Casting quality analyzing and tracking system of numerical control machine tool

Abstract

The invention relates to a numerical control machine tool casting quality analysis tracking system, which belongs to the technical field of intelligent manufacturing and quality control, and comprises a multi-physical field fusion monitoring module used for acquiring multi-element physical signals of a machine tool in real time and constructing real-time state feature vectors based on the multi-element physical signals, a defect evolution real-time inversion module and a defect evolution real-time inversion module. The real-time state feature vector calculation module is used for combining the real-time state feature vector with a preset casting initial defect factor and calculating to obtain a real-time defect activation index through a machining induced defect activation index model, and the cutting parameter dynamic optimization module is used for calculating the real-time defect activation index based on the machining induced defect activation index model and a preset process constraint by solving a normalized machining utility function. According to the method, the high-value precision casting is remarkably improved, the one-time machining qualification rate in manufacturing of core components in the aerospace field is remarkably increased, and the reworking and scrapping cost is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent manufacturing and quality control, and in particular to a quality analysis and tracking system for CNC machine tool castings. Background Art

[0002] In high-end manufacturing fields such as aerospace, precision castings of core components such as turbine blades often contain hidden metallurgical defects such as micro-porosity and inclusions in the blank stage. In high-speed CNC precision machining, the thermal field formed by the intense cutting heat and cutting force will have a complex coupling effect with these potential defects. The technical problem faced by this field is that this coupling effect will activate and induce the expansion of originally stable defects, forming secondary damage. Therefore, the pursuit of high-efficiency processing strategies often leads to a decrease in yield, while conservative processing seriously affects the production cycle and cost. Whether the existing technology adopts conservative processing parameters or relies on non-destructive testing after processing is completed, it cannot intervene in the process in real time, thereby placing processing efficiency and defect control in a dilemma that is mutually exclusive and cannot be achieved at the same time. This field urgently needs an intelligent control method that can perceive and avoid processing-induced damage in real time to achieve the unity of efficiency and quality.

[0003] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0004] The purpose of the present invention is to provide a CNC machine tool casting quality analysis and tracking system to solve the problems raised in the above background technology.

[0005] The technical solution of the present invention comprises: a multi-physics field fusion monitoring module for acquiring multivariate physical signals of a machine tool in real time and constructing a real-time state feature vector based on the multivariate physical signals;

[0006] The real-time inversion module for defect evolution is used to combine the real-time state feature vector with the preset initial defect factor of the casting and calculate the real-time defect activation index through the processing-induced defect activation index model;

[0007] The cutting parameter dynamic optimization module is used to determine the optimal cutting parameter set by solving the normalized machining utility function based on the machining-induced defect activation index model and preset process constraints;

[0008] The damage suppression mode switching module is used to compare the real-time defect activation index with the preset safety threshold and critical threshold, and based on the comparison results, combined with the optimal cutting parameter set, to determine the final processing instructions sent to the CNC system.

[0009] Preferably, the multivariate physical signals include: high-frequency acoustic emission sensor signals, three-force measuring instrument signals, high-precision accelerometer signals and non-contact infrared thermometer signals.

[0010] Preferably, the steps for constructing the real-time state feature vector are as follows:

[0011] S11, extracting a root mean square value of an acoustic emission signal from a high-frequency acoustic emission sensor signal;

[0012] S12, extracting the cutting force from the three-force measuring instrument signal;

[0013] S13, extracting the kurtosis of the vibration signal from the high-precision accelerometer signal;

[0014] S14, extracting cutting temperature from the non-contact infrared thermometer signal;

[0015] S15. Combine the root mean square value of the acoustic emission signal, the cutting force, the kurtosis of the vibration signal, and the cutting temperature to construct a real-time state feature vector.

[0016] Preferably, the calculation steps of the processing-induced defect activation index are specifically as follows:

[0017] S21. Determine the thermodynamic effect term based on the cutting temperature;

[0018] S22. determining a quasi-static mechanical effect term based on the root mean square value of the acoustic emission signal;

[0019] S23. determining a dynamic impact effect term based on the kurtosis of the vibration signal and the cutting resultant force;

[0020] S24, performing weighted summation on the thermodynamic effect term and the quasi-static mechanical effect term to obtain a combined effect term;

[0021] S25. Generate a real-time defect activation index based on the combined effect term, the dynamic impact effect term, and the casting initial defect factor.

[0022] Preferably, the steps for determining the optimal cutting parameter set are as follows:

[0023] S31. Calculating a material removal rate based on the cutting parameters to be optimized;

[0024] S32, predicting a state feature vector corresponding to the cutting parameter to be optimized using a preset cutting process proxy model;

[0025] S33, substituting the predicted state feature vector into the processing-induced defect activation index model to obtain a predicted defect activation index;

[0026] S34. Combining the material removal rate with the predicted defect activation index, the maximum value of the normalized machining utility function is solved to determine the optimal cutting parameter set.

[0027] Preferably, the cutting process proxy model is obtained by establishing a mapping relationship from cutting parameters to state feature vectors through an off-line calibration experiment.

[0028] Preferably, the steps for determining the processing instructions are as follows:

[0029] If the real-time defect activation index is less than or equal to the safety threshold, the system enters the high-efficiency mode and uses the optimal cutting parameter set calculated by the cutting parameter dynamic optimization module with the main goal of maximizing machining efficiency as the machining instruction;

[0030] If the real-time defect activation index is greater than the safety threshold and less than the critical threshold, the system enters the adaptive mode and adopts the optimal cutting parameter set calculated by the cutting parameter dynamic optimization module to balance machining efficiency and risk suppression as the machining instruction;

[0031] If the real-time defect activation index is greater than or equal to the critical threshold, the preset emergency plan is used as the processing instruction.

[0032] Preferably, the safety threshold is derived from a real-time defect activation index value that is calibrated through a trial cutting experiment and ensures that the microcrack density is lower than the material allowable standard under a preset confidence level.

[0033] Preferably, the critical threshold value is derived from: a statistical lower limit value of the instantaneous peak value of the real-time defect activation index of all failed samples immediately before failure, which is calibrated through destructive cutting experiments.

[0034] The present invention provides a CNC machine tool casting quality analysis and tracking system through improvements, which has the following improvements and advantages compared with the prior art:

[0035] First, the present invention achieves the coordinated optimization of machining efficiency and workpiece quality, transforming the traditional opposing constraint relationship between the two into a goal of coordinated improvement. The multi-physics field fusion monitoring module obtains the real-time state feature vector that characterizes the machining process, and the defect evolution real-time inversion module converts it into a quantitative assessment of internal damage risk. For the evolution behavior of microscopic defects in the material that cannot be directly observed, a reliable quantitative indicator associated with external measurable physical signals is established. The system can proactively determine an optimal cutting parameter set, which is essentially the highest efficiency solution that can be achieved without triggering damage risk. This mechanism enables the system to dynamically and adaptively operate on the optimal boundary between efficiency and quality throughout the entire machining process.

[0036] Secondly, the present invention changes the quality control mode from the passive response of post-inspection to the active avoidance of in-process prevention; the existing technology for controlling processing-induced defects mainly relies on the inspection link after the completion of processing. Once defects are found, the workpiece may have been scrapped, causing huge economic losses; the present invention changes this situation through the design of its damage suppression mode switching module; it provides clear and physically clear boundary conditions for the system's autonomous decision-making; when the index is lower than the safety threshold, the system can boldly pursue efficiency; when the index enters between the safety and critical thresholds, the optimal cutting parameter set executed by the system is a prudent choice after balancing risks; and when the index approaches or exceeds the critical threshold, the system will take the top priority emergency plan to actively suppress the occurrence of damage; this layered, closed-loop real-time intervention capability avoids the occurrence of processing-induced defects, and its direct effect is to improve the first-time processing pass rate of high-value precision castings, especially in the manufacturing of core components in the aerospace field, and significantly reduce rework and scrap costs;

[0037] Third, the overall technical solution proposed in the present invention is highly feasible and engineering robust; each key custom parameter and model in the present invention has a clear physical connotation and a definite calibration method; the various weight coefficients in the processing-induced defect activation index model, as well as the safety threshold and critical threshold on which the damage suppression mode switching module relies, the specification provides specific determination steps based on physical experiments and data statistics; similarly, the cutting process proxy model used by the cutting parameter dynamic optimization module also clarifies its technical path for establishing a mapping relationship through offline calibration experiments; this comprehensive explanation of the implementation details ensures that technical personnel in this field can reproduce and deploy the system without creative work, thereby ensuring the implementation capability of the technical solution; the four modules are tightly coupled through the two core data streams of real-time state feature vector and real-time defect activation index, forming a closed-loop system with strict logic and clear responsibilities, ensuring the stability and effectiveness of the overall operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further explained below in conjunction with the accompanying drawings and Examples:

[0039] Figure 1 It is a flow chart of a quality analysis and tracking system for CNC machine tool castings of the present invention;

[0040] Figure 2 This is a diagram of the steps for constructing the real-time state feature vector of the present invention;

[0041] Figure 3 1 is a diagram of the steps for calculating the activation index of processing-induced defects according to the present invention. DETAILED DESCRIPTION

[0042] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0043] Example 1:

[0044] See also Figure 1 ,The present invention provides a technical solution for a CNC machine tool casting quality analysis and tracking system, comprising: a multi-physics field fusion monitoring module for acquiring multivariate physical signals of the machine tool in real time and constructing a real-time state feature vector based on the multivariate physical signals;

[0045] The real-time inversion module for defect evolution is used to combine the real-time state feature vector with the preset initial defect factor of the casting and calculate the real-time defect activation index through the processing-induced defect activation index model;

[0046] The cutting parameter dynamic optimization module is used to determine the optimal cutting parameter set by solving the normalized machining utility function based on the machining-induced defect activation index model and preset process constraints;

[0047] The damage suppression mode switching module is used to compare the real-time defect activation index with the preset safety threshold and critical threshold, and based on the comparison results, combined with the optimal cutting parameter set, to determine the final processing instructions sent to the CNC system.

[0048] This embodiment provides a quality analysis and tracking system for CNC machine tool castings; the system solves the technical dilemma of the mutual constraint between processing efficiency and finished product quality caused by the inability to predict and avoid the deterioration of internal defects in real time during the finishing process of high-end castings; the architecture of this system, through the coordinated operation of its four internal modules, constitutes a complete closed-loop control logic of perception-cognition-decision-execution; the multi-physics field fusion monitoring module serves as the system's perception unit, capturing the processing status in real time; the defect evolution real-time inversion module serves as the cognitive core, interpreting the status and assessing risks; the cutting parameter dynamic optimization module acts as the decision-making brain, planning the optimal strategy; and finally, the damage suppression mode switching module serves as the execution supervision to ensure the safe issuance of instructions; this design makes the system no longer passively adapt, but proactively control, thereby transforming processing efficiency and quality control from a traditional opposing relationship to a unified goal that can be collaboratively optimized.

[0049] Example 2

[0050] Multivariate physical signals include: high-frequency acoustic emission sensor signals, three-force measuring instrument signals, high-precision accelerometer signals, and non-contact infrared thermometer signals;

[0051] refer to Figure 2 As shown in Figure 2, the steps for constructing the real-time state feature vector are as follows:

[0052] S11, extracting a root mean square value of an acoustic emission signal from a high-frequency acoustic emission sensor signal;

[0053] S12, extracting the cutting force from the three-force measuring instrument signal;

[0054] S13, extracting the kurtosis of the vibration signal from the high-precision accelerometer signal;

[0055] S14, extracting cutting temperature from the non-contact infrared thermometer signal;

[0056] S15. Combine the root mean square value of the acoustic emission signal, the cutting force, the kurtosis of the vibration signal, and the cutting temperature to construct a real-time state feature vector.

[0057] In this implementation, the function of its multi-physics field fusion monitoring module is refined; the module captures a set of carefully selected multi-physics signals through the sensor array deployed at key positions of the machine tool, which covers multiple dimensions of heat, force, vibration and microscopic activities inside the material, ensuring the comprehensiveness of the information input; after obtaining the original signal, the module constructs the real-time state feature vector through a series of clear steps ; It extracts the root mean square value of microcrack energy release from the high-frequency acoustic emission sensor signal ; Calculate the cutting force reflecting the total level of cutting load from the three-force measuring instrument signal Extracting the kurtosis of vibration signals, which are extremely sensitive to shock events, from high-precision accelerometer signals ; and obtain the cutting temperature of the tool tip-workpiece contact area from the non-contact infrared thermometer signal ; These refined key features are combined into a time-varying Normalized real-time state feature vector of changes ;

[0058] Among them, the standardization of the real-time state feature vector can adopt the minimum-maximum standardization method to convert each original feature value Linearly mapped to the interval [0, 1]; the calculation formula is: ,in and are the minimum and maximum values ​​of the feature observed in offline calibration or historical data, respectively. This standardization process can eliminate the differences in dimensions and numerical ranges between different physical signals, which is beneficial to the calculation stability of subsequent models.

[0059] The construction of this vector provides a quantitative, multi-dimensional data foundation for the subsequent analysis of the entire system, which can comprehensively depict the physical portrait of the cutting process at the current moment. Its direct benefit is to ensure the accuracy and reliability of the subsequent defect evolution inversion.

[0060] Example 3

[0061] refer to Figure 3 As shown in Figure 2, the calculation steps of the processing-induced defect activation index are as follows:

[0062] S21. Determine the thermodynamic effect term based on the cutting temperature;

[0063] S22. determining a quasi-static mechanical effect term based on the root mean square value of the acoustic emission signal;

[0064] S23. determining a dynamic impact effect term based on the kurtosis of the vibration signal and the cutting resultant force;

[0065] S24, performing weighted summation on the thermodynamic effect term and the quasi-static mechanical effect term to obtain a combined effect term;

[0066] S25. Generate a real-time defect activation index based on the combined effect term, the dynamic impact effect term, and the casting initial defect factor.

[0067] In the core of this system, the real-time inversion module of defect evolution, the above steps are processed to induce the defect activation index The model carries;

[0068] Origin and technical motivation: The original concept of this model is derived from the damage superposition principle in material damage mechanics and the energy release rate theory in fracture mechanics. The core technical motivation is to deconstruct the complex internal defect activation process driven by multi-physical field coupling in the cutting process into a linear and nonlinear combination of three key physical dimensions: thermodynamic effect, quasi-static mechanical effect, and dynamic impact effect, and to create a single calculable indicator that can quantitatively characterize the risk. The essence of this approach is to establish a mathematical probe that correlates the internal damage evolution, which cannot be directly observed, with external measurable physical signals.

[0069] The core calculation of this module follows the following formula: ;

[0070] in, is the dimensionless processing-induced defect activation index at time The instantaneous value of is the thermodynamic effect term, and the calculation formula is , is the real-time cutting temperature, is the creep initiation temperature of the material, It is the temperature at which the material undergoes a critical phase transition or a significant drop in strength. The latter two can be obtained from material handbooks or experimentally determined by differential scanning calorimetry. is the quasi-static mechanical effect term, and the calculation formula is , is the RMS value of the acoustic emission signal, It is the critical RMS value that the acoustic emission signal can reach before macroscopic fracture of the material occurs. It is derived from experimental calibration of standard specimens stretched to fracture. is the dynamic impact effect term, and the calculation formula is , is the kurtosis of the vibration signal, is the reference kurtosis under steady cutting state, the theoretical value is 3, is the cutting force, is the benchmark cutting force during steady cutting, and the latter two are derived from the statistical average of a large amount of normal machining data; They are the physical weight coefficients of thermodynamic effect, quasi-static mechanical effect and dynamic impact effect, which are obtained by back-calculating a series of test pieces through metallographic analysis with the goal of maximizing the goodness of fit and satisfying ; The initial defect factor of the casting characterizes the inherent quality of the workpiece, which is calculated based on the non-destructive testing results of the blank when it enters the factory;

[0071] The back-calculation process can be constructed as an optimization problem; through metallographic analysis and other means, an actual damage metric is quantified for each test piece i. , for example, the microcrack density per unit area; substitute the characteristic vector of the test piece during processing into the model and calculate the corresponding processing-induced defect activation index ; The goal of optimization is to find a set of weights , so that the sum of squared errors between the predicted values ​​and the actual damage metrics of all samples is minimized; the objective function J can be expressed as: ,in is the total number of trial cut pieces, For a general and physical damage metrics The associated calibration function can be a linear proportional relationship in the simplest case; this optimization problem can be solved by a nonlinear least squares method, such as the Levenberg-Marquardt algorithm;

[0072] The initial defect factor of the casting characterizes the inherent quality of the workpiece, which is calculated based on the non-destructive testing results of the blank when it enters the factory;

[0073] In the application, this module receives the real-time status feature vector from the monitoring module , and substitute its components into the above formula, combined with the preset , real-time calculation outputs a defect activation index curve The direct technical effect of this calculation process is to achieve a precise mapping from external physical representation to internal damage risk. It transforms the originally vague and qualitative problem of the quality of the processing state into a specific and continuously changing risk value, providing a quantitative basis for subsequent optimization decisions and safety control.

[0074] Example 4

[0075] The specific steps for determining the optimal cutting parameter set are:

[0076] S31. Calculating a material removal rate based on the cutting parameters to be optimized;

[0077] S32, predicting a state feature vector corresponding to the cutting parameter to be optimized using a preset cutting process proxy model;

[0078] S33, substituting the predicted state feature vector into the processing-induced defect activation index model to obtain a predicted defect activation index;

[0079] S34, combining the material removal rate and the predicted defect activation index to solve the maximum value of the normalized machining utility function to determine the optimal cutting parameter set;

[0080] The cutting process proxy model is obtained by establishing a mapping relationship from cutting parameters to state feature vectors through offline calibration experiments.

[0081] The dynamic optimization module of cutting parameters in this system, as an intelligent decision-making unit, is not a passive response, but an active planning; in order to predict the different cutting parameter sets To determine the impact of defect activation risk, this system first constructs a cutting process proxy model through offline calibration experiments; The essence of the method is to establish a feature vector from cutting parameters to predicted state by using machine learning methods such as support vector machines or neural networks. This step ensures the physical basis of the prediction; among them, the cutting parameter set Including spindle speed , feed rate and cutting depth ; The proxy model It refers to the process of cutting process proxy model, from the input cutting parameter set Mapping to obtain the predicted state feature vector The predicted state feature vector specifically includes the predicted root mean square value of the acoustic emission signal , predicted cutting force , the predicted vibration signal kurtosis , and the predicted cutting temperature ;

[0082] After having the prediction capability, the module solves a normalized processing utility function The construction of this function originates from the multi-objective optimization theory in operations research. The technical motivation is to combine the machining efficiency represented by the material removal rate MRR with the defect activation index represented by the prediction. The two essentially conflicting goals of the penalty term and the processing quality are cleverly integrated into a unified, dimensionless utility function, so that the optimization problem has a unique and solvable optimal solution.

[0083] The utility function has the form: ;

[0084] in, is the dimensionless comprehensive processing utility value; For cutting parameters Directly related material removal rate; is the reference material removal rate, which comes from the process manual or trial cutting database and is used as the benchmark for efficiency normalization; is the predicted defect activation index obtained by the surrogate model and the defect activation index model; is the efficiency-quality balance factor ( ), which is derived from the work order requirements issued by the upper-level manufacturing execution system and is used to adjust the priority weights of efficiency and quality; and are the preset safety threshold and critical threshold respectively; it is worth noting that the square form of the penalty term ensures that when the predicted defect activation index Once the safety threshold is exceeded, its negative impact on total utility will be amplified sharply at a quadratic level;

[0085] This module uses intelligent optimization algorithms such as genetic algorithms or particle swarm optimization to solve the maximum value of the function within the process constraints, thereby determining the optimal cutting parameter set. ;

[0086] Taking the particle swarm optimization algorithm as an example, as a non-limiting implementation method, its parameters can be set as: particle swarm size is 50, maximum number of iterations is 100, inertia weight is The cognitive factor decreases linearly from 0.9 to 0.4. and social factors The search space is composed of the cutting parameters preset by the process knowledge base, spindle speed , feed rate , cutting depth Defined by the upper and lower limits of

[0087] The technical effect of this process is the realization of forward-looking adaptive optimization; the system can predict the optimal parameter combination that maximizes the material removal rate while strictly controlling the risk of defect activation within a safe range before or during processing; this enables the dynamic operation of the processing process to the highest efficiency limit allowed, achieving synergy between efficiency and quality.

[0088] Example 5

[0089] The specific steps for determining processing instructions are:

[0090] If the real-time defect activation index is less than or equal to the safety threshold, the system enters the high-efficiency mode and uses the optimal cutting parameter set calculated by the cutting parameter dynamic optimization module with the main goal of maximizing machining efficiency as the machining instruction;

[0091] If the real-time defect activation index is greater than the safety threshold and less than the critical threshold, the system enters the adaptive mode and adopts the optimal cutting parameter set calculated by the cutting parameter dynamic optimization module to balance machining efficiency and risk suppression as the machining instruction;

[0092] If the real-time defect activation index is greater than or equal to the critical threshold, the preset emergency plan is used as the processing instruction;

[0093] The safety threshold is derived from a real-time defect activation index value calibrated through trial cutting experiments to ensure that the microcrack density is lower than the material allowable standard under a preset confidence level.

[0094] The source of the critical threshold is the statistical lower limit of the instantaneous peak value of the real-time defect activation index of all failed samples before failure, which is calibrated through destructive cutting experiments.

[0095] As the final execution supervisory layer of the system, the damage suppression mode switching module provides a guarantee for the stability and security of the entire system; the effectiveness of this module is highly dependent on two key thresholds and The two thresholds are not arbitrary assumptions, but are obtained through rigorous experimental calibration, ensuring the anchoring of model calculations with physical reality.

[0096] Safety threshold The determination process is to perform a series of test cutting experiments covering a wide range of parameters, followed by metallographic analysis of the test pieces, to establish the process of machining The corresponding relationship between the peak value and the subsurface microcrack density will eventually ensure that the microcrack density is lower than the material allowable standard with a 99% confidence level. Value, defined as ;

[0097] Critical threshold The determination of the failure rate is more rigorous: a set of destructive cutting tests designed to induce failure are performed, and the fracture dynamics of all samples are recorded just before the occurrence of failure events such as macro cracks or chipping. The instantaneous peak value is taken as the statistical lower limit of these peak data, such as the 10% quantile, and is defined as ; This value represents the edge of the cliff where damage is about to get out of control;

[0098] After obtaining these two anchor points, the module determines the final processing instructions sent to the CNC system based on a clear three-stage rule logic; when the real-time monitoring Less than or equal to When the system is in high efficiency mode, it fully adopts the optimization module's In pursuit of maximum production efficiency; Between and When the system enters the adaptive mode, the This is the result of the optimization module weighing the risks and efficiency, achieving a dynamic balance; and once breakthrough , the highest priority damage suppression mode is triggered, and the module will immediately reject any optimization suggestions and enforce a set of extremely conservative emergency plans, such as suspending feed and activating high-pressure cooling;

[0099] At the CNC system level, this emergency plan can be broken down into the following series of command actions: Immediately send a feed hold signal to the CNC system to pause feed, instructing the tool to rapidly lift along the Z axis or other safety coordinate axes, for example, G00Z+10.0, to free it from the workpiece; send an M code to activate or maximize the flow of high-pressure coolant to forcefully cool the cutting area; the system records all status data of the triggering event in the background and issues an audible and visual alarm, prompting the operator to intervene and inspect. This series of actions is designed to instantly halt further damage and provide safety for subsequent manual processing or replanning.

[0100] This entire set of switching logic based on physical calibration thresholds has the technical effect of giving the system clear safety boundaries and multi-level response strategies; it enables the system to boldly improve efficiency in the safe zone, prudently adapt and adjust in the risk zone, and decisively intervene before danger occurs, thereby putting proactive prevention into practice.

[0101] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A CNC machine tool casting quality analysis and tracking system, characterized in that: include: Multi-physics field fusion monitoring module, used to obtain multi-dimensional physical signals of machine tools in real time and construct real-time state feature vectors based on multi-dimensional physical signals; The real-time inversion module for defect evolution is used to combine the real-time state feature vector with the preset initial defect factor of the casting and calculate the real-time defect activation index through the processing-induced defect activation index model; The cutting parameter dynamic optimization module is used to determine the optimal cutting parameter set by solving the normalized machining utility function based on the machining-induced defect activation index model and preset process constraints; The damage suppression mode switching module is used to compare the real-time defect activation index with the preset safety threshold and critical threshold, and based on the comparison results, combined with the optimal cutting parameter set, to determine the final processing instructions sent to the CNC system.

2. A CNC machine tool casting quality analysis and tracking system according to claim 1, characterized in that: The multivariate physical signals include: high-frequency acoustic emission sensor signals, three-force measuring instrument signals, high-precision accelerometer signals and non-contact infrared thermometer signals.

3. A CNC machine tool casting quality analysis and tracking system according to claim 1, characterized in that: The specific steps for constructing the real-time state feature vector are: S11, extracting a root mean square value of an acoustic emission signal from a high-frequency acoustic emission sensor signal; S12, extracting the cutting force from the three-force measuring instrument signal; S13, extracting the kurtosis of the vibration signal from the high-precision accelerometer signal; S14, extracting cutting temperature from the non-contact infrared thermometer signal; S15. Combine the root mean square value of the acoustic emission signal, the cutting force, the kurtosis of the vibration signal, and the cutting temperature to construct a real-time state feature vector.

4. A CNC machine tool casting quality analysis and tracking system according to claim 1, characterized in that: The calculation steps of the processing-induced defect activation index are as follows: S21. Determine the thermodynamic effect term based on the cutting temperature; S22. determining a quasi-static mechanical effect term based on the root mean square value of the acoustic emission signal; S23. determining a dynamic impact effect term based on the kurtosis of the vibration signal and the cutting resultant force; S24, performing weighted summation on the thermodynamic effect term and the quasi-static mechanical effect term to obtain a combined effect term; S25. Generate a real-time defect activation index based on the combined effect term, the dynamic impact effect term, and the casting initial defect factor.

5. The CNC machine tool casting quality analysis and tracking system according to claim 1, characterized in that: The specific steps for determining the optimal cutting parameter set are: S31. Calculating a material removal rate based on the cutting parameters to be optimized; S32, predicting a state feature vector corresponding to the cutting parameter to be optimized using a preset cutting process proxy model; S33, substituting the predicted state feature vector into the processing-induced defect activation index model to obtain a predicted defect activation index; S34. Combining the material removal rate with the predicted defect activation index, the maximum value of the normalized machining utility function is solved to determine the optimal cutting parameter set.

6. A CNC machine tool casting quality analysis and tracking system according to claim 5, characterized in that: The cutting process proxy model is obtained by establishing a mapping relationship from cutting parameters to state feature vectors through offline calibration experiments.

7. The CNC machine tool casting quality analysis and tracking system according to claim 1, characterized in that: The specific steps for determining processing instructions are: If the real-time defect activation index is less than or equal to the safety threshold, the system enters the high-efficiency mode and uses the optimal cutting parameter set calculated by the cutting parameter dynamic optimization module with the main goal of maximizing machining efficiency as the machining instruction; If the real-time defect activation index is greater than the safety threshold and less than the critical threshold, the system enters the adaptive mode and adopts the optimal cutting parameter set calculated by the cutting parameter dynamic optimization module to balance machining efficiency and risk suppression as the machining instruction; If the real-time defect activation index is greater than or equal to the critical threshold, the preset emergency plan is used as the processing instruction.

8. The CNC machine tool casting quality analysis and tracking system according to claim 1, characterized in that: The source of the safety threshold is: the value of the real-time defect activation index calibrated through trial cutting experiments to ensure that the microcrack density is lower than the material allowable standard under a preset confidence level.

9. The CNC machine tool casting quality analysis and tracking system according to claim 1, characterized in that: The source of the critical threshold is the statistical lower limit of the instantaneous peak value of the real-time defect activation index of all failed samples before failure, which is calibrated through destructive cutting experiments.

Images