Method and system for monitoring the machining of alloy parts for aeronautics

Abstract

The application provides an alloy spare part machining monitoring method and system for aviation, and relates to the technical field of machining monitoring.The method comprises the following steps: for a target alloy spare part of an aviation structural part, quality requirement information and design information are determined; a target machining process and a machining control parameter sequence are determined; during a trial machining process, a plurality of machining monitoring data sets of a plurality of machining nodes are collected through pre-arranged machining monitoring units; machining error analysis is performed to obtain trial machining error monitoring results; after the trial machining process ends, quality evaluation is performed to obtain trial machining quality monitoring results; quality defect analysis is performed to obtain trial machining quality defect monitoring results; and the machining monitoring results are integrated. The application solves the technical problem that the machining monitoring of the prior art cannot provide real-time and accurate machining error monitoring, which leads to error accumulation and finally affects the machining quality of the aviation alloy spare part.

Description

Technical Field

[0001] This invention relates to the field of processing monitoring technology, and specifically to a method and system for monitoring the processing of alloy parts for aviation. Background Technology

[0002] As a crucial component of aircraft, the quality and precision of aerospace alloy parts directly impact the aircraft's performance, safety, and durability. With continuous advancements in aviation technology, the quality requirements for aircraft parts are increasingly stringent, particularly in the manufacturing of alloy parts, where high geometric accuracy, structural strength, and durability must be ensured. However, current technologies for monitoring the processing of aerospace alloy parts often focus on post-processing quality inspection or error analysis through intermittent data collection. This leads to difficulties in timely detection of accumulated processing errors. For instance, traditional error detection methods cannot provide real-time feedback on accuracy issues at each processing node, or errors are detected too late, often impacting subsequent processes and even the efficiency of the entire production cycle, ultimately affecting the quality of aerospace alloy parts. Summary of the Invention

[0003] This application provides a method and system for monitoring the processing of alloy parts for aviation, aiming to solve the technical problem that existing processing monitoring focuses on quality inspection after processing or error analysis through intermittent data collection, which cannot provide real-time and accurate processing error monitoring, leading to error accumulation and ultimately affecting the processing quality of aviation alloy parts.

[0004] The first aspect disclosed in this application provides a method for monitoring the processing of alloy parts for aerospace applications. The method includes: determining quality requirement information and design information for a target alloy part of an aerospace structural component; determining a target processing flow and a sequence of processing control parameters based on the design information; during trial processing of the target alloy part based on the processing control parameter sequence, acquiring multiple processing monitoring data sets for multiple processing nodes through pre-deployed processing monitoring units; performing processing error analysis based on the multiple processing monitoring data sets to obtain trial processing error monitoring results; after the trial processing is completed, performing a quality assessment on the acquired trial-processed alloy part to obtain trial processing quality monitoring results; performing quality defect analysis on the trial processing quality monitoring results based on the quality requirement information to obtain trial processing quality defect monitoring results; and integrating the trial processing error monitoring results and the trial processing quality defect monitoring results as the processing monitoring results for the target alloy part.

[0005] The second aspect of this application discloses a system for monitoring the processing of alloy parts for aviation. This system is used in the aforementioned method for monitoring the processing of alloy parts for aviation. The system includes: a design information acquisition module for determining quality requirements and design information for a target alloy part of an aviation structural component; a control parameter determination module for determining a target processing flow and a sequence of processing control parameters based on the design information; a monitoring data acquisition module for collecting multiple processing monitoring data sets from multiple processing nodes through pre-deployed processing monitoring units during trial processing of the target alloy part based on the processing control parameter sequence; a processing error analysis module for performing processing error analysis based on the multiple processing monitoring data sets to obtain trial processing error monitoring results; a quality assessment module for performing a quality assessment on the acquired trial-processed alloy parts after the trial processing is completed to obtain trial processing quality monitoring results; a quality defect analysis module for performing quality defect analysis on the trial processing quality monitoring results based on the quality requirements information to obtain trial processing quality defect monitoring results; and a processing monitoring result acquisition module for integrating the trial processing error monitoring results and the trial processing quality defect monitoring results as the processing monitoring results for the target alloy part.

[0006] One or more technical solutions provided in this application have at least the following beneficial effects:

[0007] By determining the quality requirements and design information of the target alloy parts, a data foundation was laid for quality and precision control during processing. Based on the design information, the processing flow and control parameter sequence were determined, further optimizing the processing technology and related parameter settings. This ensured that the processing flow did not deviate from the design requirements throughout the entire production cycle, improving production efficiency and product consistency. Monitoring data from multiple processing nodes through a processing monitoring unit allowed for real-time monitoring of the workpiece status during processing. This precise data acquisition ensured that every node in the processing received sufficient attention, enabling timely detection of processing problems. Utilizing data from different processing nodes for processing error analysis accurately identified the sources of deviation during processing, preventing errors from gradually escalating in subsequent processes. Large errors can affect the final product quality. Real-time error monitoring and analysis improve the controllability of the processing process. Based on the collected monitoring data, the quality of trial-processed alloy parts is evaluated. This not only helps to identify potential quality defects in alloy parts during trial processing, but also allows for timely adjustment of processing parameters and necessary compensation, helping to adjust the process to ensure the quality of subsequent parts. Integrating the processing error monitoring results with the quality defect monitoring results, the final processing monitoring results provide comprehensive feedback for the entire processing process. By comprehensively analyzing processing errors and quality defects, potential problems in the production process can be identified more accurately, and timely adjustments can be made through data-driven optimization measures. This allows problems in the production process to be discovered and resolved earlier, further ensuring the precision and quality of parts.

[0008] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of a process for monitoring the processing of alloy parts for aviation, provided in an embodiment of this application.

[0010] Figure 2 This is a schematic diagram of the structure of an alloy parts processing monitoring system for aviation provided in an embodiment of this application.

[0011] Explanation of reference numerals in the attached drawings: 10 for design information acquisition module, 20 for control parameter determination module, 30 for monitoring data acquisition module, 40 for processing error analysis module, 50 for quality assessment module, 60 for quality defect analysis module, and 70 for processing monitoring result acquisition module. Detailed Implementation

[0012] This application provides a method and system for monitoring the processing of alloy parts for aviation, which solves the technical problem that the existing processing monitoring focuses on quality inspection after processing or error analysis through intermittent data collection, which cannot provide real-time and accurate processing error monitoring, leading to error accumulation and ultimately affecting the processing quality of aviation alloy parts.

[0013] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0014] Example 1, as Figure 1 As shown in the embodiments of this application, a method for monitoring the processing of alloy parts for aviation is provided, the method comprising:

[0015] For target alloy components of aerospace structural parts, determine quality requirements and design information.

[0016] Based on the functional requirements of aerospace structural components, the physical, chemical, and mechanical properties of the target alloy parts after processing are clearly defined. For example, for aerospace structural components, this involves fatigue strength, weight, and stiffness, and also requires consideration of operating conditions such as ambient temperature and workload. The obtained quality requirement information includes the design standards, functional requirements, dimensional tolerances, surface quality requirements, and strength requirements of the parts; these requirements serve as the benchmark for quality defect analysis. Design information includes the geometry, dimensions, material properties, and tolerance requirements of the target alloy parts. The structure and geometric characteristics of the target alloy parts are determined using CAD or other design software.

[0017] Based on the design information, the target processing flow and the sequence of processing control parameters are determined.

[0018] Based on the design requirements of the target alloy parts, determine the required machining types, such as milling, turning, and drilling, as well as the machining sequence. For example, for high-strength alloy materials, select special heat treatment or hard machining methods, and identify the key equipment and tools in each machining step, such as CNC machine tools and laser cutting equipment. In each machining step, define the corresponding machining control parameters. For example, for milling, it is necessary to control the cutting speed, feed rate, and tool angle, establishing a sequence of machining control parameters.

[0019] During the trial processing of the target alloy parts based on the processing control parameter sequence, multiple processing monitoring data sets of multiple processing nodes are collected by pre-deployed processing monitoring units.

[0020] During the trial machining phase of the target alloy parts, preliminary trial machining operations are initiated based on the determined machining process and machining control parameter sequence. Trial machining is typically conducted before actual production to verify the adaptability of the process and equipment. During the trial machining phase, machining monitoring units, such as sensors and data acquisition systems, are installed to collect parameters at each machining node in real time. The collected data includes different types of data at each machining node, such as workpiece displacement, vibration sensor feedback, and temperature sensor data. These data are then aggregated to generate multiple machining monitoring datasets, facilitating subsequent error analysis and quality assessment.

[0021] Based on the multiple processing monitoring data sets, processing error analysis is performed to obtain the trial processing error monitoring results.

[0022] The processing monitoring dataset includes workpiece processing data and processing equipment status data. Based on the workpiece's geometry and dimensional requirements, it analyzes geometric deviations during actual processing. For example, it analyzes whether the workpiece's shape and dimensions conform to design specifications using sensor data. It monitors the processing equipment's status, such as wear and vibration, based on the processing equipment status data, analyzing whether there is equipment performance degradation that affects processing accuracy. The dataset analyzes the sources and influencing factors of errors to obtain trial processing error monitoring results.

[0023] After the trial processing is completed, the quality of the obtained trial-processed alloy parts is evaluated, and the trial processing quality monitoring results are obtained.

[0024] After the trial processing is completed, the quality of the finished alloy parts is evaluated to check whether they meet the quality standards. Specifically, visual inspection systems, such as machine vision and image processing technology, are used to inspect the surface of the alloy parts to confirm the presence of surface defects such as cracks, scratches, and dents. Non-destructive testing techniques, such as ultrasonic testing, X-ray testing, and three-dimensional geometric measurement, are used to evaluate the internal quality of the parts, checking for defects such as porosity, cracks, and inclusions, and confirming their geometric accuracy and dimensions. The parts are then tested for mechanical properties such as hardness, tensile strength, and fatigue to ensure they can withstand the required loads during use. The results of the appearance quality, structural quality, and mechanical performance evaluations are integrated to generate a comprehensive trial processing quality monitoring result.

[0025] Based on the quality requirement information, a quality defect analysis is performed on the trial processing quality monitoring results to obtain the trial processing quality defect monitoring results.

[0026] Quality requirements information includes design standards, functional requirements, dimensional tolerances, surface quality requirements, and strength requirements for components. These requirements serve as the benchmark for quality defect analysis. By comparing and analyzing the trial processing quality monitoring results, the actual results are compared with the design requirements. Parts that exceed tolerance ranges, have defects, or do not meet design standards are marked. Based on the severity, type, and impact on overall function of the defects, the need for correction or whether they can be resolved through other compensation methods is assessed. Trial processing quality defect monitoring results are generated, listing all discovered defects, defect types, defect locations, and defect severity.

[0027] The results of the trial processing error monitoring and the results of the trial processing quality defect monitoring are integrated as the processing monitoring results of the target alloy parts.

[0028] The integrated trial processing error monitoring results and trial processing quality defect monitoring results are used as the processing monitoring results for the target alloy parts. All potential quality problems before and after trial processing can be identified in a timely manner, and necessary feedback can be provided for the subsequent production process, ultimately ensuring that the parts meet the stringent standards of the aviation industry.

[0029] Furthermore, the step of performing processing error analysis based on the multiple processing monitoring data sets to obtain trial processing error monitoring results includes:

[0030] Extract the first processing monitoring data set of the first processing node, wherein the first processing monitoring data set includes workpiece processing monitoring data and processing equipment status monitoring data; perform workpiece processing deviation analysis based on the workpiece processing monitoring data to generate workpiece processing deviation monitoring results; perform processing equipment wear analysis based on the processing equipment status monitoring data to generate processing equipment fatigue monitoring results; add the workpiece processing deviation monitoring results and the processing equipment fatigue monitoring results to the trial processing error monitoring results.

[0031] Based on the machining process design, the first machining node in the trial machining process is selected, for example, at the start of cutting, the moment the tool contacts the workpiece, or the moment the feed rate changes. Workpiece machining monitoring data includes the actual machining position of the workpiece, typically acquired using laser displacement sensors, optical sensors, etc.; and geometric data, such as dimensional errors, angular deviations, and shape deviations, acquired using equipment such as a 3D coordinate measuring machine. Machining equipment status monitoring data includes machining equipment temperature, such as machine tool spindle temperature and cutting tool temperature, which have a significant impact on machining accuracy; machining equipment vibration data, captured by accelerometers or vibration monitoring systems to assess whether there is resonance or unstable operation of the equipment; and tool wear data, typically detected by sensors or cameras.

[0032] Workpiece machining deviation analysis is performed. For example, the geometric features of the workpiece, such as straightness, flatness, and roundness, are analyzed. The design dimensions are compared with the actual dimensions to identify dimensional deviations. Based on the above analysis results, specific workpiece machining deviation monitoring results are generated, including the specific value of the deviation (such as tolerance exceeding the tolerance), the location and type of the defect, and the possible impact on product performance.

[0033] Analyzing the condition monitoring data of the machining equipment identifies wear or fatigue phenomena, thereby generating fatigue monitoring results. Specifically, temperature analysis is performed on the machining equipment, as overheating may be caused by tool wear or equipment failure. Analyzing temperature changes can help determine if excessive wear or failure exists. Vibration analysis is also conducted, as vibration data can reveal the operating status of the machining equipment, such as whether there is imbalance, resonance, or other fault phenomena. Tool wear analysis is performed, by monitoring the degree of tool wear or service life, to analyze whether the tool needs to be replaced. If the tool is excessively worn, it may lead to increased machining errors.

[0034] By combining the workpiece machining deviation monitoring results with the machining equipment fatigue monitoring results, a comprehensive trial machining error monitoring result is formed, which shows the geometric and dimensional deviations of the workpiece itself, as well as the potential impact of the health status and fatigue level of the machining equipment on the error.

[0035] Furthermore, the step of performing workpiece machining deviation analysis based on the workpiece machining monitoring data to generate workpiece machining deviation monitoring results includes:

[0036] Multi-source feature extraction is performed on the workpiece processing monitoring data to obtain multi-source processing characterization parameters. These parameters are then combined to obtain the target workpiece processing feature vector. The standard alloy parts and the target alloy parts are locally retrieved, and their processing deviations are analyzed to obtain a workpiece processing deviation vector. The standard alloy parts and the target alloy parts originate from the same production batch and are prepared according to a standard process flow. Multiple sets of sample workpiece processing deviation vectors from multiple sample workpiece processing deviation records are locally retrieved and clustered to obtain multiple sample workpiece processing deviation vector domains. Trigger confidence evaluation is performed on these multiple sample workpiece processing deviation vector domains to obtain multiple sample trigger confidence evaluation coefficients. The multiple sample workpiece processing deviation vector domains and the multiple sample trigger confidence evaluation coefficients are associated and stored to construct a deviation distribution embedding space. The workpiece processing deviation vectors are projected onto this deviation distribution embedding space, and a workpiece processing deviation probability distribution is generated based on vector similarity measurement and probability mapping inference. Confidence is calculated based on the workpiece processing deviation probability distribution to obtain a deviation grading confidence level, which is then added to the workpiece processing deviation monitoring results.

[0037] Workpiece machining monitoring data comes from multiple sensors and measuring devices, such as laser displacement sensors, force sensors, and temperature sensors. Each sensor collects data providing different information about the workpiece machining process. For each data source, different feature extraction methods are used to extract representative information from the raw data; for example, Fourier transform is used to extract vibration or oscillation information at different frequencies during machining. The features extracted from multiple data sources are combined to form a target workpiece machining feature vector, which contains information reflecting the workpiece machining state from different perspectives.

[0038] A standard alloy component is selected from the same production batch and processed according to a standard process flow. Through monitoring and measurement, a standard workpiece processing feature vector is generated, including all processing features such as size, shape, and surface quality. The target workpiece's processing feature vector is compared with the standard workpiece's processing feature vector to identify differences in the processing. Specifically, this is done by calculating metrics such as Euclidean distance, Manhattan distance, or cosine similarity between the feature vectors to determine the deviation between the target workpiece's processing and the standard workpiece. This results in a workpiece processing deviation vector, which represents the differences between the target workpiece and the standard workpiece in terms of size, shape, and surface quality. Furthermore, the degree of deviation for each feature can be quantified.

[0039] Multiple sets of processing deviation records for sample workpieces were extracted from historical data. These records originated from different batches of spare parts, and their processing processes and results exhibited a certain degree of diversity. The processing deviation of each sample workpiece was analyzed using the methods described in the preceding steps to obtain a corresponding sample workpiece processing deviation vector. Clustering algorithms, such as K-means clustering and DBSCAN, were used to perform cluster analysis on the multiple sets of sample workpiece processing deviation vectors. The goal of cluster analysis was to group sample workpieces with similar deviation patterns into the same category, identifying potential processing problems. Based on the clustering results, multiple sample workpiece processing deviation vector domains were obtained. These domains represent a specific processing deviation pattern. Each sample workpiece processing deviation vector domain contained multiple workpieces with similar processing deviations, indicating that their processing processes and results had similar characteristics.

[0040] Trigger credibility evaluation is performed on the machining deviation vector domain of each sample workpiece. This involves assessing the authenticity and stability of the machining deviation pattern represented by each vector domain. By using the sample trigger credibility evaluation coefficient, it is possible to more accurately identify which machining deviation patterns are valid and which may be errors or noise. The trigger credibility evaluation is based on factors such as the machining quality of the sample workpieces, the reliability of historical data, and the number of samples. For example, if the number of samples in a certain deviation vector domain is small or the deviation is large, the credibility of that domain will decrease; if the sample deviation distribution in a certain deviation vector domain is relatively consistent, the credibility is high. Weighted averages and confidence intervals are used to calculate the sample trigger credibility evaluation coefficient for the machining deviation vector domain of each sample workpiece.

[0041] The processing deviation vector domain of each sample workpiece is associated with its corresponding sample trigger credibility evaluation coefficient to form a one-to-one storage structure. This provides a credibility label for each processing deviation pattern. The deviation distribution embedding space is a high-dimensional space, where each dimension represents a different processing feature of the workpiece. The position (or embedding) of the processing deviation vector domain of each sample workpiece in this space is determined according to its sample trigger credibility evaluation coefficient and the actual processing deviation pattern. In this space, processing deviation vectors with higher credibility will be given greater weight, while vectors with lower credibility will be given smaller weight or be dimensionality reduced.

[0042] The workpiece machining deviation vector of the target alloy part is projected into the constructed deviation distribution embedding space. Through projection, the workpiece machining deviation vector is mapped to a specific position in the space. By measuring the similarity between the target workpiece deviation vector and the existing sample vectors in the embedding space, it is possible to infer which deviation patterns the machining deviation of the target alloy part is most similar to. Based on this similarity measure, probabilistic mapping inference is performed to calculate the probability that the machining deviation of the target alloy part belongs to different deviation patterns. Finally, the workpiece machining deviation probability distribution is generated, which is the probability of the occurrence of the target alloy part under different deviation patterns.

[0043] Based on the generated probability distribution of workpiece machining deviations, the confidence level is calculated for each possible deviation pattern. The confidence level reflects the reliability of a particular deviation pattern; a higher probability implies a higher confidence level. Machining deviations are categorized into different levels, and a corresponding confidence level value is assigned to each level. For example, low deviations correspond to low probabilities and confidence levels, medium deviations to high probabilities and confidence levels, and high deviations to very high probabilities and confidence levels, indicating that the deviation pattern has a significant impact on the workpiece machining accuracy. Finally, the calculated deviation grading confidence levels are added to the workpiece machining deviation monitoring results and used as a basis for subsequent workpiece machining quality evaluation.

[0044] Furthermore, the step of calculating the confidence level based on the probability distribution of the workpiece processing deviation to obtain the deviation grading confidence level includes:

[0045] Multidimensional covariance modeling is performed based on the processing deviation vectors of the multiple sets of sample workpieces to obtain the processing deviation correlation coefficient matrix between each deviation dimension; joint probability inference is performed on the probability distribution of the workpiece processing deviation based on the processing deviation correlation coefficient matrix and the single-dimensional probability value is corrected to obtain a weighted corrected multidimensional joint deviation probability distribution; mutual information difference test and conditional independence test are performed on the multidimensional joint deviation probability distribution, and the deviation classification confidence level is generated based on the test results.

[0046] The sample workpiece processing deviation vector describes different processing characteristics in a multidimensional space. Therefore, the correlation between deviations in different dimensions can be measured by covariance. Covariance is a statistical measure used to measure the strength and direction of the relationship between two variables. Multidimensional covariance modeling involves analyzing the relationships between multiple deviation dimensions (such as size, surface quality, shape, etc.) to establish a statistical relationship model between them. The covariance matrix contains the covariance between each deviation dimension, while the processing deviation correlation coefficient matrix is ​​a standardized form of the covariance matrix, which can quantify the correlation strength between different deviation dimensions. The correlation coefficient ranges from -1 to 1, where -1 indicates a perfect negative correlation, 0 indicates no linear correlation, and 1 indicates a perfect positive correlation. The processing deviation correlation coefficient matrix can better illustrate the interaction between different processing deviations. For example, a deviation in one processing step may affect the result of another processing step; this influence can be quantified using the covariance matrix.

[0047] By performing joint probabilistic inference on each deviation dimension based on the processing deviation correlation coefficient matrix, the individual probability distribution of each deviation dimension is corrected. This is because the correlation between deviation dimensions implies that they are not independent, thus requiring consideration of their joint distribution. Specifically, the original single-dimensional probability distribution cannot fully reflect the mutual influence between deviation dimensions. By introducing joint probabilistic inference, the probability of each deviation dimension can be corrected, ensuring that their distribution better reflects the actual situation of the overall deviation. Weighted correction means weighting the deviation values ​​of different dimensions to reflect their contribution to the overall deviation. For example, a deviation in one dimension has a greater impact on the quality of the final workpiece, so its corrected probability value will be higher. The corrected multidimensional joint deviation probability distribution integrates the interactions of all deviation dimensions, and is therefore more comprehensive and accurate than individual probability distributions.

[0048] Mutual information is used to quantify the degree of information sharing between two variables. For processing bias, the mutual information difference test can be used to determine the strength of information association between different processing bias dimensions. By calculating the information difference between each dimension, redundant or unnecessary dimensions can be eliminated, thereby improving the accuracy of the model. If the mutual information difference between two dimensions is high, it indicates that the information association between them is weak, and vice versa. This helps to determine which dimensions have higher value for the assessment of the overall processing bias.

[0049] The conditional independence test is used to determine whether two bias dimensions are independent of each other given a third bias dimension. If the two dimensions are conditionally independent, their changes are not affected by each other. In multidimensional space, mutually exclusive biases (i.e., completely opposite bias patterns) may interfere with each other and affect the reliability of the biases. By using the conditional independence test, these irrelevant interference terms can be eliminated, ensuring the independence between biases.

[0050] By using mutual information difference tests and conditional independence tests, the weights of each deviation dimension can be further determined, thereby assigning confidence levels to different deviation patterns in the multidimensional joint deviation probability distribution. These confidence levels reflect the reliability and impact of a certain deviation pattern. Generally, higher mutual information difference or stronger conditional independence will lead to higher confidence levels.

[0051] Furthermore, this includes:

[0052] Based on the workpiece machining deviation monitoring results, machining error matching is performed to generate a subset of workpiece deviation compensation parameters; based on the machining equipment fatigue monitoring results, machining error matching is performed to generate a subset of equipment fatigue compensation parameters; modeling, calculation, and parameter fusion are performed on the subset of workpiece deviation compensation parameters and the subset of equipment fatigue compensation parameters respectively to generate a dynamic compensation parameter group; based on the dynamic compensation parameter group, the machining control parameter sequence is periodically iteratively optimized.

[0053] Workpiece machining deviation monitoring results reflect the deviation data of the workpiece during the machining process, including dimensional deviations, shape deviations, surface quality deviations, etc. By comparing the actual measured workpiece machining deviations with the standard values ​​of the ideal workpiece, the errors that occur during machining are determined. Error matching is the process of identifying and quantifying these differences. The comparison can use error minimization algorithms, such as the least squares method or optimization methods, to determine the deviations in workpiece machining. Through error matching, workpiece deviation compensation parameters are calculated, which are the machining adjustment values ​​required to correct these errors. According to different types of deviations, such as size, shape, and surface quality, multiple workpiece deviation compensation parameters are generated. Each workpiece deviation compensation parameter corresponds to a different type of deviation. For example, dimensional deviations require adjustment of the cutting depth or feed rate, while surface quality deviations require adjustment of the cutting tool state.

[0054] During the manufacturing process, fatigue of machining equipment can lead to a decrease in machining accuracy. Fatigue manifests as wear, vibration, or changes in mechanical precision of mechanical parts. Based on the fatigue monitoring results of machining equipment, the error caused by fatigue of the machining equipment is compared with the expected machining result through an error matching method. This allows for the identification of errors caused by equipment fatigue and the generation of corresponding equipment fatigue compensation parameters. These parameters are used to adjust the working state or machining control settings of the machining equipment, such as adjusting the feed rate and tool path, to reduce the impact of fatigue on machining quality.

[0055] For each compensation subset, different mathematical modeling techniques are used for parameter calculation. For workpiece deviation compensation, an error propagation model can be used to calculate how to adjust machining parameters to counteract the workpiece deviation. For equipment fatigue compensation, a dynamic model can be used to predict how the fatigue state of the equipment affects machining accuracy and derive corresponding compensation measures. These compensation measures are then integrated to generate a dynamic compensation parameter set, which is a set of parameters that changes with the machining process and equipment state. It adjusts the parameters in real time during the machining process to optimize machining accuracy.

[0056] The purpose of the dynamic compensation parameter set is to adjust and optimize the machining control parameter sequence in real time to ensure that the impact of workpiece machining error and equipment fatigue is minimized. Through iterative optimization, the machining control parameter sequence will be continuously adjusted to adapt to changes in the workpiece and equipment. For example, as equipment fatigue accumulates, the control parameters need to be adjusted step by step.

[0057] Furthermore, the quality assessment of the obtained trial-processed alloy parts and the acquisition of trial-processing quality monitoring results include:

[0058] The trial-processed alloy parts are transferred to the appearance quality assessment zone, where appearance defects are detected by an appearance defect detection module to obtain appearance quality assessment results. This appearance quality assessment zone is equipped with an all-around image acquisition device. The trial-processed alloy parts are then transferred to the structural quality assessment zone, where internal defects and geometric accuracy are detected by a non-destructive testing module to obtain structural quality assessment results. This non-destructive testing module includes an ultrasonic flaw detector, an X-ray imaging device, and a three-dimensional geometric measurement device. The trial-processed alloy parts are then transferred to the mechanical performance assessment zone, where mechanical performance is assessed by a mechanical performance testing module to obtain mechanical performance assessment results. This mechanical performance testing module includes a hardness testing device, a tensile testing device, and a fatigue testing machine. The appearance quality assessment results, structural quality assessment results, and mechanical performance assessment results are integrated to obtain the trial processing quality monitoring results.

[0059] The appearance quality assessment zone is used for a comprehensive inspection of the appearance of trial-processed alloy parts to identify surface defects such as scratches, cracks, porosity, and surface roughness. This zone is equipped with an omnidirectional image acquisition device to capture images of the parts from all angles. After processing by the appearance defect detection module, the appearance quality assessment results are generated, listing all detected defects, their location, size, and type, and assessing their impact on the quality of the parts.

[0060] The structural quality assessment zone is used to inspect the internal defects and geometric accuracy of test-machined alloy components. The key to structural quality assessment is detecting internal defects in alloy components, such as cracks, porosity, and inclusions, as well as whether the external geometry meets design requirements. Non-destructive testing (NDT) technology is a method to detect internal defects and geometry without damaging the material. Ultrasonic testing devices use ultrasonic waves to propagate through the material and utilize reflected waves to detect internal defects such as cracks, porosity, and inclusions. The time and intensity of the reflected waves can reveal the size, location, and type of defects. X-ray imaging devices use X-rays to penetrate the components and utilize different absorption rates to display internal defects such as porosity, cracks, and inclusions. X-ray images can provide two-dimensional or three-dimensional views to help determine whether the internal structure of the components is up to standard. Three-dimensional geometric measurement devices use laser scanning, contact probes, or optical sensors to scan the surface of the components and construct a three-dimensional model. By comparing this model with the design model, the geometric accuracy of the components is evaluated. After inspection by the NDT module, a structural quality assessment result is generated, listing the presence of internal defects and deviations in geometric accuracy of the components.

[0061] The mechanical performance evaluation zone applies a series of mechanical performance tests to trial-machined alloy parts to ensure they can withstand corresponding physical loads in actual use, maintain sufficient strength and toughness, and avoid failure. Specifically, the hardness testing device presses a hard indenter with a known load into the surface of the alloy part to test its surface hardness. Hardness is a material's ability to resist localized plastic deformation, and Rockwell hardness, Vickers hardness, and Brinell hardness are commonly used for testing. The tensile testing device fixes the sample in a tensile testing machine and gradually increases the tensile force until the material fractures. During the test, the stress-strain curve of the material is recorded, thus obtaining key mechanical performance indicators such as tensile strength, yield strength, and elongation. The fatigue testing machine periodically loads the alloy parts to simulate the cyclic stress they may experience in actual working environments until fatigue fracture occurs. Fatigue testing is used to evaluate the durability of parts under long-term repeated stress, ensuring their reliability in practical applications, especially in high-load, repeated-load environments, such as in demanding fields like aero-engines.

[0062] Different types of assessment results have different importance to the final quality assessment. Therefore, when integrating them, a weighted approach can be used to reflect the relative importance of each assessment. For example, appearance quality has a greater impact on the aesthetics and functionality of parts, but for some simpler mechanical parts, structural quality and mechanical properties have a more significant impact. The integrated trial processing quality monitoring results provide a basis for subsequent production decisions, including whether rework, repair or replacement of parts is needed, or whether further quality optimization is needed to improve the production process.

[0063] Furthermore, obtaining the appearance quality assessment result includes:

[0064] When the trial-processed alloy parts pass through the omnidirectional image acquisition device, omnidirectional image acquisition results are acquired; standard image acquisition results of standard alloy parts are locally retrieved; semantic segmentation of the standard image acquisition results and the omnidirectional image acquisition results is performed by the appearance defect detection module to obtain appearance defect detection results, wherein the appearance defect detection results have defect area identifiers.

[0065] The omnidirectional image acquisition device consists of multiple cameras or image sensors, capable of capturing surface images of trial-processed alloy parts from different angles and orientations. These image data cover the entire surface of the parts, including hidden micro-defect areas. Comprehensive image acquisition provides complete image data support for subsequent appearance defect inspection, resulting in omnidirectional image acquisition results.

[0066] The standard image acquisition results are derived from standard alloy parts that have undergone rigorous testing and are free of defects. They are acquired under the same environmental and equipment conditions and can be used as comparison templates to evaluate the appearance quality of currently trial-processed parts.

[0067] The appearance defect detection module is specifically designed to analyze acquired images, automatically identifying and marking defective regions within them. Semantic segmentation technology assigns each pixel in the image to a category, including defective and non-defective areas. After processing, all detected defective regions are marked. The appearance defect detection results display the specific defective regions, including their exact location, size, type, and other information. These markings can be displayed in the form of color coding, bounding boxes, or pixel labels, facilitating rapid location and judgment.

[0068] Furthermore, the semantic segmentation of the standard image acquisition results and the omnidirectional image acquisition results using the appearance defect detection module includes:

[0069] The appearance defect detection module includes a first convolutional neural network and a second convolutional neural network. The first and second convolutional neural networks have symmetrical network structures, share parameters, and have consistent feature spaces. The first convolutional neural network extracts features from the standard image acquisition results to obtain first image pixel features, and the second convolutional neural network extracts features from the omnidirectional image acquisition results to obtain second image pixel features. The first and second image pixel features are used to calculate the difference based on pixel coordinates, and the defect area is identified and marked based on the difference threshold to generate the appearance defect detection result.

[0070] Convolutional neural networks (CNNs) are used for image recognition tasks. In appearance defect detection, CNNs can effectively extract hierarchical features from images and identify surface defects. The first and second CNNs have symmetrical network structures, meaning they have similar network architectures and layers. This symmetrical structure maintains consistent learning and feature extraction capabilities when processing two different input images (standard image and trial processing image). Parameter sharing means that the two networks share parameters of certain convolutional layers or kernels, which effectively reduces computational resources and ensures that both networks learn image features from the same perspective. Feature space consistency means that the features extracted by the two networks are spatially aligned, i.e., their extracted feature vectors have similar dimensions and representations. Through this structure, the appearance defect detection module can ensure that the features obtained when analyzing two types of images are compatible, which is helpful for subsequent defect detection.

[0071] The standard image acquisition results are input into a first convolutional neural network (CNN). The first CNN extracts low-level to high-level features from the image, such as edges, textures, and shapes, forming the first image pixel features. The omnidirectional image acquisition results are input into a second CNN, which extracts features from the image through a similar process, forming the second image pixel features. The first and second image pixel features contain key information about the image.

[0072] The pixel features of the first image (features of the standard image) and the pixel features of the second image (features of the trial-processed parts) are compared. Based on the pixel coordinates, the difference between the two at each pixel position is calculated. The difference can be measured in various ways, such as pixel value difference, texture difference, or shape difference. A difference threshold is set. When the difference between a certain pixel point of the two images exceeds the difference threshold, it is determined that there is a defect at that position. The detected defect area is marked, usually using bounding boxes, outlines, or other visual marking methods to indicate the location, size, and shape of the defect.

[0073] Furthermore, the difference threshold includes a pixel grayscale difference threshold, a local texture difference threshold, and a difference pixel ratio threshold.

[0074] The pixel grayscale difference threshold refers to the difference in grayscale values ​​between two image pixels at the same location. Grayscale value is the brightness value of each pixel in an image, usually between 0 (black) and 255 (white). If the grayscale difference at a certain location exceeds the pixel grayscale difference threshold, it indicates that there is a large brightness difference at that pixel location, which may be caused by surface defects such as scratches or cracks.

[0075] The local texture difference threshold refers to the difference in surface structure features between two images within a certain local area. Texture features are usually related to patterns, shapes, edges, etc. in an image. When the local texture difference in an image exceeds the local texture difference threshold, it usually indicates that there are surface changes or defects in that area. For example, surface irregularities, cracks, or wear of a material can change its texture pattern.

[0076] The difference pixel ratio threshold refers to the proportion of pixels in a certain area of ​​an image that meet a certain threshold in terms of grayscale difference or texture difference. By setting the difference pixel ratio threshold, it is possible to avoid judging the entire area as a defect based on only a few abnormal pixels, ensuring that the detection results are more stable and accurate. This threshold helps to filter out noise or small errors and focus on a larger range of defect areas.

[0077] By combining these three methods, the differences between trial-processed alloy parts and standard alloy parts can be identified more accurately, and possible defect areas can be automatically located. This multi-level and multi-dimensional analysis method can better adapt to various defect types in actual production and improve the accuracy and reliability of detection.

[0078] Example 2, based on the same inventive concept as the alloy parts processing monitoring method for aviation described in the foregoing examples, such as... Figure 2 As shown in the embodiment of this application, a monitoring system for the processing of alloy parts for aviation is provided, the system comprising:

[0079] The design information acquisition module 10 is used to determine quality requirements and design information for target alloy parts of aerospace structural components; the control parameter determination module 20 is used to determine the target processing flow and processing control parameter sequence based on the design information; the monitoring data acquisition module 30 is used to collect multiple processing monitoring data sets of multiple processing nodes through pre-deployed processing monitoring units during the trial processing of the target alloy parts based on the processing control parameter sequence; the processing error analysis module 40 is used to perform processing error analysis based on the multiple processing monitoring data sets and obtain trial processing error monitoring results; the quality assessment module 50 is used to perform quality assessment on the acquired trial-processed alloy parts after the trial processing process is completed and obtain trial processing quality monitoring results; the quality defect analysis module 60 is used to perform quality defect analysis on the trial processing quality monitoring results based on the quality requirements information and obtain trial processing quality defect monitoring results; the processing monitoring result acquisition module 70 is used to integrate the trial processing error monitoring results and the trial processing quality defect monitoring results as the processing monitoring results of the target alloy parts.

[0080] Furthermore, the machining error analysis module 40 is used to perform the following operation steps:

[0081] Extract the first processing monitoring data set of the first processing node, wherein the first processing monitoring data set includes workpiece processing monitoring data and processing equipment status monitoring data; perform workpiece processing deviation analysis based on the workpiece processing monitoring data to generate workpiece processing deviation monitoring results; perform processing equipment wear analysis based on the processing equipment status monitoring data to generate processing equipment fatigue monitoring results; add the workpiece processing deviation monitoring results and the processing equipment fatigue monitoring results to the trial processing error monitoring results.

[0082] Furthermore, the machining error analysis module 40 is used to perform the following operation steps:

[0083] Extract the first processing monitoring data set of the first processing node, wherein the first processing monitoring data set includes workpiece processing monitoring data and processing equipment status monitoring data; perform workpiece processing deviation analysis based on the workpiece processing monitoring data to generate workpiece processing deviation monitoring results; perform processing equipment wear analysis based on the processing equipment status monitoring data to generate processing equipment fatigue monitoring results; add the workpiece processing deviation monitoring results and the processing equipment fatigue monitoring results to the trial processing error monitoring results.

[0084] Furthermore, the machining error analysis module 40 is used to perform the following operation steps:

[0085] Multidimensional covariance modeling is performed based on the processing deviation vectors of the multiple sets of sample workpieces to obtain the processing deviation correlation coefficient matrix between each deviation dimension; joint probability inference is performed on the probability distribution of the workpiece processing deviation based on the processing deviation correlation coefficient matrix and the single-dimensional probability value is corrected to obtain a weighted corrected multidimensional joint deviation probability distribution; mutual information difference test and conditional independence test are performed on the multidimensional joint deviation probability distribution, and the deviation classification confidence level is generated based on the test results.

[0086] Furthermore, the machining error analysis module 40 is used to perform the following operation steps:

[0087] Based on the workpiece machining deviation monitoring results, machining error matching is performed to generate a subset of workpiece deviation compensation parameters; based on the machining equipment fatigue monitoring results, machining error matching is performed to generate a subset of equipment fatigue compensation parameters; modeling, calculation, and parameter fusion are performed on the subset of workpiece deviation compensation parameters and the subset of equipment fatigue compensation parameters respectively to generate a dynamic compensation parameter group; based on the dynamic compensation parameter group, the machining control parameter sequence is periodically iteratively optimized.

[0088] Furthermore, the quality assessment module 50 is used to perform the following operational steps:

[0089] The trial-processed alloy parts are transferred to the appearance quality assessment zone, where appearance defects are detected by an appearance defect detection module to obtain appearance quality assessment results. This appearance quality assessment zone is equipped with an all-around image acquisition device. The trial-processed alloy parts are then transferred to the structural quality assessment zone, where internal defects and geometric accuracy are detected by a non-destructive testing module to obtain structural quality assessment results. This non-destructive testing module includes an ultrasonic flaw detector, an X-ray imaging device, and a three-dimensional geometric measurement device. The trial-processed alloy parts are then transferred to the mechanical performance assessment zone, where mechanical performance is assessed by a mechanical performance testing module to obtain mechanical performance assessment results. This mechanical performance testing module includes a hardness testing device, a tensile testing device, and a fatigue testing machine. The appearance quality assessment results, structural quality assessment results, and mechanical performance assessment results are integrated to obtain the trial processing quality monitoring results.

[0090] Furthermore, the quality assessment module 50 is used to perform the following operational steps:

[0091] When the trial-processed alloy parts pass through the omnidirectional image acquisition device, omnidirectional image acquisition results are acquired; standard image acquisition results of standard alloy parts are locally retrieved; semantic segmentation of the standard image acquisition results and the omnidirectional image acquisition results is performed by the appearance defect detection module to obtain appearance defect detection results, wherein the appearance defect detection results have defect area identifiers.

[0092] Furthermore, the quality assessment module 50 is used to perform the following operational steps:

[0093] The appearance defect detection module includes a first convolutional neural network and a second convolutional neural network. The first and second convolutional neural networks have symmetrical network structures, share parameters, and have consistent feature spaces. The first convolutional neural network extracts features from the standard image acquisition results to obtain first image pixel features, and the second convolutional neural network extracts features from the omnidirectional image acquisition results to obtain second image pixel features. The first and second image pixel features are used to calculate the difference based on pixel coordinates, and the defect area is identified and marked based on the difference threshold to generate the appearance defect detection result.

[0094] Furthermore, the difference threshold includes a pixel grayscale difference threshold, a local texture difference threshold, and a difference pixel ratio threshold.

[0095] Through the foregoing detailed description of the monitoring method for the processing of alloy parts for aviation, those skilled in the art can clearly understand the monitoring system for the processing of alloy parts for aviation in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0096] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for monitoring the processing of alloy parts for aviation, characterized in that, The method includes: For target alloy components of aerospace structural parts, determine quality requirements and design information; Based on the design information, the target processing flow and the sequence of processing control parameters are determined; During the trial processing of the target alloy parts based on the processing control parameter sequence, multiple processing monitoring data sets of multiple processing nodes are collected by pre-deployed processing monitoring units. Based on the multiple processing monitoring data sets, processing error analysis is performed to obtain the trial processing error monitoring results; After the trial processing is completed, the quality of the obtained trial-processed alloy parts is evaluated, and the trial processing quality monitoring results are obtained. Based on the quality requirement information, a quality defect analysis is performed on the trial processing quality monitoring results to obtain the trial processing quality defect monitoring results. The results of the trial processing error monitoring and the results of the trial processing quality defect monitoring are integrated as the processing monitoring results of the target alloy parts; The step of performing processing error analysis based on the multiple processing monitoring data sets to obtain trial processing error monitoring results includes: Extract the first processing monitoring data set of the first processing node, wherein the first processing monitoring data set includes workpiece processing monitoring data and processing equipment status monitoring data; Based on the workpiece processing monitoring data, workpiece processing deviation analysis is performed to generate workpiece processing deviation monitoring results; Based on the condition monitoring data of the processing equipment, a loss analysis of the processing equipment is performed to generate fatigue monitoring results of the processing equipment. Add the workpiece machining deviation monitoring results and the machining equipment fatigue monitoring results to the trial machining error monitoring results; The step of performing workpiece machining deviation analysis based on the workpiece machining monitoring data and generating workpiece machining deviation monitoring results includes: Multi-source feature extraction is performed on the workpiece processing monitoring data to obtain multi-source processing characterization parameters, and the multi-source processing characterization parameters are combined to obtain the target workpiece processing feature vector; The standard alloy parts are locally called to retrieve the standard workpiece machining feature vector. The workpiece machining feature vector of the target workpiece is analyzed to obtain the workpiece machining deviation vector. The standard alloy parts and the target alloy parts are from the same production batch and are prepared according to the standard process flow. The system locally retrieves multiple sets of sample workpiece processing deviation records, along with their processing deviation vectors, and performs cluster analysis to obtain multiple sample workpiece processing deviation vector domains. Trigger credibility evaluation is performed on the multiple sample workpiece processing deviation vector domains to obtain multiple sample trigger credibility evaluation coefficients; The processing deviation vector domains of the multiple sample workpieces and the trigger confidence evaluation coefficients of the multiple samples are associated and stored to construct a deviation distribution embedding space; The workpiece processing deviation vector is projected onto the deviation distribution embedding space, and a workpiece processing deviation probability distribution is generated based on vector similarity measurement and probability mapping inference. The confidence level is calculated based on the probability distribution of the workpiece processing deviation to obtain the deviation level confidence level, and the deviation level confidence level is added to the workpiece processing deviation monitoring result.

2. The method for monitoring the processing of alloy parts for aviation as described in claim 1, characterized in that, The step of calculating the confidence level based on the probability distribution of the workpiece processing deviation to obtain the deviation grading confidence level includes: Based on the processing deviation vectors of the multiple sets of sample workpieces, multidimensional covariance modeling is performed to obtain the processing deviation correlation coefficient matrix between each deviation dimension. Based on the machining deviation correlation coefficient matrix, joint probability inference is performed on the workpiece machining deviation probability distribution, and the single-dimensional probability value is corrected to obtain a weighted and corrected multi-dimensional joint deviation probability distribution. The mutual information difference test and conditional independence test are performed on the multidimensional joint bias probability distribution, and the bias grading confidence level is generated based on the test results.

3. The method for monitoring the processing of alloy parts for aviation as described in claim 1, characterized in that, include: Based on the workpiece machining deviation monitoring results, machining error matching is performed to generate a subset of workpiece deviation compensation parameters; Based on the fatigue monitoring results of the processing equipment, processing error matching is performed to generate a subset of equipment fatigue compensation parameters. Modeling, calculation and parameter fusion are performed on the workpiece deviation compensation parameter subset and the equipment fatigue compensation parameter subset respectively to generate a dynamic compensation parameter set; The processing control parameter sequence is periodically iteratively optimized based on the dynamic compensation parameter set.

4. The method for monitoring the processing of alloy parts for aviation as described in claim 1, characterized in that, The process of evaluating the quality of the obtained trial-processed alloy parts and obtaining the trial-processing quality monitoring results includes: The trial-processed alloy parts are transferred to the appearance quality assessment zone, where appearance defects are detected by the appearance defect detection module to obtain the appearance quality assessment results. The appearance quality assessment zone is equipped with an all-around image acquisition device. The trial-processed alloy parts are transferred to the structural quality assessment zone, where the internal defects and geometric accuracy of the trial-processed alloy parts are detected by a non-destructive testing module to obtain the structural quality assessment results. The non-destructive testing module includes an ultrasonic flaw detector, an X-ray imaging device, and a three-dimensional geometric measurement device. The trial-processed alloy parts are transferred to the mechanical performance evaluation zone, where mechanical performance is evaluated by the mechanical performance testing module to obtain the mechanical performance evaluation results. The mechanical performance testing module includes a hardness testing device, a tensile testing device, and a fatigue testing machine. By integrating the appearance quality assessment results, the structural quality assessment results, and the mechanical performance assessment results, the trial processing quality monitoring results are obtained.

5. The method for monitoring the processing of alloy parts for aviation as described in claim 4, characterized in that, The obtained appearance quality assessment results include: When the trial-processed alloy parts pass through the omnidirectional image acquisition device, omnidirectional image acquisition results are obtained; Locally retrieve standard image acquisition results of standard alloy parts; The appearance defect detection module performs semantic segmentation on the standard image acquisition results and the omnidirectional image acquisition results to obtain appearance defect detection results, wherein the appearance defect detection results have defect area identifiers.

6. The method for monitoring the processing of alloy parts for aviation as described in claim 5, characterized in that, The semantic segmentation of the standard image acquisition results and the omnidirectional image acquisition results by the appearance defect detection module includes: The appearance defect detection module includes a first convolutional neural network and a second convolutional neural network, wherein the first convolutional neural network and the second convolutional neural network have symmetrical network structures, share parameters, and have consistent feature spaces; The first image pixel features are obtained by extracting features from the standard image acquisition results using the first convolutional neural network, and the second image pixel features are obtained by extracting features from the omnidirectional image acquisition results using the second convolutional neural network. The pixel features of the first image and the pixel features of the second image are calculated based on pixel coordinates. Based on the difference threshold, the defect area is identified and marked, and the appearance defect detection result is generated.

7. The method for monitoring the processing of alloy parts for aviation as described in claim 6, characterized in that, The difference thresholds include pixel grayscale difference threshold, local texture difference threshold, and difference pixel percentage threshold.

8. A monitoring system for the machining of alloy parts for aviation, characterized in that, For implementing the method for monitoring the machining of alloy parts for aviation as described in any one of claims 1-7, the system comprises: The design information acquisition module is used to determine the quality requirements and design information for target alloy parts of aerospace structural components. The control parameter determination module is used to determine the target processing flow and the sequence of processing control parameters based on the design information. The monitoring data acquisition module is used to collect and acquire multiple processing monitoring data sets of multiple processing nodes through a pre-deployed processing monitoring unit during the trial processing of the target alloy parts based on the processing control parameter sequence. The machining error analysis module is used to perform machining error analysis based on the multiple machining monitoring data sets and obtain the trial machining error monitoring results. The quality assessment module is used to assess the quality of the obtained trial-processed alloy parts after the trial processing is completed, and to obtain the trial processing quality monitoring results. The quality defect analysis module is used to perform quality defect analysis on the trial processing quality monitoring results based on the quality requirement information, and to obtain the trial processing quality defect monitoring results. The processing monitoring result acquisition module is used to integrate the trial processing error monitoring results and the trial processing quality defect monitoring results as the processing monitoring results of the target alloy parts.

Images