Cutter health state online monitoring method and system based on workpiece machining quality data fusion

Abstract

The invention discloses a tool health state online monitoring method and system based on workpiece machining quality data fusion, and the method comprises the steps: firstly, calculating the deviation between the actual groove width and the standard groove width of a workpiece, and constructing a time sequence of the groove width deviation; then, a sliding time window mechanism is adopted, multi-dimensional time sequence characteristics including the mean value, the standard deviation, the slope, the out-of-tolerance count and the front value are dynamically extracted, and the nonlinear evolution trend in the tool abrasion process is comprehensively described; further, the fused feature vectors can be input into a random forest regression model, and high-precision prediction of the current slot width is achieved; and finally, evaluating the degradation state of the cutter in real time based on the deviation between the predicted groove width and the standard groove width. According to the method, the groove width data in the machining process is directly utilized, additional arrangement of a sensor is not needed, the method has the advantages of being low in cost, high in reliability and high in real-time performance, the rejection rate can be reduced on the premise that hardware investment is not increased, and the service life of a tool is prolonged.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of intelligent manufacturing and predictive maintenance, and particularly relates to a tool health state online monitoring method based on workpiece machining quality data fusion, which is particularly suitable for online monitoring of tool health state in precision machining. BACKGROUND

[0002] In the field of intelligent manufacturing, especially in the precision machining process of scroll compressor core components-scroll disc, real-time monitoring and accurate prediction of tool health state are the key to ensure product precision and performance. Scroll disc profile is usually made of high-hardness aluminum alloy or cast iron material by high-speed milling, and the tooth shape precision requirement is extremely high, and it is a thin-walled part. The small wear of the tool during machining will directly reflect on the change of the workpiece groove width, and then affect the sealing and efficiency of the compressor. However, the current tool state monitoring technology applied to such machine tools faces significant bottlenecks: first, the traditional indirect monitoring method relying on vibration, acoustic emission or cutting force sensor not only needs to integrate the precision sensor inside the high-speed spindle system or the tool handle, which is complex to install and high in cost, and the signal is also easily affected by the machine tool itself vibration, electromagnetic interference and cooling liquid environment, with low signal-to-noise ratio. In the scroll disc machining process, the monitoring stability and reliability are difficult to guarantee. Second, the method of directly observing the tool flank wear land by machine vision must interrupt the machining process and disassemble the tool for offline measurement, which not only reduces the comprehensive efficiency of the equipment and increases the production rhythm, but also cannot capture the dynamic wear changes in the machining process. Third, the fixed time or fixed processing amount tool changing strategy completely ignores the influence of tool material batch difference, coating uniformity and machining parameter fine-tuning, and cannot adapt to individual differences, resulting in a too conservative tool changing strategy causing a huge waste of tool remaining life, or a too late tool changing causing tool sudden collapse and even workpiece scrap, which brings huge quality risk and economic loss in scroll disc high-value workpiece machining.

[0003] To address the above challenges, the prior art solutions have inherent limitations in the context of scroll disc machining. The monitoring scheme based on cutting force is relatively direct, but it requires the installation of a force meter under the workbench, which changes the dynamics of the machine tool, and for three-axis linkage machining centers, it is difficult to accurately decouple complex spatial cutting forces, the system is complex and the modification cost is high. Although the machine vision detection method is intuitive, it is limited by the closed machining chamber and cutting fluid mist environment of the scroll disc machine tool, the imaging quality is poor, and frequent stoppage for detection is contrary to the current trend of unmanned and black-light factory intelligent manufacturing. The slot width monitoring based on simple tolerance threshold commonly used on the current production line (such as SPC statistical process control) can only passively alarm after the workpiece size is out of tolerance, which is a post remedial measure and cannot model and predict the progressive degradation process of the tool, so it cannot issue an early warning in the early stage of tool performance degradation, nor can it predict the remaining useful life, making it difficult to avoid the production of batch waste products.

[0004] In summary, the prior art has not effectively solved the three core problems in high-precision machining such as scroll discs: First, how to deeply mine and utilize the conventional detection data (such as time series data of slot width size) that continuously generates during the existing machining process, which reflects the final quality of machining, to build a low-cost monitoring solution without additional sensors; Second, how to establish an accurate, dynamic, and quantifiable mapping relationship model from "workpiece geometric precision changes" to "tool physical wear"; Third, how to build a forward-looking early warning mechanism based on the above model, realize the transition from "passive alarm" to "active prediction", and intervene in advance before the tool performance degradation affects the workpiece precision, thereby fundamentally avoiding batch waste products and improving the quality control level and predictive maintenance capability of intelligent manufacturing. SUMMARY

[0005] The purpose of the present application is to overcome the deficiencies in the prior art, provide a tool health state online monitoring method based on workpiece machining quality data fusion, to solve the three major pain points existing in the traditional monitoring method: the installation of special sensors is complex, the cost is high and the signal is easy to be disturbed; direct detection needs to interrupt the production process, affecting the machining efficiency; and the fixed life tool replacement strategy cannot adapt to the individual differences of tools, resulting in low tool utilization or batch scrap risk. The specific implementation path includes: first, continuously collecting the slot width size data of the processed workpiece through the online measurement system, and calculating the real-time deviation of the slot width value from the standard slot width value, to build a complete slot width deviation time sequence; second, using the sliding window technology to extract multi-dimensional time sequence features from the slot width deviation time sequence, including moving average reflecting systematic deviation, standard deviation reflecting processing stability, change slope indicating degradation rate and super tolerance frequency statistics; then, based on machine learning algorithms such as random forest, the mapping relationship from multi-dimensional features to tool wear state is established, and the real-time calculation and evaluation of tool health index is realized; finally, by analyzing the historical trend of the health index, the tool degradation curve is established, the remaining service life is predicted, and the warning signal is sent when the predicted life reaches the preset threshold, guiding the production plan and tool replacement decision, and finally realizing the intelligent tool management change from passive response to active prediction. In order to achieve the above purpose, the present application adopts the following technical scheme: The purpose of the present application is realized by the following technical scheme: A tool health state online monitoring method based on workpiece machining quality data fusion, comprising: S1, obtaining the slot width actual measurement value sequence of the continuously processed workpiece and the corresponding standard slot width value online or near line; calculating the deviation between the slot width actual measurement value and the standard slot width value for each workpiece to obtain the time sequence of slot width deviation; S2, for the current workpiece to be evaluated, select the slot width deviation of the previous L workpieces from the slot width deviation time sequence to form a sliding time window, and calculate each feature value in the sliding time window: The mean of the slot width deviation is used to represent the systematic deviation of the slot width size; The standard deviation of the slot width deviation is used to represent the volatility of the slot width size and the stability of the processing process; The linear change slope of the slot width deviation is used to quantitatively represent the slot width change rate caused by tool wear; The number of times the slot width deviation exceeds the preset tolerance threshold is used to capture abnormal fluctuations in the processing process; And the slot width actual measurement value of the previous workpiece is taken as a time sequence continuity feature value, together with the above-mentioned feature values to form a feature vector representing the tool state; S3, normalize the absolute value of the groove width deviation and perform exponential weighted moving average smoothing to obtain a degradation index reflecting the tool health state; input the feature vector into a pre-trained random forest regression model to output a groove width prediction value of the current workpiece or a subsequent workpiece; based on the deviation between the groove width prediction value and the standard groove width value, evaluate the tool degradation trend, and when the degradation index corresponding to the current workpiece exceeds a preset tool breaking point threshold, determine it as a tool breaking point of tool failure; S4, based on the degradation index, construct a tool degradation curve, and calculate the change slope of the degradation index within a sliding time window; when the change slope meets a preset compensation determination condition, determine the corresponding workpiece as a compensation point for performing tool parameter compensation operation; and display the tool degradation curve, the tool breaking point and the compensation point in the visual interface to assist the operator in making tool maintenance decisions.

[0006] Further, the normalization of the absolute value of the groove width deviation includes: taking the absolute value of the maximum groove width deviation at the tool breaking point as a reference, normalizing the absolute value of the groove width deviation corresponding to the subsequent workpiece, and taking the normalization result as the base value of the degradation index.

[0007] Further, the preset tolerance threshold is set according to the design tolerance range of the workpiece groove width, and the number of times of exceeding the preset tolerance threshold is the number of workpieces that meet the condition that the absolute value of the groove width deviation is greater than the preset tolerance threshold within the sliding time window.

[0008] Further, the determination of tool compensation includes: calculating the instantaneous change slope of the degradation index within the sliding time window; comparing the instantaneous change slope with a preset negative determination slope threshold; when the instantaneous change slope is less than the negative determination slope threshold, it is determined that a tool health state significantly improved event caused by external compensation operation has occurred at that moment.

[0009] Further, for the precision machining process of a compressor scroll plate, the tool health state in the machining process of the scroll plate is monitored and the remaining life is evaluated by analyzing the time sequence change of the groove width deviation of the scroll plate.

[0010] Further, based on the degradation index containing the compensation point obtained based on actual machining data, an exponential growth model is used to simulate the tool degradation process without compensation strategy to obtain a degradation index without compensation strategy; based on the degradation index containing the compensation point and the degradation index without compensation strategy, respectively, the corresponding life of the tool with compensation strategy and without compensation strategy and the life improvement multiple are calculated; The tool cost parameter and the number of processed products are combined to quantify the cost savings brought by the compensation strategy, which provides data-driven decision basis for the optimization of the tool management strategy of the production line.

[0011] Further, the exponential growth model iteratively updates the degradation index based on a degradation growth rate parameter and a random fluctuation factor, which is used to reproduce the natural degradation trend of the tool without compensation.

[0012] Preferably, the present application also provides a tool health state online monitoring system based on workpiece processing quality data fusion, comprising: A data acquisition and preprocessing module is configured to acquire a sequence of actual groove width measurement values of continuously processed workpieces and corresponding standard groove width values online or nearline, and calculate the deviation between the actual groove width measurement value and the standard groove width value for each workpiece to obtain a time series of groove width deviation; A multi-dimensional time series feature extraction module is configured to select the groove width deviation of the previous L workpieces to form a sliding time window from the time series of groove width deviation for the current workpiece to be evaluated, and sequentially calculate each feature value in the sliding time window, including the mean, standard deviation, linear change slope of the groove width deviation, and the number of times the groove width deviation exceeds the preset tolerance threshold, and take the actual groove width measurement value of the previous workpiece as a time series continuity feature value, and together with the above-mentioned feature values to form a feature vector representing the tool state; A tool change point determination module is configured to normalize and exponentially weighted moving average smooth the absolute value of the groove width deviation to obtain a degradation index reflecting the tool health state, input the feature vector into a pre-trained random forest regression model, and output the groove width prediction value of the current workpiece or subsequent workpiece, evaluate the tool degradation trend based on the deviation between the groove width prediction value and the standard groove width value, and determine the tool change point when the degradation index corresponding to the current workpiece exceeds the preset tool breaking point threshold. A compensation point identification and visualization module is configured to construct a tool degradation curve based on the degradation index, calculate the change slope of the degradation index in the sliding time window, and determine the corresponding workpiece as a compensation point for performing tool parameter compensation operation when the change slope meets the preset compensation determination condition, and display the tool degradation curve, the tool change point and the compensation point in the visualization interface to assist the operator in making tool maintenance decisions. Preferably, the present application also provides an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor is configured to execute the steps of the tool health state online monitoring method when executing the computer program.

[0013] Preferably, the present application also provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the tool health state online monitoring method.

[0014] Compared with the prior art, the technical scheme of the present application has the beneficial effects that: 1. Directly using the existing workpiece groove width detection data in the production process, constructing the groove width deviation time sequence by calculating the deviation between the actual measurement value and the standard value of the groove width, without the need for additional arrangement of special sensors such as force, vibration or acoustic emission. Using the detection data directly related to the final geometric accuracy as the monitoring signal, the data source is directly reliable, without the need for modification of the machine tool structure, significantly reducing the hardware investment and operation complexity, and avoiding the problem of indirect signal being easily disturbed, with a direct and reliable data source.

[0015] 2. A sliding time window is introduced for the groove width deviation sequence, and multi-dimensional features such as the mean, standard deviation, linear change slope, number of times exceeding the preset tolerance threshold and previous workpiece groove width actual value are extracted from the window to form a multi-dimensional feature vector. The mean reflects systematic deviation and can capture the gradual widening or narrowing trend of the groove width caused by overall tool wear; the standard deviation describes the fluctuation degree and can distinguish between stable wear and abnormal fluctuation; the slope reflects the change speed of the deviation and can describe the acceleration stage of wear; the number of times exceeding the preset tolerance threshold is sensitive to frequent impacts near the tolerance edge, which is beneficial to discovering early quality risks; the previous workpiece groove width value introduces time sequence continuity information, so that the model can better understand short-term dynamics. The above multi-features cooperate to make the tool state description more comprehensive, effectively capture the progressive degradation trend of the tool, overcome the hysteresis of simple threshold alarm, and improve the accuracy and robustness of degradation identification and trend prediction.

[0016] 3. Degradation index construction based on absolute deviation normalization and exponential weighted moving average (EWMA); taking the absolute value of the maximum groove width deviation at the tool change point as the reference to normalize the subsequent deviations, and using exponential weighted moving average to smooth the degradation index, to obtain a smoothed degradation index sequence. Normalization makes the degradation index of different batches of tools and different working conditions comparable, facilitating the setting of threshold and strategy; EWMA gives more weight to recent data, retaining the degradation trend and effectively filtering out high-frequency measurement noise, making the degradation curve smooth and having clear physical meaning.

[0017] 4. The tool change point is identified by setting a tool breakage point threshold, and a dynamic threshold detection mechanism can be formed in combination with historical statistical characteristics. Tool change decision is no longer dependent on experience, but is dynamically determined based on the actual degradation degree of the tool, which can fully exploit the remaining life of high-quality tools and identify risks in advance when degradation accelerates, improving tool utilization and reducing sudden breakage and batch rejects.

[0018] 5. Calculate the change slope of the degradation index within the sliding time window, and identify the tool parameter compensation behavior combined with the slope mutation feature; the degradation curve presents periodic changes with periodic rising and sudden falling after compensation. Through the structured analysis of the degradation curve, the typical cycle of natural wear → parameter compensation → wear re-accumulation → complete failure can be clearly presented, which helps to understand the compensation ability boundary of the control system to the machining quality, and provides a basis for optimizing the compensation frequency and compensation amount, so as to maximize the tool life under the premise of ensuring accuracy.

[0019] 6. On the basis of actual compensation data, introduce an exponential growth model to simulate the tool degradation process under non-compensation conditions, compare the tool life, number of machined parts and cost benefits under the two strategies, and calculate the life improvement multiple and unit tool cost savings. Through data-driven comparative analysis, the life can be improved by 1.91 times, and the cost can be saved by about 900 yuan per tool changing period, so that the compensation strategy can be upgraded from experience selection to measurable and optimized decision-making.

[0020] 7. Input multi-dimensional time series features into a random forest regression model to realize groove width prediction; experiments show that R 2 can be obtained close to 0.9, RMSE≈0.0020, MAE≈0.0012, with micron-level error. Random forest can capture complex nonlinear relationships and feature interactions, combined with sliding time window features, it can still maintain high prediction accuracy under fluctuating working conditions and nonlinear tool wear, so that the system can not only monitor the current state, but also make forward-looking predictions about the machining quality of the next or several workpieces, further releasing early warnings.

[0021] 8. Integration of online monitoring and visualization, avoiding downtime and offline detection; the degradation index, tool changing point and compensation point are intuitively presented in the visualization interface, realizing closed-loop online monitoring from data acquisition, feature extraction, state evaluation to result display. The operator can observe the tool state evolution trend in real time without interrupting the machining process, adjust the plan in advance before approaching the risk area, and significantly improve the comprehensive utilization rate of equipment; it meets the development of intelligent and unmanned workshops.

[0022] 9. Map the geometric accuracy (groove width) of the machining result to the tool comprehensive performance state through the degradation index, while considering factors such as machine tool compensation behavior and machining conditions. Through the method of reverse tracing from output quality to input state, realize quality reverse tracing and system-level performance evaluation, and provide a unified framework for the subsequent introduction of more quality features and the realization of "workpiece machining quality data fusion" extension.

[0023] 10. The method can be data-driven, and can automatically learn and adapt to the wear characteristics of different tools, different materials and different machining parameters through model training, overcoming the limitations of fixed life tool replacement strategy, realizing predictive maintenance based on the actual state of individual tools, and significantly improving tool utilization. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 is a flowchart of the feature extraction part.

[0025] Figure 2 is a flowchart of predicting the groove width of a workpiece using a random forest regression model.

[0026] Figure 3a is a comparison curve of the predicted groove width and the actual value, Figure 3b each subgraph in is a segmented visualization view of the comparison curve of the total ID sequence in Figure 3a

[0027] Figure 4 is a tool degradation amount trend graph with a compensation strategy.

[0028] Figure 5 is a tool degradation amount trend graph without a compensation strategy. DETAILED DESCRIPTION

[0029] The present application will be further described in detail below in conjunction with the accompanying drawings and specific examples. It should be understood that the specific examples described herein are intended to explain the present application, and are not intended to limit the present application.

[0030] Example 1 The present embodiment provides a tool health state online monitoring method based on workpiece machining quality data fusion, comprising the following steps: S1, data acquisition and preprocessing: online or nearline acquisition of groove width actual measurement value sequence of continuous machining workpiece and its corresponding standard groove width value ; calculate the groove width deviation value of each workpiece : ; get the time sequence of groove width deviation }.

[0031] S2, multi-dimensional time sequence feature extraction: for the current workpiece , select the groove width deviation data of the previous workpieces to form a sliding time window, and extract a feature vector fusing multi-dimensional degradation information from the sliding time window: , wherein,​ is the mean of groove width deviation within the sliding time window, reflecting the systematic deviation of groove width dimension; is the standard deviation of groove width deviation within the sliding time window, reflecting the fluctuation of groove width dimension and the stability of machining; is the linear change slope of groove width deviation within the sliding time window, quantitatively characterizing the change rate of groove width caused by tool wear; is the number of times exceeding the preset tolerance threshold within the sliding time window, is the preset tolerance threshold, is an indicator function, which is used to capture abnormal fluctuations in the machining process; is the actual measurement value of the groove width of the previous workpiece, which is used as a time series continuity feature value to enhance the continuity of the time series.

[0032] S3, tool degradation state evaluation based on statistical process; To evaluate the tool health state in real time and accurately detect the tool change (tool breakage) point, the embodiment proposes a statistical process method based on groove width deviation sequence, the core of which is to construct a smooth and normalized tool degradation index (the larger the value, the worse the tool health state), and a double threshold condition is set to trigger the tool change warning reliably. First, the absolute value sequence of groove width deviation obtained in the order of workpieces is monitored, and an abnormal deviation threshold based on process knowledge is set When first exceeds , the time and the corresponding absolute value of deviation are recorded, and they are used as the reference maximum value for subsequent normalization processing , thereby establishing an objective scale for index calculation. Based on the multi-dimensional feature vector of the current workpiece , it is input into the pre-trained random forest regression model for prediction. The model is composed of decision trees, and each decision tree independently outputs a prediction value , and then the average of the prediction values of all trees is taken in the aggregation layer to obtain the final prediction value , which represents the prediction of the groove width deviation of the next workpiece . Further, the reference maximum deviation defined in S3 can be used to calculate the predicted health index , which is used to evaluate the predicted health state of the next workpiece, and the closer the value is to 1, the healthier it is, and the closer the value is to 0, the more serious the degradation is.

[0033] Then, the reference value is used to normalize the historical and real-time deviations, and the sequence To suppress measurement noise and smooth the trend, an exponentially weighted moving average (EWMA) filter is applied to the sequence. The specific calculation formula is as follows: Smoothing factor The weights used to control historical information are used to obtain the continuous tool degradation index. This clearly characterizes the gradual degradation process of tool performance. Finally, tool change is determined by comprehensively considering the level of degradation indicators and their instantaneous changes: when the degradation level conditions are met simultaneously... (in For a degradation threshold close to 1 (e.g., 0.85) and mutation conditions (in When the abrupt change threshold between adjacent deviations is reached, the system determines that a tool breakage has occurred and triggers a tool change signal. This process effectively filters out measurement noise interference. By continuously depicting the progressive degradation of the tool and reliably locating the tool change point using a dual threshold mechanism, the system achieves precise tracking and status assessment of the tool's degradation process throughout its entire lifecycle.

[0034] S4. Tool status visualization and compensation point recognition: Based on tool degradation index Construct tool degradation curves by continuously The slope of change of the degradation index is calculated within a sliding time window consisting of individual workpieces. It can identify compensation events (such as tool adjustment or process parameter optimization) during tool degradation. The compensation point determination criteria are as follows: , here It is a negative decision slope threshold specifically used to determine abrupt changes in degradation trends. Its physical meaning is the same as the preset tolerance threshold used in S2 to count the number of out-of-tolerance events. different: This is used to determine whether the groove width deviation of a single workpiece exceeds the process tolerance; it falls under the dimensional threshold. This method, used to determine whether the rate of decline of the degradation curve reaches the condition for a compensation event, falls under the threshold category of trend change. It can monitor tool health online and display the tool degradation curve, tool change point, and compensation point in real time through a visual interface, providing intuitive data support for tool maintenance and process adjustment decisions during production. It achieves real-time tool status monitoring based on multi-threshold rules and statistical processes. Specifically, it includes a negative slope threshold. The slope can be statistically set based on the typical slope values ​​when compensation events occur in historical data, or determined by process engineers based on experience.

[0035] S5. Comparative Analysis and Economic Benefit Evaluation of Tool Compensation Strategies: A comparison framework between compensated and uncompensated strategies is established by simulating the tool degradation process under the uncompensated strategy.

[0036] S501. Degradation index with compensated strategy (actual monitoring): Based on the actual machining data, the degradation index under the compensated strategy is constructed . The calculation method of this index is consistent with the one defined in S3, that is, first normalize the absolute value of groove width deviation: .

[0037] Then apply the exponential weighted moving average (EWMA) filter for smoothing: .

[0038] Where is the smoothing factor, is the reference maximum deviation defined in S3. By monitoring in real time and combining the double threshold judgment mechanism in S3, the actual tool life under the compensated strategy can be obtained (in terms of the number of machined workpieces).

[0039] S502. Degradation index without compensation strategy (simulated generation): To construct a comparison benchmark, the tool degradation trend under the uncompensated strategy is simulated. This embodiment takes compressor scroll disc machining as an example, and the specific simulation method is as follows: 1) Initial state: take the ratio of the absolute value of groove width deviation of the first workpiece (k=1) to the reference maximum deviation as the initial degradation index: ; 2) Trend simulation: for subsequent products k (k≥2), use an exponential growth model and introduce random fluctuations to simulate its degradation process: ; Where is the preset degradation growth rate, which can be calibrated according to historical data. is the random fluctuation factor, which follows a normal distribution with mean 1 and standard deviation (e.g. ), that is , and is truncated to a reasonable interval to conform to physical reality. The function ensures that the index value does not exceed 1.0 (the complete degradation threshold).

[0040] When first reaches or exceeds 1.0, it is determined that the tool has failed in simulation, and the corresponding product number at this time is​​ This refers to the tool life under the simulated uncompensated strategy. .

[0041] S503. Quantitative Assessment of Technological Advantages and Economic Benefits: By comparing the tool life with compensation strategies Tool life with no compensation strategy This allows for a quantitative comparison of tool life (based on the number of products machined) under two strategies and the technical advantages of the compensation strategy. The tool life improvement factor of the compensation strategy is calculated. and the number of processed products The technical advantages of the compensation strategy are evaluated, and further combined with the cost parameter of a single tool (assuming the cost per tool is...). The direct economic benefits (tool cost savings) brought about by the compensation strategy can be calculated: .

[0042] This method uses a visual interface (such as...) Figure 4 and 5 The tool degradation state diagrams (with and without compensation strategies) visually compare the tool degradation curves under the two strategies. Figure 4 The marked compensation points and tool change points are for compensation strategies. Figure 5 The dashed lines mark the tool change points under the simulated no-compensation strategy. This provides production decision-makers with clear quantitative evidence and data support for evaluating the effectiveness of compensation strategies and optimizing tool maintenance plans.

[0043] Preferably, the sliding time window feature engineering and random forest regression model in this embodiment are used to achieve higher precision prediction of the workpiece groove width. In this embodiment, the training data comes from continuous monitoring of a tool during processing, which includes a tool change operation. Data before and after the tool change is effectively collected. A total of 1990 workpiece groove width data were monitored, and 398 effective measurement points were obtained according to the frequency of measuring every 5 workpieces. Each sample extracts a 5-dimensional feature vector through a sliding time window with a length of L (L=10 in this embodiment), including the mean, standard deviation, linear change slope, number of times exceeding the preset tolerance threshold, and actual groove width value of the previous workpiece, to integrate multi-dimensional degradation information. Before model training, all features are standardized by Z-score to improve training stability. The random forest regression model used is composed of 100 decision trees, with a maximum depth limit of 10 for each tree, and a minimum sample split number of 5 and a minimum sample leaf number of 2 to prevent overfitting. The model is optimized and evaluated for stability through 5-fold cross-validation, and the out-of-bag sample error is used for internal validation. During training, the first 278 measurement points (about 70%) are used as the training set, and the last 120 measurement points (about 30%) are used as the test set. The model achieves a prediction performance of RMSE 0.0020, MAE 0.0012, and R 2 0.94 on the test set. To further verify the generalization ability, the groove width data from another independent processing location were tested, and reliable results of RMSE about 0.0023, MAE about 0.0015, and R 2 about 0.92 were obtained, indicating that the model has good fitting ability and generalization performance. According to the specific requirements of the patent law for the description of the drawings, Figure 2 the complete process from 5-dimensional feature vector input, independent prediction by 100 decision trees, average output of prediction value in the aggregation layer, and final calculation of health index is shown; Figure 3a By plotting the comparison curve of the predicted and actual values of the compressor scroll disc obtained by processing, the predictability corresponding to the aforementioned evaluation indicators is intuitively displayed. Further, Figure 3b by segmenting and visualizing the comparison curve in Figure 3a according to the total compressor scroll disc product ID order; through error analysis, it is found that the predicted and actual values are highly consistent within the tolerance range, especially in the area with large groove width fluctuations, the random forest regression model still maintains good tracking ability, verifying its effectiveness in capturing nonlinear wear trends and small deviation evolution. The prediction model constructed in this embodiment not only performs excellently in statistical indicators, but also exhibits good stability and adaptability in engineering practice, providing reliable data support and model foundation for subsequent tool life prediction, processing quality control, and optimization of intelligent manufacturing systems.

[0044] like Figure 4 As shown, the tool degradation curve does not represent the linear accumulation of physical tool wear, but rather reflects the dynamic change process of the tool's overall machining efficiency. This curve quantifies the deviation of the machined product from the standard groove width, mapping the tool state to a normalized degradation index, thus achieving indirect perception of the tool's health status. The phased upward trend in the curve represents the accuracy decay caused by natural tool wear during machining, while the sudden downward inflection point corresponds to the machine tool control system's compensation intervention on the tool parameters. This compensation mechanism temporarily restores the tool's machining accuracy by correcting its machining parameters, reducing the product groove width deviation and consequently lowering the degradation index. It is worth noting that this decrease is not a repair of the tool's physical state, but rather an accuracy compensation effect achieved through system-level compensation. The entire change pattern reveals the typical characteristics of the tool's life cycle: after experiencing an iterative cycle of "wear-compensation-re-wear," the tool eventually enters an uncompensable failure state (marked with tool change points in the figure).

[0045] The value of this method lies in its ability to infer the input state (tool performance) from the output data (product slot width), establishing a quality reverse traceability mechanism based on machining results, and providing data support for predictive maintenance. By analyzing the changing trends and compensation frequency of the curves, the tool compensation strategy can be optimized, thereby achieving the optimal balance between machining accuracy and tool life, reducing unplanned downtime and scrap rate. After adopting the tool compensation strategy described in this invention, under the same cutting conditions, a single tool can continuously process 233 products, compared to 122 products without the compensation strategy, resulting in a 1.91-fold increase in tool life, or 111 more products processed. Based on a tool price of 1000 yuan, this translates to a cost saving of 909.84 yuan per tool change cycle. During the experiment, only one tool change was detected, at product ID=233, where the tool wear was 0.0536, fully verifying the significant effect of the compensation strategy in extending tool life, reducing tool change frequency, and saving machining costs.

[0046] Example 2 Based on the same inventive concept, this application also provides an online tool health status monitoring system based on workpiece machining quality data fusion, which can be used to implement the methods described in the above embodiments, as shown in the following embodiments. Since the principle of the online tool health status monitoring system based on workpiece machining quality data fusion is similar to that of the online monitoring method, the implementation of the online monitoring system can refer to the implementation of the online monitoring method, and repeated details will not be elaborated further. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0047] Embodiments of the present application provide a specific implementation of a tool health state online monitoring system based on workpiece machining quality data fusion capable of realizing an online monitoring method, and specifically include the following contents: A data acquisition and preprocessing module is configured to acquire a sequence of actual groove width measurement values of continuously machined workpieces and corresponding standard groove width values online or near online, and calculate the deviation between the actual groove width measurement value and the standard groove width value for each workpiece to obtain a time sequence of groove width deviation; A multi-dimensional time sequence feature extraction module is configured to select the groove width deviation of the previous L workpieces to form a sliding time window from the time sequence of groove width deviation for the current workpiece to be evaluated, and sequentially calculate each feature value in the sliding time window, including the mean, standard deviation, linear change slope of the groove width deviation, and the number of times the groove width deviation exceeds a preset tolerance threshold, and take the actual groove width measurement value of the previous workpiece as a time sequence continuity feature value, and together with the above-mentioned feature values to form a feature vector representing the tool state; A tool change point determination module is configured to normalize and perform exponential weighted moving average smoothing processing on the absolute value of the groove width deviation to obtain a degradation index reflecting the tool health state; when the degradation index corresponding to the current workpiece exceeds a preset tool change point threshold, the time point corresponding to the workpiece is determined as a tool failure tool change point; A compensation point identification and visualization module is configured to construct a tool degradation curve based on the degradation index, calculate the change slope of the degradation index in the sliding time window, and when the change slope meets a preset compensation determination condition, determine the corresponding workpiece as a compensation point for performing tool parameter compensation operation; and display the tool degradation curve, the tool change point and the compensation point in a visual interface to assist the operator in making tool maintenance decisions; A compensation strategy effect comparison module is configured to obtain the degradation index containing the compensation point based on the actual machining data, simulate the tool degradation process without compensation strategy using an exponential growth model to obtain the degradation index without compensation strategy, calculate the corresponding life and life improvement multiple of the tool under the compensation strategy and the non-compensation strategy based on the degradation index containing the compensation point and the degradation index without compensation strategy, and combine the tool cost parameter and the number of processed products to quantify the cost savings brought by the compensation strategy, thereby providing data-driven decision basis for the optimization of production line tool management strategy.

[0048] Preferably, the embodiments of the present application also provide a specific implementation of an electronic device capable of realizing all steps in the tool health state online monitoring method based on workpiece machining quality data fusion in the above-mentioned embodiments, and the electronic device specifically includes the following contents: The processor, the memory, the communications interface and the bus; The processor, the memory, the communications interface and the bus;

[0049] The processor is configured to invoke the computer program in the memory, and the processor executes the computer program to implement all the steps in the tool health state online monitoring method based on workpiece machining quality data fusion in the above embodiments.

[0050] The embodiments of the present application also provide a computer readable storage medium capable of implementing all the steps in the tool health state online monitoring method based on workpiece machining quality data fusion in the above embodiments. The computer readable storage medium stores a computer program. When the computer program is executed by the processor, all the steps in the tool health state online monitoring method based on workpiece machining quality data fusion in the above embodiments are implemented.

[0051] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment mainly describes the difference from other embodiments. In particular, for the hardware+program type embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiments.

[0052] The above describes specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the acts or steps recited in the claims can be performed in a different order than those in the embodiments and still achieve desirable results. In addition, the processes depicted in the figures do not necessarily require the particular order shown or sequential order to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous.

[0053] Although the present application provides method operation steps such as embodiments or flowcharts, more or fewer operation steps can be included based on conventional or non-creative labor. The order of steps listed in the embodiments is only one of the many execution orders of the steps, and does not represent the only execution order. When the device or client product is executed in practice, the method order shown in the embodiments or the drawings can be executed in sequence or in parallel (for example, in a parallel processor or multi-thread processing environment).

[0054] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, a system or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0055] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0056] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0057] The present application is not limited to the embodiments described above. The above description of specific embodiments is intended to describe and enable the technical solutions of the present application, and the specific embodiments described above are merely illustrative and are not restrictive. Those skilled in the art can make many forms of specific changes under the inspiration of the present application without departing from the purpose of the present application and the scope protected by the claims, and these all belong to the scope of protection of the present application.

Claims

1. A method for online monitoring of tool health status based on workpiece machining quality data fusion, characterized in that, include: S1. Obtain the actual measurement sequence of the groove width of continuously processed workpieces and the corresponding standard groove width value online or near-line; For each workpiece, the deviation between the actual measured value of the groove width and the standard groove width value is calculated to obtain a time sequence of the groove width deviation; S2. For the workpiece to be evaluated, select the slot width deviations of the first L workpieces from the time sequence of the slot width deviations to form a sliding time window, and calculate each characteristic value sequentially within the sliding time window: The mean value of the slot width deviation is used to characterize the systematic offset of the slot width dimension; The standard deviation of the groove width deviation is used to characterize the variability of the groove width dimension and the stability of the processing. The slope of the linear change in groove width deviation is used to quantitatively characterize the rate of change in groove width caused by tool wear; The number of times the groove width deviation exceeds the preset tolerance threshold is used to capture abnormal fluctuations during the processing. The actual measured value of the groove width of the previous workpiece is used as the temporal continuity feature value, which together with the above feature values ​​constitutes a feature vector characterizing the tool state. S3. The absolute value of the groove width deviation is normalized and smoothed by exponential weighted moving average to obtain a degradation index reflecting the health status of the tool. The feature vector is input into a pre-trained random forest regression model, which outputs the predicted slot width of the current workpiece or subsequent workpieces. Based on the deviation between the predicted slot width and the standard slot width, the tool degradation trend is evaluated. When the degradation index corresponding to the current workpiece exceeds the preset tool breakage threshold, it is determined to be the tool change point where the tool fails. S4. Construct a tool degradation curve based on the degradation index, calculate the slope of the degradation index within the sliding time window, and when the slope of the degradation index meets the preset compensation judgment condition, determine the corresponding workpiece as the compensation point for performing tool parameter compensation operation. The tool degradation curve, tool change point, and compensation point are realistically displayed in the visualization interface to assist operators in making tool maintenance decisions.

2. The online monitoring method for tool health status according to claim 1, characterized in that, The normalization process for the absolute value of the slot width deviation includes: taking the absolute value of the maximum slot width deviation at the tool change point as a benchmark, normalizing the absolute value of the slot width deviation corresponding to subsequent workpieces, and using the normalization result as the base value of the degradation index.

3. The online tool health status monitoring method according to claim 1 or 2, characterized in that, The preset tolerance threshold is set according to the design tolerance range of the workpiece groove width, and the number of times the preset tolerance threshold is exceeded is the number of workpieces whose absolute value of the groove width deviation is greater than the preset tolerance threshold within the sliding time window.

4. The online monitoring method for tool health status according to claim 1, characterized in that, The determination of tool compensation includes: calculating the instantaneous change slope of the degradation index within the sliding time window; comparing the instantaneous change slope with a preset negative determination slope threshold; when the instantaneous change slope is less than the negative determination slope threshold, it is determined that a significant improvement in tool health caused by external compensation operation has occurred at that moment.

5. The online monitoring method for tool health status according to claim 1, characterized in that, For the precision machining process of compressor scroll plates, the tool health status and remaining life assessment during the scroll plate machining process are carried out online by analyzing the temporal changes in the scroll plate groove width deviation.

6. The online monitoring method for tool health status according to claim 1, characterized in that, Based on the degradation index containing compensation points obtained from actual machining data, an exponential growth model is used to simulate the tool degradation process under the no-compensation strategy, and the degradation index of the no-compensation strategy is obtained. Based on the degradation index containing compensation points and the degradation index without compensation strategy, respectively, the tool life and life improvement factor under compensation strategy and without compensation strategy are calculated. By combining tool cost parameters with the number of products processed, the cost savings brought about by the compensation strategy are quantified, providing a data-driven decision-making basis for optimizing the tool management strategy of the production line.

7. The online monitoring method for tool health status according to claim 6, characterized in that, The exponential growth model iteratively updates the degradation index based on the degradation growth rate parameter and the random fluctuation factor, in order to reproduce the natural degradation trend of the tool under uncompensated conditions.

8. An online tool health status monitoring system based on workpiece machining quality data fusion, characterized in that, include: The data acquisition and preprocessing module is used to acquire the actual measurement sequence of the groove width of continuously processed workpieces and the corresponding standard groove width value online or near-line. For each workpiece, the deviation between the actual measured value of the groove width and the standard groove width value is calculated to obtain a time sequence of the groove width deviation; The multi-dimensional temporal feature extraction module is used to select the slot width deviations of the first L workpieces from the temporal sequence of the slot width deviations to form a sliding time window for the current workpiece to be evaluated. Within the sliding time window, each feature value is calculated sequentially, including the mean, standard deviation, linear change slope, and number of times the slot width deviation exceeds the preset tolerance threshold. The actual measured value of the slot width of the previous workpiece is used as the temporal continuity feature value, which together with the above feature values ​​constitutes a feature vector characterizing the tool state. The tool change point determination module is used to normalize the absolute value of the slot width deviation and perform exponential weighted moving average smoothing to obtain a degradation index reflecting the tool health status; input the feature vector into a pre-trained random forest regression model to output the predicted slot width value of the current workpiece or subsequent workpieces; based on the deviation between the predicted slot width value and the standard slot width value, evaluate the tool degradation trend; when the degradation index corresponding to the current workpiece exceeds the preset tool breakage point threshold, it is determined as a tool change point of tool failure. The compensation point identification and visualization module is used to construct a tool degradation curve based on the degradation index, calculate the slope of the degradation index within the sliding time window, and determine the corresponding workpiece as the compensation point for performing tool parameter compensation operation when the slope of the degradation index meets the preset compensation judgment conditions. The module also displays the tool degradation curve, the tool change point, and the compensation point in a visual interface to assist operators in making tool maintenance decisions.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it is configured to perform the steps of the online tool health status monitoring method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it performs the steps of the online tool health status monitoring method according to any one of claims 1 to 7.

Images