A cutting system and cutting method capable of accurately positioning chip wafer

Abstract

The present invention relates to the field of chip wafer processing technology, and in particular to a cutting system and a cutting method thereof capable of accurately positioning a chip wafer, comprising a cutting machine main body and an image processing module, a sensor module, a motion control module, a focusing module and a data processing module installed on the cutting machine main body. The present invention enables the cutting system to accurately identify the position and edge of the chip through the coordinated work of the image processing module and the sensor module, thereby achieving high-precision positioning. The dynamic adjustment capabilities of the motion control module and the focusing module enable them to adapt to wafers of different thicknesses and different cutting requirements, further improving the cutting accuracy. Real-time analysis and optimization suggestions by the data processing module enable the system to quickly respond and adjust cutting parameters, reducing unnecessary downtime and waiting time. The real-time monitoring function of the sensor module can promptly discover and correct potential problems, thereby enhancing the stability and reliability of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of chip wafer processing, and in particular relates to a cutting system and a cutting method capable of accurately positioning a chip wafer. Background Art

[0002] In the semiconductor packaging process, chip wafers need to be cut. Wafer cutting mainly uses mechanical diamond cutting technology, which uses high-speed diamond blades to make precise cuts on the wafer. At the same time, deionized water is used to rinse the silicon chips generated by cutting to keep it clean and avoid damage.

[0003] For example, the patent application document application number is 202410516344.1, which discloses a cutting device for chip wafer processing, including a cutting tool, which is installed on a base through a moving frame, and the cutting tool has a laser body. The moving frame is used to support and position the cutting tool, and control the movement of the cutting tool in the horizontal and vertical directions. The cutting tool includes a frame assembled on the moving frame, and the frame is used to assemble the cutting tool on the moving frame, wherein the sensor is installed on the cutting tool, and the cutting tool also includes a focusing tube, which is threadedly installed in the middle of the bottom end of the laser body, and the focusing tube is hollow. The cutting equipment for chip wafer processing uses multiple laser beams and a cutting head with an adjustable focal length to achieve more precise and efficient cutting of wafers, which can improve cutting accuracy, processing speed and cutting quality.

[0004] However, when the above-mentioned cutting equipment uses multiple laser beams for cutting, it is necessary to ensure the synchronization and consistency between the laser beams. However, if there is a time delay or power difference between the laser beams, it will lead to inaccurate cutting trajectory, thereby affecting the cutting accuracy of the chip wafer. Therefore, we need to propose a cutting system and a cutting method that can accurately position the chip wafer, which can effectively improve the precise positioning of the chip wafer to improve the cutting accuracy of the wafer chip. Summary of the invention

[0005] In view of the above problems, the present invention provides a cutting system capable of accurately positioning a chip wafer, comprising a cutting machine body and an image processing module, a sensor module, a motion control module, a focusing module and a data processing module installed on the cutting machine body;

[0006] The cutting machine body comprises a cutting tool and a moving frame for supporting the cutting tool, the cutting tool is mounted with a laser, and the position of the laser is controlled by a focusing module;

[0007] The image processing module collects wafer surface images through a camera and uses an image processing algorithm to analyze them to identify chip positions and edges on the wafer;

[0008] The sensor module includes a position sensor and a distance sensor, and the sensor module is used to monitor the position of the cutting tool and the state of the wafer in real time;

[0009] The motion control module is used to control the motion of the cutting tool in the horizontal and vertical directions to ensure that the cutting tool can move accurately along a predetermined trajectory;

[0010] The focusing module is used to adjust the focal length of the laser to adapt to wafers of different thicknesses and different cutting requirements;

[0011] The data processing module is used to receive data from the image processing module, the sensor module, the motion control module and the focusing module, and perform comprehensive analysis and processing to optimize cutting parameters and improve cutting efficiency;

[0012] The motion control module accurately controls the motion trajectory and speed of the cutting tool according to the data provided by the image processing module and the sensor module, and cooperates with the focusing module to adjust the distance of the laser to adapt to wafers of different thicknesses.

[0013] Furthermore, the process of the image processing module collecting and identifying the chip position and edge on the wafer is as follows:

[0014] A1. Initialize the camera settings;

[0015] A2: The camera starts to collect image data of the wafer surface in real time and performs image preprocessing on the image data;

[0016] A3 divides the processed image into chip area and non-chip area;

[0017] A4, extracting the geometric features of the chip area to describe the chip shape and size;

[0018] A5. Analyze the texture features of the chip area to distinguish different types of chips or identify specific marks on the chip;

[0019] A6. Use a template image containing chip features to match the features in the image and calculate the similarity of each position to determine the specific position and direction of the chip. The similarity calculation formula is:

[0020] ,in, is the image to be processed, are two random feature vectors in the image to be processed, is the template image, are two random feature vectors in the template image;

[0021] A7, automatically learn image features through multi-layer convolution and pooling operations, and generate feature maps in the last layer. The feature maps are converted into a one-dimensional vector through the fully connected layer, and then input into the Softmax function to convert it into a probability distribution, and the feature maps are classified and recognized;

[0022] A8. Convert the chip position in the image coordinate system to the position in the actual physical coordinate system, and then transmit the positioning result to the motion control module.

[0023] Furthermore, in step A2, during image preprocessing, the image data is first subjected to convolution processing, and the formula used for the convolution processing is:

[0024] ,

[0025] ,

[0026] in, represents the input image, Indicates that the size of the input image is X*X, F is the convolution kernel size F*F, S is the stride, and P is the padding;

[0027] The processed data is then smoothed to remove noise and reduce random fluctuations. The smoothing formula is:

[0028] ,in is the standard deviation, and are two selected feature vectors in the input image;

[0029] Finally, edge enhancement calculation is performed on the smoothed data, and the calculation formula is as follows:

[0030] Horizontal direction: ,

[0031] Vertical direction: ,

[0032] According to the calculated and Calculate the total gradient magnitude for each pixel:

[0033] , to enhance the edge information in the image.

[0034] Furthermore, in step A7, the formula for multi-layer convolution and pooling operations is:

[0035] ,in is the size and sliding range of the convolution kernel, is the pixel position currently being processed, is the convolution kernel used to scan the input image and extract features, and b is the bias term used to adjust the output of the activation function;

[0036] The Softmax function formula is:

[0037] ,in, The first elements, To normalize the sum of the indices of all categories.

[0038] Furthermore, the process of the motion control module controlling the cutting tool to move in the horizontal and vertical directions is as follows:

[0039] B1. The receiving sensor module obtains real-time data and the chip position and edge information analyzed by the image processing module. The real-time data includes information provided by the position sensor and the distance sensor, so as to help the motion control module understand the current position of the cutting tool and the status of the wafer;

[0040] B2, based on the data received in B1, calculate the optimal moving trajectory of the cutting tool, and determine the cutting speed of the cutting tool according to the material properties and thickness of the wafer and the required cutting accuracy;

[0041] B3. After calculating the optimal moving trajectory and cutting speed, the cutting tool is controlled by the motor driver to move according to the calculated trajectory and speed;

[0042] B4. During the cutting process, the feedback information from the sensor module is continuously received, such as the actual position and speed of the tool and the state change of the wafer. By comparing the actual data with the preset value, the motion control module needs to determine whether the position and speed of the cutting tool need to be adjusted. There are two situations as follows;

[0043] B4.1. When the position and speed of the cutting tool need to be adjusted, the motion control module will immediately take measures to make adjustments;

[0044] B4.2. When there is no need to adjust the position and speed of the cutting tool, the motion control module will continue to repeat the operation of S4.

[0045] Furthermore, in step B2, when calculating the optimal movement trajectory of the cutting tool, a virtual wafer surface model is first constructed based on the chip position and edge information provided by the image processing module and the real-time data provided by the sensor module. Then, the Dijkstra algorithm is used to plan an optimal path from the starting point to the end point on the model. The optimal path usually avoids obstacles as much as possible and takes into account the kinematic limitations of the tool. After the path planning is completed, the path is smoothed by interpolation and fitting algorithms.

[0046] Furthermore, in step B2, when determining the cutting speed of the cutting tool, the motion control module uses a machine learning model to predict the optimal cutting speed under different conditions. The prediction process is as follows:

[0047] C1. Collect a large amount of historical data on wafer cutting;

[0048] C2, extract key features related to cutting speed from the collected data;

[0049] C3. Select appropriate machine learning models based on the nature of the problem and the characteristics of the data. Machine learning models include linear regression, support vector machine and neural network.

[0050] C4. Use the collected data to train the selected model. During the training process, the model parameters need to be continuously adjusted to optimize its performance. At the same time, in order to evaluate the generalization ability of the model, the data set needs to be divided into a training set and a test set. The test set should be used regularly for verification during the training process.

[0051] C5. When the model training is completed, the trained model predicts the optimal cutting speed under different conditions.

[0052] Furthermore, the process of focusing the laser by the focusing module is as follows:

[0053] D1. When the laser cutting device is powered on, the focusing module will receive a focusing return signal indicating that the focusing mirror needs to return to its absolute zero position;

[0054] D2. According to the received focus return zero signal, the focus module controls the focus mirror to move to the absolute zero position at a certain speed through the servo motor;

[0055] D3. After the focusing mirror returns to the absolute zero position, the focusing module will detect whether the servo motor is working normally. If the servo motor is in a normal state, the focusing module will be ready to receive subsequent focus adjustment signals; if the servo motor is abnormal, the focusing module will send a reset signal to reset the servo motor;

[0056] D4. When receiving the focus adjustment signal, the focus adjustment module reads the analog voltage input signal containing the target focus position;

[0057] D5. Performing preset signal processing on the read analog voltage signal, and determining the output frequency and pulse number of the pulse control signal for driving the servo motor according to the processed signal;

[0058] D6. According to the determined output frequency and pulse quantity, the focusing module outputs corresponding pulse control signals, and drives the servo motor through these signals to drive the focusing mirror to move within the preset motion range.

[0059] Furthermore, the process of the data processing module comprehensively analyzing and processing the received data is as follows:

[0060] E1, the data processing module integrates the chip position and edge information provided by the image processing module, the real-time data provided by the sensor module, the path planning and speed adjustment information of the motion control module, and the focus position information of the focusing module to build a system state model;

[0061] E2. Analyze the data in depth based on the system status model, calculate the relative position relationship between the chip and the cutting tool, evaluate whether the tool's motion trajectory conforms to the predetermined plan, and determine whether the current focus position is appropriate to confirm the current status of the system and potential problems;

[0062] E3. Make corresponding decisions based on the results of data analysis;

[0063] E4. After the decision is made, the data processing module will generate corresponding instructions and send them to the relevant execution agencies;

[0064] E5. During the execution of instructions, the data processing module will continuously monitor the operating status of each module and collect feedback information. The feedback information will be used to evaluate the execution effect of the instructions and the operating status of the system. If any problems or deviations are found, the data processing module will make timely adjustments to ensure the stable operation of the system.

[0065] E6. Based on long-term operating data and feedback information, the data processing module optimizes and improves the operating strategy of the entire system.

[0066] Based on the above-described cutting system capable of accurately positioning a chip wafer, the present invention further provides a cutting method capable of accurately positioning a chip wafer, comprising the following steps:

[0067] S1, the image processing module collects the wafer surface image through the camera, and uses the image processing algorithm to analyze the image, identify the chip position and edge on the wafer, and obtain the wafer status data;

[0068] S2, the position of the cutting tool is monitored in real time by the sensor module, and the motion control module controls the motion trajectory and speed of the cutting tool in combination with the wafer status data;

[0069] S3, the motion control module controls the movement of the cutting tool according to the cutting requirements and sends an instruction to the focusing module to adjust the focal length of the laser;

[0070] S4, the data processing module receives the recognition results of the image processing module, the monitoring results of the sensor module, the control parameters of the motion control module and the focal length information of the focusing module in real time, and conducts a comprehensive analysis to evaluate whether the current cutting state meets the accuracy requirements. There are two situations:

[0071] S4.1. If the current cutting state meets the accuracy requirements, the motion control module controls the cutting tool to perform the cutting operation;

[0072] S4.2. If the current cutting state does not meet the accuracy requirements, it is necessary to predict possible problems, make optimization adjustments based on the prediction results, and then repeat S4.

[0073] The beneficial effects of the present invention are:

[0074] 1. The present invention enables the cutting system to accurately identify the chip position and edge through the cooperation of the image processing module, the sensor module, the motion control module, the focusing module and the data processing module, thereby achieving high-precision positioning. The dynamic adjustment capability of the motion control module and the focusing module enables it to adapt to wafers of different thicknesses and different cutting requirements, further improving the cutting accuracy. The real-time analysis and optimization suggestions of the data processing module enable the system to quickly respond and adjust the cutting parameters, reducing unnecessary downtime and waiting time. The real-time monitoring function of the sensor module can promptly discover and correct potential problems, thereby enhancing the stability and reliability of the system.

[0075] 2. The present invention uses the image processing module and the sensor module to cooperate with each other, using the sensor module to provide real-time data, and the image processing module uses the data provided by the sensor to perform precise calculation and analysis to ensure that the position of each chip can be accurately identified.

[0076] 3. The present invention uses the cooperation between the sensor module and the motion control module to feed back the monitored data to the motion control module in real time, so as to adjust the position and speed of the cutting tool in time to ensure cutting accuracy.

[0077] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0079] Figure 1 A block diagram of a cutting system according to an embodiment of the present invention is shown;

[0080] Figure 2 A flowchart of collecting and identifying chip positions and edges on a wafer by an image processing module according to an embodiment of the present invention is shown;

[0081] Figure 3 A flow chart showing the motion control module controlling the cutting tool to move in the horizontal and vertical directions according to an embodiment of the present invention is shown;

[0082] Figure 4 A flowchart of a motion control module predicting an optimal cutting speed according to an embodiment of the present invention is shown;

[0083] Figure 5 A cutting flow chart according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0084] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0085] The embodiment of the present invention provides a cutting system capable of accurately positioning a chip wafer, such as Figure 1-5 As shown, it includes a cutting machine body and an image processing module, a sensor module, a motion control module, a focusing module and a data processing module installed on the cutting machine body. The cutting machine body includes a cutting tool and a moving frame for supporting the cutting tool. A laser is installed on the cutting tool, and the position of the laser is controlled by the focusing module. The specific structure of the cutting tool and the moving frame is the same as the cutting tool and the moving frame structure disclosed in the prior art (the patent document application number is 202410516344.1, disclosing a cutting device for chip wafer processing), which will not be described in detail here.

[0086] The image processing module collects wafer surface images through a camera and uses an image processing algorithm to analyze them to identify chip positions and edges on the wafer;

[0087] The process of the image processing module collecting and identifying the chip position and edge on the wafer is as follows:

[0088] A1. Initialize the camera settings. The camera is set to a high-definition camera capable of shooting at multiple angles. The initialization settings include adjusting the parameters of resolution and exposure time to ensure that the quality of the collected images meets the analysis requirements.

[0089] A2: The camera starts to collect image data of the wafer surface in real time and performs image preprocessing on the image data;

[0090] In image preprocessing, the image data is first convolved. The formula used for convolution is:

[0091] ,

[0092] ,

[0093] in, represents the input image, Indicates that the size of the input image is X*X, F is the convolution kernel size F*F, S is the stride, and P is the padding;

[0094] The processed data is then smoothed to remove noise and reduce random fluctuations. The smoothing formula is:

[0095] ,in is the standard deviation, and are two selected feature vectors in the input image;

[0096] Finally, edge enhancement calculation is performed on the smoothed data, and the calculation formula is as follows:

[0097] Horizontal direction: ,

[0098] Vertical direction: ,

[0099] According to the calculated and Calculate the total gradient magnitude for each pixel:

[0100] , to enhance the edge information in the image.

[0101] A3 divides the processed image into chip area and non-chip area;

[0102] A4. Extract the geometric features of the chip area to describe the shape and size of the chip, such as perimeter, area, major axis and minor axis;

[0103] A5. Analyze the texture features of the chip area to distinguish different types of chips or identify specific marks on the chip;

[0104] A6. Use a template image containing chip features to match the features in the image and calculate the similarity of each position to determine the specific position and direction of the chip. The similarity calculation formula is:

[0105] ,in, is the image to be processed, are two random feature vectors in the image to be processed, is the template image, are two random feature vectors in the template image;

[0106] A7, automatically learn image features through multi-layer convolution and pooling operations, and generate feature maps in the last layer. The feature maps are converted into a one-dimensional vector through the fully connected layer, and then input into the Softmax function to convert into a probability distribution. The feature maps are classified and identified to improve the accuracy and robustness of positioning;

[0107] The formula for multi-layer convolution and pooling operations is:

[0108] ,in is the size and sliding range of the convolution kernel, is the pixel position currently being processed, is the convolution kernel used to scan the input image and extract features, and b is the bias term used to adjust the output of the activation function;

[0109] The Softmax function formula is:

[0110] ,in, The first elements, To normalize the sum of indices for all categories;

[0111] A8. Convert the chip position in the image coordinate system to the position in the actual physical coordinate system so that the motion control module can perform precise cutting. Then transmit the positioning result to the motion control module as the basis for the movement of the cutting tool.

[0112] The sensor module includes a position sensor and a distance sensor, and the sensor module is used to monitor the position of the cutting tool and the wafer status in real time. The position sensor uses a light sensor to measure the light changes between the object and the light sensor, so as to know the position of the chip; the distance sensor determines the distance by emitting a light beam and receiving the reflected laser, calculating the time difference, and the position sensor and the distance sensor use common knowledge in the field, which will not be described in detail here.

[0113] Through the cooperation between the image processing module and the sensor module, the sensor module provides real-time data, and the image processing module uses the data provided by the sensor to perform precise calculations and analysis to ensure that the position of each chip can be accurately identified.

[0114] The motion control module is used to control the motion of the cutting tool in the horizontal and vertical directions, ensuring that the cutting tool can move accurately according to a predetermined trajectory.

[0115] The process of the motion control module controlling the cutting tool to move in the horizontal and vertical directions is as follows:

[0116] B1. The receiving sensor module obtains real-time data and the chip position and edge information analyzed by the image processing module. The real-time data includes information provided by the position sensor and the distance sensor, so as to help the motion control module understand the current position of the cutting tool and the status of the wafer;

[0117] B2, based on the data received in B1, calculate the optimal moving trajectory of the cutting tool, and determine the cutting speed of the cutting tool according to the material properties and thickness of the wafer and the required cutting accuracy;

[0118] B3. After calculating the optimal moving trajectory and cutting speed, the cutting tool is controlled by the motor driver to move according to the calculated trajectory and speed;

[0119] B4. During the cutting process, the feedback information from the sensor module is continuously received, such as the actual position and speed of the tool and the state change of the wafer. By comparing the actual data with the preset value, the motion control module needs to determine whether the position and speed of the cutting tool need to be adjusted. There are two situations as follows;

[0120] B4.1. When the position and speed of the cutting tool need to be adjusted, the motion control module will immediately take measures to make adjustments;

[0121] B4.2, when there is no need to adjust the position and speed of the cutting tool, the motion control module will continue to repeat the operation of S4;

[0122] In step B4, when determining whether the position and speed of the cutting tool need to be adjusted, if a deviation or abnormality is detected, the motion control module will immediately take measures to make adjustments. For example, if the tool deviates from the predetermined trajectory, the motion control module will send instructions to the motor driver to adjust the direction of the tool and return to the correct trajectory. Similarly, if it is detected that the cutting speed is too fast or too slow and affects the cutting quality, the motion control module will adjust the cutting speed as needed.

[0123] Among them, in step B2, when calculating the optimal movement trajectory of the cutting tool, a virtual wafer surface model is first constructed based on the chip position and edge information provided by the image processing module and the real-time data provided by the sensor module. Then, an optimal path from the starting point to the end point is planned on the model through the Dijkstra algorithm. The optimal path usually avoids obstacles (such as gaps between chips) as much as possible and considers the kinematic limitations of the tool (such as maximum speed, acceleration, etc.). After the path planning is completed, the path is smoothed by the interpolation and fitting algorithm to reduce the jitter and vibration of the tool during movement, thereby improving the cutting accuracy and ensuring that the tool can move stably along the planned trajectory;

[0124] Dijkstra is a greedy algorithm for calculating the single-source shortest path problem. It maintains a distance array to record the shortest distance from the source point to each vertex, and selects the vertex with the shortest distance for expansion at each step. The interpolation and fitting algorithm is used to generate a continuous and smooth trajectory to reduce the jitter and vibration of the path. Linear interpolation is the simplest interpolation method, which estimates new data points by connecting the straight line between two points. Cubic spline interpolation uses a cubic polynomial to fit the curve between each two data points to obtain a smoother trajectory. The least squares method is a fitting method that finds the best fitting curve by minimizing the sum of squared errors. The Dijkstra algorithm and the interpolation and fitting algorithm are both common knowledge technologies in the field, and will not be described in detail here.

[0125] When determining the cutting speed of the cutting tool, the motion control module uses a machine learning model to predict the optimal cutting speed under different conditions. The prediction process is as follows:

[0126] C1. Collect a large amount of historical data about wafer cutting. The historical data includes information about material hardness, toughness, thickness, required cutting accuracy, and actual cutting speed used. These data will serve as the basis for training the machine learning model.

[0127] C2. Extract key features related to cutting speed from the collected data; key features include physical properties of wafer material type, thickness, hardness, toughness, and process parameters such as required cutting accuracy. Feature selection can reduce the complexity of the model and improve its generalization ability.

[0128] C3. Select appropriate machine learning models based on the nature of the problem and the characteristics of the data. Machine learning models include linear regression, support vector machine and neural network.

[0129] C4. Use the collected data to train the selected model. During the training process, the model parameters need to be continuously adjusted to optimize its performance. At the same time, in order to evaluate the generalization ability of the model, the data set needs to be divided into a training set and a test set, and the test set should be used regularly for verification during the training process.

[0130] C5. After the model training is completed, the trained model predicts the optimal cutting speed under different conditions. Specifically, the current conditions (such as material properties, thickness, required cutting accuracy, etc.) are input into the model, and then the corresponding cutting speed prediction value is obtained. This prediction value can be used as a reference value in the actual cutting process.

[0131] Through the continuous cycle of the above process, the motion control module can monitor and adjust the movement of the cutting tool in real time, thereby ensuring high precision throughout the cutting process. This dynamic adjustment mechanism enables the system to adapt to various complex situations and improve cutting accuracy and efficiency.

[0132] Through the cooperation of the sensor module and the motion control module, the monitored data is fed back to the motion control module in real time, so as to adjust the position and speed of the cutting tool in time to ensure cutting accuracy.

[0133] The focusing module is used to adjust the focal length of the laser to adapt to wafers of different thicknesses and different cutting requirements.

[0134] The process of focusing the laser by the focusing module is as follows:

[0135] D1. When the laser cutting device is powered on, the focusing module will receive a focusing return signal indicating that the focusing mirror needs to return to its absolute zero position;

[0136] D2. According to the received focus return zero signal, the focus module controls the focus mirror to move to the absolute zero position at a certain speed through the servo motor. During this process, the focus module will detect whether it has received the zero point sensor signal fed back by the mechanical zero point sensor device at the absolute zero point position. If the signal is not received, the focus module will continue to control the focus mirror to move at the same speed until the zero point sensor signal is received;

[0137] D3. After the focusing mirror returns to the absolute zero position, the focusing module will detect whether the servo motor is working normally. If the servo motor is in a normal state, the focusing module will be ready to receive subsequent focus adjustment signals; if the servo motor is abnormal, the focusing module will send a reset signal to reset the servo motor;

[0138] D4. When receiving the focus adjustment signal, the focus adjustment module reads the analog voltage input signal including the target focus position; the analog voltage input signal is provided by the sensor module to indicate the desired focus position;

[0139] D5. Perform preset signal processing on the read analog voltage signal, and determine the output frequency and pulse number of the pulse control signal for driving the servo motor according to the processed signal.

[0140] D6. According to the determined output frequency and number of pulses, the focusing module outputs corresponding pulse control signals, and drives the servo motor through these signals to drive the focusing mirror to move within the preset motion range. The movement of the servo motor will change the focusing position of the laser beam, thereby achieving precise adjustment of the focus point.

[0141] The motion control module accurately controls the motion trajectory and speed of the cutting tool based on the data provided by the image processing module and the sensor module, and cooperates with the focusing module to adjust the distance of the laser to adapt to wafers of different thicknesses.

[0142] The data processing module is used to receive data from the image processing module, the sensor module, the motion control module and the focusing module, and perform comprehensive analysis and processing to optimize cutting parameters and improve cutting efficiency.

[0143] The process of comprehensive analysis and processing of received data by the data processing module is as follows:

[0144] E1. The data processing module integrates the chip position and edge information provided by the image processing module, the real-time data (such as tool position and speed) provided by the sensor module, the path planning and speed adjustment information of the motion control module, and the focus position information of the focusing module to build a system state model, which provides a basis for subsequent analysis.

[0145] E2. Analyze the data in depth based on the system status model, calculate the relative position relationship between the chip and the cutting tool, evaluate whether the tool's motion trajectory conforms to the predetermined plan, and determine whether the current focus position is appropriate to confirm the current status of the system and potential problems;

[0146] E3. Make corresponding decisions based on the results of data analysis. For example, adjust the tool's motion trajectory according to the relative position between the chip and the cutting tool; or adjust the laser's focal length according to the current focal position. These decisions will directly affect the efficiency and quality of the cutting process.

[0147] E4. After the decision is made, the data processing module will generate corresponding instructions and send them to the relevant actuators. The corresponding instructions include adjusting the movement speed of the tool, changing the focal length of the laser, recording the current working status, etc. Through precise instruction generation and transmission, it can ensure that the system works according to the predetermined plan.

[0148] E5. During the execution of instructions, the data processing module will continuously monitor the operating status of each module and collect feedback information. The feedback information will be used to evaluate the execution effect of the instructions and the operating status of the system. If any problems or deviations are found, the data processing module will make timely adjustments to ensure the stable operation of the system.

[0149] E6. Based on long-term operating data and feedback information, the data processing module optimizes and improves the operating strategy of the entire system, which may include adjusting algorithm parameters, updating software versions, introducing new functions, and improving system performance and reliability through continuous optimization and improvement.

[0150] Through the coordinated work of the image processing module and the sensor module, the cutting system can accurately identify the chip position and edge, thereby achieving high-precision positioning. The dynamic adjustment capabilities of the motion control module and the focusing module enable it to adapt to wafers of different thicknesses and different cutting requirements, further improving the cutting accuracy. The real-time analysis and optimization suggestions of the data processing module enable the system to respond quickly and adjust the cutting parameters, reducing unnecessary downtime and waiting time. The real-time monitoring function of the sensor module can promptly detect and correct potential problems, enhancing the stability and reliability of the system.

[0151] Based on the above-described cutting system capable of accurately positioning a chip wafer, the present invention further provides a cutting method capable of accurately positioning a chip wafer, comprising the following steps:

[0152] S1, the image processing module collects the wafer surface image through the camera, and uses the image processing algorithm to analyze the image, identify the chip position and edge on the wafer, and obtain the wafer status data;

[0153] S2, the position of the cutting tool is monitored in real time by the sensor module, and the motion control module controls the motion trajectory and speed of the cutting tool in combination with the wafer status data;

[0154] S3, the motion control module controls the movement of the cutting tool according to the cutting requirements and sends an instruction to the focusing module to adjust the focal length of the laser;

[0155] S4, the data processing module receives the recognition results of the image processing module, the monitoring results of the sensor module, the control parameters of the motion control module and the focal length information of the focusing module in real time, and conducts a comprehensive analysis to evaluate whether the current cutting state meets the accuracy requirements. There are two situations:

[0156] S4.1. If the current cutting state meets the accuracy requirements, the motion control module controls the cutting tool to perform the cutting operation;

[0157] S4.2. If the current cutting state does not meet the accuracy requirements, it is necessary to predict possible problems, make optimization adjustments based on the prediction results, and then repeat S4.

[0158] The above steps achieve precise positioning and efficient cutting of chip wafers, reduce production costs and enhance system stability.

[0159] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cutting system capable of accurately positioning a chip wafer, characterized in that: It includes a cutting machine body and an image processing module, a sensor module, a motion control module, a focusing module and a data processing module installed on the cutting machine body; The cutting machine body comprises a cutting tool and a moving frame for supporting the cutting tool, the cutting tool is mounted with a laser, and the position of the laser is controlled by a focusing module; The image processing module collects wafer surface images through a camera and uses an image processing algorithm to analyze them to identify chip positions and edges on the wafer; The sensor module includes a position sensor and a distance sensor, and the sensor module is used to monitor the position of the cutting tool and the state of the wafer in real time; The motion control module is used to control the motion of the cutting tool in the horizontal and vertical directions to ensure that the cutting tool can move accurately along a predetermined trajectory; The focusing module is used to adjust the focal length of the laser to adapt to wafers of different thicknesses and different cutting requirements; The data processing module is used to receive data from the image processing module, the sensor module, the motion control module and the focusing module, and perform comprehensive analysis and processing to optimize cutting parameters and improve cutting efficiency; The motion control module accurately controls the motion trajectory and speed of the cutting tool according to the data provided by the image processing module and the sensor module, and cooperates with the focusing module to adjust the distance of the laser to adapt to wafers of different thicknesses; The process of the image processing module collecting and identifying the chip position and edge on the wafer is as follows: A1. Initialize the camera settings; A2: The camera starts to collect image data of the wafer surface in real time and performs image preprocessing on the image data; A3 divides the processed image into chip area and non-chip area; A4, extracting the geometric features of the chip area to describe the chip shape and size; A5. Analyze the texture features of the chip area to distinguish different types of chips or identify specific marks on the chip; A6. Use a template image containing chip features to match the features in the image and calculate the similarity of each position to determine the specific position and direction of the chip. The similarity calculation formula is: ,in , is the image to be processed, are two random feature vectors in the image to be processed, is the template image, are two random feature vectors in the template image; A7, automatically learn image features through multi-layer convolution and pooling operations, and generate feature maps in the last layer. The feature maps are converted into a one-dimensional vector through the fully connected layer, and then input into the Softmax function to convert it into a probability distribution, and the feature maps are classified and recognized; A8, convert the chip position in the image coordinate system into the position in the actual physical coordinate system, and then transmit the positioning result to the motion control module; The process of the motion control module controlling the cutting tool to move in the horizontal and vertical directions is as follows: B1. The receiving sensor module obtains real-time data and the chip position and edge information analyzed by the image processing module. The real-time data includes information provided by the position sensor and the distance sensor; B2, based on the data received in B1, calculate the optimal moving trajectory of the cutting tool, and determine the cutting speed of the cutting tool according to the material properties and thickness of the wafer and the required cutting accuracy; B3. After calculating the optimal moving trajectory and cutting speed, the cutting tool is controlled by the motor driver to move according to the calculated trajectory and speed; B4. During the cutting process, the feedback information from the sensor module is continuously received, including the actual position and speed of the tool and the state change of the wafer. By comparing the actual data with the preset value, the motion control module needs to determine whether the position and speed of the cutting tool need to be adjusted. There are two situations as follows; B4.

1. When the position and speed of the cutting tool need to be adjusted, the motion control module will immediately take measures to make adjustments; B4.2, when there is no need to adjust the position and speed of the cutting tool, the motion control module will continue to repeat the operation of S4; In step B2, when determining the cutting speed of the cutting tool, the motion control module uses a machine learning model to predict the optimal cutting speed under different conditions. The prediction process is as follows: C1. Collect a large amount of historical data on wafer cutting; C2, extract key features related to cutting speed from the collected data; C3. Select appropriate machine learning models based on the nature of the problem and the characteristics of the data. Machine learning models include linear regression, support vector machine and neural network. C4. Use the collected data to train the selected model. During the training process, the model parameters need to be continuously adjusted to optimize its performance. At the same time, in order to evaluate the generalization ability of the model, the data set needs to be divided into a training set and a test set. The test set should be used regularly for verification during the training process. C5. After the model training is completed, the trained model predicts the optimal cutting speed under different conditions; The process of focusing the laser by the focusing module is as follows: D1. When the laser cutting device is powered on, the focusing module will receive a focusing return signal indicating that the focusing mirror needs to return to its absolute zero position; D2. According to the received focus return zero signal, the focus module controls the focus mirror to move to the absolute zero position at a certain speed through the servo motor; D3. After the focusing mirror returns to the absolute zero position, the focusing module will detect whether the servo motor is working normally. If the servo motor is in a normal state, the focusing module will be ready to receive subsequent focus adjustment signals; if the servo motor is abnormal, the focusing module will send a reset signal to reset the servo motor; D4. When receiving the focus adjustment signal, the focus adjustment module reads the analog voltage input signal containing the target focus position; D5. Performing preset signal processing on the read analog voltage signal, and determining the output frequency and pulse number of the pulse control signal for driving the servo motor according to the processed signal; D6. According to the determined output frequency and pulse quantity, the focusing module outputs corresponding pulse control signals, and drives the servo motor through these signals to drive the focusing mirror to move within the preset motion range.

2. A cutting system capable of accurately positioning a chip wafer according to claim 1, characterized in that: In step A2, during image preprocessing, the image data is first subjected to convolution processing, and the formula used for the convolution processing is: ， ， in, represents the input image, Indicates that the size of the input image is X*X, F is the convolution kernel size F*F, S is the stride, and P is the padding; The processed data is then smoothed to remove noise and reduce random fluctuations. The smoothing formula is: ,in is the standard deviation, and are two selected feature vectors in the input image; Finally, edge enhancement calculation is performed on the smoothed data, and the calculation formula is as follows: Horizontal direction: ; Vertical direction: ; According to the calculated and Calculate the total gradient magnitude for each pixel: , to enhance the edge information in the image.

3. A cutting system capable of accurately positioning a chip wafer according to claim 2, characterized in that: In step A7, the formula for multi-layer convolution and pooling operations is: ,in is the size and sliding range of the convolution kernel, is the pixel position currently being processed, is the convolution kernel used to scan the input image and extract features, and b is the bias term used to adjust the output of the activation function; The Softmax function formula is: in, The first elements, To normalize the sum of the indices of all categories.

4. A cutting system capable of accurately positioning a chip wafer according to claim 3, characterized in that: In step B2, when calculating the optimal movement trajectory of the cutting tool, a virtual wafer surface model is first constructed based on the chip position and edge information provided by the image processing module and the real-time data provided by the sensor module. Then, the Dijkstra algorithm is used to plan an optimal path from the starting point to the end point on the model. The optimal path will try to avoid obstacles and take into account the kinematic limitations of the tool. After the path planning is completed, the path is smoothed by interpolation and fitting algorithms.

5. A cutting system capable of accurately positioning a chip wafer according to claim 4, characterized in that: The process of comprehensive analysis and processing of received data by the data processing module is as follows: E1, the data processing module integrates the chip position and edge information provided by the image processing module, the real-time data provided by the sensor module, the path planning and speed adjustment information of the motion control module, and the focus position information of the focusing module to build a system state model; E2. Analyze the data in depth based on the system status model, calculate the relative position relationship between the chip and the cutting tool, evaluate whether the tool's motion trajectory conforms to the predetermined plan, and determine whether the current focus position is appropriate to confirm the current status of the system and potential problems; E3. Make corresponding decisions based on the results of data analysis; E4. After the decision is made, the data processing module will generate corresponding instructions and send them to the relevant execution agencies; E5. During the execution of instructions, the data processing module will continuously monitor the operating status of each module and collect feedback information. The feedback information will be used to evaluate the execution effect of the instructions and the operating status of the system. If any problems or deviations are found, the data processing module will make timely adjustments to ensure the stable operation of the system. E6. Based on long-term operating data and feedback information, the data processing module optimizes and improves the operating strategy of the entire system.

6. A cutting method capable of accurately positioning a chip wafer, based on a cutting system capable of accurately positioning a chip wafer as claimed in any one of claims 1 to 5, characterized in that: The steps include: S1, the image processing module collects the wafer surface image through the camera, and uses the image processing algorithm to analyze the image, identify the chip position and edge on the wafer, and obtain the wafer status data; S2, the position of the cutting tool is monitored in real time by the sensor module, and the motion control module controls the motion trajectory and speed of the cutting tool in combination with the wafer status data; S3, the motion control module controls the movement of the cutting tool according to the cutting requirements and sends an instruction to the focusing module to adjust the focal length of the laser; S4, the data processing module receives the recognition results of the image processing module, the monitoring results of the sensor module, the control parameters of the motion control module and the focal length information of the focusing module in real time, and conducts a comprehensive analysis to evaluate whether the current cutting state meets the accuracy requirements. There are two situations: S4.

1. If the current cutting state meets the accuracy requirements, the motion control module controls the cutting tool to perform the cutting operation; S4.

2. If the current cutting state does not meet the accuracy requirements, it is necessary to predict possible problems, make optimization adjustments based on the prediction results, and then repeat S4.

Images