Intelligent Monitoring System and Method Based on Production Line Processing

Abstract

The present invention discloses an intelligent monitoring system and method based on production line processing, which relates to the technical field of data processing and management, and includes: a workpiece scanning module, which uses a flatness detection device and a hardness detection device to scan the surface of the workpiece; a region division module, for each scanning point, calculates the local density of the scanning point, and divides the surface of the workpiece into several independently controlled partitions through a clustering algorithm according to the local density values of each scanning point; a cutting parameter optimization module, which uses a random forest algorithm to establish a tool life model, takes the flatness and hardness of the workpiece as inputs, and outputs the optimal cutting parameter combination corresponding to each flatness and hardness interval; a partition control module, calculates the average flatness and average hardness of each partition, determines the optimal cutting parameter combination according to the average flatness and average hardness, and independently controls each partition, significantly improving the machining accuracy and prolonging the service life of the tool.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing management, and particularly to an intelligent monitoring system and method based on production line processing. Background Art

[0002] In modern manufacturing, the quality and production efficiency of workpieces are directly related to the competitiveness of enterprises. As a key link in workpiece production, the monitoring and optimization of the manufacturing and machining production line are particularly important, especially for the processing and management of processing data. Traditional processing monitoring methods mainly rely on manual inspection and empirical judgment. This method is not only inefficient but also difficult to ensure the stability and consistency of workpiece quality.

[0003] During the workpiece processing, due to the differences in workpiece shape, size, and material, as well as improper selection of cutting parameters, problems such as excessive tool wear, low processing efficiency, and unstable workpiece quality often occur. Traditional solutions mainly rely on the experience of operators and the trial-and-error method. This method is not only inefficient but also difficult to find the optimal combination of cutting parameters. In addition, for large or complex-shaped workpieces, there are often multiple regions with different flatness and hardness on their surfaces. Traditional processing methods often use unified cutting parameters for processing, which is likely to cause over-cutting or under-cutting in some regions, affecting the overall quality of the workpiece. Summary of the Invention

[0004] (I) Technical Problems to be Solved

[0005] Aiming at the deficiencies of the prior art, the present invention provides an intelligent monitoring system and method based on production line processing. Through comprehensive detection of flatness and hardness, local density clustering and zoning, prediction of the best cutting parameters, and independent control of each zone, it realizes the overall control of workpiece processing quality, intelligent optimization of cutting parameters, and precise control of the processing process, significantly improving the processing accuracy and extending the service life of the tool.

[0006] (II) Technical Solutions

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions: An intelligent monitoring system based on production line processing, including:

[0008] A workpiece scanning module, which uses a flatness detection device and a hardness detection device to scan the surface of the workpiece, obtains the flatness data and hardness data of each scanning point, aggregates and constructs a scanning data set;

[0009] Region division module: For each scanning point, calculate the local flatness density and local hardness density of the scanning point within its surrounding neighborhood. Based on the local flatness density and local hardness density of the scanning point within its surrounding neighborhood, calculate the local density of the scanning point. According to the local density values of each scanning point, divide the workpiece surface into several independently controlled partitions using the K-means++ clustering algorithm;

[0010] Cutting parameter optimization module: Obtain the historical machining data of workpieces with different flatness and hardness under various cutting parameter combinations. Use the random forest algorithm to establish a tool life model, taking the flatness and hardness of the workpiece as inputs and outputting the optimal cutting parameter combination corresponding to each flatness and hardness interval;

[0011] Partition control module: For each partition on the workpiece surface, obtain the flatness data and hardness data in the scanning dataset, calculate the average flatness and average hardness of each partition, and determine the optimal cutting parameter combination based on the average flatness and average hardness to independently control each partition.

[0012] Furthermore, traverse each scanning point, compare the flatness of each scanning point with a preset flatness threshold. If the flatness of the scanning point reaches the preset flatness threshold, record the coordinates of the scanning point. Based on the recorded coordinates of the scanning points, perform pre-processing on the workpiece surface until the flatness of each scanning point is lower than the preset flatness threshold. After pre-processing, re-obtain the flatness data of each scanning point.

[0013] Furthermore, for each scanning point, use a square window with a side length of 2w + 1 centered on the scanning point as the surrounding neighborhood of the scanning point, where w is the window width. Calculate the local flatness density and local hardness density of each scanning point within its surrounding neighborhood based on the flatness and hardness of each scanning point. The calculation formulas are as follows:

[0014] ;

[0015] Where and represent the local flatness density and local hardness density respectively, and represent the flatness of the i-th scanning point and the j-th scanning point respectively, and represent the hardness of the i-th scanning point and the j-th scanning point respectively, represents the set of scanning points within the surrounding neighborhood of scanning point i, and σ is the width parameter of the Gaussian kernel.

[0016] Furthermore, calculate the local density of the scanning point based on the local flatness density and local hardness density of the scanning point within its surrounding neighborhood: , where represents the local density, and represent the weights of flatness and hardness respectively, and ;

[0017] Among them, the weight of flatness: , the weight of hardness: , where , and N is the number of all scanned points.

[0018] Furthermore, the surface of the workpiece is divided into several independently controlled partitions by the K-means++ clustering algorithm, specifically including:

[0019] S1: Initialize the clustering centers using the K-means++ algorithm;

[0020] S2: For each scanned point, calculate the Euclidean distance from the scanned point to each clustering center, and assign each scanned point to the category to which the nearest clustering center belongs;

[0021] S3: For each category, recalculate the clustering center of the category, and the new clustering center is the average value of the local density values of all scanned points in the category;

[0022] S4: Repeat S2 to S3 until the clustering centers no longer change (i.e., converge), or the preset number of iterations is reached;

[0023] S5: Obtain k clustering centers and the clustering category to which each scanned point belongs according to the final clustering result, and connect adjacent scanned points belonging to the same cluster into a continuous area to form several independently controlled partitions.

[0024] Furthermore, initializing the clustering centers using the K-means++ algorithm includes:

[0025] S11: Randomly select a scanned point as the first clustering center;

[0026] S12: For each scanned point, calculate the Euclidean distance from the scanned point to the nearest clustering center: , where represents the Euclidean distance from the i-th scanned point to the k-th clustering center, represents the local density of the i-th scanned point, represents the local density of the k-th clustering center;

[0027] S13: For each scanning point, calculate the square of the Euclidean distance from the scanning point to the nearest cluster center as the weight for being selected as the next cluster center, and select the next cluster center according to the weights of each scanning point by the roulette wheel selection method;

[0028] S14: Repeat S12 - S13 until k cluster centers are selected.

[0029] Furthermore, output the optimal cutting parameter combinations corresponding to each flatness and hardness interval, specifically including:

[0030] Take the flatness and hardness of the workpiece as feature variables, the tool life as the target variable, and the cutting parameters as optimization variables. Use the random forest algorithm to establish a tool life model, and use historical machining data to train the tool life model to establish the mapping relationship between flatness and hardness and tool life.

[0031] Furthermore, within the preset range of cutting parameters, traverse all possible cutting parameter combinations through grid search. Take the flatness and hardness of the workpiece as inputs, use the trained tool life model to predict the tool life under each cutting parameter combination, and calculate the mean squared error MSE between the predicted tool life and the actual value under each cutting parameter combination;

[0032] Pre - set an error threshold, screen out the cutting parameter combinations with the mean squared error MSE less than the error threshold, sort the screened cutting parameter combinations, and select the cutting parameter combination with the longest tool life as the optimal cutting parameter combination, and output the optimal cutting parameter combinations corresponding to each flatness and hardness interval.

[0033] Furthermore, for each partition on the workpiece surface, according to the mean flatness and mean hardness, find the flatness and hardness interval in the output of the tool life model that is closest to the mean flatness and mean hardness of this partition, and select the optimal cutting parameter combination corresponding to this interval.

[0034] The intelligent monitoring method based on production line machining includes the following steps:

[0035] Step 1: Use a flatness detection device and a hardness detection device to scan the workpiece surface, obtain the flatness data and hardness data of each scanning point, and summarize and construct a scanning data set;

[0036] Step 2: For each scanning point, calculate the local flatness density and local hardness density of the scanning point within its surrounding neighborhood. Through the local flatness density and local hardness density of the scanning point within its surrounding neighborhood, calculate the local density of the scanning point. According to the local density values of each scanning point, divide the workpiece surface into several independently controlled partitions by the K - means++ clustering algorithm;

[0037] Step 3: Obtain the historical processing data of workpieces with different flatness and hardness under various combinations of cutting parameters. Use the random forest algorithm to establish a tool life model, taking the flatness and hardness of the workpiece as inputs, and outputting the optimal combination of cutting parameters corresponding to each flatness and hardness interval.

[0038] Step 4: For each partition on the workpiece surface, obtain the flatness data and hardness data in the scan dataset, calculate the average flatness and average hardness of each partition, and determine the optimal combination of cutting parameters according to the average flatness and average hardness, and perform independent control on each partition.

[0039] (III) Beneficial Effects

[0040] The present invention provides an intelligent monitoring system and method based on production line processing, having the following beneficial effects:

[0041] (1) By using the flatness detection device and hardness detection device to scan the workpiece surface point by point, the flatness and hardness data of each scan point can be accurately obtained. Through pre-processing, the high points and uneven parts on the workpiece surface can be removed, thereby reducing tool wear during the cutting process. A flat surface helps to reduce the friction and vibration between the tool and the workpiece, and further extends the service life of the tool.

[0042] (2) By calculating the local density for each scan point and clustering according to the local density, the workpiece surface can be divided into regions with similar characteristics, and differential processing is performed on each region, thereby improving the overall processing accuracy. The surface characteristics (such as flatness and hardness) of the workpiece in different regions may be different, so the required tools and processing parameters may also be different. Through partition control, appropriate tools and processing parameters can be selected for each region, thereby extending the service life of the tool.

[0043] (3) By obtaining the historical processing data of workpieces with different flatness and hardness under various combinations of cutting parameters and establishing a tool life model, the optimal combination of cutting parameters can be selected according to the specific characteristics of the workpiece surface, thereby optimizing the processing process, selecting the cutting parameter combination with the least tool wear, and effectively extending the service life of the tool.

[0044] (4) By calculating the average flatness and average hardness of each partition and determining the optimal combination of cutting parameters based on these averages, it can be ensured that each partition is processed using the cutting parameters most suitable for its characteristics, fully considering the differences on the workpiece surface, thereby significantly improving the processing accuracy and reducing errors caused by improper cutting parameters. By considering the influence of flatness and hardness on tool life, the optimal combination of cutting parameters selected for each partition can maximize the service life of the tool. Description of the Drawings

[0045] Figure 1 Schematic diagram of the intelligent monitoring system based on production line processing of the present invention;

[0046] Figure 2 Cutting schematic diagram of the workpiece processed on the production line of the present invention;

[0047] Figure 3 Schematic diagram of the steps of the intelligent monitoring method based on production line processing of the present invention. Detailed implementation manners

[0048] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0049] Please refer to Figure 1 - Figure 2 , the present invention provides an intelligent monitoring system based on production line processing, including: a workpiece scanning module, a region division module, a cutting parameter optimization module, and a partition control module; wherein,

[0050] The workpiece scanning module uses a flatness detection device and a hardness detection device to scan the surface of the workpiece, obtain the flatness data and hardness data of each scanning point, summarize and construct a scanning data set;

[0051] Place the workpiece on the detection platform, use a fixture or other fixing device to fix the workpiece at the detection position to prevent movement or deformation during the detection process, set the scanning mode and scanning parameters, such as scanning speed, scanning depth, and scanning interval, and use the path planning function of the scanning software or the device itself to generate the final scanning path according to the set scanning parameters and the selected scanning mode.

[0052] It should be noted that according to the shape and size of the workpiece, select a suitable scanning mode, including grid scanning, spiral scanning, linear scanning, etc. Grid scanning is suitable for workpieces with regular shapes and flat surfaces, spiral scanning is more suitable for cylindrical or conical workpieces, and linear scanning is often used for long strip or ribbon workpieces;

[0053] Taking a rectangular workpiece as an example, when performing flatness detection on it, select the grid scanning mode, divide the surface of the workpiece into multiple small grids, each grid as a scanning point, set the grid size and scanning speed according to the size of the workpiece and the expected detection accuracy, use the scanning device to generate a scanning path, and perform point-by-point scanning on the surface of the workpiece according to the scanning path.

[0054] When determining the scanning point spacing, it is necessary to make a reasonable selection according to the actual situation. A smaller scanning point spacing will improve the detection accuracy, but also increase the detection time and cost. While a larger scanning point spacing will reduce the detection accuracy, but can reduce the detection time and cost;

[0055] Start the flatness detection device (such as a laser rangefinder, displacement sensor, etc.), move the detection probe along the surface of the workpiece, and perform point-by-point scanning on the surface of the workpiece according to the scanning path. At each scanning point, the detection probe measures the distance from the surface of the workpiece to the probe and records the flatness data of each scanning point;

[0056] It should be noted that flatness is the data difference between the surface of an object during processing or production and the absolute horizontal, which reflects the flatness of the object's surface. Specifically, it is evaluated by measuring the distances from various points on the surface of the object to a reference plane;

[0057] Traverse each scanning point, compare the flatness of each scanning point with the preset flatness threshold. If the flatness of the scanning point reaches the preset flatness threshold, record the coordinates of this scanning point. According to the recorded coordinates of the scanning points, perform pre-processing on the surface of the workpiece, such as grinding, polishing, etc., to remove the protruding parts, or perform other necessary pre-treatment operations until the flatness of each scanning point is lower than the preset flatness threshold. After the pre-processing, re-obtain the flatness data of each scanning point;

[0058] It should be noted that the setting of the flatness threshold includes: initially setting the flatness threshold, performing trial processing on a small batch of workpieces. During the trial processing, monitor the processing quality and tool life. If the processing quality does not meet the requirements or the tool life is too short, adjust the flatness threshold to ensure improving the processing quality and extending the tool life. After multiple trial processing and adjustments, determine the final flatness threshold;

[0059] Start the hardness detection device (such as a hardness tester, indentation instrument, etc.), move the detection probe along the surface of the workpiece, and perform point-by-point scanning on the surface of the workpiece according to the scanning path. At each scanning point, press the test head into the surface of the workpiece to a predetermined depth or apply a predetermined test force, and record the hardness data of each scanning point;

[0060] Summarize the flatness data and hardness data and construct a scanning data set. The scanning data set includes the coordinates, flatness values, and hardness values of each scanning point;

[0061] By using a flatness detection device and a hardness detection device to scan the workpiece surface point by point, the flatness and hardness data of each scan point can be accurately obtained. Through preprocessing, the high points and uneven parts on the workpiece surface can be removed, thereby reducing tool wear during the cutting process. A flat surface helps to reduce the friction and vibration between the tool and the workpiece, and further extends the service life of the tool.

[0062] The region division module calculates the local flatness density and local hardness density of each scan point within its surrounding neighborhood. Based on the local flatness density and local hardness density of the scan point within its surrounding neighborhood, the local density of the scan point is calculated. According to the local density values of each scan point, the workpiece surface is divided into several independently controlled partitions by using the K-means++ clustering algorithm.

[0063] Preprocess the flatness data and hardness data in the scan dataset, such as removing outliers and filling in missing values, and then normalize the preprocessed flatness data and hardness data, mapping the flatness data and hardness data to the interval [0, 1]: , where represents the normalized data, represents the original data, represents the minimum value, represents the maximum value;

[0064] For each scan point, a square window with a side length of 2w + 1 centered on the scan point is used as the surrounding neighborhood of the scan point, where w is the window width. Obtain the flatness and hardness of each scan point, and calculate the local flatness density and local hardness density of each scan point within its surrounding neighborhood. The calculation formulas are as follows:

[0065] ;

[0066] where and represent the local flatness density and local hardness density respectively, and represent the flatness of the i-th scan point and the j-th scan point respectively, and represent the hardness of the i-th scan point and the j-th scan point respectively, represents the set of scan points within the surrounding neighborhood of scan point i, and σ is the width parameter of the Gaussian kernel.

[0067] It should be noted that the window size (i.e., the side length or radius of the square window) should be determined according to the size, shape, and expected detection accuracy of the workpiece. A larger window can capture more extensive spatial features but may reduce the resolution of local features, while a smaller window can capture finer local features but may be affected by noise;

[0068] Calculate the local density of each scan point based on the local density of flatness and the local density of hardness in its surrounding neighborhood: , where, represents the local density, and represent the weights of flatness and hardness respectively, and ;

[0069] Among them, the weight of flatness: , the weight of hardness:

[0070] , where, , and N is the number of all scan points.

[0071] According to the local density values of each scan point, divide the surface of the workpiece into several independently controlled partitions through the K-means++ clustering algorithm, specifically including:

[0072] S1: Initialize the clustering centers using the K-means++ algorithm, including:

[0073] S11: Randomly select a scan point as the first clustering center;

[0074] S12: For each scan point, calculate the Euclidean distance from this scan point to the nearest clustering center: ,

[0075] where, represents the Euclidean distance from the i-th scan point to the k-th clustering center, represents the local density of the i-th scan point, represents the local density of the k-th clustering center;

[0076] S13: For each scan point, calculate the square of the Euclidean distance from this scan point to the nearest clustering center as the weight for being selected as the next clustering center, and select the next clustering center according to the weights of each scan point through the roulette wheel selection method;

[0077] S14: Repeat S12~S13 until k clustering centers are selected;

[0078] It should be noted that the k value (i.e., the number of clustering categories) in the K-means++ algorithm can be determined by the elbow method. The elbow method is to draw the sum of squared errors (SSE) curve corresponding to different k values. As the k value increases, the SSE usually gradually decreases, but after a certain k value, the decrease amplitude of the SSE will significantly slow down. This point is the so-called "elbow". The optimal k value is determined by observing the position of the "elbow" of the curve;

[0079] S2: For each scanning point, calculate the Euclidean distance from this scanning point to each clustering center, and assign each scanning point to the category to which the nearest clustering center belongs;

[0080] S3: For each category, recalculate the clustering center of this category. The new clustering center is the average value of the local density values of all scanning points in this category;

[0081] S4: Repeat S2~S3 until the clustering center no longer changes (i.e., converges), or the preset number of iterations is reached;

[0082] S5: Obtain k clustering centers and the clustering category to which each scanning point belongs according to the final clustering result. Connect adjacent scanning points belonging to the same cluster into continuous regions to form several independently controlled partitions;

[0083] It should be noted that since the clustering algorithm is based on local density values and does not directly consider the physical positions of scanning points, the partitions divided may not be continuous. Therefore, it is necessary to judge whether the scanning points of the same cluster are adjacent according to the physical positions of the scanning points, so as to perform partitioning;

[0084] By calculating the local density of each scanning point and clustering according to the local density, the workpiece surface can be divided into regions with similar characteristics, and differential processing is performed on each region, thereby improving the overall processing accuracy. The workpiece surface characteristics (such as flatness and hardness) in different regions may be different, so the required tools and processing parameters may also be different. Through partition control, appropriate tools and processing parameters can be selected for each region, thereby prolonging the service life of the tool.

[0085] The cutting parameter optimization module obtains the historical processing data of workpieces with different flatness and hardness under various cutting parameter combinations, uses the random forest algorithm to establish a tool life model, takes the flatness and hardness of the workpiece as inputs, and outputs the optimal cutting parameter combination corresponding to each flatness and hardness interval;

[0086] Obtain historical machining data of workpieces with different flatness and hardness under various combinations of cutting parameters from the historical database. The data includes the flatness and hardness of the workpieces, cutting parameters (such as cutting speed, feed rate, cutting depth, etc.), and the corresponding tool life.

[0087] Take the flatness and hardness of the workpiece as feature variables, the tool life as the target variable, and the cutting parameters as optimization variables. Use the random forest algorithm to establish a tool life model, and use the historical machining data to train the tool life model to establish the mapping relationship between flatness and hardness and tool life.

[0088] Within the preset range of cutting parameters, traverse all possible combinations of cutting parameters through grid search. Take the flatness and hardness of the workpiece as the input, and use the trained tool life model to predict the tool life under each combination of cutting parameters. Calculate the mean squared error MSE between the predicted tool life and the actual value under each combination of cutting parameters.

[0089] Preset an error threshold in advance, screen out the combinations of cutting parameters with the mean squared error MSE less than the error threshold, sort the screened combinations of cutting parameters, and select the combination of cutting parameters with the longest tool life as the optimal combination of cutting parameters. Output the optimal combination of cutting parameters corresponding to each flatness and hardness interval.

[0090] It should be noted that a suitable range of error thresholds is determined through experiments or experience, and further adjustment and optimization are carried out within this range. For example, first set an error threshold, screen out a certain number of combinations of cutting parameters for actual machining tests, and gradually adjust the error threshold according to the test results and feedback until the best combination that meets the machining requirements and prediction accuracy is found.

[0091] By obtaining the historical machining data of workpieces with different flatness and hardness under various combinations of cutting parameters and establishing a tool life model, it is possible to select the optimal combination of cutting parameters according to the specific characteristics of the workpiece surface, thereby optimizing the machining process, selecting the combination of cutting parameters with the least tool wear, and effectively extending the service life of the tool.

[0092] Partition control module. For each partition on the workpiece surface, obtain the flatness data and hardness data in the scan dataset, calculate the mean flatness and mean hardness of each partition, and determine the optimal combination of cutting parameters according to the mean flatness and mean hardness, and perform independent control on each partition.

[0093] For each partition on the workpiece surface, obtain the flatness data and hardness data in the scan dataset, and calculate the mean flatness and mean hardness of each partition.

[0094] For each partition on the workpiece surface, based on the mean flatness and mean hardness, by looking up the flatness and hardness intervals in the output of the tool life model that are closest to the mean flatness and mean hardness of this partition, select the optimal cutting parameter combination corresponding to this interval;

[0095] During the production line machining process, according to the selected optimal cutting parameter combination, adjust parameters such as cutting speed, feed rate, and cutting depth, and independently control each partition;

[0096] By calculating the mean flatness and mean hardness of each partition and determining the optimal cutting parameter combination based on these means, it can be ensured that each partition is machined using the cutting parameters most suitable for its characteristics, fully considering the differences on the workpiece surface, thereby significantly improving the machining accuracy and reducing the errors caused by inappropriate cutting parameters. By considering the influence of flatness and hardness on tool life, the optimal cutting parameter combination selected for each partition can maximize the tool life.

[0097] Please refer to Figure 3 This invention also provides an intelligent monitoring method based on production line machining, including the following steps:

[0098] Step 1: Use a flatness detection device and a hardness detection device to scan the workpiece surface, obtain the flatness data and hardness data of each scan point, summarize and construct a scan data set;

[0099] Step 2: For each scan point, calculate the local flatness density and local hardness density of this scan point within the surrounding neighborhood. Through the local flatness density and local hardness density of this scan point within the surrounding neighborhood, calculate the local density of this scan point. According to the local density values of each scan point, divide the workpiece surface into several independently controlled partitions by the K-means++ clustering algorithm;

[0100] Step 3: Obtain the historical machining data of workpieces with different flatness and hardness under various cutting parameter combinations, use the random forest algorithm to establish a tool life model, take the flatness and hardness of the workpiece as inputs, and output the optimal cutting parameter combination corresponding to each flatness and hardness interval;

[0101] Step 4: For each partition on the workpiece surface, obtain the flatness data and hardness data in the scan data set, calculate the mean flatness and mean hardness of each partition, determine the optimal cutting parameter combination according to the mean flatness and mean hardness, and independently control each partition;

[0102] In the application, several formulas involved are calculated by taking their numerical values after dimensionlessization. The formula is obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation, and the coefficients in the formula are set by those skilled in the art according to the actual situation.

[0103] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by the combination of electronic hardware, computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution.

[0104] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, and may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0105] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all of them should be covered by the protection scope of this application.

Claims

1. An intelligent monitoring system based on production line processing, characterized in that: Including: A workpiece scanning module, which uses a flatness detection device and a hardness detection device to scan the surface of the workpiece, obtains the flatness data and hardness data of each scanning point, summarizes and constructs a scanning data set; The region division module, for each scanning point, takes a square window with a side length of 2 w +1 centered at the scanning point as the surrounding neighborhood of the scanning point, where w is the window width. By the flatness and hardness of each scanning point, calculate the local flatness density and local hardness density of each scanning point within the surrounding neighborhood. The calculation formula is as follows: ； Among them, and respectively represent the local density of flatness and the local density of hardness, and respectively represent the flatness of the i th scanning point and the j th scanning point, and respectively represent the hardness of the i th scanning point and the j th scanning point, represents the set of scanning points in the surrounding neighborhood of the scanning point i , σ is the width parameter of the Gaussian kernel; Calculate the local density of the scan point based on the local density of flatness and the local density of hardness in the surrounding neighborhood of the scan point: , where represents the local density, and represent the weights of flatness and hardness respectively, and ; Among them, the weight of flatness: , the weight of hardness: , where , N is the number of all scanning points; According to the local density value of each scanning point, the surface of the workpiece is divided into several independently controlled partitions by the K-means++ clustering algorithm; A cutting parameter optimization module, which obtains the historical processing data of workpieces with different flatness and hardness under various cutting parameter combinations, uses the random forest algorithm to establish a tool life model, takes the flatness and hardness of the workpiece as inputs, and outputs the optimal cutting parameter combination corresponding to each flatness and hardness interval; A partition control module, for each partition on the surface of the workpiece, obtains the flatness data and hardness data in the scanning data set, calculates the average flatness and average hardness of each partition, determines the optimal cutting parameter combination according to the average flatness and average hardness, and independently controls each partition.

2. The intelligent monitoring system based on production line processing according to claim 1, characterized in that: Traverse each scanning point, compare the flatness of each scanning point with a preset flatness threshold. If the flatness of the scanning point reaches the preset flatness threshold, record the coordinates of the scanning point, and perform pre-processing on the surface of the workpiece according to the recorded coordinates of the scanning point until the flatness of each scanning point is lower than the preset flatness threshold. After the pre-processing, re-obtain the flatness data of each scanning point.

3. The intelligent monitoring system based on production line processing according to claim 1, wherein: Dividing the surface of the workpiece into several independently controlled partitions by the K-means++ clustering algorithm specifically includes: S1: Initialize the clustering centers using the K-means++ algorithm; S2: For each scanning point, calculate the Euclidean distance from the scanning point to each clustering center, and assign each scanning point to the category to which the nearest clustering center belongs; S3: For each category, recalculate the clustering center of the category, and the new clustering center is the average value of the local density values of all scanning points in the category; S4: Repeat S2~S3 until the clustering centers no longer change or reach the preset number of iterations; S5: Obtain according to the final clustering result k cluster centers and the cluster categories to which each scan point belongs, connect adjacent scan points belonging to the same cluster into continuous regions, and form several independently controlled partitions.

4. The intelligent monitoring system based on production line processing according to claim 1, wherein: Initializing the clustering centers using the K-means++ algorithm includes: S11: Randomly select a scanning point as the first clustering center; S12: For each scan point, calculate the Euclidean distance from the scan point to the nearest cluster center: , where represents the Euclidean distance from the i -th scan point to the k -th cluster center, represents the local density of the i-th scan point, represents the local density of the k -th cluster center; S13: For each scanning point, calculate the square of the Euclidean distance from the scanning point to the nearest clustering center as the weight for being selected as the next clustering center, and select the next clustering center by the roulette wheel selection method according to the weight of each scanning point; S14: Repeat S12~S13 until k cluster centers are selected.

5. The intelligent monitoring system based on production line processing according to claim 1, characterized in that: Outputting the optimal cutting parameter combination corresponding to each flatness and hardness interval specifically includes: Taking the flatness and hardness of the workpiece as feature variables, the tool life as the target variable, and the cutting parameters as optimization variables, using the random forest algorithm to establish a tool life model, training the tool life model with historical processing data, and establishing a mapping relationship between flatness and hardness and tool life.

6. The intelligent monitoring system based on production line processing according to claim 5, characterized in that: Within the preset cutting parameter range, all possible combinations of cutting parameters are traversed through grid search. Taking the flatness and hardness of the workpiece as inputs, the trained tool life model is used to predict the tool life under each combination of cutting parameters, and the mean square error MSE between the predicted tool life and the actual value under each combination of cutting parameters is calculated. An error threshold is preset, and the combinations of cutting parameters with the mean square error MSE less than the error threshold are screened out. The screened combinations of cutting parameters are sorted, and the combination of cutting parameters with the longest tool life is selected as the optimal combination of cutting parameters, and the optimal combination of cutting parameters corresponding to each flatness and hardness interval is output.

7. The intelligent monitoring system based on production line processing according to claim 1, wherein: For each partition on the surface of the workpiece, according to the mean flatness and mean hardness, by looking up the flatness and hardness intervals in the output of the tool life model that are closest to the mean flatness and mean hardness of this partition, the optimal combination of cutting parameters corresponding to this interval is selected.

8. An intelligent monitoring method based on production line processing, using the system according to any one of claims 1 to 7, characterized in that: It includes the following steps: Step 1: Use a flatness detection device and a hardness detection device to scan the surface of the workpiece, obtain the flatness data and hardness data of each scan point, summarize and construct a scan data set. Step 2: For each scanning point, take this scanning point as the center, and a square window with a side length of 2 w +1 as the surrounding neighborhood of this scanning point, where, w is the window width. Calculate the flatness local density and hardness local density of each scanning point within the surrounding neighborhood based on the flatness and hardness of each scanning point. The calculation formulas are as follows: ； Among them, and respectively represent the local density of flatness and the local density of hardness, and respectively represent the flatness of the i th scanning point and the j th scanning point, and respectively represent the hardness of the i th scanning point and the j th scanning point, represents the set of scanning points of the scanning point i in the surrounding neighborhood, σ is the width parameter of the Gaussian kernel; Calculate the local density of the scan point based on the local density of flatness and the local density of hardness in the surrounding neighborhood of the scan point: , where represents the local density, and represent the weights of flatness and hardness respectively, and ; Among them, the weight of flatness: , the weight of hardness: , where , N is the number of all scanned points; According to the local density value of each scan point, the surface of the workpiece is divided into several independently controlled partitions by the K-means++ clustering algorithm. Step 3: Obtain the historical processing data of workpieces with different flatness and hardness under various combinations of cutting parameters, use the random forest algorithm to establish a tool life model, take the flatness and hardness of the workpiece as inputs, and output the optimal combination of cutting parameters corresponding to each flatness and hardness interval. Step 4: For each partition on the surface of the workpiece, obtain the flatness data and hardness data in the scan data set, calculate the mean flatness and mean hardness of each partition, determine the optimal combination of cutting parameters according to the mean flatness and mean hardness, and perform independent control on each partition.

Images