Intelligent precision parts full-process processing control system

Abstract

The present invention discloses an intelligent full-process processing control system for precision parts, which relates to the field of processing control technology. Based on the priority diagram method, the system analyzes the various processing-influencing parameters of precision parts and determines the weight values ​​of each processing-influencing parameter for processing monitoring and control. The system monitors the various processing-influencing parameters of precision parts in real time and, based on the weight values ​​of each processing-influencing parameter, provides real-time early warning for precision part processing. A precision part processing qualification detection model is constructed based on historical processing samples of precision parts, and a qualified detection threshold for the processing qualification detection model is identified. The processing qualification detection model and the qualified detection threshold are used to intelligently detect the processing results of each batch of precision parts, and then output a precision part processing detection report. Through intelligent full-process control, the processing process is optimized, processing accuracy is improved, processing quality is ensured, and potential problems are promptly warned and corrected.

Description

Technical Field

[0001] The present invention relates to the field of processing control technology, and in particular to a full-process processing control system for intelligent precision parts. Background Art

[0002] With the rapid development of high-tech industries such as aerospace, medical equipment, and precision instruments, the requirements for the processing accuracy of parts are becoming increasingly higher. In modern manufacturing, production efficiency is one of the important indicators to measure the competitiveness of an enterprise. Improving processing efficiency and processing quality can not only shorten the production cycle, but also reduce production costs, thereby improving the profitability of the enterprise.

[0003] However, traditional precision machining processes mostly rely on manual monitoring, or use relatively simple sensors for data collection, which lacks real-time performance and depth of data analysis, making it difficult to detect potential problems in a timely manner. Quality inspection often relies on manual inspection, or functional testing based on simple equipment, which is subject to problems such as human error, incomplete detection range, and low machining efficiency. At the same time, existing machining process control often lacks accurate analysis and weight allocation of various machining parameters. The adjustment process is more dependent on experience, and ignores the role of historical data in the machining process in improving quality, lacking data-driven intelligent optimization methods.

[0004] Therefore, in order to solve the above problems, there is an urgent need for a full-process processing control system for intelligent precision parts. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides an intelligent precision parts full-process processing control system, which solves the problem of lack of comprehensive analysis and real-time monitoring of multiple influencing factors in traditional processing control.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent precision parts full-process processing control system, including: an influencing parameter analysis module, which is used to analyze the various processing influencing parameters of precision parts based on the priority diagram method, and determine the weight value of each processing influencing parameter of precision parts for processing monitoring and control; a real-time monitoring module, which is used to monitor the various processing influencing parameters of precision parts in real time, and provide real-time early warning for precision parts processing in combination with the weight value of each processing influencing parameter of precision parts; a detection model construction module, which is used to construct a processing qualification detection model for precision parts based on historical processing samples of precision parts, and identify the qualified detection threshold of the processing qualification detection model; a processing detection module, which is used to use the processing qualification detection model and the qualified detection threshold to perform intelligent detection on the processing results of each batch of precision parts, and then output a precision parts processing detection report.

[0007] Furthermore, based on the priority diagram method, each processing influencing parameter of the precision parts is analyzed, and the specific analysis of the weight value of each processing influencing parameter of the precision parts for processing monitoring and control is determined as follows: based on the historical processing fault records, the historical record ratio of each processing influencing parameter of the precision parts is retrieved, and the specific historical record ratio is the ratio of the historical record number of processing faults caused by each processing influencing parameter to the total historical record number of processing faults; each processing influencing parameter of the precision parts is monitored, and then the rate of change of each processing influencing parameter per unit time is calculated; the historical record ratio and the rate of change per unit time of each processing influencing parameter of the precision parts are normalized, and then the normalized historical record ratio and the rate of change per unit time of each processing influencing parameter are summed up respectively to obtain the impact score of each processing influencing parameter; the impact score of each processing influencing parameter is weighted according to the priority diagram method to determine the weight value of each processing influencing parameter of the precision parts for processing monitoring and control.

[0008] Furthermore, the various processing influencing parameters of the precision parts specifically include cutting parameters, machine tool parameters, vibration parameters and tool parameters. The specific cutting parameter is cutting speed, the specific machine tool parameter is machine tool temperature, the specific vibration parameter is vibration amplitude, and the specific tool parameter is tool wear.

[0009] Furthermore, each processing influencing parameter of precision parts is monitored in real time, and the specific analysis of real-time early warning of precision parts processing is carried out in combination with the weight value of each processing influencing parameter of precision parts is as follows: each processing influencing parameter of precision parts is obtained in real time, each processing influencing parameter is combined with the corresponding weight value, and then the sum is obtained to obtain the processing monitoring index of precision parts; the processing monitoring threshold is obtained based on the historical processing fault record, and the processing monitoring index of precision parts is compared with the processing monitoring threshold in real time. When the processing monitoring index exceeds the processing monitoring threshold, the alarm mechanism is triggered.

[0010] Furthermore, a processing qualification detection model for precision parts is constructed based on historical processing samples of precision parts, and the specific analysis of identifying the qualified detection threshold of the processing qualification detection model is as follows: obtaining historical processing sample parameters of precision parts, the historical processing sample parameters include the dimensional accuracy of each precision part, the surface roughness of each precision part, the residual stress of each precision part and the hardness of each precision part; obtaining the unique characteristic index of each precision part based on dimensional accuracy, surface roughness, residual stress and hardness; using Logistic regression, taking the unique characteristic index of each precision part as the model input, to obtain the processing qualification detection model for precision parts, and the objective function output by the processing qualification detection model is specifically the probability that the precision parts are detected as qualified products; classifying the historical processing samples of precision parts, and then outputting the qualified detection threshold of the processing qualification detection model.

[0011] Furthermore, the historical processing samples of precision parts are classified, and the specific analysis of the qualified detection threshold of the processing qualified detection model is output as follows: the historical processing samples of precision parts are classified according to the detection of qualified precision parts and the detection of unqualified precision parts, and qualified precision parts historical samples and unqualified precision parts historical samples are obtained; the qualified precision parts historical sample results and unqualified precision parts historical sample results are output according to the precision parts processing qualified detection model, and the qualified precision parts historical sample results and unqualified precision parts historical sample results are combined into probability samples; based on the random generation of multiple candidate thresholds between 0 and 1, the probability samples are compared according to different candidate thresholds, and then the comparison results are output, and the comparison results are combined according to the comparison results. Confusion matrix; the confusion matrix includes the number of historical samples of qualified precision parts that are determined to be qualified based on the candidate threshold comparison, the number of historical samples of qualified precision parts that are determined to be unqualified based on the candidate threshold comparison, the number of historical samples of unqualified precision parts that are determined to be qualified based on the candidate threshold comparison, and the number of historical samples of unqualified precision parts that are determined to be unqualified based on the candidate threshold comparison; different confusion matrices obtained based on different candidate threshold comparison result combinations, the harmonic means of the precision and recall rate corresponding to different candidate thresholds are obtained respectively; the harmonic means of the precision and recall rate corresponding to different candidate thresholds are compared, and the candidate threshold corresponding to the maximum harmonic mean of the precision and recall rate is marked as the qualified detection threshold of the processing qualified detection model.

[0012] Furthermore, the qualified processing detection model and the qualified detection threshold are used to perform intelligent detection on the processing results of each batch of precision parts, and then the specific analysis of the precision parts processing detection report is output as follows: randomly extract processing samples of each batch of precision parts processing, and then obtain the detection parameters of each batch of processing samples, the detection parameters include the dimensional accuracy of each batch of processing samples, the surface roughness of each batch of processing samples, the residual stress of each batch of processing samples and the hardness of each batch of processing samples; input the detection parameters of each batch of processing samples into the qualified processing detection model respectively, and output the objective function of each batch of processing samples; compare the objective function of each batch of processing samples with the qualified detection threshold respectively, when the objective function is lower than the qualified detection threshold, mark the batch of processing samples as unqualified in the intelligent detection, when the objective function is greater than or equal to the qualified detection threshold, mark the batch of processing samples as qualified in the intelligent detection; count the intelligent detection results of each batch of processing samples, and then output the precision parts processing detection report based on the intelligent detection results of each batch of processing samples, the precision parts processing detection report provides the batch information of the processing samples that fail the intelligent detection.

[0013] The present invention has the following beneficial effects:

[0014] By analyzing the various processing influencing parameters through the priority diagram method, the degree of influence and weight of each parameter on the component processing process can be systematically determined, which can help accurately evaluate the key influencing factors in the processing process and improve the predictability of processing quality; real-time monitoring of various parameters in the processing process and dynamic analysis combined with their weight values ​​can be used to warn of potential problems in real time, making the manufacturing process more intelligent and able to issue warnings before abnormalities occur, avoiding processing defects and waste of resources; using historical processing sample data, targeted processing qualification detection models can be constructed according to the processing characteristics of specific components. Personalized modeling based on data can more accurately identify qualified and unqualified components and improve the accuracy of detection; intelligent detection of processing results through processing qualification detection models and qualified detection thresholds can automatically evaluate the processing quality of each batch of components, and ultimately generate detailed processing inspection reports, which provide strong support for subsequent quality management and feedback adjustments. By continuously monitoring and analyzing the processing process, various parameters can be adjusted in a timely manner, and the processing process can be optimized, thereby continuously improving the processing accuracy and quality of precision components.

[0015] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a structural diagram of the full-process processing control system of the intelligent precision parts of the present invention.

[0017] Figure 2This is a method flow chart of the full-process processing control system of intelligent precision parts of the present invention. DETAILED DESCRIPTION

[0018] The embodiment of the present application uses an intelligent full-process processing control system for precision parts to optimize the processing process, improve processing accuracy, ensure processing quality, and promptly warn and correct potential problems through intelligent full-process control.

[0019] The overall idea of ​​the embodiment of this application is: by integrating modules such as influencing parameter analysis, real-time monitoring, detection model construction and processing detection, an intelligent, closed-loop full-process processing control system for precision parts is formed, aiming to improve processing accuracy, optimize the processing process and realize real-time monitoring and intelligent detection.

[0020] See also Figure 1 、 Figure 2 The embodiment of the present invention provides a technical solution: an intelligent precision component full-process processing control system, including: an influencing parameter analysis module, which is used to analyze various processing influencing parameters of precision components based on the priority diagram method, and determine the weight value of each processing influencing parameter of precision components for processing monitoring and control; a real-time monitoring module, which is used to monitor various processing influencing parameters of precision components in real time, and provide real-time early warning for precision component processing in combination with the weight value of each processing influencing parameter of precision components; a detection model construction module, which is used to construct a processing qualification detection model of precision components based on historical processing samples of precision components, and identify the qualified detection threshold of the processing qualification detection model; a processing detection module, which is used to use the processing qualification detection model and the qualified detection threshold to perform intelligent detection on the processing results of each batch of precision components, and then output a precision component processing detection report.

[0021] Specifically, based on the priority diagram method, each processing influencing parameter of precision parts is analyzed to determine the specific analysis of the weight value of each processing influencing parameter of precision parts for processing monitoring and control: based on the historical processing fault records, the historical record ratio of each processing influencing parameter of precision parts is retrieved. The specific historical record ratio is the ratio of the number of historical records of processing faults caused by each processing influencing parameter to the total number of historical records of processing faults. The historical record ratio is obtained in the following way: all processing fault historical records within a period of time are collected, including detailed information of each fault, such as fault type, time, cause analysis, etc. From these historical records, fault information related to each processing influencing parameter (cutting parameters, machine tool parameters, vibration parameters and tool parameters) is extracted. For example, a certain fault is caused by too high a cutting parameter, and another fault may be caused by too large a tool parameter. For each processing influencing parameter, the historical record of the fault is counted. times. For example, if the machine tool parameters are too high and cause 20 faults, and the total number of fault history records is 100, then the historical record proportion of the machine tool parameters is 0.2; monitor the various processing influencing parameters of precision parts, and then calculate the rate of change of each processing influencing parameter per unit time; normalize the historical record proportion and the rate of change per unit time of each processing influencing parameter of precision parts to eliminate the differences in dimensions and magnitudes of different parameters, and then sum the normalized historical record proportion and the rate of change per unit time of each processing influencing parameter to obtain the impact score of each processing influencing parameter, that is, the impact score acquisition expression is: P=Z+B, P represents the impact score, Z represents the historical record proportion, and B represents the rate of change per unit time; assign weights to the impact scores of each processing influencing parameter according to the priority diagram method to determine the weight value of each processing influencing parameter of precision parts for processing monitoring and control.

[0022] In this implementation scheme, the various processing influencing parameters of precision parts specifically include cutting parameters, machine tool parameters, vibration parameters and tool parameters. The specific cutting parameter is the cutting speed, the specific machine tool parameter is the machine tool temperature, the specific vibration parameter is the vibration amplitude, and the specific tool parameter is the tool wear.

[0023] Cutting parameters (cutting speed) refer to the relative speed between the tool and the contact surface of the workpiece, usually expressed in cutting length per minute (m / min). Cutting speed is a key parameter affecting processing efficiency and surface quality. Real-time cutting speed data is obtained through the machine tool control system or online measuring devices (such as sensors) and monitored in combination with processing technology settings; machine tool parameters (machine tool temperature) refer to the temperature of key machine tool components (such as spindles, guide rails, etc.). Excessive temperature may cause reduced machine tool accuracy, increased component wear or increased processing errors. By installing temperature sensors (such as thermocouples or infrared temperature sensors) at key positions on the machine tool, the temperature of each machine tool component can be monitored in real time. ; Vibration parameter (vibration amplitude) refers to the vibration intensity generated by the machine tool during the processing. Vibration may cause processing instability, produce unqualified parts or shorten the tool life. Vibration sensors (such as accelerometers) are installed on different parts of the machine tool (such as the spindle, bed, etc.) to monitor the vibration amplitude in real time. Tool parameters (tool wear) refer to the degree of reduction in the cutting performance of the tool during use. Excessive tool wear may lead to reduced processing quality and production efficiency. Tool wear can be indirectly judged through tool wear sensors or with the help of cutting force, vibration, temperature and other data, or the degree of wear can be estimated through visual inspection of the tool using a visual inspection system.

[0024] The cutting parameters, machine parameters, vibration parameters and tool parameters are weighted according to the priority diagram method as shown in Table 1 below:

[0025]

[0026]

[0027] Table 1

[0028] It should be noted that in Table 1, the impact scores of H, I, J, and K correspond one-to-one to the impact scores of cutting parameters, machine tool parameters, vibration parameters, and tool parameters in this embodiment. Specifically, these impact scores are sorted by value and then corresponded to the impact scores of H, I, J, and K, respectively, in descending order. By considering the impact scores of cutting parameters, machine tool parameters, vibration parameters, and tool parameters, the machining status of precision parts can be comprehensively monitored, avoiding the one-sidedness of single-factor evaluations.

[0029] In Table 1, the number 0 indicates relatively unimportant, the number 1 indicates relatively more important, and the number 0.5 indicates equally important. For example: the impact score of H is 70, the impact score of I is 68, the impact score of J is 65, and the impact score of K is 60. Then, the importance of the impact score of H relative to the impact score of I, the impact score of J, and the impact score of K is higher; the impact score of I is less important than the impact score of H, but is important relative to the impact score of J and the impact score of K; the impact score of J is less important than the impact score of H and the impact score of I, but is important relative to the impact score of K; the impact score of K is less important than the impact score of H, the impact score of I, and the impact score of J. For example, since the impact score of H is greater than the impact score of I, the importance of the impact score of H relative to the impact score of I is recorded as 1.

[0030] The TTL indicator specifically represents the score of the relative importance of each impact score. For example, the TTL indicator of the relative importance of H's impact score is specifically obtained as: TTL = 0.5 + 1 + 1 + 1 = 3.5. The weight corresponding to the relative importance of H's impact score is specifically obtained as the ratio of the TTL indicator of H's relative importance of impact score to the sum of the TTL indicators of H's relative importance of impact score, I's relative importance of impact score, J's relative importance of impact score, and K's relative importance of impact score.

[0031] By combining historical data analysis with real-time monitoring, it is possible to dynamically adjust the focus on each influencing parameter, improve the monitoring accuracy during the processing, and reduce potential risks in production; evaluating the importance and real-time changes of each processing influencing parameter can effectively warn of factors that may cause processing quality problems, ensuring the stable processing quality of precision parts; using the priority diagram method, historical records and real-time monitoring to calculate the weight value, automatically adjust the monitoring strategy, reduce manual intervention and human errors, and improve production efficiency; with the continuous accumulation of historical data, the weight value of each processing influencing parameter can be continuously optimized, thereby continuously improving the reliability and accuracy of the processing process.

[0032] Specifically, each processing influencing parameter of precision parts is monitored in real time, and the specific analysis of real-time early warning of precision parts processing is carried out in combination with the weight value of each processing influencing parameter of precision parts: each processing influencing parameter of precision parts is obtained in real time, each processing influencing parameter is combined with the corresponding weight value, and then the processing monitoring index of precision parts is obtained by summing up. An example of the specific processing monitoring index acquisition expression is: PMI=CP*γ1+MTP*γ2+VP*γ3+TWL*γ4, where PMI represents processing monitoring index, CP represents cutting parameter, MTP represents machine tool parameter, VP represents vibration parameter, TWL represents tool parameter, γ1, γ2, γ3, and γ4 represent the weight values ​​of cutting parameter, machine tool parameter, vibration parameter and tool parameter respectively; the processing monitoring threshold is obtained based on historical processing fault records, and the processing monitoring index of precision parts is compared with the processing monitoring threshold in real time. When the processing monitoring index exceeds the processing monitoring threshold, the alarm mechanism is triggered to remind the operator to make corrections or enable protection measures to prevent further fault expansion.

[0033] In this implementation plan, the specific steps for obtaining the processing monitoring threshold are as follows: collect relevant data generated during the historical processing process, including various processing parameters (cutting parameters, machine tool parameters, vibration parameters and tool parameters) and quality feedback of the final product, and continuously record the data of each processing cycle through the data acquisition system; based on historical data, analyze the normal range and tolerance limit of each parameter in the processing process. These ranges can be statistically analyzed by comparing the processing data of qualified parts with the data of defective parts to find the normal fluctuation range of each parameter; use statistical tools (such as mean, standard deviation, control chart, etc.) to analyze the fluctuation range of each influencing parameter. Specifically, the processing monitoring threshold can be set as the upper and lower limits of the normal fluctuation range of each parameter, such as exceeding ±3 The part that is more than times the standard deviation may be regarded as abnormal. A preliminary threshold range can also be set based on experience and expert knowledge, and continuously optimized and adjusted in actual production; then, each parameter is weighted to obtain a processing monitoring threshold. The processing monitoring threshold is not static and can be dynamically adjusted based on real-time processing data and fault history records. For example, when the system finds that the fluctuation pattern of certain parameters has changed, the threshold can be re-evaluated and adjusted to maintain the flexibility and adaptability of the system. In actual applications, since there are inevitably some uncertainties in the processing process, a reasonable fault tolerance interval can be set near the threshold. When the processing monitoring indicator approaches the threshold, you can consider taking early warning instead of immediately triggering an alarm to avoid unnecessary downtime caused by overly sensitive alarms.

[0034] Real-time monitoring of various influencing parameters during the machining of precision parts enables timely identification of potential problems, significantly reducing the occurrence of faults and avoiding quality issues caused by machining errors or environmental changes. By combining the weights of various machining-influencing parameters, the system accurately reflects the degree of impact of each parameter on machining quality, making the monitoring system more refined. The contribution of each influencing factor to the final machining quality can be adjusted according to actual conditions, improving overall monitoring and control accuracy. When machining monitoring indicators exceed preset thresholds, an alarm mechanism is triggered to promptly alert the operator or automatically intervene to prevent the problem from escalating. This is crucial for avoiding major machining errors in precision parts and improving production efficiency and quality reliability. Automated monitoring and early warning systems reduce manual errors and omissions, ensuring a more stable and controllable machining process. Real-time feedback on machining status allows for the timely correction of potential problems, reducing equipment downtime and repair time, and improving production efficiency and production line utilization. Over long-term monitoring, the system accumulates a large amount of machining data. Analysis of this data can identify potential optimization areas, such as machining parameters and equipment maintenance cycles, to further improve machining quality and efficiency.

[0035] Specifically, a processing qualification detection model for precision parts is constructed based on historical processing samples of precision parts, and the specific analysis of identifying the qualified detection threshold of the processing qualification detection model is as follows: obtaining the historical processing sample parameters of precision parts, which include the dimensional accuracy of each precision part, the surface roughness of each precision part, the residual stress of each precision part, and the hardness of each precision part; obtaining the unique characteristic index of each precision part based on dimensional accuracy, surface roughness, residual stress, and hardness; using Logistic regression, taking the unique characteristic index of each precision part as the model input, and obtaining the processing qualification detection model of precision parts. The specific processing qualification detection model expression is: Where P(y=qualified) is the probability of qualified precision parts, and UFI represents the unique characteristic index. The historical processing samples of precision parts are classified, and the qualified detection threshold of the processing qualified detection model is output.

[0036] Dimensional accuracy is obtained through precision measuring tools (such as three-dimensional measuring machines, laser rangefinders, etc.). By measuring the various important dimensions of the parts, it is determined whether they meet the design specifications. Surface roughness is measured using a surface roughness meter to obtain the roughness parameters of the part surface, such as Ra value and Rz value. Residual stress is obtained by measuring the residual stress of the parts through methods such as X-ray diffraction, stress sensors, or forging tests. By measuring the micro-stress state inside the part, the distribution and magnitude of the residual stress are obtained. Hardness is measured using a hardness tester. Common methods include Rockwell hardness, Vickers hardness, and Brinell hardness. The hardness value is closely related to the mechanical properties of the part material and is an important indicator for judging part quality. The unique characteristic index is based on the various parameters provided in the historical processing samples (dimensional accuracy, surface roughness, residual stress, hardness). Through feature engineering (such as normalization and standardization), a "unique characteristic index" is constructed for each component. Multiple input features are combined into a comprehensive value as the input of the model. The specific unique characteristic index expression example is as follows: Wherein, UFI represents the unique characteristic index, DA represents dimensional accuracy, SR represents surface roughness, RS represents residual stress, HA represents hardness, ω1, ω2, ω3, and ω4 represent the weight values ​​of dimensional accuracy, surface roughness, residual stress, and hardness, respectively. The weight values ​​of dimensional accuracy, surface roughness, residual stress, and hardness are obtained as follows: the correlation between different features (dimensional accuracy, surface roughness, residual stress, and hardness) and the target variable is evaluated using statistical methods such as the Pearson correlation coefficient or mutual information. The correlation coefficient between each feature and the target variable is calculated, and the feature is assigned a weight according to the size of the correlation. The stronger the correlation, the greater the weight. The principal component analysis method can also be used to project multiple features onto a few principal components through linear transformation. The variance contribution rate of the principal components is calculated to evaluate the contribution of each original feature to the total variance. Features with high variance contribution rates can be considered more important to the model and thus assigned larger weights. Alternatively, the hierarchical analysis method can be used to compare the importance of each feature pair by pair based on expert judgment, and finally a weight matrix is ​​obtained. The standard weight is obtained using matrix operations.

[0037] In this implementation scheme, the historical processing samples of precision parts are classified, and the specific analysis of the qualified detection threshold of the qualified processing detection model is output as follows: the historical processing samples of precision parts are classified according to the qualified precision parts and the unqualified precision parts, and the qualified precision parts historical samples and the unqualified precision parts historical samples are obtained; the qualified precision parts historical sample results and the unqualified precision parts historical sample results are output according to the qualified processing detection model of precision parts, and the qualified precision parts historical sample results and the unqualified precision parts historical sample results are combined into probability samples; a plurality of candidate thresholds are randomly generated between 0 and 1, and the probability samples are compared according to different candidate thresholds, and then the comparison results are output, and the comparison results are combined according to the comparison results. The confusion matrix includes the number of qualified historical samples of precision parts that are determined to be qualified based on the candidate threshold comparison, the number of qualified historical samples of precision parts that are determined to be unqualified based on the candidate threshold comparison, the number of qualified historical samples of unqualified precision parts that are determined to be qualified based on the candidate threshold comparison, and the number of unqualified historical samples of unqualified precision parts that are determined to be unqualified based on the candidate threshold comparison; different confusion matrices are obtained based on the combination of different candidate threshold comparison results, and the harmonic means of the precision and recall rate corresponding to different candidate thresholds are obtained respectively; the harmonic means of the precision and recall rate corresponding to different candidate thresholds are compared, and the candidate threshold corresponding to the maximum harmonic mean of the precision and recall rate is marked as the qualified detection threshold of the qualified processing detection model.

[0038] The candidate threshold represents multiple candidate thresholds used to evaluate the model classification results, ranging from 0 to 1. Each candidate threshold indicates that samples with a probability value greater than this value are judged as qualified precision parts, and samples with a probability value less than this value are judged as unqualified precision parts. Multiple thresholds are randomly generated between 0 and 1 by setting a step size, or these thresholds are defined according to the needs of actual applications.

[0039] The confusion matrix is ​​a matrix used to evaluate model classification performance. It records the comparison between true and predicted labels. It includes: true positives (TP), which are the number of qualified precision parts from historical samples that were deemed qualified based on the candidate threshold; false negatives (FN), which are the number of qualified precision parts from historical samples that were deemed unqualified based on the candidate threshold; false positives (FP), which are the number of unqualified precision parts from historical samples that were deemed qualified based on the candidate threshold; and true negatives (TN), which are the number of unqualified precision parts from historical samples that were deemed unqualified based on the candidate threshold. The confusion matrix is ​​generated by comparing the prediction results of all samples using different thresholds.

[0040] Precision is specifically used to measure the accuracy of the model in determining qualified precision parts. It is calculated by the true positive and false positive examples in the confusion matrix. The calculation formula is: P recision Indicates accuracy.

[0041] The recall rate is used to measure the model's ability to identify qualified precision parts. It is calculated by the true positive and false negative examples in the confusion matrix. The calculation formula is: R ecall Represents the recall rate.

[0042] The harmonic mean (F1-Score) is used to comprehensively evaluate the classification effect of the model. It is obtained by calculating the harmonic mean of precision and recall. The calculation formula is: F1 represents the harmonic mean.

[0043] By constructing a machining qualification detection model based on historical machining samples, the machining qualification of precision parts can be determined more accurately. The model uses multiple machining features (dimensional accuracy, surface roughness, residual stress, hardness) for comprehensive analysis, which can improve the accuracy of qualified detection and reduce the errors and subjective judgments of manual detection. Using historical machining sample data, a machining qualification detection model is generated through machine learning methods (Logistic regression). This not only reduces manual intervention, but also continuously improves the model's predictive ability as sample data accumulates. Based on the model, through testing and comparison of different thresholds, the optimal qualified judgment threshold can be found to ensure that the model can make the best decision in various situations. The confusion matrix is ​​used to measure the harmonic mean of the model's precision and recall rate at different thresholds. It can scientifically select appropriate thresholds, improve the practical application effect of the model, and significantly improve production efficiency and detection speed. In addition, the automated system can process large amounts of data and provide more stable output results.

[0044] Specifically, the processing results of each batch of precision parts are intelligently detected using the processing qualification detection model and the qualified detection threshold, and the specific analysis of the precision parts processing detection report is output as follows: randomly extract processing samples of each batch of precision parts processing, and then obtain the detection parameters of each batch of processing samples, which include the dimensional accuracy of each batch of processing samples, the surface roughness of each batch of processing samples, the residual stress of each batch of processing samples and the hardness of each batch of processing samples; input the detection parameters of each batch of processing samples into the processing qualification detection model respectively, and output the objective function of each batch of processing samples; compare the objective function of each batch of processing samples with the qualified detection threshold respectively, when the objective function is lower than the qualified detection threshold, mark the batch of processing samples as unqualified in the intelligent detection, and when the objective function is greater than or equal to the qualified detection threshold, mark the batch of processing samples as qualified in the intelligent detection; count the intelligent detection results of each batch of processing samples, and then output the precision parts processing detection report based on the intelligent detection results of each batch of processing samples, and the precision parts processing detection report provides the batch information of the processing samples that fail the intelligent detection.

[0045] In this implementation, traditional component inspection often requires manual intervention, especially in mass production. Manual inspection is not only time-consuming but also susceptible to fatigue and subjective factors. However, an intelligent qualified inspection model can automatically process and analyze a large number of processing samples, reducing manual inspection steps and improving production efficiency. The intelligent inspection model analyzes multiple inspection parameters (dimensional accuracy, surface roughness, residual stress, and hardness) in real time, achieving a comprehensive assessment without the need for tedious manual inspections each time. Through the systematic analysis of the qualified inspection model, the bias and error associated with manual inspection can be avoided. The model compares the objective function value of each batch of processing samples with a preset qualified threshold, ensuring more objective and consistent inspection results for each sample. The qualified inspection threshold can be adjusted based on actual processing technology and empirical data, ensuring that the inspection standards are more consistent with actual production requirements. Unqualified processing samples are accurately marked and further localized to specific batches. Quality differences and potential problems between batches can be quickly identified, helping to promptly detect and resolve problems that arise during the processing process (such as equipment failure, operational errors, or material problems) and prevent them from spreading.

[0046] In summary, this application has at least the following effects:

[0047] Through the influencing parameter analysis module, the system can accurately analyze and determine the weight value of each processing influencing parameter, which helps to optimize the processing process and ensure that key parameters are more strictly controlled, thereby improving processing accuracy, timely discovering anomalies in the processing process, avoiding the production of defective products, and thus improving processing efficiency; using the processing qualification detection model and the qualification detection threshold for intelligent detection, it can quickly and accurately determine whether the processing results are qualified, reducing the error and time cost of manual detection; through full-process monitoring and control, the processing process is made more transparent and controllable, and the collaborative work between various modules ensures the stability and consistency of the processing process; through precise analysis and real-time monitoring of processing influencing parameters, it can ensure that the processing process meets the design requirements, thereby improving product quality and reliability, ensuring that each batch of products meets quality standards, and improving the overall quality level of the product.

[0048] Those skilled in the art will appreciate that embodiments of the present invention may be provided as systems. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0049] The present invention is described with reference to a block diagram of a system according to an embodiment of the present invention. It should be understood that the combination of each structure in the block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, such that the instructions, executed by the processor of the computer or other programmable data processing device, produce a device for implementing the functions specified in each structure in the block diagram.

[0050] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in each structure of the structural diagram.

[0051] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in each structure in the structural diagram.

[0052] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0053] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. Intelligent precision parts full-process processing control system, characterized by: include: The influencing parameter analysis module is used to analyze the various processing influencing parameters of precision parts based on the priority diagram method, and determine the weight value of each processing influencing parameter of precision parts for processing monitoring and control; The real-time monitoring module is used to monitor the various processing influencing parameters of precision parts in real time, and to provide real-time early warning for precision parts processing based on the weight values ​​of the various processing influencing parameters of precision parts; A detection model building module is used to build a qualified processing detection model for precision parts based on historical processing samples of precision parts, and to identify a qualified detection threshold of the qualified processing detection model; The processing inspection module is used to intelligently inspect the processing results of each batch of precision parts using the processing qualification inspection model and qualification inspection threshold, and then output the precision parts processing inspection report; Based on the priority diagram method, the various processing influencing parameters of precision parts are analyzed, and the specific analysis of the weight values ​​of various processing influencing parameters of precision parts for processing monitoring and control is determined as follows: Retrieve the historical record percentage of each processing influencing parameter of precision parts based on historical processing failure records. The specific historical record percentage is the ratio of the number of historical records of processing failures caused by each processing influencing parameter to the total number of historical records of processing failures; Monitor various processing influencing parameters of precision parts, and then calculate the rate of change of each processing influencing parameter per unit time; Normalize the historical record proportion and the rate of change per unit time of each processing influencing parameter of precision parts, and then sum the normalized historical record proportion and the rate of change per unit time of each processing influencing parameter to obtain the impact score of each processing influencing parameter; The influence scores of various processing influencing parameters are weighted according to the priority diagram method to determine the weight value of each processing influencing parameter of precision parts for processing monitoring and control.

2. The intelligent precision parts full-process processing control system according to claim 1 is characterized in that: The various processing influencing parameters of the precision parts specifically include cutting parameters, machine tool parameters, vibration parameters and tool parameters. The specific cutting parameter is cutting speed, the specific machine tool parameter is machine tool temperature, the specific vibration parameter is vibration amplitude, and the specific tool parameter is tool wear.

3. The intelligent precision parts full-process processing control system according to claim 1 is characterized in that: Real-time monitoring of various processing influencing parameters of precision parts is carried out, and the specific analysis of real-time early warning of precision parts processing is carried out in combination with the weight values ​​of various processing influencing parameters of precision parts as follows: Acquire various processing influencing parameters of precision parts in real time, combine each processing influencing parameter with the corresponding weight value, and then sum them up to obtain the processing monitoring index of precision parts; The processing monitoring threshold is obtained based on historical processing fault records, and the processing monitoring indicators of precision parts are compared with the processing monitoring threshold in real time. When the processing monitoring indicator exceeds the processing monitoring threshold, the alarm mechanism is triggered.

4. The intelligent precision parts full-process processing control system according to claim 1 is characterized in that: The specific analysis of building a qualified processing detection model for precision parts based on historical processing samples of precision parts and identifying the qualified detection threshold of the qualified processing detection model is as follows: Obtaining historical processing sample parameters of precision parts, wherein the historical processing sample parameters include dimensional accuracy of each precision part, surface roughness of each precision part, residual stress of each precision part, and hardness of each precision part; Obtain unique characteristic indicators for each precision component based on dimensional accuracy, surface roughness, residual stress and hardness; Using logistic regression, the unique characteristic index of each precision component is used as the model input to obtain a precision component processing qualification detection model. The objective function output by the processing qualification detection model is specifically the probability that the precision component is detected as qualified. The historical processing samples of precision parts are classified, and then the qualified detection threshold of the processing qualified detection model is output.

5. The intelligent precision parts full-process processing control system according to claim 4 is characterized in that: The specific analysis of the qualified detection threshold of the processing qualified detection model obtained by classifying the historical processing samples of precision parts is as follows: The historical processing samples of precision parts are classified according to the qualified precision parts and the unqualified precision parts, and the qualified precision parts historical samples and the unqualified precision parts historical samples are obtained; Outputting qualified precision parts historical sample results and unqualified precision parts historical sample results based on the precision parts processing qualified inspection model, and combining the qualified precision parts historical sample results and the unqualified precision parts historical sample results into a probability sample; Based on the random generation of multiple candidate thresholds between 0 and 1, the probability samples are compared according to different candidate thresholds, and then the comparison results are output, which are combined into a confusion matrix according to the comparison results; The confusion matrix includes the number of samples of qualified precision parts history samples determined to be qualified based on the candidate threshold comparison, the number of samples of qualified precision parts history samples determined to be unqualified based on the candidate threshold comparison, the number of samples of unqualified precision parts history samples determined to be qualified based on the candidate threshold comparison, and the number of samples of unqualified precision parts history samples determined to be unqualified based on the candidate threshold comparison; According to the different confusion matrices obtained by combining the results of different candidate thresholds, the harmonic means of precision and recall corresponding to different candidate thresholds are obtained respectively; The harmonic means of precision and recall corresponding to different candidate thresholds are compared, and the candidate threshold corresponding to the maximum harmonic mean of precision and recall is marked as the qualified detection threshold of the processing qualified detection model.

6. The intelligent precision parts full-process processing control system according to claim 1 is characterized in that: The specific analysis of the precision parts processing inspection report is output by intelligently inspecting the processing results of each batch of precision parts using the processing qualification inspection model and the qualified inspection threshold. Randomly sample processing samples from each batch of precision parts to obtain the test parameters of each batch of processing samples, including the dimensional accuracy of each batch of processing samples, the surface roughness of each batch of processing samples, the residual stress of each batch of processing samples, and the hardness of each batch of processing samples; Input the inspection parameters of each batch of processing samples into the processing qualification inspection model respectively, and output the objective function of each batch of processing samples; Compare the objective function of each batch of processed samples with the qualified detection threshold. When the objective function is lower than the qualified detection threshold, the batch of processed samples is marked as unqualified in the intelligent detection. When the objective function is greater than or equal to the qualified detection threshold, the batch of processed samples is marked as qualified in the intelligent detection. The intelligent detection results of each batch of processing samples are counted, and then a precision parts processing inspection report is output based on the intelligent detection results of each batch of processing samples. The precision parts processing inspection report provides batch information of the processing samples that fail the intelligent detection.

Images