An intelligent control system for industrial production line automation

Abstract

This invention relates to the field of intelligent control technology and discloses an automated intelligent control system for industrial production lines, comprising: a surface data acquisition module for real-time acquisition of surface roughness data and surface image data of workpieces of the same type; a data processing and analysis module for preprocessing and feature extraction of the acquired data, calculating comprehensive surface roughness indices and dynamic defect indices, and simultaneously acquiring process health scores fed back by quality inspection of workpieces of the same type; and an evaluation model construction module for constructing an evaluation model based on comprehensive surface roughness indices of workpieces of the same type in a historical database. This invention, by establishing a complete "perception-analysis-decision-control" closed loop, achieves real-time monitoring and automatic adjustment of the production line's process status, significantly improving the automation level and response efficiency of process control, reducing reliance on human experience, and effectively reducing the defect rate and production interruption risk caused by process fluctuations.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to an intelligent control system for industrial production line automation. Background Technology

[0002] In existing technologies, industrial production lines rely heavily on manual sampling or discrete offline testing equipment to monitor the surface quality of workpieces. This approach is characterized by significant lag and subjectivity, making it difficult to capture continuous fluctuations in process conditions and early defect trends in a timely manner.

[0003] Although some automated inspection solutions exist, they are often limited to static measurement of a single parameter, lacking comprehensive quantitative analysis of the multidimensional characteristics of surface roughness and the dynamic evolution of defects. Furthermore, they fail to link real-time evaluation results with production line control commands in a closed loop, resulting in slow process adjustments, passive quality control, and an inability to effectively prevent the generation of batch defective products, thus hindering further improvements in production efficiency and product consistency. Summary of the Invention

[0004] This invention provides an automated intelligent control system for industrial production lines to solve existing technical problems. It addresses the issues of poor process stability, lagging quality control, and low production efficiency caused by the lack of real-time, comprehensive evaluation and closed-loop control of workpiece surface conditions in existing technologies.

[0005] To solve the above-mentioned technical problems, according to one aspect of the present invention, more specifically, an automated intelligent control system for industrial production lines, comprising:

[0006] The surface data acquisition module is used to acquire surface roughness data and surface image data of workpieces of the same model in real time.

[0007] The data processing and analysis module is used to preprocess the collected data, extract features, calculate the comprehensive surface roughness index and the dynamic defect index, and obtain the process health score fed back by the quality inspection of the same type of workpiece.

[0008] The evaluation model construction module is used to construct a process status evaluation model based on the comprehensive surface roughness index, defect dynamic index, and process health score feedback from the quality inspection department for the same type of workpiece in the historical database.

[0009] The process status assessment model outputs the process health score of the newly acquired workpiece of the same model based on the comprehensive surface roughness index and the dynamic defect index of the newly acquired workpiece of the same model.

[0010] The process status assessment module is used to determine the process stability based on the process health score and generate a deceleration control command when the process is determined to be unstable.

[0011] The control command execution module is used to send deceleration commands to the production line equipment controller to adjust the equipment operating speed;

[0012] The specific working steps of the automated intelligent control system for this industrial production line are as follows:

[0013] S1. Real-time acquisition of surface condition data of the same type of workpiece in the production line; preprocessing and feature extraction of the acquired surface condition data, and calculation of comprehensive surface roughness index and dynamic defect index, while obtaining the process health score fed back by the quality inspection of the same type of workpiece;

[0014] S2. Construct a process status assessment model based on the comprehensive surface roughness index, dynamic defect index, and process health score of the same type of workpiece in the historical database.

[0015] S3. Input the comprehensive surface roughness index and dynamic defect index of the newly collected workpiece of the same model into the process condition evaluation model, and output the process health score of the newly collected workpiece of the same model from the process condition evaluation model.

[0016] S4. Determine whether the current process status is stable based on the health score; when the health score is lower than the stability threshold, it is determined that the process status is unstable, and a production line speed reduction command is automatically generated.

[0017] Furthermore, in step S1, the surface state data includes surface roughness data obtained by an online laser confocal microscope or a high-resolution optical scattering instrument, as well as surface image data obtained by a machine vision system.

[0018] Furthermore, the specific process for obtaining the comprehensive surface roughness index is as follows:

[0019] (1) Three-dimensional surface morphology data acquired by laser confocal microscopy Perform a two-dimensional wavelet transform to decompose it into subbands with different spatial frequencies;

[0020] (2) Calculate the roughness parameters of each sub-band, including the arithmetic mean deviation. Root mean square deviation skewness and kurtosis ;

[0021] (3) Extract the power spectral characteristics representing the surface texture direction and calculate the main texture direction angle. and its consistency coefficient .

[0022] Furthermore, the comprehensive surface roughness index The calculation formula is:

[0023] ;

[0024] In the above formula, , , , These are reference values ​​for the corresponding roughness parameters; to Let be the weight coefficients of each feature term, and satisfy . .

[0025] Furthermore, the specific process for defect detection and classification of surface image data is as follows:

[0026] (1) Use convolutional neural networks to perform semantic segmentation on surface images to identify and locate defective areas;

[0027] (2) Perform morphological and textural analysis on the segmented defective areas and classify them into at least one of the following: bubbles, scratches, pits, and impurities;

[0028] (3) Count the number of various defects per unit area. and its total area percentage .

[0029] Furthermore, the aforementioned defect dynamic index The calculation formula is:

[0030] ;

[0031] In the above formula, This represents the total number of defect categories. , where represents the severity weight of the i-th type of defect; For the first The reference threshold for the number of defects; This represents the change in the total number of defects within the most recent time window. The length of the time window; This is the rate of change weighting coefficient.

[0032] Furthermore, the specific steps for the evaluation model construction module to construct the process status evaluation model are as follows:

[0033] (1) Obtain workpieces of the same model whose process health scores, as reported by the quality inspection institute, show an arithmetic distribution;

[0034] (2) The relationship between the comprehensive surface roughness index, the dynamic defect index and the process health score reported by the quality inspection institute was obtained by controlling the variable method.

[0035] (3) Construct a process status evaluation model based on the correlation between the comprehensive surface roughness index and the dynamic defect index.

[0036] Furthermore, the process status evaluation model is based on a mathematical relationship model between the comprehensive surface roughness index and the dynamic defect index.

[0037] It is constructed by analyzing the correlation between the above indicators and quality inspection feedback scores in historical data, and can automatically output the corresponding process health score based on the surface feature data of newly collected workpieces.

[0038] Furthermore, the process health score reported by the quality inspection institute is a quantitative evaluation given by professional quality inspectors who conduct item-by-item testing on workpieces of the same model based on a set of pre-set quality inspection items and then synthesize the scores from each item. This score reflects the overall process quality level of the workpiece.

[0039] The present invention provides an automated intelligent control system for industrial production lines. Compared with existing technologies, the advantages achieved by this method are as follows:

[0040] 1. By establishing a complete "perception-analysis-decision-control" closed loop, this invention enables real-time monitoring and automatic adjustment of the production line process status, significantly improving the automation level and response efficiency of process control, reducing reliance on human experience, and effectively reducing the defect rate and production interruption risk caused by process fluctuations.

[0041] 2. This invention is based on an evaluation method that combines high-precision surface feature extraction with dynamic defect analysis. This system can accurately capture subtle changes and defect trends on the workpiece surface, thereby achieving quantitative evaluation and early warning of process health status, improving the accuracy and foresight of quality control, and providing reliable data support for continuous optimization of production processes.

[0042] 3. The process status evaluation model constructed by integrating historical data and real-time indicators in this invention has good adaptability and generalization ability. It can maintain the consistency of evaluation under different production batches and operating conditions, enhance the objectivity and scientific nature of the quality control process, and provide an effective tool for standardized and reusable intelligent monitoring.

[0043] 4. The adjustment mechanism of the present invention can automatically trigger a speed reduction command when the process state tends to be unstable, realize flexible adjustment of the production rhythm, help maintain the equipment in stable operating conditions, extend the service life of key components, and avoid production capacity loss caused by forced shutdown, thereby improving overall production efficiency and equipment utilization rate while ensuring quality. Attached Figure Description

[0044] Figure 1 This is a flowchart of the present invention;

[0045] Figure 2This is a graph showing the relationship between the comprehensive surface roughness index and the process health score reported by the quality inspection institute in this invention.

[0046] Figure 3 This is a graph showing the relationship between the comprehensive surface roughness index and the process health score reported by the quality inspection institute in this invention.

[0047] Figure 4 This is a diagram showing the relationship between the comprehensive surface roughness index and the dynamic defect index in this invention. Detailed Implementation

[0048] To make the technical solution of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] Example 1

[0050] like Figure 1 As shown, according to one aspect of the present invention, an intelligent control system for industrial production line automation is provided, comprising:

[0051] The surface data acquisition module is used to acquire surface roughness data and surface image data of workpieces of the same model in real time.

[0052] The data processing and analysis module is used to preprocess the collected data, extract features, calculate the comprehensive surface roughness index and the dynamic defect index, and obtain the process health score fed back by the quality inspection of the same type of workpiece.

[0053] The evaluation model construction module is used to construct a process status evaluation model based on the comprehensive surface roughness index, defect dynamic index, and process health score feedback from the quality inspection department for the same type of workpiece in the historical database.

[0054] Among them, the process status assessment model outputs the process health score of the newly acquired workpiece of the same model based on the comprehensive surface roughness index and the dynamic defect index of the newly acquired workpiece of the same model.

[0055] The process status assessment module is used to determine the process stability based on the process health score and generate a deceleration control command when the process is determined to be unstable.

[0056] The control command execution module is used to send deceleration commands to the production line equipment controller to adjust the equipment operating speed;

[0057] The specific process by which the data processing and analysis module obtains the comprehensive surface roughness index is as follows:

[0058] (1) Three-dimensional surface morphology data acquired by laser confocal microscopy Perform a two-dimensional wavelet transform to decompose it into subbands with different spatial frequencies;

[0059] (2) Calculate the roughness parameters of each sub-band, including the arithmetic mean deviation. Root mean square deviation skewness and kurtosis ;

[0060] (3) Extract the power spectral characteristics representing the surface texture direction and calculate the main texture direction angle. and its consistency coefficient .

[0061] Among them, the comprehensive index of surface roughness The calculation formula is:

[0062] ;

[0063] In the above formula, , , , These are reference values ​​for the corresponding roughness parameters; to Let be the weight coefficients of each feature term, and satisfy . .

[0064] The specific process by which the data processing and analysis module detects and classifies defects in surface image data is as follows:

[0065] (1) Use convolutional neural networks to perform semantic segmentation on surface images to identify and locate defective areas;

[0066] (2) Perform morphological and textural analysis on the segmented defective areas and classify them into at least one of the following: bubbles, scratches, pits, and impurities;

[0067] (3) Count the number of various defects per unit area. and its total area percentage .

[0068] Among them, the dynamic indicators of defects The calculation formula is:

[0069] ;

[0070] In the above formula, This represents the total number of defect categories. , where represents the severity weight of the i-th type of defect; For the first The reference threshold for the number of defects; This represents the change in the total number of defects within the most recent time window. The length of the time window; This is the rate of change weighting coefficient.

[0071] By constructing a closed-loop intelligent control system consisting of surface data acquisition, data processing and analysis, evaluation model construction, process status assessment, and control command execution, real-time monitoring and automatic control of the production line process status are achieved.

[0072] By combining high-precision surface feature extraction with dynamic defect analysis and establishing a process status assessment model based on historical data, a deceleration command is automatically triggered when the process health score is detected to be below a threshold, forming a complete automated "sensing-analysis-decision-control" chain. The advantages of this design are that it significantly improves the adaptability and response speed of process stability control, reduces manual intervention, lowers the defect rate, and extends the stable operation cycle of the equipment through real-time feedback adjustment.

[0073] Example 2

[0074] like Figure 1 As shown, the specific working steps of an automated intelligent control system for an industrial production line are as follows:

[0075] Step 1: Collect surface condition data of the same type of workpiece in the production line in real time; preprocess and extract features from the collected surface condition data, and calculate the comprehensive surface roughness index and the dynamic defect index, while obtaining the process health score fed back by the quality inspection of the same type of workpiece.

[0076] The surface condition data includes surface roughness data obtained through online laser confocal microscopy or high-resolution optical scattering instrument, as well as surface image data obtained through machine vision system.

[0077] Step 2: Construct a process status assessment model based on the comprehensive surface roughness index, dynamic defect index, and process health score of the same type of workpiece in the historical database.

[0078] Step 3: Input the comprehensive surface roughness index and dynamic defect index of the newly collected workpiece of the same model into the process condition assessment model, and output the process health score of the newly collected workpiece of the same model from the process condition assessment model.

[0079] Step 4: Determine whether the current process status is stable based on the health score; when the health score is lower than the stability threshold, the process status is determined to be unstable, and a production line speed reduction command is automatically generated.

[0080] In this embodiment, the specific steps for the evaluation model construction module to construct the process status evaluation model are as follows:

[0081] (1) Obtain workpieces of the same model whose process health scores, as reported by the quality inspection institute, show an arithmetic distribution;

[0082] Based on the historical database of workpieces of the same model that have been processed and inspected, surface roughness data was obtained through an online laser confocal microscope or a high-resolution optical scattering instrument, as well as surface image data obtained through a machine vision system.

[0083] Among them, the process health score reported by the quality inspection department for the same type of workpiece is:

[0084] ;

[0085] In the above formula, This indicates the items that need to be inspected (including density inspection, coating uniformity inspection, and inspection for bubbles, missing materials, scratches, etc.). This represents the quality inspection score of the i-th item (the quality inspector scores the items according to their quality, with scores ranging from 0 to 100%).

[0086] Furthermore, the process health scores of these workpieces of the same model that have already undergone quality inspection are used as a large sample in an arithmetic progression. For example, workpieces of the same model with process health scores of 0.21, 0.23, 0.25, ..., 0.75 are selected as a large sample (including qualified and unqualified workpieces).

[0087] (2) The relationship between the comprehensive surface roughness index, the dynamic defect index and the process health score reported by the quality inspection institute was obtained by controlling the variable method.

[0088] For example, screening dynamic indicators of defects from samples A subsample with all values ​​of 0.5 was used. A mathematical formula was constructed based on the relationship between the comprehensive surface roughness index and the process health score reported by the quality inspection institute in this subsample. The process health score reported by the quality inspection institute in this case is denoted as... Therefore, we have:

[0089] ;

[0090] In the above formula, , Used for control Process health score reported by the quality inspection institute A constant that approaches an approximation.

[0091] Therefore, based on this subsample, the comprehensive surface roughness index and the process health score reported by the quality inspection institute are analyzed. For example... Figure 2 As shown, it can be determined , When, formula Process health score reported by the quality inspection institute Approaching approximation.

[0092] For example, selecting comprehensive surface roughness indicators from samples. A subsample with all values ​​of 0.5 was used. A mathematical formula was constructed based on the relationship between the comprehensive surface roughness index and the process health score reported by the quality inspection institute in this subsample. The process health score reported by the quality inspection institute in this case is denoted as... Therefore, we have:

[0093] ;

[0094] In the above formula, , Used for control Process health score reported by the quality inspection institute A constant that approaches an approximation.

[0095] Therefore, based on this subsample, the comprehensive surface roughness index and the process health score reported by the quality inspection institute are analyzed. For example... Figure 3 As shown, it can be determined , When, formula Process health score reported by the quality inspection institute Approaching approximation.

[0096] (3) Construct a process status evaluation model based on the correlation between the comprehensive surface roughness index and the dynamic defect index.

[0097] From a large sample, arbitrarily obtain the comprehensive surface roughness index, dynamic defect index, and process health score (referred to as...) of workpieces of the same model. Based on the correlation between the comprehensive surface roughness index and the dynamic defect index, the mathematical formula for constructing the process condition evaluation model is as follows:

[0098] ;

[0099] In the above formula, , Used for control Process health score reported by the quality inspection institute A constant that approaches an approximation. For example... Figure 4 As shown, it can be determined , When, formula Process health score reported by the quality inspection institute Approaching approximation.

[0100] Therefore, the specific formula for the process status assessment model constructed using a large sample is as follows:

[0101] ;

[0102] In the above formula, Indicates dynamic indicators of defects; represents the comprehensive index of surface roughness; h represents the process health score of the newly acquired workpiece of the same model output by the process condition assessment model.

[0103] Table 1. Raw data for constructing process status assessment models in a portion of large samples.

[0104]

[0105] (4) Set the evaluation threshold of the process status evaluation model.

[0106] For example, from a large sample (including qualified and unqualified workpieces), the process health score at the critical point between qualified and unqualified workpieces of the same model is used as the evaluation threshold of the process status assessment model. If the value output by the process status assessment model for a newly acquired workpiece of the same model is lower than this evaluation threshold, it indicates that the newly acquired workpiece of the same model is unqualified.

[0107] By employing the controlled variable method and combining large-sample historical data, the mathematical relationships between the comprehensive surface roughness index, the dynamic defect index, and the manual quality inspection score are analyzed, and then an assessment model that can accurately reflect the health status of the process is constructed.

[0108] This model can not only output corresponding health scores based on real-time collected workpiece surface features, but also automatically determine process compliance by setting dynamic thresholds. Its advantages lie in improving the scientific rigor and objectivity of condition assessment, enhancing the system's adaptability and generalization capabilities under different production batches and operating conditions, and providing reliable data-driven support for process optimization and quality control.

[0109] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. An automated intelligent control system for an industrial production line, characterized in that, include: The surface data acquisition module is used to acquire surface roughness data and surface image data of workpieces of the same model in real time. The data processing and analysis module is used to preprocess the collected data, extract features, calculate the comprehensive surface roughness index and the dynamic defect index, and obtain the process health score fed back by the quality inspection of the same type of workpiece. The evaluation model construction module is used to construct a process status evaluation model based on the comprehensive surface roughness index, defect dynamic index, and process health score feedback from the quality inspection department for the same type of workpiece in the historical database. The process status assessment model outputs the process health score of the newly acquired workpiece of the same model based on the comprehensive surface roughness index and the dynamic defect index of the newly acquired workpiece of the same model. The process status assessment module is used to determine the process stability based on the process health score and generate a deceleration control command when the process is determined to be unstable. The control command execution module is used to send deceleration commands to the production line equipment controller to adjust the equipment operating speed; The specific working steps of the automated intelligent control system for this industrial production line are as follows: S1. Real-time acquisition of surface condition data of the same type of workpiece in the production line; preprocessing and feature extraction of the acquired surface condition data, and calculation of comprehensive surface roughness index and dynamic defect index, while obtaining the process health score fed back by the quality inspection of the same type of workpiece; S2. Construct a process status assessment model based on the comprehensive surface roughness index, dynamic defect index, and process health score of the same type of workpiece in the historical database. S3. Input the comprehensive surface roughness index and dynamic defect index of the newly collected workpiece of the same model into the process condition evaluation model, and output the process health score of the newly collected workpiece of the same model from the process condition evaluation model. S4. Determine whether the current process status is stable based on the health score; when the health score is lower than the stability threshold, it is determined that the process status is unstable, and a production line speed reduction command is automatically generated.

2. The industrial production line automation intelligent control system according to claim 1, characterized in that: In step S1, the surface state data includes surface roughness data obtained by an online laser confocal microscope or a high-resolution optical scattering instrument, and surface image data obtained by a machine vision system.

3. The industrial production line automation intelligent control system according to claim 1, characterized in that: The specific process for obtaining the comprehensive surface roughness index is as follows: (1) Three-dimensional surface morphology data acquired by laser confocal microscopy Perform a two-dimensional wavelet transform to decompose it into subbands with different spatial frequencies; (2) Calculate the roughness parameters of each sub-band, including the arithmetic mean deviation. Root mean square deviation skewness and kurtosis ; (3) Extract the power spectral characteristics representing the surface texture direction and calculate the main texture direction angle. and its consistency coefficient .

4. The industrial production line automation intelligent control system according to claim 3, characterized in that: The comprehensive index of surface roughness The calculation formula is: ； In the above formula, , , , These are reference values ​​for the corresponding roughness parameters; to Let be the weight coefficients of each feature term, and satisfy . .

5. The industrial production line automation intelligent control system according to claim 1, characterized in that: The specific process of defect detection and classification for surface image data is as follows: (1) Use convolutional neural networks to perform semantic segmentation on surface images to identify and locate defective areas; (2) Perform morphological and textural analysis on the segmented defective areas and classify them into at least one of the following: bubbles, scratches, pits, and impurities; (3) Count the number of various defects per unit area. and its total area percentage .

6. The industrial production line automation intelligent control system according to claim 5, characterized in that: The defect dynamic index The calculation formula is: ； In the above formula, This represents the total number of defect categories. , where represents the severity weight of the i-th type of defect; For the first The reference threshold for the number of defects; This represents the change in the total number of defects within the most recent time window. The length of the time window; This is the rate of change weighting coefficient.

7. The industrial production line automation intelligent control system according to claim 1, characterized in that: The specific steps for constructing the process status evaluation model using the evaluation model construction module are as follows: (1) Obtain workpieces of the same model whose process health scores, as reported by the quality inspection institute, show an arithmetic distribution; (2) The relationship between the comprehensive surface roughness index, the dynamic defect index and the process health score reported by the quality inspection institute was obtained by controlling the variable method. (3) Construct a process status evaluation model based on the correlation between the comprehensive surface roughness index and the dynamic defect index.

8. The industrial production line automation intelligent control system according to claim 7, characterized in that: The process status evaluation model is based on a mathematical relationship model between the comprehensive surface roughness index and the dynamic defect index. It is constructed by analyzing the correlation between the above indicators and quality inspection feedback scores in historical data, and can automatically output the corresponding process health score based on the surface feature data of newly collected workpieces.

9. The industrial production line automation intelligent control system according to claim 1, characterized in that: The process health score reported by the quality inspection institute is a quantitative evaluation given by professional quality inspectors based on multiple preset quality inspection items, which are tested item by item on the same type of workpiece, and the scores are combined.

Images