masterbatch ratio And produce accordingly; among them, for materials that are not yet predictable before feeding... The historical steady-state mean is used as the default value in the solution; The real-time control module is used to monitor the actual production process. Replace the default values ​​used in the proportioning solution module, and re-solve using the prediction model to obtain the corrected values. and And the feeding ratio is adjusted; The model update module is used to compare the output of the prediction model in real time. Compared with actual measurement When the deviation between the two exceeds a threshold, the regression coefficients of the prediction model are incrementally updated based on new sample data, wherein the new sample data includes data from the period in question. and corresponding actual measurements .

[0035] The functions of each module described above correspond one-to-one with the execution process of each step in the aforementioned method embodiment, and will not be repeated here.

[0036] The present invention also provides an electronic device, which includes a processor, a communication interface, a memory, and a communication bus. The processor calls program instructions stored in the memory via the communication bus to execute the precise mixing and proportioning method for cable sheath masterbatch described in any of the foregoing embodiments.

[0037] The present invention also provides a computer program product comprising a computer program that, when executed by a processor, implements all the steps of the aforementioned method for precise mixing and proportioning of cable sheath masterbatch.

[0038] The present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements all the steps of the aforementioned method for precise mixing and proportioning of cable sheath masterbatch.

[0039] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it.

Claims

1. A method for precise mixing and proportioning of cable sheath masterbatch, characterized in that, The steps include the following: S1, Model Establishment: A predictive model is established based on historical production data. The predictive model is as follows: in, Oxygen index, For surface roughness, This refers to the percentage of flame retardant masterbatch by mass. This refers to the percentage of color masterbatch by mass. For the effective particle size of flame retardant masterbatch, The moisture content of the flame retardant masterbatch, For extruder specific energy consumption, , and , , , , , , These are the regression coefficients; S2, Data Acquisition: Obtain the current production batch and and collect data in real time. ; S3, Proportioning solution: based on the current batch , and target oxygen index Target surface roughness The initial proportion of flame retardant masterbatch that minimizes performance deviation is determined using the prediction model. masterbatch ratio And produce accordingly; among them, for materials that are not yet predictable before feeding... The historical steady-state mean is used as the default value in the solution; S4, Real-time Control: During the production process, based on actual measurements Replace the default value used in step S3, and re-solve using the prediction model to obtain the corrected value. and And the feeding ratio is adjusted; S5, Model Update: Real-time comparison of the prediction model output Compared with actual measurement When the deviation between the two exceeds a threshold, the regression coefficients of the prediction model are incrementally updated based on new sample data, wherein the new sample data includes data from the period in question. and corresponding actual measurements .

2. The method for precise mixing and proportioning of cable sheath masterbatch according to claim 1, characterized in that, In step S1, the prediction model described in step S1 is trained using partial least squares regression. The training samples are historical batches accumulated from the same production line under steady-state conditions, and each sample contains data from that batch. and corresponding actual measurements .

3. The method for precise mixing and proportioning of cable sheath masterbatch according to claim 1, characterized in that, In step S3, the initial solution that minimizes the performance deviation is obtained. and Specifically: Based on the proportion of flame retardant masterbatch masterbatch ratio Let the variable be to be determined, provided that the following conditions are met simultaneously: and The sum shall not exceed 50% to ensure that the proportion of base material is not less than 50%; Between 25% and 42%; Between 3% and 8%; Find the set that minimizes the weighted sum of the following two items. and : First item: Predicting the oxygen index With target oxygen index The square of the difference, multiplied by the weight ; Second item: When predicting surface roughness Greater than the target value At that time, take and The square of the difference, multiplied by the weight ;when Less than or equal to When this condition is met, the term is zero. in, and Substitute the prediction model from step S1 into the current batch as well as The default value is calculated. and The preset target value, and These are the preset weighting coefficients.

4. The method for precise mixing and proportioning of cable sheath masterbatch according to claim 1, characterized in that, In step S3, the The default value is the number of recent identical or similar formulations under steady-state production conditions. Statistical mean.

5. The method for precise mixing and proportioning of cable sheath masterbatch according to claim 1, characterized in that, In step S4, for and The correction is made to transition smoothly in a ramp manner.

6. The method for precise mixing and proportioning of cable sheath masterbatch according to claim 1, characterized in that, In step S5, the threshold is: continuous One production batch The absolute value of the prediction deviation is greater than or equal to 0.5; Or, continuous One production batch The absolute value of the prediction deviation is greater than or equal to 0.3 μm; in .

7. The method for precise mixing and proportioning of cable sheath masterbatch according to claim 1, characterized in that, In step S5, the incremental update employs a recursive partial least squares algorithm with a forgetting factor.

8. A precise mixing and proportioning system for cable sheath masterbatch, characterized in that, include: The model building module is used to build a predictive model based on historical production data. The predictive model is as follows: in, Oxygen index, For surface roughness, This refers to the percentage of flame retardant masterbatch by mass. This refers to the percentage of color masterbatch by mass. For the effective particle size of flame retardant masterbatch, The moisture content of the flame retardant masterbatch, For extruder specific energy consumption, , and , , , , , , These are the regression coefficients; The data acquisition module is used to obtain the current production batch's... and and collect data in real time. ; The proportioning solution module is used to calculate the proportions based on the current batch. , and target oxygen index Target surface roughness The initial proportion of flame retardant masterbatch that minimizes performance deviation is determined using the prediction model. masterbatch ratio And produce accordingly; among them, for materials that are not yet predictable before feeding... The historical steady-state mean is used as the default value in the solution; The real-time control module is used to monitor the actual production process. Replace the default values ​​used in the proportioning solution module, and re-solve using the prediction model to obtain the corrected values. and And the feeding ratio is adjusted; The model update module is used to compare the output of the prediction model in real time. Compared with actual measurement When the deviation between the two exceeds a threshold, the regression coefficients of the prediction model are incrementally updated based on new sample data, wherein the new sample data includes data from the period in question. and corresponding actual measurements .

Images