Cable production quality control method based on multi-source data

Abstract

The invention relates to the technical field of production quality control, in particular to a cable production quality control method based on multi-source data. The method comprises the following steps: collecting multi-source production data on a cable production line; based on the environment state data and the process parameter data, calculating an environment comprehensive cooling coefficient at each acquisition moment; calculating a dynamic cooling rate factor at each acquisition moment based on the environment comprehensive cooling coefficient and the process parameter data; calculating a proportion of a difference value between the dynamic cooling rate factor and a preset optimal dynamic cooling rate factor in the optimal dynamic cooling rate factor, and obtaining a non-equilibrium crystallization risk index at each acquisition moment; inputting the non-equilibrium crystallization risk index and the multi-source production data into a pre-trained dynamic Bayesian network model, obtaining a predicted value of a cable quality parameter at a next acquisition moment, and controlling the cable production quality; the problem that the cable production quality control effect is poor is solved.

Description

Technical Field

[0001] This application relates to the field of production quality control technology, and in particular to a method for cable production quality control based on multi-source data. Background Technology

[0002] Cables are a fundamental infrastructure of modern industry and social life, and their quality directly affects the stability and safety of power transmission and information communication. During cable production, especially in the extrusion molding process of the insulation and sheath layers, product quality is influenced by numerous factors. These factors include production process parameters such as extruder temperature, screw speed, and traction speed, as well as environmental factors in the production workshop, such as ambient temperature, humidity, and air pressure.

[0003] To achieve precise control over production quality, data-driven models have begun to be applied in existing technologies for quality prediction and anomaly diagnosis. Among these, Dynamic Bayesian Networks (DBNs), as probabilistic graphical models capable of processing time-series data and expressing uncertain causal relationships between variables, have shown potential in monitoring complex industrial processes. These models use collected environmental sensor data and process parameter data as input nodes and the final quality inspection results as output nodes, building the network model by learning from historical data to predict potential product quality under specific process and environmental conditions. However, existing methods based on Dynamic Bayesian Networks have significant limitations. These methods typically treat environmental factors as independent, parallel input variables, directly inputting them into the model along with process parameters. They lack analysis of the physical processes and cannot reflect the combined impact of multiple environmental factors on product quality, resulting in poor quality control in cable production. Summary of the Invention

[0004] To address the problem of poor quality control in cable production due to the inability to reflect the combined effects of multiple environmental factors on product quality, this application provides a cable production quality control method based on multi-source data.

[0005] This application provides a cable production quality control method based on multi-source data, employing the following technical solution: A cable production quality control method based on multi-source data includes the following steps: Collect multi-source production data from the cable production line, including environmental status data, process parameter data, and product quality data; Based on environmental condition data and process parameter data, the comprehensive environmental cooling coefficient at each acquisition time is calculated. The comprehensive environmental cooling coefficient represents the comprehensive cooling capacity of the environment on the cable. Based on the comprehensive environmental cooling coefficient and process parameter data, the dynamic cooling rate factor at each acquisition time is calculated. The difference between the dynamic cooling rate factor and the preset optimal dynamic cooling rate factor is calculated as a percentage of the optimal dynamic cooling rate factor to obtain the non-equilibrium crystallization risk index at each acquisition time. The non-equilibrium crystallization risk index and the multi-source production data are input into a pre-trained dynamic Bayesian network model to obtain the predicted value of the cable quality parameters at the next acquisition time, thereby controlling the cable production quality.

[0006] The beneficial effects are as follows: By periodically collecting environmental data, process parameter data, and product quality data, and gradually constructing an environmental comprehensive cooling coefficient, a dynamic cooling rate factor, and an unbalanced crystallization risk index after data preprocessing, a physically enhanced dynamic Bayesian network model is finally used for prediction and alarm. The causal chain between environmental disturbances, process changes, and final quality is clearly quantified, enabling the model to not only predict cable quality in real time, but also provide early warnings in the early stages of risk, thereby improving production consistency and quality controllability.

[0007] Furthermore, the environmental condition data includes ambient temperature data, relative humidity data, and atmospheric pressure data; the process parameter data includes cable traction speed data, extruder melt temperature data, and cooling water tank temperature data; and the product quality data includes cable outer diameter data.

[0008] The beneficial effects are: avoiding the one-sidedness of describing a single environmental variable, fully reflecting the comprehensive effect of the environment on cable cooling, transforming the environmental impact from fuzzy statistical characteristics into clear physical indicators, thereby improving the model's adaptability to complex environments.

[0009] Furthermore, the method for obtaining the overall environmental cooling coefficient is as follows: The saturated water vapor pressure of ambient temperature data was obtained using the Antoine equation. The formula for calculating the overall environmental cooling coefficient is as follows: In the formula, The overall environmental cooling coefficient at each data collection time; The data includes ambient temperature data at each collection time. This contains extruder melt temperature data at each acquisition time. This contains relative humidity data at each data collection time. For ambient temperature data The corresponding saturated water vapor pressure, This contains atmospheric pressure data at each data collection time.

[0010] The beneficial effects are: by combining the environmental comprehensive cooling coefficient and the traction speed to construct a dynamic cooling rate factor, the material cooling rate can be accurately reflected; not only the external environment is considered, but also internal process parameters are integrated, which can directly correspond to the cooling process of cable insulation materials, making the subsequent risk assessment highly consistent with the physical reality.

[0011] Furthermore, the method for obtaining the dynamic cooling rate factor is as follows: The product of the cable traction speed data, the specific heat capacity of the cable insulation material, and the density of the cable insulation material at each acquisition time is calculated, and the ratio of the environmental comprehensive cooling coefficient to the product is used as the dynamic cooling rate factor at each acquisition time.

[0012] The beneficial effects are as follows: by using the deviation of the dynamic cooling rate factor from the optimal value as a risk measure, the abstract risk of crystallization defects is transformed into a quantifiable index, and the magnitude of the risk is presented intuitively through numerical values, which helps process engineers to clarify the direction of adjustment.

[0013] Furthermore, the method for obtaining the optimal dynamic cooling rate factor is as follows: Select the cable batch with the highest quality pass rate from historical data; The distribution of dynamic cooling rate factors during the production process of the aforementioned cable batches was statistically analyzed, and the dynamic cooling rate factors were... The distribution center value is used as the optimal dynamic cooling rate factor. .

[0014] The beneficial effects are: the optimal cooling rate factor is determined by the statistical mean of historical high-quality batch data, and the optimal process conditions are extracted by using production experience data, so as to realize the combination of data-driven and physical mechanism and ensure that the source of risk assessment standards is true and reliable.

[0015] Furthermore, the network structure of the dynamic Bayesian network model is as follows: Using the environmental comprehensive cooling coefficient and process parameter data as the mother node, the non-equilibrium crystallization risk index as the intermediate hidden variable node, and the cable quality parameters as the child nodes.

[0016] Furthermore, the mother node also includes cable traction speed data, extruder melt temperature data, and cooling water tank temperature data.

[0017] Furthermore, the cable quality parameter is the cable outer diameter deviation.

[0018] Furthermore, the manufacturing quality of the control cable includes: An alarm is triggered when the predicted value of the cable quality parameter exceeds the preset acceptable range.

[0019] Furthermore, the preset qualified range is a preset percentage of the cable outer diameter data.

[0020] This application has the following technical effects: This application proposes a cable production quality control method based on multi-source data. This method sequentially constructs an environmental comprehensive cooling coefficient, a dynamic cooling rate factor, a non-equilibrium crystallization risk index, and an optimal dynamic cooling rate factor, and combines this with a physically enhanced dynamic Bayesian network to control cable production quality. First, the environmental comprehensive cooling coefficient integrates environmental temperature, humidity, and air pressure variables to uniformly reflect the potential of convective heat transfer and evaporative heat dissipation, avoiding the one-sidedness of single-variable descriptions and transforming the impact of environmental disturbances on the cooling process into clear and quantifiable indicators. Second, the dynamic cooling rate factor combines the environmental comprehensive cooling coefficient with traction speed and material thermophysical parameters to reflect the actual rate of temperature decrease in the insulation material, thus obtaining a dynamic quantitative result directly reflecting the physical process under the combined effects of environment and process. Furthermore, the non-equilibrium crystallization risk index reflects the degree to which the crystallization process deviates from the ideal state by comparing the relative deviation between the actual cooling rate and the optimal cooling rate. Whether the cooling is too fast or too slow, it linearly amplifies the risk level, providing process engineers with an intuitive quality risk signal. Meanwhile, the optimal dynamic cooling rate factor is obtained by statistically analyzing the average cooling rate of historical high-yield batches, ensuring not only the scientific rigor and objectivity of the evaluation benchmark but also making the parameter operable and practically instructive. Furthermore, the dynamic Bayesian network uses process parameters and environmental factors as parent nodes, risk indices as latent variables, and quality indicators as output nodes to explicitly construct a physical causal chain, improving prediction accuracy and robustness while enhancing the model's interpretability and diagnosability. This application, through the deep integration of physical mechanisms and data-driven approaches, can provide early warnings and derive process optimization strategies before quality defects occur, addressing the problem of poor quality control in cable production due to the inability to reflect the combined impact of multiple environmental factors on product quality, thereby significantly improving the stability of cable production process quality. Attached Figure Description

[0021] Figure 1 This is a flowchart of the cable production quality control method based on multi-source data in the embodiments of this application. Detailed Implementation

[0022] This application discloses a cable production quality control method based on multi-source data. The method involves collecting multi-source production data from the cable production line; calculating the comprehensive environmental cooling coefficient at each collection time based on environmental state data and process parameter data; calculating the dynamic cooling rate factor at each collection time based on the comprehensive environmental cooling coefficient and process parameter data; calculating the proportion of the difference between the dynamic cooling rate factor and the preset optimal dynamic cooling rate factor within the optimal dynamic cooling rate factor to obtain the non-equilibrium crystallization risk index at each collection time; and inputting the non-equilibrium crystallization risk index and the multi-source production data into a pre-trained dynamic Bayesian network model to obtain predicted values ​​of cable quality parameters at the next collection time, thereby controlling cable production quality. This method solves the problem of poor cable production quality control.

[0023] Reference Figure 1 The cable production quality control method based on multi-source data includes steps S1-S4.

[0024] Step S1: Collect multi-source production data from the cable production line, including environmental status data, process parameter data, and product quality data.

[0025] This application periodically acquires multi-source production data at various times from multiple data sources on the cable production line. In one embodiment of this application, the acquired multi-source production data includes: environmental condition data, process parameter data, and product quality data.

[0026] The environmental status data is acquired by setting up sensors near the cooling section of the production workshop to collect ambient temperature, relative humidity, and atmospheric pressure data. The process parameter data is acquired by obtaining cable traction speed data, extruder melt temperature data, and cooling water tank temperature data from the production line control system. In one embodiment of this application, the process parameter data is acquired from a programmable logic controller (PLC). The implementer can select other production line control systems based on actual conditions. The product quality data is acquired by using a laser diameter gauge to obtain cable outer diameter data. This cable outer diameter data serves as the benchmark for calculating cable outer diameter deviation. By judging whether the cable outer diameter deviation exceeds the preset acceptable range, accurate quality warnings and production control can be achieved, ensuring the stability of cable production quality. In one embodiment of this application, the interval between any two adjacent acquisition times is 2 seconds. The implementer can select other values ​​based on actual conditions.

[0027] The collected environmental status data, process parameter data, and product quality data undergo preprocessing, including timestamp alignment, outlier removal, and normalization. The preprocessing process is well-known and will not be elaborated upon in this application.

[0028] Step S2: Based on environmental condition data and process parameter data, calculate the comprehensive environmental cooling coefficient at each acquisition time. The comprehensive environmental cooling coefficient represents the comprehensive cooling capacity of the environment on the cable. Based on the comprehensive environmental cooling coefficient and process parameter data, calculate the dynamic cooling rate factor at each acquisition time.

[0029] Furthermore, ambient temperature, humidity, and air pressure collectively determine the heat exchange capacity of air, and no single variable can fully describe this. Therefore, to characterize the combined impact of multiple environmental factors on the cooling process after cable extrusion, this application constructs a comprehensive environmental cooling coefficient at each data collection time based on environmental state data and process parameter data. Considering the combined effects of convective heat transfer and evaporative heat dissipation, the calculation formula is as follows: In the formula, The overall environmental cooling coefficient at each data collection time; The data includes ambient temperature data at each collection time. This contains extruder melt temperature data at each acquisition time. This contains relative humidity data at each data collection time. For ambient temperature data The corresponding saturated water vapor pressure, The data includes atmospheric pressure at each acquisition time. It should be noted that the saturated water vapor pressure is obtained using the Antoine equation, a well-known technique, and will not be elaborated upon in this application.

[0030] It should be noted that the overall environmental cooling coefficient The higher the value, the stronger the environment's ability to remove heat. This reflects the convective heat transfer potential driven by temperature difference, when the ambient temperature data... When the temperature decreases, the temperature difference increases, and the value of this term increases; This reflects the evaporative heat dissipation potential related to humidity and air pressure, when the ambient humidity... Reduce or atmospheric pressure As the temperature decreases, water evaporates more easily, increasing this factor. Saturated water vapor pressure reflects the ability of water molecules to change from a liquid to a gaseous state at the current temperature, while atmospheric pressure reflects the degree to which the external environment inhibits evaporation. The larger the ratio of the two, the easier it is for water molecules to escape from the environment, the stronger the evaporation potential, and the more significant the cooling effect. Therefore, when the environment becomes colder, drier, or the air pressure is lower, the overall environmental cooling coefficient increases. The value increases, indicating that the cooling effect of the environment on the cable is enhanced. It unifies convective heat transfer and evaporative heat dissipation into a cooling potential, avoiding the one-sidedness of a single variable description and establishing a comprehensive index coupled with multiple environmental factors.

[0031] The cooling effect of the environment ultimately affects the cable, and its effect is also related to the time the cable is exposed to the environment, which is determined by the traction speed. To reflect the actual cooling intensity experienced by the cable in the cooling zone, this application uses the comprehensive environmental cooling coefficient. and cable traction speed data Construct the dynamic cooling rate factor at each acquisition time. The calculation formula is: In the formula, This represents the dynamic cooling rate factor at each acquisition time. The overall environmental cooling coefficient at each data collection time. This provides data on cable traction speed at each acquisition time. , where is the specific heat capacity of the cable's insulation material, is an inherent property of the material, and is a known parameter; The density of the cable insulation material is an inherent property of the material and is a known parameter; the methods for obtaining the specific heat capacity and density of the insulation material are not described in detail in this application.

[0032] It should be noted that the dynamic cooling rate factor Reflects the actual rate of temperature decrease in cable manufacturing materials; dynamic cooling rate factor Overall Cooling Coefficient with Environment It is directly proportional, meaning the stronger the environmental cooling capacity, the faster the material cools; dynamic cooling rate factor Cable traction speed data Inversely proportional, meaning the faster the traction speed, the shorter the time the cable spends in the cooling zone, the weaker the cooling effect per unit length of material, and the slower the cooling rate. Dynamic cooling rate factor. By combining external environmental influences with internal process parameters, dynamic indicators that directly describe the physical processes of materials were obtained.

[0033] Step S3: Calculate the proportion of the difference between the dynamic cooling rate factor and the preset optimal dynamic cooling rate factor in the optimal dynamic cooling rate factor, and obtain the non-equilibrium crystallization risk index at each collection time.

[0034] The cooling rate of the insulation material in a cable directly determines the crystallization process of its polymer chains. Cooling that is too fast or too slow will result in an undesirable crystal structure, generating internal stress, and ultimately leading to abnormal dimensional shrinkage and decreased mechanical properties. To quantify these risks, this application uses a material dynamic cooling rate factor. Constructing the non-equilibrium crystallization risk index at each collection time The calculation formula is: In the formula, The non-equilibrium crystallization risk index at each collection time; This represents the dynamic cooling rate factor at each acquisition time. This refers to the optimal dynamic cooling rate factor for cable materials; the optimal dynamic cooling rate factor indicates that cooling at this rate yields the best crystal structure and the minimum internal stress. This application obtains the optimal dynamic cooling rate factor by analyzing the distribution of dynamic cooling rate factors in historical production data. The specific method is as follows: Select the cable batches with the highest quality pass rate from historical data; statistically analyze the dynamic cooling rate factor during the production process of these cable batches. The distribution of all dynamic cooling rate factors The distribution center value is used as the optimal cooling rate factor .

[0035] It should be noted that the non-equilibrium crystallization risk index reflects the relative deviation between the actual cooling rate and the optimal cooling rate of the cable manufacturing materials. The higher the value of the non-equilibrium crystallization risk index, the higher the risk of quality defects in the cable. Specifically, when the dynamic cooling rate factor... Equal to the optimal dynamic cooling rate factor At that time, the non-equilibrium crystallization risk index A value of 0 indicates that the crystallization process is in an ideal state, with the lowest risk; when the dynamic cooling rate factor is 0, it .... Deviation from the optimal dynamic cooling rate factor due to changes in environmental or process parameters At that time, regardless of whether the deviation from the direction is too fast or too slow, the non-equilibrium crystallization risk index The value increases linearly with the degree of deviation, intuitively quantifying the quality risk introduced by improper cooling.

[0036] Step S4: Input the non-equilibrium crystallization risk index and the multi-source production data into the pre-trained dynamic Bayesian network model to obtain the predicted values ​​of the cable quality parameters at the next acquisition time and control the cable production quality.

[0037] For the non-equilibrium crystallization risk index obtained at each collection time As the core latent variable node, a physically enhanced dynamic Bayesian network model is constructed. In this application, the network structure of the dynamic Bayesian network model is explicitly designed to reflect physical causal chains, specifically: The mother node layer includes directly controllable process parameter data nodes and environmental comprehensive cooling coefficients. In one embodiment of this application, the directly controllable process parameter data nodes included in the mother node layer are: cable traction speed data, extruder melt temperature data, and cooling water tank temperature data.

[0038] Intermediate hidden variable node: i.e., the non-equilibrium crystallization risk index Its conditional probability distribution is determined by its parent node.

[0039] Sub-node: Cable quality parameter node. In one embodiment of this application, the final quality index node is the cable outer diameter deviation.

[0040] The conditional probability tables (CPTs) of the dynamic Bayesian network model are trained using historical multi-source production data. Specifically, in actual production, the system collects multi-source production data in real time and calculates the non-equilibrium crystallization risk index at each collection time. Furthermore, the non-equilibrium crystallization risk index Multi-source production data is input into a trained dynamic Bayesian network to predict the cable outer diameter deviation at the next data collection moment in real time. An alarm is triggered when the predicted cable outer diameter deviation exceeds a preset acceptable range; conversely, no alarm is triggered when the predicted deviation does not exceed the preset acceptable range. In one embodiment of this application, the preset acceptable range is set to 0.1% of the cable outer diameter data; implementers can select other values ​​based on actual conditions.

[0041] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. In some implementations, multitasking and parallel processing are also possible or may be advantageous.

[0042] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some of the technical features, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A cable production quality control method based on multi-source data, characterized in that, Including the following steps: Collect multi-source production data from the cable production line, including environmental status data, process parameter data, and product quality data; Based on environmental condition data and process parameter data, the comprehensive environmental cooling coefficient at each data collection time is calculated. The comprehensive environmental cooling coefficient characterizes the overall cooling capacity of the environment on the cable. Based on the comprehensive environmental cooling coefficient and process parameter data, the dynamic cooling rate factor at each data acquisition time is calculated. Calculate the proportion of the difference between the dynamic cooling rate factor and the preset optimal dynamic cooling rate factor in the optimal dynamic cooling rate factor, and obtain the non-equilibrium crystallization risk index at each collection time. The non-equilibrium crystallization risk index and the multi-source production data are input into a pre-trained dynamic Bayesian network model to obtain the predicted values ​​of cable quality parameters at the next acquisition time, thereby controlling the cable production quality.

2. The cable production quality control method based on multi-source data according to claim 1, characterized in that, The environmental status data includes ambient temperature data, relative humidity data, and atmospheric pressure data; the process parameter data includes cable traction speed data, extruder melt temperature data, and cooling water tank temperature data; the product quality data includes cable outer diameter data.

3. The cable production quality control method based on multi-source data according to claim 2, characterized in that, The method for obtaining the overall environmental cooling coefficient is as follows: The saturated water vapor pressure of ambient temperature data was obtained using the Antoine equation. The formula for calculating the overall environmental cooling coefficient is as follows: In the formula, The overall environmental cooling coefficient at each data collection time; The data includes ambient temperature data at each collection time. This contains extruder melt temperature data at various acquisition times. This contains relative humidity data at each data collection time. For ambient temperature data The corresponding saturated water vapor pressure, This contains atmospheric pressure data at each data collection time.

4. The cable production quality control method based on multi-source data according to claim 2, characterized in that, The method for obtaining the dynamic cooling rate factor is as follows: The product of the cable traction speed data, the specific heat capacity of the cable insulation material, and the density of the cable insulation material at each acquisition time is calculated, and the ratio of the environmental comprehensive cooling coefficient to the product is used as the dynamic cooling rate factor at each acquisition time.

5. The cable production quality control method based on multi-source data according to claim 1, characterized in that, The method for obtaining the optimal dynamic cooling rate factor is as follows: Select the cable batch with the highest quality pass rate from historical data; The distribution of dynamic cooling rate factors during the production process of the aforementioned cable batches was statistically analyzed, and the dynamic cooling rate factors were... The distribution center value is used as the optimal dynamic cooling rate factor. .

6. The cable production quality control method based on multi-source data according to claim 1, characterized in that, The network structure of the dynamic Bayesian network model is as follows: Using the environmental comprehensive cooling coefficient and process parameter data as the mother node, the non-equilibrium crystallization risk index as the intermediate hidden variable node, and the cable quality parameters as the child nodes.

7. The cable production quality control method based on multi-source data according to claim 6, characterized in that, The mother node also includes cable traction speed data, extruder melt temperature data, and cooling water tank temperature data.

8. The cable production quality control method based on multi-source data according to claim 1, characterized in that, The cable quality parameter is the cable outer diameter deviation.

9. The cable production quality control method based on multi-source data according to claim 1, characterized in that, The manufacturing quality of the control cable includes: An alarm is triggered when the predicted value of the cable quality parameter exceeds the preset acceptable range.

10. The cable production quality control method based on multi-source data according to claim 9, characterized in that, The preset qualified range is set as a preset percentage of the cable outer diameter data.

Images