Production process method for improving chip removal of PCB micro drill

Abstract

The invention relates to the technical field of machining tools, and particularly discloses a production process method for improving PCB micro-drill chip removal, and the method comprises the steps: S1, drill bit pretreatment; s2, designing a plurality of sections of chip grooves; s3, hole drilling parameters are preset, hole drilling experiments are conducted at different hole drilling speeds and feeding speeds, and the hole drilling parameters are determined by detecting the hole diameter, the hole depth and the hole thickness; s4, establishing a drilling process monitoring model based on deep learning; and S5, after drilling is completed, surface cleaning is conducted on the PCB, residual cuttings and cooling liquid are removed, the hole diameter, the hole depth and the hole thickness are detected, data analysis and quality evaluation are conducted on the detection result, and unqualified drilled holes are automatically marked and repaired or re-drilled. Through precise drilling parameter setting, chip groove design and optimization adjustment of a deep learning model, the drilling precision is remarkably improved, the deviation of the hole diameter, the hole depth and the hole thickness is reduced, and high-precision PCB manufacturing is ensured.

Description

Technical Field

[0001] The invention relates to the technical field of machining tools, and in particular to a production process method for improving chip removal of PCB micro-drills. Background Art

[0002] The chip removal problem of PCB micro-drilling during the drilling process has become a key factor affecting processing efficiency and precision. Although the existing PCB micro-drilling technology has made many improvements in structural design, there are still problems such as poor chip removal, high drill breakage rate and poor hole roughness performance in practical applications, which affect the quality and efficiency of PCB drilling. After searching, a PCB micro-drill and its processing method with a publication number of CN102922007B were disclosed, and the publication date was August 24, 2016. The patent designs multiple chip grooves and gradually increases the helix angle to improve the arrangement density and chip holding capacity of the chip grooves. However, this technical solution mainly relies on the structural optimization of the chip grooves, and fails to make systematic improvements to the chip removal process from the perspective of production technology. In addition, although the change in the helix angle of the chip groove increases the chip removal capacity, it may cause the drill bit rigidity to decrease in actual processing and increase the risk of drill breakage. Summary of the invention

[0003] In order to overcome the above technical problems existing in the prior art, an embodiment of the present invention provides a production process method for improving PCB micro-drill chip removal, the method comprising the following steps:

[0004] S1: drill bit pretreatment, heat treatment and surface chemical treatment of the drill bit substrate;

[0005] S2: Design multiple chip flutes, with the helix angle of each chip flute gradually increasing, and use high-precision CNC machine tools to process the chip flutes;

[0006] S3: Preset drilling parameters, use different drilling speeds and feed speeds to conduct drilling experiments, and determine drilling parameters by testing hole diameter, hole depth and hole roughness. Drilling parameters include drilling speed, feed speed and drilling depth. The drilling speed is adjusted according to the drill bit diameter and material, the feed speed is adjusted according to the hardness and thickness of the PCB material, and the drilling depth is adjusted according to the number of PCB layers and design requirements.

[0007] S4: Establish a drilling process monitoring model based on deep learning. By collecting real-time data during the drilling process, including drilling speed, feed speed, drill temperature and cutting force, the deep neural network model is used to analyze the collected real-time data. According to the analysis results of the deep learning model, the drilling parameters are automatically adjusted to optimize the drilling process.

[0008] S5: After drilling is completed, the surface of the PCB board is cleaned to remove residual chips and coolant, and the hole diameter, hole depth and hole roughness are detected. The test results are analyzed and evaluated for quality. For unqualified drilling holes, they are automatically marked and repaired or re-drilled.

[0009] Preferably, in S1, the following steps are specifically included:

[0010] S1.1: The heat treatment process specifically includes quenching and tempering. The quenching temperature is 1050℃, the quenching time is 1 hour, the tempering temperature is 620℃, and the tempering time is 2 hours.

[0011] S1.2: The drill bit substrate is surface treated by a chemical nickel plating process to form a uniform nickel coating. The chemical nickel plating process specifically includes a plating solution formula of 50g / L NiSO4, 15g / L NiCl2, 30g / L NaH2PO2, 20g / L H3BO3, and 10g / L C6H8O7. The plating solution temperature is 90°C, the plating solution pH value is 4.8, and the coating thickness is 5μm.

[0012] Preferably, in S2, the following steps are specifically included:

[0013] S2.1: The drill bit substrate is designed with multiple chip grooves, specifically, the helix angle of the first chip groove gradually increases from 15° to 25°, the helix angle of the second chip groove gradually increases from 25° to 35°, and the helix angle of the third chip groove gradually increases from 35° to 45°. The depth of the chip groove is 0.2-0.3mm and the width is 0.5-0.7mm.

[0014] S2.2: A coolant circulation system is used during the machining process. The specific parameters are that the machining accuracy of the CNC machine tool is 0.005 mm and the surface finish of the drill substrate is 0.4 μm.

[0015] Preferably, in S4, the following steps are specifically included:

[0016] S4.1: The collected real-time data include drilling speed, feed speed, drill temperature and cutting force. After preprocessing these real-time data, feature extraction and time series analysis are performed on the processed real-time data through the combination of convolutional neural network (CNN) and long short-term memory network (LSTM);

[0017] S4.2: According to the analysis results obtained in S4.1, the drilling parameters of S3 are automatically adjusted to achieve the goal of automatically reducing the drilling speed or increasing the coolant flow rate when the model detects an abnormal increase in cutting force;

[0018] S4.3: Establish a prediction model to predict the wear degree and life of the drill bit, and predict the replacement time of the drill bit by real-time monitoring of the drill bit usage.

[0019] Preferably, in S4.1, the collected drilling speed, feed speed, drill bit temperature and cutting force are preprocessed, including normalization processing and missing value filling. The normalization processing adopts the Min-Max normalization method, and the specific formula is:

[0020]

[0021] Wherein, x′ is the normalized data, x is the original data, min(x) is the minimum value of the data, min(x) is the maximum value of the data, and the original data includes drilling speed, feed speed, drill temperature and cutting force.

[0022] Preferably, the preprocessed real-time data is input into a convolutional neural network (CNN) for feature extraction. The structure of the CNN includes a convolutional layer, a pooling layer and a fully connected layer. The activation function of the convolutional layer adopts ReLU, the pooling layer adopts maximum pooling, and the activation function of the fully connected layer adopts tanh.

[0023] Next, the features extracted by CNN are input into the long short-term memory network (LSTM) for time series analysis. The structure of LSTM includes input gate, forget gate and output gate. The specific formula is as follows:

[0024] Input Gate:

[0025] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0026] Forget Gate:

[0027] f t =σ(W f ·[h t-1 ,x t ]+b f );

[0028] Cell Status:

[0029] c t =f t ·c t-1 +i t tanh(W c ·[h t-1 ,x t ]+b c );

[0030] Output Gate:

[0031] o t =σ(Wo ·[h t-1 ,x t ]+b o );

[0032] Hidden state:

[0033] h t =o t ·tanh(c t );

[0034] Among them, i t 、f t , c t , o t ,h t are the input gate, forget gate, output gate and hidden state respectively, σ is the Sigmoid activation function, tanh is the hyperbolic tangent activation function, W i , W f , W c , and W o is the weight matrix, b i , b f , b c , and b o is the bias term, h t-1 is the hidden state of the previous time step, x t is the input for the current time step.

[0035] Preferably, in S4.3, the specific steps of predicting the wear degree and life of the drill bit are:

[0036] S4.3.1: Collect drill bit temperature, cutting force, and drilling time data during the drilling process;

[0037] S4.3.2: Use the ARIMA model to analyze the collected real-time data and predict the wear and life of the drill bit;

[0038] S4.3.3: When the prediction model detects that the wear of the drill bit exceeds 20%, an instruction to replace the drill bit is automatically issued;

[0039] S4.3.4: When the prediction model detects that the life of the drill bit is lower than the preset threshold, the drilling parameters are automatically adjusted to reduce the drilling speed and feed rate to extend the service life of the drill bit.

[0040] Preferably, the formula of the ARIMA model is:

[0041] W t =c+φ1W t-1 +φ2W t-2 +…+φ p W t-p +θ1∈ t-1+θ2∈ t-2 +…+θ q ∈ t-q +∈ t +β0+β1T t +β2F t +β3Life t ;

[0042] Among them, W t is the predicted value of the wear degree or life of the drill bit at time point t;

[0043] T t and F t are the actual values ​​of drill temperature and cutting force at time point t, respectively;

[0044] Life t is the cumulative working time of the drill bit;

[0045] c is a constant term, which represents the mean of the time series;

[0046] φ i is the coefficient of the autoregressive term, which indicates the influence of the wear degree or life of the drill bit at the past i time points on the current time point;

[0047] p is the order of the autoregressive term, that is, the number of past time points considered;

[0048] W t-i is the actual value of the wear degree or life of the drill bit at time point ti;

[0049] θ i is the coefficient of the moving average term, which represents the impact of the error term at the past i time points on the current time point;

[0050] q is the order of the moving average term, that is, the error term of how many past time points are considered;

[0051] ∈ t is the white noise error term at time point t, representing random fluctuations that cannot be explained by the model;

[0052] ∈ t-i White noise error term at time point ti;

[0053] β0 is the intercept term of the linear regression equation;

[0054] β1 The influence coefficient of drill bit temperature on the wear degree or life of the drill bit;

[0055] β2 is the influence coefficient of cutting force on the wear degree or life of the drill bit;

[0056] β3 is the influence coefficient of the drill working time on the wear degree or life of the drill.

[0057] Preferably, in S4.3, when predicting the wear degree and life of the drill bit, fault diagnosis is also included, and the specific steps are:

[0058] S4.3.5: Establish a fault diagnosis model based on support vector machine (SVM) to classify abnormal conditions of the drill bit;

[0059] S4.3.6: When the fault diagnosis model detects an abnormal state of the drill bit, it automatically issues an alarm and records the specific parameters of the abnormal state;

[0060] S4.3.7: Adjust drilling parameters or replace the drill bit based on the specific parameters of the recorded abnormal condition.

[0061] Preferably, in S5, the following steps are specifically included:

[0062] S5.1: After drilling is completed, use an ultrasonic cleaning machine to clean the surface of the PCB board. The ultrasonic cleaning machine sets multiple cleaning points on the surface of the PCB board to remove residual chips and coolant through ultrasonic vibration, so that the surface cleanliness of the PCB board reaches the micron level;

[0063] S5.2: Use high-precision optical testing equipment to test the hole diameter, hole depth and hole roughness, and perform data analysis and quality assessment on the test results;

[0064] S5.3: For unqualified drill holes, they are automatically marked and repaired or re-drilled. During the repair process, steps S3-S4 are repeated.

[0065] Through the technical solution provided by the present invention, the present invention has at least the following technical effects:

[0066] 1. Through precise drilling parameter setting, chip groove design and optimization adjustment of deep learning model, the drilling accuracy is significantly improved, the deviation of hole diameter, hole depth and hole roughness is reduced, and high-precision PCB manufacturing is ensured.

[0067] 2. The combination of heat treatment process and surface chemical nickel plating process in S1 significantly improves the hardness, wear resistance, heat dissipation performance and service life of the drill bit, thus playing a vital role in the entire PCB micro-drilling chip removal process and ensuring the efficiency, precision and quality of processing.

[0068] 3. The combination of multi-segment chip groove design and coolant circulation system in step S2 optimizes the chip removal effect, cutting performance and processing accuracy in PCB micro-drilling processing, which not only improves production efficiency and processing quality, but also effectively extends the service life of the drill bit, with significant economic benefits.

[0069] 4. The data analysis and automatic adjustment mechanism in step S4, combined with advanced artificial intelligence technology, further enhances the intelligent control capability of the PCB micro-drilling chip removal process, optimizes the use effect and life prediction of the drill bit, and ultimately achieves multiple beneficial effects of improving processing accuracy, production efficiency and reducing costs.

[0070] 5. The data preprocessing steps in S4.1 (including normalization and missing value filling) provide a solid foundation for subsequent intelligent analysis and automated adjustment, improve the intelligence level of the entire PCB micro-drilling chip removal production process method, and make the production process more efficient, stable and precise.

[0071] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0073] Figure 1 It is a specific flow chart of a production process method for improving PCB micro-drill chip removal provided by an embodiment of the present application;

[0074] Figure 2 A flowchart of S1 in a production process method for improving PCB micro-drilling chip removal provided in an embodiment of the present application;

[0075] Figure 3 A flowchart of S4 in a production process method for improving PCB micro-drill chip removal provided in an embodiment of the present application. DETAILED DESCRIPTION

[0076] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0077] The terms "system" and "network" in the embodiments of the present invention can be used interchangeably. "Multiple" means two or more than two. In view of this, "multiple" can also be understood as "at least two" in the embodiments of the present invention. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / ", unless otherwise specified, generally indicates that the previous and subsequent associated objects are in an "or" relationship. In addition, it should be understood that in the description of the embodiments of the present invention, the words "first", "second", etc. are only used to distinguish the purpose of description, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying order.

[0078] See also Figure 1-3 The embodiment of the present invention provides a production process method for improving PCB micro-drill chip removal, the method comprising the following steps:

[0079] S1: drill bit pretreatment, heat treatment and surface chemical treatment of the drill bit substrate;

[0080] S2: Design multiple chip flutes, with the helix angle of each chip flute gradually increasing, and use high-precision CNC machine tools to process the chip flutes;

[0081] S3: Preset drilling parameters, use different drilling speeds and feed speeds to conduct drilling experiments, and determine drilling parameters by testing hole diameter, hole depth and hole roughness. Drilling parameters include drilling speed, feed speed and drilling depth. The drilling speed is adjusted according to the drill bit diameter and material, the feed speed is adjusted according to the hardness and thickness of the PCB material, and the drilling depth is adjusted according to the number of PCB layers and design requirements.

[0082] S4: Establish a drilling process monitoring model based on deep learning. By collecting real-time data during the drilling process, including drilling speed, feed speed, drill temperature and cutting force, the deep neural network model is used to analyze the collected real-time data. According to the analysis results of the deep learning model, the drilling parameters are automatically adjusted to optimize the drilling process.

[0083] S5: After drilling is completed, the surface of the PCB board is cleaned to remove residual chips and coolant, and the hole diameter, hole depth and hole roughness are detected. The test results are analyzed and evaluated for quality. For unqualified drilling holes, they are automatically marked and repaired or re-drilled.

[0084] In one possible implementation, the pretreatment of the drill bit improves the hardness and wear resistance of the drill bit through heat treatment and surface chemical treatment, which provides a solid foundation for the subsequent drilling process. Heat treatment improves the hardness of the drill bit and reduces wear; chemical treatment (such as nickel plating) can reduce friction and enhance cutting performance, providing technical support for more efficient chip removal and precision drilling.

[0085] Furthermore, the chip groove design is the core of the entire process. The chip groove with a gradually increasing segmented helix angle can effectively guide the discharge of chips, avoiding the drill bit from getting stuck or wearing due to chip accumulation. The chip groove is processed by high-precision CNC machine tools to ensure the accuracy of the size and angle of the chip groove, further improving the chip removal efficiency.

[0086] Furthermore, before actual drilling, the most suitable drilling parameters, including drilling speed, feed speed and drilling depth, are determined through experiments to ensure that the drilling process is neither too fast to cause overheating of the drill bit nor too slow to affect production efficiency. The drilling parameters are optimized according to the hardness, number of layers and design requirements of the PCB material to improve the accuracy and consistency of drilling. Through precise drilling parameter setting, chip groove design and optimization adjustment of the deep learning model, the drilling accuracy is significantly improved, the deviation of hole diameter, hole depth and hole roughness is reduced, and high-precision PCB manufacturing is ensured.

[0087] Furthermore, the innovation of this method is to use deep learning technology to monitor and adjust the drilling process in real time. By analyzing the real-time data of the drilling process (such as drilling speed, feed speed, drill temperature, cutting force, etc.) through the deep neural network model, the drilling parameters can be adjusted dynamically, and abnormal drilling status can be corrected in time, avoiding quality problems and drill loss caused by improper parameters.

[0088] Furthermore, after drilling is completed, ultrasonic cleaning is used to remove chips and coolant to ensure that the PCB surface is clean, which is conducive to subsequent quality inspection. High-precision optical inspection equipment is used to check the hole diameter, hole depth and hole roughness, and unqualified holes are repaired or re-drilled to ensure that the final product meets the design requirements and quality standards.

[0089] The heat treatment and surface chemical treatment of the drill bit improve the hardness and wear resistance of the drill bit. In particular, the optimized chip groove design and effective application of coolant effectively reduce the wear of the drill bit and extend the service life of the drill bit. At the same time, the multi-stage helical angle chip groove design and high-precision processing ensure the smooth discharge of chips, avoiding overheating or damage of the drill bit caused by chip blockage, thereby further improving the stability and safety of processing.

[0090] In the embodiment of the present invention, in S1, the following steps are specifically included:

[0091] S1.1: The heat treatment process specifically includes quenching and tempering. The quenching temperature is 1050℃, the quenching time is 1 hour, the tempering temperature is 620℃, and the tempering time is 2 hours.

[0092] S1.2: The drill bit substrate is surface treated by a chemical nickel plating process to form a uniform nickel coating. The chemical nickel plating process specifically includes a plating solution formula of 50g / L NiSO4, 15g / L NiCl2, 30g / L NaH2PO2, 20g / L H3BO3, and 10g / L C6H8O7. The plating solution temperature is 90°C, the plating solution pH value is 4.8, and the coating thickness is 5μm.

[0093] In one possible implementation, the main purpose of heat treatment is to improve the hardness and wear resistance of the drill bit and reduce drill bit failure due to wear during the drilling process. Specifically, the quenching process is carried out at 1050°C. After 1 hour of high-temperature heating, rapid cooling forms a high-hardness martensitic structure, which greatly improves the wear resistance of the drill bit. The tempering process is carried out at 620°C for 2 hours. The purpose is to reduce the internal stress formed during the quenching process and moderately restore the toughness of the material to prevent the drill bit from brittle cracking during use. The combination of quenching and tempering improves the wear resistance of the drill bit and ensures that it can maintain a longer service life during high-speed and high-load drilling.

[0094] Through the heat treatment process (quenching and tempering) in S1.1, the hardness of the drill bit is significantly improved, which can effectively resist the friction and wear during PCB processing. The chemical nickel plating process in S1.2 provides additional protection for the drill bit. The low friction performance of the nickel layer greatly reduces the friction coefficient between the drill bit and the material, thereby reducing the wear of the drill bit. The combination of the two effectively improves the durability of the drill bit and extends the service life of the drill bit.

[0095] Furthermore, after the heat treatment is completed, the surface of the drill bit substrate is subjected to chemical nickel plating. The nickel layer has good corrosion resistance and low friction coefficient, which can effectively reduce the friction when the drill bit contacts the PCB material and reduce the wear of the drill bit. In addition, during the chemical nickel plating process, the formulation and temperature control of the plating solution (such as NiSO4, NiCl2, NaH2PO2, etc.) ensure that the coating is uniform and dense, and the coating thickness is 5μm, which can provide good surface protection and cutting performance. Through this surface treatment, the surface quality and cutting efficiency of the drill bit are significantly improved, which helps to reduce heat accumulation during the drilling process, thereby avoiding poor cutting or drill bit damage due to overheating.

[0096] In the embodiment of the present invention, in S2, the following steps are specifically included:

[0097] S2.1: The drill bit substrate is designed with multiple chip grooves, specifically, the helix angle of the first chip groove gradually increases from 15° to 25°, the helix angle of the second chip groove gradually increases from 25° to 35°, and the helix angle of the third chip groove gradually increases from 35° to 45°. The depth of the chip groove is 0.2-0.3mm and the width is 0.5-0.7mm.

[0098] S2.2: A coolant circulation system is used during the machining process. The specific parameters are that the machining accuracy of the CNC machine tool is 0.005 mm and the surface finish of the drill substrate is 0.4 μm.

[0099] In a possible implementation, in step S2.1, a multi-stage helix angle design is performed for the chip flute of the drill. This design divides the chip flute into three sections, and the helix angle of each section gradually increases, namely: the first section gradually increases from 15° to 25°, the second section gradually increases from 25° to 35°, and the third section gradually increases from 35° to 45°. Such a design can effectively guide and improve the chip discharge efficiency, avoid chip accumulation or jamming, while reducing the cutting resistance of the drill bit, reducing heat accumulation, and improving the service life and processing stability of the drill bit. The depth (0.2-0.3mm) and width (0.5-0.7mm) of the chip flute are set to further optimize the space of the chip removal channel, so that the chips can flow out smoothly, effectively improving the chip removal effect.

[0100] In step S2.2, a coolant circulation system is used to further improve the cooling effect and chip removal performance during the cutting process. The coolant can not only reduce the cutting temperature, but also take away the heat generated during the cutting process, reduce the overheating of the drill bit, and thus maintain the cutting stability of the drill bit. The efficient circulation system of the coolant ensures that the coolant can effectively flow to the drill bit and the processing part, maintain the processing accuracy (0.005mm), and provide sufficient lubrication to avoid excessive friction between the drill bit and the PCB surface. The flow of coolant also helps to remove chips from the drill hole and prevent chips from flowing back to the tip of the drill bit, further improving the chip removal effect.

[0101] In the embodiment of the present invention, in S4, the following steps are specifically included:

[0102] S4.1: The collected real-time data include drilling speed, feed speed, drill temperature and cutting force. After preprocessing these real-time data, feature extraction and time series analysis are performed on the processed real-time data through the combination of convolutional neural network (CNN) and long short-term memory network (LSTM);

[0103] S4.2: According to the analysis results obtained in S4.1, the drilling parameters of S3 are automatically adjusted to achieve the goal of automatically reducing the drilling speed or increasing the coolant flow rate when the model detects an abnormal increase in cutting force;

[0104] S4.3: Establish a prediction model to predict the wear degree and life of the drill bit, and predict the replacement time of the drill bit by real-time monitoring of the drill bit usage.

[0105] In one possible implementation, in step S4.1, key parameters of the drilling process are first collected in real time through sensors, including drilling speed, feed speed, drill bit temperature, and cutting force. These data are the basis for understanding the processing status, judging tool wear, and optimizing processing parameters. Then, by preprocessing the collected data, convolutional neural network (CNN) and long short-term memory network (LSTM) are used for feature extraction and time series analysis. CNN is good at extracting spatial features from raw data, and LSTM can effectively capture long-term dependencies when processing time series data, helping to analyze dynamic changes in the drilling process, thereby achieving more accurate anomaly detection and performance prediction.

[0106] In step S4.2, combined with the analysis results of S4.1, the system can automatically and intelligently adjust the drilling parameters by detecting abnormal increases in cutting force (such as overheating, overload, etc.). For example, when the cutting force is too large, the model will automatically reduce the drilling speed or increase the flow of coolant to avoid overheating and excessive wear of the drill bit, ensuring the stability and quality of the machining process. This automated adjustment mechanism not only improves machining efficiency, but also effectively reduces the need for human intervention, enhancing the automation and intelligence of the production process.

[0107] Furthermore, in step S4.3, based on the data analysis results in S4.1 and S4.2, a prediction model is established to monitor the wear and life of the drill bit. By real-time monitoring of the use status of the drill bit, the system can issue an advance warning when the drill bit is close to replacement, reducing the risk of reduced processing accuracy or processing failure due to excessive wear of the drill bit. The use of the prediction model helps to achieve accurate tool management, avoid excessive or premature tool replacement, thereby saving costs and optimizing production plans.

[0108] By combining CNN and LSTM to analyze the collected data, various dynamic changes in the processing process, such as changes in cutting force, temperature and other factors, can be reflected in real time and accurately. The system automatically adjusts the drilling parameters based on this information, thereby realizing intelligent control of the processing process. In particular, when the cutting force increases, the drilling speed or coolant flow rate is automatically adjusted, which can effectively avoid overheating and wear of the drill bit and maintain the stability and efficiency of the processing process.

[0109] In the embodiment of the present invention, in S4.1, the collected drilling speed, feed speed, drill bit temperature and cutting force are preprocessed, including normalization processing and missing value filling. The normalization processing adopts the Min-Max normalization method, and the specific formula is:

[0110]

[0111] Wherein, x′ is the normalized data, x is the original data, min(x) is the minimum value of the data, max(x) is the maximum value of the data, and the original data includes drilling speed, feed speed, drill temperature and cutting force.

[0112] In a possible implementation, in step S4.1, data collection mainly includes key information such as drilling speed, feed speed, drill temperature and cutting force. Since these data are usually obtained from different sensors or devices, their dimensions and numerical ranges may be different. In order to ensure that the subsequent analysis model can effectively process these data, the data must first be preprocessed to eliminate the scale differences between different data so that all input data are within a unified numerical range.

[0113] Furthermore, in the normalization process, the Min-Max standardization method is used. Through this method, the data value of each variable can be mapped to the interval [0,1]. In this way, all the collected raw data will be converted to the same standard range, avoiding analysis errors caused by different dimensions.

[0114] In addition, missing value filling is also an important part of data preprocessing. In the actual production process, sensor data may be missing due to various reasons. If these missing data are not processed, it will affect the accuracy and reliability of subsequent analysis models. Missing value filling is usually supplemented by interpolation or other intelligent filling algorithms to ensure data integrity.

[0115] After normalization and missing value filling, the data can more accurately reflect the real-time status of the drilling process, providing a more accurate basis for the system to intelligently adjust the drilling parameters. For example, after processing, the cutting force, temperature and other data can effectively help the system determine whether the drill bit is overheating, overloading and other problems, thereby achieving timely parameter adjustment and optimizing the processing process.

[0116] In an embodiment of the present invention, the preprocessed real-time data is input into a convolutional neural network (CNN) for feature extraction. The structure of the CNN includes a convolutional layer, a pooling layer and a fully connected layer. The activation function of the convolutional layer adopts ReLU, the pooling layer adopts maximum pooling, and the activation function of the fully connected layer adopts tanh.

[0117] Next, the features extracted by CNN are input into the long short-term memory network (LSTM) for time series analysis. The structure of LSTM includes input gate, forget gate and output gate. The specific formula is as follows:

[0118] Input Gate:

[0119] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0120] Forget Gate:

[0121] f t =σ(W f ·[h t-1 ,x t ]+b f );

[0122] Cell Status:

[0123] c t =f t ·c t-1 +i t tanh(W c ·[h t-1 ,x t ]+b c );

[0124] Output Gate:

[0125] o t =σ(W o ·[h t-1 ,x t ]+b o );

[0126] Hidden state:

[0127] h t =o t ·tanh(c t );

[0128] Among them, i t 、f t , c t , o t ,h t are the input gate, forget gate, output gate and hidden state respectively, σ is the Sigmoid activation function, tanh is the hyperbolic tangent activation function, W i , W f , W c , and W o is the weight matrix, b i , b f , b c , and b o is the bias term, h t-1 is the hidden state of the previous time step, x tis the input for the current time step.

[0129] In one possible implementation, first, feature extraction is performed on CNN. Specifically, preprocessed data (such as drilling speed, feed speed, temperature, cutting force, etc.) is input into a convolutional neural network (CNN) for feature extraction. The structure of CNN includes convolutional layers, pooling layers, and fully connected layers. Its main task is to automatically learn effective features that can describe the chip removal process from the input data.

[0130] In the convolution layer, the ReLU activation function is used to extract the local features of the input data through the convolution operation. The convolution operation performs a weighted summation of the input through the weight matrix and adds a bias, and then obtains the output through the ReLU activation function to ensure the learning of nonlinear features.

[0131] Subsequently, the pooling layer processes the output of the convolutional layer and uses the maximum pooling method to retain the maximum value of each local area, thereby reducing the dimension of the feature and enhancing the stability of the feature.

[0132] Next, the fully connected layer uses the tanh activation function to obtain the output features through weighting and biasing, and performs high-dimensional mapping on the features extracted by the convolutional layer and the pooling layer.

[0133] Furthermore, LSTM is used for time series analysis. Specifically, the features extracted by CNN are used as input and sent to the long short-term memory network (LSTM) for time series analysis. LSTM can effectively capture the time dependency and dynamic changes in the data and is an ideal model for processing data with time series characteristics.

[0134] LSTM transmits and updates information through the input gate, forget gate, cell state and output gate. The input gate determines the impact of the current input on the cell state, the forget gate controls the degree of forgetting of the information at the previous moment, and the cell state retains and updates long-term memory. Finally, the output gate determines the hidden state at the current moment, reflecting the model output at the current moment.

[0135] Furthermore, the preprocessed data is first subjected to CNN for feature extraction, and then the extracted features are input into LSTM for time series modeling. LSTM analyzes the time series dependency of data through the temporal recursive structure, and finally generates a prediction output for future drilling parameters or processing status.

[0136] The combination of CNN and LSTM not only improves the data processing capability in the chip removal process of PCB micro-drilling, but also optimizes the production process and improves production efficiency and processing accuracy through intelligent prediction and real-time adjustment.

[0137] In the embodiment of the present invention, in S4.3, the specific steps of predicting the wear degree and life of the drill bit are:

[0138] S4.3.1: Collect drill bit temperature, cutting force, and drilling time data during the drilling process;

[0139] S4.3.2: Use the ARIMA model to analyze the collected real-time data and predict the wear and life of the drill bit;

[0140] S4.3.3: When the prediction model detects that the wear of the drill bit exceeds 20%, an instruction to replace the drill bit is automatically issued;

[0141] S4.3.4: When the prediction model detects that the life of the drill bit is lower than the preset threshold, the drilling parameters are automatically adjusted to reduce the drilling speed and feed rate to extend the service life of the drill bit.

[0142] In one possible implementation, during the data collection phase of step S4.3.1, the key variables in the production process—drill temperature, cutting force, and drilling time—are collected in real time. These data are crucial for subsequent wear prediction, because the temperature and cutting force of the drill directly reflect its working state, while the drilling time can reflect the usage time of the drill. By accurately collecting these parameters, the model can analyze the health of the drill more comprehensively.

[0143] Furthermore, in the ARIMA model analysis stage, the ARIMA (Autoregressive Integrated Moving Average) model is used to analyze the collected time series data. The ARIMA model is suitable for processing time-dependent sequence data. By establishing a time series relationship between data, it can predict the wear degree and remaining service life of the drill bit. The ARIMA model will use the historical data of drill bit temperature, cutting force, and drilling time to identify trends and seasonal fluctuations, thereby effectively predicting future wear.

[0144] Furthermore, when the ARIMA model predicts that the wear of the drill bit exceeds 20%, the system will automatically issue an instruction to replace the drill bit. This prediction is not only based on the trend analysis of historical data, but also takes into account the potential risks in the use of the drill bit. It can provide timely warnings and take measures to avoid the decline in processing quality or equipment failure caused by excessive wear of the drill bit.

[0145] Furthermore, if the ARIMA model predicts that the life of the drill is approaching a preset threshold (i.e., the life of the drill will not be enough to complete the current task), the system will automatically adjust the drilling parameters, such as reducing the drilling speed and feed rate, to extend the life of the drill. By adjusting these parameters, the load on the drill is reduced, and the heat accumulation during the cutting process is reduced, thereby delaying the wear of the drill and ensuring the continuity of the production process.

[0146] In the embodiment of the present invention, the formula of the ARIMA model is:

[0147] W t =c+φ1W t-1 +φ2W t-2 +…+φ p W t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ q ∈ t-q +∈ t +β0+β1T t +β2F t +β3Life t ;

[0148] Among them, W t is the predicted value of the wear degree or life of the drill bit at time point t;

[0149] T t and F t are the actual values ​​of drill temperature and cutting force at time point t, respectively;

[0150] Life t is the cumulative working time of the drill bit;

[0151] c is a constant term, which represents the mean of the time series;

[0152] φ i is the coefficient of the autoregressive term, which indicates the influence of the wear degree or life of the drill bit at the past i time points on the current time point;

[0153] p is the order of the autoregressive term, that is, the number of past time points considered;

[0154] W t-i is the actual value of the wear degree or life of the drill bit at time point ti;

[0155] θ i is the coefficient of the moving average term, which represents the impact of the error term at the past i time points on the current time point;

[0156] q is the order of the moving average term, that is, the error term of how many past time points are considered;

[0157] ∈ t is the white noise error term at time point t, representing random fluctuations that cannot be explained by the model;

[0158] ∈ t-i White noise error term at time point ti;

[0159] β0 is the intercept term of the linear regression equation;

[0160] β1 The influence coefficient of drill bit temperature on the wear degree or life of the drill bit;

[0161] β2 is the influence coefficient of cutting force on the wear degree or life of the drill bit;

[0162] β3 is the influence coefficient of the drill working time on the wear degree or life of the drill.

[0163] In one possible implementation, during the production process, drill bit temperature and cutting force are two key parameters that can reflect the working state and wear of the drill bit. The data is collected continuously and forms a time series data set as the input of the ARIMA model. Through this data, the ARIMA model can use the past temperature and cutting force data to predict the working state and wear of the drill bit in the future.

[0164] Furthermore, the autoregressive part uses historical data to predict the current drill temperature or cutting force, assuming that there is a certain linear relationship between the current value and the values ​​at previous time points. Therefore, this part reflects the dependence between drill wear and the working state at previous moments, and can capture the trend of gradual changes in the drill during the production process.

[0165] Furthermore, the moving average part uses the error terms of the previous time points to correct the drill temperature or cutting force at the current moment. The error term represents the deviation of the model prediction and reflects the unpredictable random fluctuations in the system. The moving average part is used to reduce the impact of these random fluctuations on the model prediction results.

[0166] The error term is a random variable that represents the difference between the actual observed value and the value predicted by the model. This term helps the model adjust to factors that cannot be directly predicted from past data, such as sudden production process changes or external disturbances.

[0167] Through the combination of the above autoregressive and moving average parts, the ARIMA model can accurately predict the future trends of drill temperature and cutting force, thus providing a scientific basis for the optimization of production processes.

[0168] In the embodiment of the present invention, in S4.3, when predicting the wear degree and life of the drill bit, fault diagnosis is also included, and the specific steps are:

[0169] S4.3.5: Establish a fault diagnosis model based on support vector machine (SVM) to classify abnormal conditions of the drill bit;

[0170] S4.3.6: When the fault diagnosis model detects an abnormal state of the drill bit, it automatically issues an alarm and records the specific parameters of the abnormal state;

[0171] S4.3.7: Adjust drilling parameters or replace the drill bit based on the specific parameters of the recorded abnormal condition.

[0172] In one possible implementation, first, the system uses the working parameters of the drill bit (such as temperature, cutting force, etc.) and historical data to build a support vector machine model for real-time monitoring of the working status of the drill bit. SVM is an algorithm that can effectively process high-dimensional data and has strong classification capabilities, and can distinguish between normal and abnormal working conditions of the drill bit.

[0173] Furthermore, during the fault diagnosis process, the SVM model will classify the data collected in real time, and when it detects that the drill bit is in an abnormal state (such as excessive wear, high temperature, etc.), the model will issue an alarm. The alarm not only reminds the operator to pay attention to the possible abnormality of the drill bit, but also records the specific parameters of the abnormal state (such as temperature, instantaneous value of cutting force), providing a basis for subsequent adjustments.

[0174] Furthermore, once an abnormality occurs, the system will automatically adjust the drilling process parameters (such as feed speed, rotation speed, etc.) to slow down the wear of the drill bit according to the recorded specific parameters (such as abnormal temperature or cutting force value); or when the wear degree is too high, the system will automatically instruct to replace the drill bit. This link can realize real-time intelligent adjustment of the production process, avoid excessive wear of the drill bit, and ensure production efficiency and product quality.

[0175] The fault diagnosis model combined with SVM can effectively improve the accuracy of drill bit status monitoring. By real-time classification of the drill bit working status, potential fault hazards can be discovered in advance. Different from the traditional experience-based fault detection method, the SVM model can automatically classify abnormal conditions according to multi-dimensional features by learning historical data, reducing human omissions and ensuring timely detection of faults.

[0176] In the embodiment of the present invention, in S5, the following steps are specifically included:

[0177] S5.1: After drilling is completed, use an ultrasonic cleaning machine to clean the surface of the PCB board. The ultrasonic cleaning machine sets multiple cleaning points on the surface of the PCB board to remove residual chips and coolant through ultrasonic vibration, so that the surface cleanliness of the PCB board reaches the micron level;

[0178] S5.2: Use high-precision optical testing equipment to test the hole diameter, hole depth and hole roughness, and perform data analysis and quality assessment on the test results;

[0179] S5.3: For unqualified drill holes, they are automatically marked and repaired or re-drilled. During the repair process, steps S3-S4 are repeated.

[0180] In a possible implementation, after drilling is completed, an ultrasonic cleaning machine is first used to clean the surface of the PCB. The purpose of this step is to remove the chips and coolant remaining on the surface of the PCB to ensure that the board surface is free of contamination. By setting multiple cleaning points, ultrasonic cleaning can remove tiny particles that are difficult to remove at the micron level, avoiding the influence of residues on subsequent processes.

[0181] Ultrasonic cleaning can effectively clean fine residues while removing chips and coolant, ensuring the cleanliness of the PCB surface and preventing dirt from affecting subsequent inspections and quality assessments. This cleaning method is more efficient and accurate than traditional physical cleaning methods and can achieve micron-level cleanliness.

[0182] Furthermore, after cleaning, high-precision optical inspection equipment is used to perform precision inspection on the drilled holes. This step mainly detects important parameters such as the hole diameter, hole depth and hole roughness of the drilled holes. The eligibility of these parameters is directly related to the performance and reliability of the PCB, so they need to be evaluated through precise measurement methods. This link provides a data basis for subsequent quality judgment and repair.

[0183] High-precision optical inspection equipment can efficiently and accurately measure the diameter, depth and roughness of the drilled holes to ensure that each drilled hole meets the technical requirements. Through data analysis and quality assessment, the quality status of each drilled hole can be fully understood, providing a basis for subsequent repairs or adjustments. The automated inspection of this link not only improves the stability of product quality, but also reduces the error rate of manual inspection.

[0184] Furthermore, for unqualified holes detected, the system will automatically mark and start the repair or re-drilling process. If the hole diameter, hole depth or hole roughness does not meet the requirements, the production line will automatically return to the S3-S4 step to diagnose drill wear, adjust process parameters or re-drill to ensure that the quality of the final hole meets the standard requirements. Through automated processing, unqualified holes are repaired in a timely manner, avoiding errors and production delays caused by manual intervention.

[0185] For unqualified drilling, the automatic marking and repair mechanism reduces the need for manual intervention, shortens the repair time, and ensures the high quality of the final PCB by re-drilling and adjusting process parameters. In the automated repair process, the feedback control system is combined with the previous wear prediction model (such as the ARIMA and SVM models in the S3-S4 steps) to optimize the repair process and avoid recurrence of similar problems.

[0186] The optional implementation modes of the embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above implementation modes. Within the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical scheme of the embodiments of the present invention, and these simple modifications all belong to the protection scope of the embodiments of the present invention.

[0187] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe various possible combinations.

[0188] Those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions to enable a single-chip microcomputer, a chip or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0189] In addition, various implementation modes of the embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the embodiments of the present invention, they should also be regarded as the contents disclosed by the embodiments of the present invention.

Claims

1. A production process method for improving PCB micro-drilling chip removal, characterized in that: The method comprises: S1: drill bit pretreatment, heat treatment and surface chemical treatment of the drill bit substrate; S2: Design multiple chip flutes, with the helix angle of each chip flute gradually increasing, and use high-precision CNC machine tools to process the chip flutes; S3: Preset drilling parameters, use different drilling speeds and feed speeds to conduct drilling experiments, and determine drilling parameters by testing hole diameter, hole depth and hole roughness. Drilling parameters include drilling speed, feed speed and drilling depth. The drilling speed is adjusted according to the drill bit diameter and material, the feed speed is adjusted according to the hardness and thickness of the PCB material, and the drilling depth is adjusted according to the number of PCB layers and design requirements. S4: Establish a drilling process monitoring model based on deep learning. By collecting real-time data during the drilling process, including drilling speed, feed speed, drill temperature and cutting force, the deep neural network model is used to analyze the collected real-time data. According to the analysis results of the deep learning model, the drilling parameters are automatically adjusted to optimize the drilling process. S5: After drilling is completed, the surface of the PCB board is cleaned to remove residual chips and coolant, and the hole diameter, hole depth and hole roughness are detected. The test results are analyzed and evaluated for quality. For unqualified drilling holes, they are automatically marked and repaired or re-drilled.

2. A production process method for improving PCB micro-drilling chip removal according to claim 1, characterized in that: In S1, the following steps are specifically included: S1.1: The heat treatment process specifically includes quenching and tempering. The quenching temperature is 1050℃, the quenching time is 1 hour, the tempering temperature is 620℃, and the tempering time is 2 hours. S1.2: The drill bit substrate is surface treated by a chemical nickel plating process to form a uniform nickel coating. The chemical nickel plating process specifically includes a plating solution formula of 50g / L NiSO4, 15g / L NiCl2, 30g / L NaH2PO2, 20g / L H3BO3, and 10g / L C6H8O7. The plating solution temperature is 90°C, the plating solution pH value is 4.8, and the coating thickness is 5μm.

3. The production process method for improving PCB micro-drilling chip removal according to claim 1, characterized in that: In S2, the following steps are specifically included: S2.1: The drill bit substrate is designed with multiple chip grooves, specifically, the helix angle of the first chip groove gradually increases from 15° to 25°, the helix angle of the second chip groove gradually increases from 25° to 35°, and the helix angle of the third chip groove gradually increases from 35° to 45°. The depth of the chip groove is 0.2-0.3mm and the width is 0.5-0.7mm. S2.2: A coolant circulation system is used during the machining process. The specific parameters are that the machining accuracy of the CNC machine tool is 0.005 mm and the surface finish of the drill substrate is 0.4 μm.

4. The production process method for improving PCB micro-drilling chip removal according to claim 1, characterized in that: In S4, the following steps are specifically included: S4.1: The collected real-time data include drilling speed, feed speed, drill temperature and cutting force. After preprocessing these real-time data, feature extraction and time series analysis are performed on the processed real-time data through the combination of convolutional neural network (CNN) and long short-term memory network (LSTM); S4.2: According to the analysis results obtained in S4.1, the drilling parameters of S3 are automatically adjusted to achieve the goal of automatically reducing the drilling speed or increasing the coolant flow rate when the model detects an abnormal increase in cutting force; S4.3: Establish a prediction model to predict the wear degree and life of the drill bit, and predict the replacement time of the drill bit by real-time monitoring of the drill bit usage.

5. A production process method for improving PCB micro-drilling chip removal according to claim 4, characterized in that: In S4.1, the collected drilling speed, feed speed, drill temperature and cutting force are preprocessed, including normalization and missing value filling. The normalization method uses the Min-Max normalization method. The specific formula is: Wherein, x′ is the normalized data, x is the original data, min(x) is the minimum value of the data, max(x) is the maximum value of the data, and the original data includes drilling speed, feed speed, drill temperature and cutting force.

6. The production process method for improving PCB micro-drilling chip removal according to claim 1, characterized in that: The preprocessed real-time data is input into a convolutional neural network (CNN) for feature extraction. The structure of the CNN includes a convolutional layer, a pooling layer, and a fully connected layer. The activation function of the convolutional layer uses ReLU, the pooling layer uses maximum pooling, and the activation function of the fully connected layer uses tanh. Next, the features extracted by CNN are input into the long short-term memory network (LSTM) for time series analysis. The structure of LSTM includes input gate, forget gate and output gate. The specific formula is as follows: Input Gate: i t =σ(W i ·[h t-1 ,x t ]+b i ); Forget Gate: f t =σ(W f ·[h t-1 ,x t ]+b f ); Cell Status: c t =f t ·c t-1 +i t ·tanh(W c ·[h t-1 ,x t ]+b c ); Output Gate: the t =σ(W o ·[h t-1 ,x t ]+b o ); Hidden state: h t =o t ·tanh(c t ); Among them, i t 、f t , c t , o t ,h t are the input gate, forget gate, output gate and hidden state respectively, σ is the Sigmoid activation function, tanh is the hyperbolic tangent activation function, W i , W f , W c , and W o is the weight matrix, b i , b f , b c , and b o is the bias term, h t-1 is the hidden state of the previous time step, x t is the input for the current time step.

7. The production process method for improving chip removal of PCB micro-drilling according to claim 4, characterized in that: In S4.3, the specific steps for predicting the wear degree and life of the drill bit are: S4.3.1: Collect drill bit temperature, cutting force, and drilling time data during the drilling process; S4.3.2: Use the ARIMA model to analyze the collected real-time data and predict the wear and life of the drill bit; S4.3.3: When the prediction model detects that the wear of the drill bit exceeds 20%, an instruction to replace the drill bit is automatically issued; S4.3.4: When the prediction model detects that the life of the drill bit is lower than the preset threshold, the drilling parameters are automatically adjusted to reduce the drilling speed and feed rate to extend the service life of the drill bit.

8. The production process method for improving chip removal of PCB micro-drilling according to claim 7, characterized in that: The formula for the ARIMA model is: W t =c+φ1W t-1 +φ2W t-2 +…+φ p W t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ q ∈ t-q +∈ t +β0+β1T t +β2F t +β3Life t ; Among them, W t is the predicted value of the wear degree or life of the drill bit at time point t; T t and F t are the actual values ​​of drill temperature and cutting force at time point t, respectively; Life t is the cumulative working time of the drill bit; c is a constant term, which represents the mean of the time series; φ i is the coefficient of the autoregressive term, which indicates the influence of the wear degree or life of the drill bit at the past i time points on the current time point; p is the order of the autoregressive term, that is, the number of past time points considered; W t-i is the actual value of the wear degree or life of the drill bit at time point ti; θ i is the coefficient of the moving average term, which represents the impact of the error term at the past i time points on the current time point; q is the order of the moving average term, that is, the error term of how many past time points are considered; ∈ t is the white noise error term at time point t, representing random fluctuations that cannot be explained by the model; ∈ t-i White noise error term at time point ti; β0 is the intercept term of the linear regression equation; β1 The influence coefficient of drill bit temperature on the wear degree or life of the drill bit; β2 is the influence coefficient of cutting force on the wear degree or life of the drill bit; β3 is the influence coefficient of the drill working time on the wear degree or life of the drill.

9. The production process method for improving chip removal of PCB micro-drilling according to claim 4, characterized in that: In S4.3, the prediction of the wear degree and life of the drill bit also includes fault diagnosis. The specific steps are as follows: S4.3.5: Establish a fault diagnosis model based on support vector machine (SVM) to classify abnormal conditions of the drill bit; S4.3.6: When the fault diagnosis model detects an abnormal state of the drill bit, it automatically issues an alarm and records the specific parameters of the abnormal state; S4.3.7: Adjust drilling parameters or replace the drill bit based on the specific parameters of the recorded abnormal condition.

10. The production process method for improving chip removal of PCB micro-drilling according to claim 1, characterized in that: In S5, the following steps are specifically included: S5.1: After drilling is completed, use an ultrasonic cleaning machine to clean the surface of the PCB board. The ultrasonic cleaning machine sets multiple cleaning points on the surface of the PCB board to remove residual chips and coolant through ultrasonic vibration, so that the surface cleanliness of the PCB board reaches the micron level; S5.2: Use high-precision optical testing equipment to test the hole diameter, hole depth and hole roughness, and perform data analysis and quality assessment on the test results; S5.3: For unqualified drill holes, they are automatically marked and repaired or re-drilled. During the repair process, steps S3-S4 are repeated.

Images