Alloy cutter wear prediction method and system based on machine vision

Abstract

The invention provides an alloy cutter wear prediction method and system based on machine vision. The method comprises the following steps: acquiring initial surface image data of a cutter; performing optimization enhancement processing aiming at tool wear detection on the initial surface image data to generate an optimized target surface image; extracting contour image data of the cutter from the target surface image, and generating a comprehensive feature data set capable of representing edge contour features and texture features of the cutter based on texture analysis of the contour image data; performing comprehensive quantification on the edge contour features and the texture features represented by the comprehensive feature data set to obtain a first wear index prediction value representing the current wear condition of the tool; and monitoring the first wear index prediction value so as to trigger a real-time monitoring system to carry out abnormity alarm under the condition that the first wear index prediction value exceeds a preset prediction threshold value. The invention provides an efficient, accurate and practical tool wear prediction mode.

Description

Technical Field

[0001] The present application relates to the field of visual inspection, and more specifically, to a method and system for predicting alloy tool wear based on machine vision. Background Art

[0002] The study of tool wear is a crucial area in the manufacturing industry, directly impacting production efficiency, product quality, and equipment safety. With the advancement of industrial intelligence, real-time monitoring and accurate diagnosis of tool conditions are crucial for ensuring production continuity and reducing costs. However, traditional tool wear detection methods often rely on manual experience or simple measurement tools, resulting in inefficiency, insufficient accuracy, and an inability to adapt to complex working conditions. These limitations make it difficult for companies to effectively intervene in the early stages of tool wear, leading to production interruptions and quality issues.

[0003] Against this backdrop, the accurate assessment and prediction of tool wear faces numerous challenges. Foremost among these challenges is acquiring real-time information on subtle changes in the tool surface within a dynamic machining environment. Due to the high speed and complexity of machining processes, traditional image acquisition methods often struggle to capture clear and comprehensive data. The application of visual intelligence technology offers a new solution to this problem. By employing advanced visual inspection methods, tool surface information can be more accurately captured within a dynamic machining environment. However, the industry has yet to propose a robust solution using visual intelligence technology to address this issue, and the precise assessment and prediction of tool wear remains a significant challenge.

[0004] Therefore, the present application provides a method and system for predicting alloy tool wear based on machine vision to solve one of the above technical problems. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for predicting alloy tool wear based on machine vision, which can solve at least one of the technical problems mentioned above. The specific solution is as follows: According to a specific embodiment of the present application, in a first aspect, the present application provides a method for predicting alloy tool wear based on machine vision, comprising: Acquire initial surface image data of a tool; perform optimization and enhancement processing on the initial surface image data for tool wear detection to generate an optimized target surface image; extract contour image data of the tool from the target surface image, and generate a comprehensive feature data set capable of characterizing edge contour features and texture features of the tool based on texture analysis of the contour image data; comprehensively quantify the edge contour features and texture features characterized by the comprehensive feature data set to obtain a first wear index prediction value that characterizes the current wear condition of the tool; monitor the first wear index prediction value to trigger a real-time monitoring system to issue an abnormal alarm when the first wear index prediction value exceeds a preset prediction threshold.

[0006] According to a specific embodiment of the present application, in a second aspect, the present application provides an alloy tool wear prediction system based on machine vision, comprising: An acquisition unit is used to acquire initial surface image data of a tool; a generation unit is used to perform optimization and enhancement processing on the initial surface image data for tool wear detection to generate an optimized target surface image; and to extract contour image data of the tool from the target surface image, and based on texture analysis of the contour image data, to generate a comprehensive feature data set capable of characterizing edge contour features and texture features of the tool; a processing unit is used to comprehensively quantify the edge contour features and texture features characterized by the comprehensive feature data set to obtain a first wear index prediction value that characterizes the current wear condition of the tool; and to monitor the first wear index prediction value to trigger a real-time monitoring system to issue an abnormal alarm when the first wear index prediction value exceeds a preset prediction threshold.

[0007] Compared with the prior art, the above solution of the embodiment of the present application has at least the following beneficial effects: The present application provides a method and system for predicting alloy tool wear based on machine vision. It uses high-speed and high-resolution imaging equipment in a dynamic processing environment to capture tool surface details in real time, and combines image preprocessing algorithms to optimize the extraction of edge contours and texture features, significantly improving detection accuracy and data reliability. At the same time, by constructing a wear index prediction value mapping model and a time series dynamic evolution model, it realizes quantitative analysis and trend prediction of tool wear degree, overcoming the limitations of traditional methods that rely on manual experience or static analysis. Furthermore, based on real-time monitoring and abnormal alarm mechanisms, it can actively intervene in the early stages of wear, combine long-term and short-term memory networks to model future wear trends, and extend tool service life by dynamically adjusting processing parameters. Finally, a closed-loop data flow is formed to optimize the production process, effectively solving the problems of insufficient accuracy, delayed response, and high maintenance costs of tool wear monitoring under complex working conditions, and providing an efficient and accurate solution for the intelligent upgrade of the manufacturing industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A flow chart of a method for predicting alloy tool wear based on machine vision is shown; Figure 2 A flow chart of a method for obtaining initial surface image data of a tool is shown; Figure 3 A flow chart of a method for generating an optimized target surface image is shown; Figure 4 A flow chart of a method for generating a comprehensive feature dataset is shown; Figure 5 A flow chart of a method for obtaining a first wear indicator prediction value is shown; Figure 6 A flow chart of a method for predicting the future service life of a tool is shown; Figure 7 A flow chart of a method for obtaining a second wear index prediction value is shown; Figure 8 A flow chart of a method for adjusting processing parameter instructions and updating a database is shown; Figure 9 A flow chart of a method for obtaining risk assessment results is shown; Figure 10 A unit block diagram of an alloy tool wear prediction system based on machine vision is shown. DETAILED DESCRIPTION

[0009] To make the objectives, technical solutions, and advantages of this application more clear, this application will be further described in detail below with reference to the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0010] The terms used in the examples of this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms "a," "the," and "the" used in the examples of this application and the appended claims are also intended to include plural forms, and unless the context clearly indicates otherwise, "a plurality" generally includes at least two.

[0011] It should be understood that the term "and / or" as used herein is merely a description of the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0012] It should be understood that although the terms first, second, third, etc. may be used to describe in the embodiments of the present application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, without departing from the scope of the embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first.

[0013] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0014] It should also be noted that the terms "include," "comprises," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a product or system comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such product or system. In the absence of further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the product or system comprising the element.

[0015] It should be noted in particular that any symbols and / or numbers in the specification that are not marked in the accompanying drawings are not drawing marks.

[0016] The optional embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0017] The embodiment provided in this application is an embodiment of a method for predicting alloy tool wear based on machine vision.

[0018] The following combination Figure 1 The embodiments of the present application are described in detail.

[0019] Figure 1 A flow chart of a method for predicting alloy tool wear based on machine vision is shown in FIG. Figure 1 As shown, the following steps are included.

[0020] Step S101: acquiring initial surface image data of a tool.

[0021] Step S102 : performing optimization and enhancement processing for tool wear detection on the initial surface image data to generate an optimized target surface image.

[0022] Step S103 : extracting the contour image data of the tool from the target surface image, and generating a comprehensive feature data set capable of characterizing the edge contour features and texture features of the tool based on texture analysis of the contour image data.

[0023] Step S104 , comprehensively quantifying the edge profile features and texture features represented by the comprehensive feature data set to obtain a first wear index prediction value representing the current wear condition of the tool.

[0024] Step S105 : monitoring the first wear index prediction value, so as to trigger the real-time monitoring system to issue an abnormality alarm when the first wear index prediction value exceeds a preset prediction threshold.

[0025] The machine vision-based alloy tool wear prediction method provided in this application captures tool surface details in real time through high-speed and high-resolution imaging equipment in a dynamic processing environment, and optimizes the extraction of edge contours and texture features in combination with image preprocessing algorithms, significantly improving detection accuracy and data reliability. At the same time, by constructing a wear index prediction value mapping model and a time series dynamic evolution model, quantitative analysis and trend prediction of tool wear degree are achieved, overcoming the limitations of traditional methods that rely on manual experience or static analysis. Furthermore, based on real-time monitoring and abnormal alarm mechanisms, active intervention can be made at the early stage of wear, and future wear trends can be modeled by combining long-short-term memory networks. The tool life can be extended by dynamically adjusting processing parameters, ultimately forming a closed-loop data flow to optimize the production process, effectively solving the problems of insufficient accuracy, delayed response, and high maintenance costs of tool wear monitoring under complex working conditions, and providing an efficient and accurate solution for the intelligent upgrade of the manufacturing industry.

[0026] Figure 2 A flow chart of a method for obtaining initial surface image data of a tool is shown, as shown in FIG. Figure 2 As shown, the following steps are included.

[0027] Step S201 : photographing the surface of the tool in real time during the machining process to obtain initial image data.

[0028] Step S202 : extracting multiple frames of image data from the initial image data at preset time intervals, performing denoising processing on the multiple frames of image data, and determining whether the clarity of the multiple frames of image data reaches a preset clarity threshold.

[0029] Step S203 : If the clarity of the multiple frames of image data does not reach the preset clarity threshold, the brightness and contrast of the multiple frames of image data are adjusted to obtain optimized image data.

[0030] For example, if the clarity of the multiple frames of image data reaches a preset clarity threshold, step S203 is skipped after step S202 and step S204 is directly executed.

[0031] Step S204: using the optimized image data as the initial surface image data of the tool.

[0032] To address the challenge of capturing tool surface information in dynamic machining environments, some embodiments employ high-speed, high-resolution imaging equipment to capture the tool surface in real time. Specifically, the tool surface is continuously scanned at a fixed frequency during machining to acquire initial surface image data, ensuring clear, detailed information on the tool surface is captured even at high speeds.

[0033] As a feasible embodiment, based on the characteristics of high-speed imaging and high-resolution technology, a high-speed camera device is used to capture the tool surface in real time during the machining process. By adjusting the camera parameters to adapt to the motion state in a dynamic environment, initial image data containing surface details is obtained. In response to the requirements of continuous scanning and fixed frequency, multiple frames of image data are extracted from the initial image data at preset time intervals, and these image data are denoised to determine whether their clarity meets the preset threshold. If it is detected that the image clarity does not meet the preset threshold, the brightness and contrast of these image data are adjusted, and the characteristics of the surface details are combined to finally obtain optimized image data. The optimized image data is then input into a pre-established feature extraction tool to determine the wear characteristics of the tool surface, thereby obtaining surface information reflecting the tool status.

[0034] For example, in a turning task, a high-speed camera captures the tool surface at a speed of 5,000 frames per second. By adjusting the focal length and exposure time to adapt to vibrations and light changes during machining, initial image data containing the tool surface texture and tiny defects is obtained. The advantage of this method is that it can monitor the tool status in real time, providing a reliable data basis for subsequent analysis.

[0035] As a specific example, assume that during a continuous scanning process, one image frame is captured every 0.2 seconds, for a total of 10 frames. These images are then denoised to remove noise caused by vibration or light interference. If the image clarity does not meet a preset threshold (e.g., a clarity score below 80), the brightness and contrast are adjusted, for example by increasing the brightness by 20% and the contrast by 15%, to better highlight the detailed features of the tool surface. When the clarity of the multiple frames of image data reaches the preset clarity threshold, the brightness and contrast adjustment step is skipped and the next step is performed, where the optimized image data is used as the initial surface image data for the tool.

[0036] Figure 3 A flow chart of a method for generating an optimized target surface image is shown. Figure 3 As shown, the following steps are included.

[0037] Step S301 : performing preliminary filtering processing on the initial surface image data by adopting a mean filtering method to obtain a first processed image.

[0038] Step S302 , performing denoising on pixels in the first processed image whose pixel difference values ​​with adjacent areas are greater than a preset difference threshold, and performing secondary filtering based on filtering parameters matching the machining environment of the tool to obtain a second processed image.

[0039] Step S303: If the clarity of the second processed image does not reach the preset clarity threshold, brightness adjustment is performed on the second processed image to obtain a third processed image.

[0040] For example, if the clarity of the second processed image reaches a preset clarity threshold, step S303 is skipped after step S302 and step S304 is directly executed.

[0041] Step S304: performing edge enhancement processing on the second processed image or the third processed image to obtain a target surface image.

[0042] In some embodiments, in order to solve the problem of difficulty in feature extraction of tool surface images due to fuzzy interference in dynamic machining environments, an image preprocessing algorithm is used to optimize the initial surface image data to generate a target surface image that can reflect the edge contour and texture characteristics of the tool surface.

[0043] As a feasible embodiment, the preliminary filtering process in the above steps is implemented using a mean filter method. For example, in high-speed milling, when the initial image is slightly blurred due to machine tool vibration, a 3x3 area mean filter is used to smooth the image to obtain the first processed image. This method can effectively reduce the interference of random noise on subsequent analysis. As a specific embodiment, deep denoising can adapt to complex interference in the processing environment by dynamically adjusting filter parameters. For example, in the presence of changing lighting or unstable vibration frequency, a non-local means denoising method is selected to remove noise by comparing pixel values ​​in similar areas of the image, thereby obtaining a second processed image. Preferably, the filter parameters can be dynamically adjusted based on the vibration frequency of the processing environment, for example, increasing the filter strength when the vibration frequency is high to ensure that image details are not over-smoothed. For example, if the clarity of the second processed image does not meet the preset threshold (e.g., a clarity score below 85), brightness adjustment is required. If the image is generally dark with unclear edge details, the brightness can be increased by 15%, while the contrast can be enhanced by 10% to produce the third processed image. This processing method can significantly improve the visualization of subtle textures and defects on the tool surface, providing a reliable data foundation for subsequent analysis. Furthermore, to enhance the details of the third processed image, edge enhancement is performed to enhance the gradient differences in pixel values. For example, if subtle wear marks are present on the tool surface during machining, Laplace sharpening is used to enhance the edge regions to generate the final target surface image. Specifically, if the texture features of a region in the image are complex, the sharpening intensity can be appropriately reduced to avoid introducing excessive noise. It's important to note that the aforementioned processing steps closely align with the optimization of the tool surface image in a dynamic machining environment, forming a complete chain from initial filtering to deep denoising, brightness adjustment, and edge enhancement. Each step specifically addresses specific issues presented by the dynamic environment (such as blur, noise, and unclear details). The resulting target surface image more realistically reflects the tool surface condition, providing a reliable basis for subsequent wear analysis and machining parameter adjustments based on the predicted value of the first wear indicator. For example, in a dynamic machining environment, the target surface image generated through this multi-step process can effectively preserve key feature information on the tool surface. For example, in high-speed cutting, after optimization, the edge contours and texture features in the image become clearer, laying the foundation for subsequent extraction of tool profile image data and calculation of the first wear index prediction value. This method has the advantage of being able to cope with complex environmental interference while ensuring the integrity of key feature information, thereby improving the accuracy of tool condition monitoring and the stability of the machining process. In summary, this embodiment solves the problem of feature extraction of tool surface images caused by interference in dynamic machining environments through a multi-level image preprocessing algorithm, and provides high-quality basic data for subsequent wear quantification analysis based on the first wear index prediction value.

[0044] Figure 4 A flow chart of a method for generating a comprehensive feature dataset is shown in FIG. Figure 4 As shown, the following steps are included.

[0045] Step S401 , performing spatial relationship analysis between pixels based on grayscale changes between pixels in the contour image data, and obtaining comprehensive data for describing contours and textures.

[0046] Step S402: Perform brightness enhancement processing on the integrated data to obtain brightness-enhanced integrated data.

[0047] Step S403 , performing feature fusion on the comprehensive data after brightness enhancement to obtain a comprehensive feature data set capable of characterizing the edge profile features and texture features of the tool.

[0048] As a feasible embodiment, contour extraction can be achieved through edge detection tools. For example, in high-speed cutting, if the tool surface image has blurred boundaries due to subtle vibrations, the edge detection threshold can be adjusted (for example, to 0.2) to capture more subtle edge information and generate contour image data. This approach can better adapt to complex situations in dynamic environments. As a specific example, for texture analysis of contour image data, tools such as gray-level co-occurrence matrices can be used to extract features. For example, when periodic wear textures are present on a tool surface, the frequency and directionality of grayscale changes can be analyzed to extract texture details and generate comprehensive data. This approach can facilitate subsequent in-depth assessment of the tool's condition. In some specific embodiments, if the clarity score of the integrated data falls below 80, brightness adjustment is required. The purpose of brightness adjustment is to enhance image visualization and highlight key areas of the tool surface. For example, if the image is generally dark, the brightness value can be increased by 20%, while the contrast parameter can be adjusted to make detailed areas more visible. This enhanced integrated data provides a clearer foundation for subsequent feature analysis. Furthermore, during the feature fusion stage, multidimensional features such as contour and texture can be categorized and organized. For example, if both the integrity of the tool edge and the degree of surface wear need to be considered, feature data can be integrated using a weighted approach (e.g., setting the edge feature weight to 0.6 and the texture feature weight to 0.4) to generate a comprehensive feature dataset. This integration approach can support a comprehensive assessment of the tool's condition. For example, when determining whether the integrated data meets the requirements for subsequent processing, feature completeness or data consistency requirements can be set. For example, if features are missing in certain areas of the integrated data during processing monitoring, the dataset can be improved through supplementary processing or re-collection to ensure quality. This meticulous judgment mechanism helps improve the reliability of subsequent analysis. For example, the final comprehensive feature dataset can be further refined and categorized. For example, features can be divided into structural features (such as edge integrity) and surface quality features (such as crack distribution). For example, if a localized crack is present on the tool surface, the relevant features can be classified as surface quality features for subsequent targeted analysis. This classification and organization approach can provide more targeted data support for tool condition monitoring, thereby optimizing the decision-making process for machining parameter adjustments. In summary, the comprehensive feature dataset generated through the above steps comprehensively reflects the edge profile and texture characteristics of the tool surface, providing a reliable basis for subsequent wear quantification analysis based on the predicted value of the first wear indicator. This method is highly robust and adaptable in dynamic machining environments, helping to improve the accuracy and efficiency of tool condition monitoring.

[0049] Figure 5A flow chart of a method for obtaining a first wear index prediction value is shown. Figure 5 As shown, the following steps are included.

[0050] Step S501 : performing splitting processing on edge contour features and texture features based on the comprehensive feature data set to obtain split contour data set and texture data set.

[0051] Step S502 : obtaining a quantization benchmark from a preset mapping library, and performing a quantization on the contour dataset and the texture dataset based on the quantization benchmark to obtain a quantization value group.

[0052] Step S503 : performing integrity analysis on the value distribution of the quantized value group, and if the integrity of the value distribution does not reach a completeness threshold, performing value adjustment processing on the quantized value group to obtain an adjusted quantized value group.

[0053] Step S504 , performing a fluctuation analysis on the adjusted quantized value group, and calculating various index values ​​of the preset wear index based on the quantized value group when the fluctuation of the quantized value group meets the preset fluctuation range.

[0054] Step S505 : performing weighted processing on each indicator value according to the preset weight corresponding to each indicator value to obtain a first wear indicator prediction value.

[0055] In some embodiments, by integrating feature data sets and utilizing a pre-established wear index prediction value mapping model, edge profile and texture features are converted into quantifiable wear degree values ​​to obtain a first wear index prediction value that characterizes the current wear state of the tool, providing a basis for subsequent dynamic evolution analysis.

[0056] As a feasible embodiment, the splitting of the comprehensive feature dataset can be achieved through separation tools. For example, in high-speed cutting, when the tool surface image contains complex edge information and wear textures, the edge profile data can be first extracted to analyze geometric changes, and then the texture data can be extracted to assess surface wear. This layered processing method can more accurately reflect the tool status. As a specific example, the quantitative benchmark matching process relies on a pre-set mapping library. For example, if a contour dataset indicates a slight defect on the tool edge, the matching tool can retrieve the corresponding quantitative benchmark from the mapping library (e.g., categorizing the defect severity into three levels: slight, moderate, and severe) to initially generate a set of quantitative values. This approach provides a standardized reference for subsequent analysis. For example, if the completeness of a quantized value set does not meet a preset threshold, numerical adjustments may be necessary. For example, if the values ​​of certain edge features in a quantized value set are too low, failing to meet the completeness threshold of 0.5, correction tools can be used to smooth the values ​​or supplement missing data to generate an adjusted quantized value set. This adjustment can improve data reliability and provide a more accurate basis for subsequent calculations. Furthermore, when determining whether a quantitative value group meets the calculation conditions, a consistency standard can be set. For example, a fluctuation range of the quantitative value group is required to not exceed 0.1 to ensure data stability. In some specific embodiments, when calculating indicators based on the adjusted quantitative value group, indicator weights can be set based on business needs. For example, if edge wear and surface roughness are the two indicators that need to be focused on in tool wear assessment, the edge wear weight can be set to 0.7 and the surface roughness weight can be set to 0.3. This weighting method can highlight the importance of key indicators and ensure that the final value is more in line with actual processing needs. In some specific embodiments, after the first wear index prediction value is generated, it can be compared and analyzed with historical data. For example, if the final wear index value is 0.8, which exceeds the preset safety threshold of 0.6, it can be used as a basis for triggering tool replacement or adjusting processing parameters. This approach can effectively extend tool life and optimize processing efficiency. In summary, the first wear index prediction generated through the above steps can fully reflect the current wear state of the tool and provide reliable input for the subsequent time-series-based dynamic evolution model. This method significantly improves the accuracy and practicality of wear assessment through hierarchical feature analysis, standardized quantitative benchmark matching, and weighted calculation.

[0057] Figure 6 A flow chart of a method for predicting the future service life of a tool is shown in FIG. Figure 6 As shown, the following steps are included.

[0058] Step S601: Determine the change of the first wear index prediction value over time based on historical data, and generate a wear evolution curve of the tool.

[0059] Step S602 , predicting the future wear trend of the tool in combination with the wear evolution curve, obtaining a second wear index prediction value, and determining a reference range of the remaining service life of the tool based on the second wear index prediction value.

[0060] The second wear index prediction value represents the wear condition of the tool after a specified period of time in the future.

[0061] In some embodiments, for the predicted value of the first wear index, combined with historical processing time data, a dynamic evolution model based on time series is constructed, the changing trend of the predicted value of the first wear index over time is analyzed, and a wear evolution curve of the tool is generated to reflect the cumulative law of the wear process.

[0062] As a feasible embodiment, the integration of the first wear index prediction value and historical machining time data can be achieved through time series matching. For example, if the first wear index prediction value of the tool is collected over the past 100 hours during a machining task, it can be matched with the corresponding time points to form a continuous indicator sequence. If data is missing at certain time points, interpolation methods are used to supplement the data to ensure the integrity of the sequence. This processing method can provide a reliable data foundation for subsequent analysis. As a specific example, to segment indicator sequence data, time series processing tools can be used to divide the data into fixed time intervals (e.g., every 10 hours). For example, within a 100-hour machining task, if the predicted value of the first wear indicator within a certain time segment rises from 0.2 to 0.5, exceeding the preset threshold of 0.1, further processing is required. This segmentation method helps to discover the changing patterns of wear indicators in different time periods and supports dynamic evolution analysis. In some specific embodiments, when indicator fluctuations exceed a threshold, a data smoothing tool can be used to adjust the segments with large fluctuations. For example, if the predicted value of the first wear indicator within a certain time segment jumps from 0.3 to 0.6 in a short period of time, which is obviously discontinuous, a moving average method can be used to smooth the data so that the adjusted value stabilizes at around 0.4. This processing method can reduce noise interference and ensure the continuity and reliability of subsequent analysis. In some embodiments, a curve generation tool can be used to visualize the adjusted stationary data sequence, generating a wear evolution curve that reflects the changing trend of the predicted first wear indicator value. For example, in a machining task, the generated curve shows that the predicted first wear indicator value changes slowly initially, then gradually accelerates in the later stages, indicating that tool wear exhibits a dynamic characteristic of cumulative intensification. This visualization method can intuitively demonstrate the wear evolution pattern, allowing operators to quickly determine whether intervention is necessary. For example, when determining whether a wear evolution curve exhibits dynamic characteristics, one can focus on the slope change and inflection point of the curve. For example, if the curve shows a clear inflection point at 80 hours of machining time, the predicted value of the first wear index will rapidly increase from 0.4 to 0.7, indicating that the tool may be entering a stage of accelerated wear. This characteristic identification can help plan maintenance in advance and avoid quality issues caused by tool failure during machining. Furthermore, based on the wear evolution curve, the predicted value of the second wear indicator can be combined with a time series model to predict the value and generate a reference range for the tool's remaining service life. For example, the Long Short-Term Memory (LSTM) model can be used to analyze the historical trend of the predicted value of the first wear indicator, predict the predicted value of the second wear indicator for a period of time in the future, and calculate the tool's remaining service life based on preset evaluation criteria. This approach can provide a scientific basis for machining parameter adjustments and maintenance decisions. In summary, the wear evolution curve generated through the above steps fully reflects the cumulative pattern of tool wear and provides key data support for generating the second wear index prediction value and remaining service life assessment. This method significantly improves the accuracy and practicality of tool wear trend prediction through time series modeling and dynamic feature analysis.

[0063] Figure 7 A flow chart of a method for obtaining a second wear index prediction value is shown. Figure 7 As shown, the following steps are included.

[0064] Step S701 : According to the time series information corresponding to the wear evolution curve, the wear data of the tool is obtained in the real-time monitoring system, and the wear data and the wear evolution curve are combined into a complete data set.

[0065] Step S702 : For the complete data set, fluctuation features are extracted in combination with the wear evolution curve and the wear data, so as to mark the fluctuation features exceeding the preset fluctuation threshold as potential risk points.

[0066] Step S703: predicting the wear trend of the potential risk point to obtain a second wear index prediction value.

[0067] In some embodiments, based on the risk assessment results, a long short-term memory network (LSTM) model is used to predict the future trend of tool wear. The first wear indicator prediction value and the wear evolution curve are combined to generate a second wear indicator prediction value for a period of time in the future, and the reference range of the remaining service life of the tool is determined based on the second wear indicator prediction value.

[0068] As a feasible embodiment, real-time data acquisition and organization can be achieved through high-frequency sampling. For example, during a machining task, the system collects the first wear index prediction value data every 30 seconds to ensure real-time data. The collected data is stored in chronological order, forming a complete time series from the start of the machining process to the current moment. Furthermore, by retrieving tool wear records under the same machining conditions over the past 500 hours and merging them with the current data, a comprehensive dataset is generated, providing the basis for subsequent feature extraction. As a specific example, fluctuation feature extraction can be achieved by comparing the current wear evolution curve with historical data. For example, if the predicted value of the first wear indicator increases from 0.2 to 0.5 after 40 hours of processing, while the historical data only increases from 0.2 to 0.3 during the same period, the system will mark this discrepancy as an abnormal fluctuation point. Further analysis of the fluctuation amplitude and duration will reveal that if it exceeds a preset threshold (such as a fluctuation amplitude limit of 0.25), it will be identified as a potential risk point, providing a basis for subsequent predictions. For example, the LSTM model can be used to predict future wear trends at potential risk points and generate a predicted value for the second wear index. Assuming the current predicted value of the first wear index is 0.5 and its fluctuations show a clear upward trend, the system can predict that the second wear index will reach 0.7 within the next five hours. This prediction can provide operators with an early warning reference, allowing them to adjust machining parameters or plan tool changes in a timely manner, reducing the risk of sudden failures. Furthermore, a comprehensive analysis of the remaining useful life requires combining the predicted value of the first wear indicator, its fluctuation characteristics, and pre-set evaluation criteria. For example, if the current predicted value of the first wear indicator is 0.5, and the predicted value of the second wear indicator reaches 0.7 within the next five hours, and the evaluation criteria stipulate that the indicator value of 0.8 is the critical lifespan, the system may determine that the tool's remaining useful life is approximately eight hours. This analysis method can provide important reference for production planning, ensure the continuity of machining tasks, and avoid quality issues caused by excessive tool wear. For example, in the complete monitoring chain from data collection to life prediction, the system discovers abnormalities in the predicted value of the first wear indicator through real-time collection, confirms risk points through historical data extraction and fluctuation analysis, and predicts future trends to generate a second wear indicator predicted value. The final output is an analysis result containing a reference range for the remaining life. This closed-loop process can fully grasp the tool status, ensure processing stability, and effectively extend the tool life cycle. In summary, through the above steps, this embodiment, based on the LSTM model and dynamic data analysis, accurately predicts the future wear trend of cutting tools and provides a reliable basis for scientifically evaluating their remaining service life. This method significantly enhances the intelligent level of tool condition monitoring and provides strong support for efficient production in the manufacturing industry.

[0069] Figure 8 A flow chart of a method for adjusting the processing parameter instructions and updating the database is shown. Figure 8 As shown, the following steps are included.

[0070] Step S801 : performing wear trend analysis on the tool according to the wear evolution curve, and outputting a risk assessment result indicating whether the tool is in an accelerated wear stage.

[0071] Step S802 : adjusting the tool's machining parameter optimization strategy according to the risk assessment result, the second wear index prediction value, and the reference range of the tool's remaining service life, to obtain an adjusted machining parameter instruction.

[0072] Step S803 : updating the first wear index prediction value, the second wear index prediction value, the wear evolution curve, the adjusted machining parameter instruction, and the reference range of the remaining service life of the tool to the tool status monitoring database in real time.

[0073] In some embodiments, the machining parameter optimization strategy is automatically adjusted based on the second wear indicator prediction value and a reference range of the tool's remaining service life. Specifically, if the second wear indicator prediction value indicates a trend of accelerated wear, an adjusted machining parameter instruction is generated by reducing the machining speed or depth of cut, thereby extending the tool's service life.

[0074] As a feasible embodiment, the real-time monitoring system can acquire current machining status data to analyze the predicted value of the first wear indicator and the wear trend. For example, during a machining task, the system collects data on the predicted value of the first wear indicator every minute. The current value is 0.6, and the predicted value of the second wear indicator indicates that it may reach 0.75 within the next three hours, while the preset threshold is 0.8. In this case, the system detects an accelerating wear trend through comparison and generates a preliminary adjustment plan, recommending a 10% to 15% reduction in machining speed and a 0.2 to 0.3 mm reduction in cutting depth to slow wear. As a specific example, during the refinement of the initial adjustment plan, data feedback tools can be used to obtain real-time information on changes in machining status. For example, assuming that after adjustment, the machining speed is reduced to 90% and the cutting depth is reduced to 2.5 mm, the system detects a slowdown in the increase in the predicted value of the first wear indicator and predicts that the predicted value of the second wear indicator will only increase to 0.68 within the next three hours. Based on the remaining tool life reference range (the tool life threshold is 0.8), the system further adjusts the machining speed to 88% and the cutting depth to 2.4 mm, generating specific parameter instructions. This refinement ensures that the adjustment range is more closely aligned with actual machining needs. In some specific embodiments, a parameter instruction generation tool can be used to dynamically match the final parameter instructions to executable processing control signals. For example, suppose the equipment is currently operating at 500 rpm, and the instruction calls for an adjustment to 440 rpm. However, the actual output signal is 450 rpm, which deviates from the preset state. In this case, the system recalculates the adjustment range and generates a new control signal (435 rpm) to ensure that the equipment's operating state is consistent with the instruction. This dynamic matching mechanism can promptly correct deviations and improve processing control accuracy. In some specific embodiments, automated control tools can be used to update processing equipment parameters based on the application of new control signals. Assuming that, after the update, the machining speed stabilizes at 435 rpm and the cutting depth is 2.4 mm, the system, through real-time monitoring, discovers that the rate of increase of the predicted value of the first wear indicator has further slowed, indicating that the machining state is stabilizing. Based on this monitoring information, the system determines that the machining state is highly stable and reports a normal result. This stable feedback provides a reliable basis for subsequent machining tasks, ensuring the continuity of the production process and the rational use of tools. For example, the aforementioned steps, from data comparison to parameter adjustment, signal conversion, and equipment updates, form a complete closed-loop control process. Imagine, for example, that during a machining task, the system detects a trend of accelerated wear using the predicted value of the second wear indicator, promptly generates an adjustment plan and refines the parameters. Ultimately, automated tools are used to update the equipment and ensure stable machining conditions. This process effectively addresses the potential risks associated with tool wear and ensures the smooth execution of machining tasks. Furthermore, the tool status monitoring database is updated in real time according to the adjusted processing parameter instructions.

[0075] For example, in the field of tool condition monitoring, a real-time monitoring system can continuously track the tool's operating status by collecting the predicted first wear index and wear evolution curve data from machining equipment. Suppose that during a machining task, the system collects the predicted first wear index data every five minutes. The current value is 0.65, while the preset threshold is 0.7. After comparison, if the index approaches the critical value, the system generates preliminary status feedback information, marking the acquisition time and index value as the updated content. This preliminary feedback provides a basis for subsequent analysis. For example, to integrate data and generate a complete data package, a data storage tool can combine preliminary feedback with the predicted value of the first wear indicator. For example, if the prediction model indicates that the predicted value of the second wear indicator may reach 0.72 within the next two hours, the system will integrate this predicted data with the current measured value of 0.65 to form a data package containing a timestamp, indicator value, and predicted trend. An automated script tool then formats the data package, ensuring that fields are standardized (for example, time formatting is adjusted to year, month, day, hour, minute, and second) for easier database recognition and storage. This approach improves data processing efficiency. In some embodiments, during data upload to the tool condition monitoring database, the database management tool verifies data integrity. If the timestamp of a field in an uploaded data packet conflicts with existing closed-loop data, the system automatically adjusts the format (for example, by adding a unique identifier) ​​to ensure the data is not overwritten and determines whether it meets the entry requirements. This verification mechanism prevents data loss or duplication and ensures database accuracy. For example, dynamic modeling tools can be used to extract key data from tool status reports generated based on uploaded database information. For example, if the report shows the current predicted value of the first wear indicator is 0.65 and the predicted value of the second wear indicator is trending towards accelerated wear, the system will extract key fields from the report (such as wear value and predicted time period) based on the machining parameter adjustment requirements and determine the focus of subsequent monitoring as adjustments to machining speed and cutting depth. This dynamic report generation method provides clear guidance for subsequent decision-making. For example, in specific implementations, scenarios where the predicted value of the first wear indicator exceeds the preset range can be analyzed from multiple perspectives. For example, if the predicted value of the first wear indicator reaches 0.68 during a particular run, the system will not only generate status feedback but also analyze the causes of accelerated wear (such as the hardness of the processed material) based on historical data and provide recommendations to focus on equipment operational stability. Furthermore, the system can combine the predicted value of the second wear indicator to assess risks within the next hour and provide early warnings. This multi-faceted analysis provides a more comprehensive basis for parameter adjustments. In some embodiments, to address the business difficulties of database updates and report generation, the system uses automated scripts and verification mechanisms to resolve data conflicts and inconsistent formats. At the same time, the application of dynamic modeling tools makes report generation more flexible and adaptable to the needs of different processing scenarios. This approach not only ensures data accuracy, but also provides reliable support for determining monitoring priorities. In summary, through the above steps, this embodiment achieves dynamic adjustment of machining parameters based on the first and second wear indicator prediction values, and updates the tool status monitoring database via a closed-loop data flow. This method not only improves the real-time and accuracy of tool status monitoring but also extends tool life by optimizing machining strategies, providing a highly efficient solution for the intelligent upgrade of the manufacturing industry.

[0076] Figure 9 A flow chart of a method for obtaining risk assessment results is shown in FIG. Figure 9 As shown, the following steps are included.

[0077] Step S901 : Match the wear evolution curve with the historical wear evolution curve point by point, and extract the fluctuation characteristics of key time nodes based on the matching results.

[0078] Step S902 : performing wear trend analysis on the tool according to the fluctuation characteristics, and outputting a risk assessment result indicating whether the tool is in an accelerated wear stage.

[0079] In this embodiment, the historical wear evolution curve is the data collected by the system from the time a previously used tool of the same model was newly assembled until it was damaged. Comparing the current wear evolution curve with the historical wear evolution curve is equivalent to analyzing the current tool usage based on the historical usage of the same tool.

[0080] As a feasible embodiment, the current wear evolution curve can be compared with the historical wear evolution curve through point-by-point matching. For example, if the current wear evolution curve shows a rapid increase in the predicted value of the first wear index from 0.3 to 0.6 after 50 hours of machining time, while the historical wear evolution curve only increases from 0.3 to 0.4 at the same time point, this indicates that the current tool may be experiencing accelerated wear. This comparison can more clearly identify abnormal fluctuations and provide a basis for subsequent judgment. As a specific example, if fluctuation characteristics indicate a trend toward accelerated wear, a stage identification tool can be used to mark the tool's status. For example, if the predicted value of the first wear indicator for the current tool continues to rise rapidly after 50 hours, the system will automatically mark it as an accelerated wear stage and generate a stage identifier. This marking method helps clarify the tool's current status and facilitates the implementation of targeted measures (such as adjusting machining parameters or planning replacement). For example, when generating a risk assessment report, the report generation tool can integrate the predicted value of the first wear indicator and the fluctuation characteristic data. If the report shows the predicted value of the first wear indicator is 0.6 and the fluctuation characteristics indicate a clear trend of accelerated wear, the system will determine the risk level as high and store the report for later review. This report generation method provides an intuitive basis for management decision-making and ensures the stability of the processing process. For example, the implementation of each of the aforementioned steps can form a complete chain, from real-time data collection to final report generation. For example, in a single machining task, the system seamlessly integrates the entire process, from capturing the first wear indicator prediction value, comparing historical data, marking stages, to generating a risk assessment report. This complete system ensures comprehensive monitoring of tool status and significantly improves machining reliability through timely alarms and risk assessments. For example, in practice, the definition of critical time nodes requires incorporating significant changes in the predicted value of the first wear index. For example, if the predicted value of the first wear index rapidly increases from 0.3 to 0.6 at a certain point in time, indicating that the tool may be entering a stage of accelerated wear, this point in time is marked as a critical time node. By analyzing the fluctuation characteristics of such time nodes, potential risks can be more accurately identified. In summary, through the above steps, this embodiment dynamically assesses tool wear trends by comparing the current wear evolution curve with historical wear evolution curves, combined with analysis of the fluctuation characteristics of the predicted value of the first wear indicator. This provides a scientific basis for risk warning and machining parameter optimization. This method significantly enhances the intelligent level of tool condition monitoring through closed-loop data analysis and phased judgment.

[0081] The present application also provides a system embodiment that is consistent with the above embodiment, which is used to implement the method steps of the above embodiment. The explanation based on the same name meaning is the same as the above embodiment, and has the same technical effect as the above embodiment, which will not be repeated here.

[0082] like Figure 10 As shown, the present application provides an alloy tool wear prediction system 1000 based on machine vision, comprising: The acquisition unit 1001 is used to acquire initial surface image data of a tool.

[0083] The generating unit 1002 is configured to perform optimization and enhancement processing on the initial surface image data for tool wear detection to generate an optimized target surface image. The generating unit 1002 is configured to extract the tool's contour image data from the target surface image and, based on texture analysis of the contour image data, generate a comprehensive feature dataset capable of characterizing the tool's edge contour features and texture features.

[0084] Processing unit 1003 comprehensively quantifies the edge profile features and texture features represented by the comprehensive feature dataset to obtain a first wear index prediction value representing the current wear condition of the tool. The processing unit 1003 monitors the first wear index prediction value to trigger an abnormality alarm in the real-time monitoring system when the first wear index prediction value exceeds a preset prediction threshold.

[0085] Regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0086] Although operations are described in a particular order in the drawings, this should not be understood as requiring that the operations be performed in the particular order shown or in serial order, or that all shown operations be performed to obtain the desired results. In certain circumstances, multitasking and parallel processing may be advantageous.

[0087] The methods and systems of the present application can be implemented using standard programming techniques, using rule-based logic or other logic to implement the various method steps. It should also be noted that the terms "system" and "module" as used herein and in the claims are intended to include implementations using one or more lines of software code and / or hardware implementations and / or devices for receiving input.

[0088] Any steps, operations or procedures described herein may be performed or implemented using one or more hardware or software modules, either alone or in combination with other devices. In one embodiment, the software modules are implemented using a computer program product comprising a computer-readable medium containing computer program code, which can be executed by a computer processor to perform any or all of the steps, operations or procedures described.

[0089] The foregoing description of the implementation of the present application has been provided for purposes of illustration and description. The foregoing description is not intended to be exhaustive or to limit the present application to the precise form disclosed, and various variations and modifications are possible in accordance with the above teachings or may result from the practice of the present application. These embodiments have been selected and described in order to illustrate the principles of the present application and its practical application, so as to enable those skilled in the art to utilize the present application in various embodiments and modifications as appropriate for the particular use contemplated.

[0090] Regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0091] It is further understood that, unless otherwise specified, “connection” includes a direct connection where there are no other components between the two elements, and also includes an indirect connection where there are other elements between the two elements.

[0092] It should be further understood that although operations are described in a particular order in the drawings in the embodiments of the present application, this should not be construed as requiring that these operations be performed in the particular order shown or in a serial order, or that all of the illustrated operations be performed to obtain the desired results. In certain circumstances, multitasking and parallel processing may be advantageous.

[0093] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to encompass any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the field of the present application that are not disclosed herein. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the scope of claims below.

[0094] It should be understood that the present application is not limited to the precise structures described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the scope of the appended claims.

[0095] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application 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 replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for predicting alloy tool wear based on machine vision, characterized in that: include: Acquire initial surface image data of the tool; performing optimization and enhancement processing for tool wear detection on the initial surface image data to generate an optimized target surface image; Extracting the contour image data of the tool from the target surface image, and generating a comprehensive feature data set capable of characterizing the edge contour features and texture features of the tool based on texture analysis of the contour image data; Comprehensively quantifying the edge profile features and texture features represented by the comprehensive feature data set to obtain a first wear index prediction value representing the current wear condition of the tool; The first wear index prediction value is monitored to trigger a real-time monitoring system to issue an abnormal alarm when the first wear index prediction value exceeds a preset prediction threshold.

2. The method according to claim 1, characterized in that The obtaining of initial surface image data of the tool comprises: During the machining process, the surface of the tool is photographed in real time to obtain initial image data; extracting multiple frames of image data from the initial image data at preset time intervals, performing denoising on the multiple frames of image data, and determining whether the clarity of the multiple frames of image data reaches a preset clarity threshold; If the clarity of the multiple frames of image data does not reach a preset clarity threshold, adjusting the brightness and contrast of the multiple frames of image data to obtain optimized image data; The optimized image data is used as initial surface image data of the tool.

3. The method according to claim 1, characterized in that The performing optimization and enhancement processing on the initial surface image data for tool wear detection to generate an optimized target surface image includes: Performing preliminary filtering processing on the initial surface image data by using a mean filtering method to obtain a first processed image; Denoising the pixels in the first processed image whose pixel difference values ​​with adjacent areas are greater than a preset difference threshold, and performing secondary filtering based on filtering parameters matching the machining environment of the tool to obtain a second processed image; If the clarity of the second processed image does not reach a preset clarity threshold, performing brightness adjustment on the second processed image to obtain a third processed image; Perform edge enhancement processing on the second processed image or the third processed image to obtain the target surface image.

4. The method according to claim 1, wherein The generating of a comprehensive feature data set capable of characterizing edge profile features and texture features of the tool based on texture analysis of the profile image data comprises: Performing spatial relationship analysis between pixels based on grayscale changes between pixels in the contour image data to obtain comprehensive data for describing contours and textures; Performing brightness enhancement processing on the comprehensive data to obtain brightness-enhanced comprehensive data; Feature fusion is performed on the brightness-enhanced comprehensive data to obtain a comprehensive feature data set capable of characterizing the edge profile features and texture features of the tool.

5. The method according to claim 1, wherein The step of comprehensively quantifying the edge profile features and texture features represented by the comprehensive feature data set to obtain the first wear index prediction value includes: According to the comprehensive feature data set, splitting processing is performed on the edge contour feature and the texture feature to obtain the split contour data set and texture data set; Acquire a quantization benchmark from a preset mapping library, and obtain a quantization value group for the contour dataset and the texture dataset based on the quantization benchmark; performing integrity analysis on the numerical distribution of the quantized numerical value group, and if the integrity of the numerical distribution does not reach a completeness threshold, performing numerical adjustment processing on the quantized numerical value group to obtain an adjusted quantized numerical value group; performing a volatility analysis on the adjusted quantized value group, and calculating various index values ​​of the preset wear index based on the quantized value group when the volatility of the quantized value group conforms to a preset fluctuation range; The various index values ​​are weighted according to the preset weights corresponding to the various index values ​​to obtain a first wear index prediction value.

6. The method according to claim 1, characterized in that The method further comprises: Determine, based on historical data, how the predicted value of the first wear index changes over time, and generate a wear evolution curve for the tool; In combination with the wear evolution curve, the future wear trend of the tool is predicted to obtain a second wear index prediction value, and a reference range of the remaining service life of the tool is determined based on the second wear index prediction value; The second wear index prediction value represents the wear condition of the tool after a specified period of time in the future.

7. The method according to claim 6, characterized in that The method of predicting the future wear trend of the tool in combination with the wear evolution curve to obtain a second wear index prediction value includes: According to the time series information corresponding to the wear evolution curve, the wear data of the tool is correspondingly acquired in the real-time monitoring system, and the wear data and the wear evolution curve are combined into a complete data set; For the complete data set, fluctuation features are extracted in combination with the wear evolution curve and the wear data, so as to mark fluctuation features exceeding a preset fluctuation threshold as potential risk points; A wear trend prediction is performed on the potential risk point to obtain a second wear index prediction value.

8. The method according to claim 6, characterized in that The method further comprises: performing a wear trend analysis on the tool according to the wear evolution curve, and outputting a risk assessment result indicating whether the tool is in an accelerated wear stage; Adjusting a machining parameter optimization strategy for the tool according to the risk assessment result, the second wear index prediction value, and a reference range of the remaining service life of the tool to obtain an adjusted machining parameter instruction; The first wear index prediction value, the second wear index prediction value, the wear evolution curve, the adjusted processing parameter instruction, and the reference range of the remaining service life of the tool are updated in real time to a tool status monitoring database.

9. The method according to claim 8, characterized in that The step of performing wear trend analysis on the tool according to the wear evolution curve and outputting a risk assessment result indicating whether the tool is in an accelerated wear stage includes: Matching the wear evolution curve with the historical wear evolution curve point by point, and extracting the fluctuation characteristics of key time nodes based on the matching results; According to the fluctuation characteristics, a wear trend analysis is performed on the tool, and a risk assessment result is outputted indicating whether the tool is in an accelerated wear stage.

10. A machine vision-based alloy tool wear prediction system, characterized in that: include: an acquisition unit, configured to acquire initial surface image data of a tool; a generating unit, configured to perform optimization and enhancement processing on the initial surface image data for tool wear detection to generate an optimized target surface image; and extracting contour image data of the tool from the target surface image, and generating a comprehensive feature data set capable of characterizing edge contour features and texture features of the tool based on texture analysis of the contour image data; A processing unit comprehensively quantifies the edge profile features and texture features represented by the comprehensive feature data set to obtain a first wear index prediction value representing the current wear condition of the tool; and monitors the first wear index prediction value to trigger a real-time monitoring system to issue an abnormal alarm when the first wear index prediction value exceeds a preset prediction threshold.

Images