A time-varying chatter stability adaptive control method for gear form grinding

Abstract

The application relates to a gear forming grinding time-varying chatter stability adaptive control method, which comprises the following steps: in the gear forming grinding process, vibration acceleration signals, grinding wheel spindle speed signals, acoustic emission signals and grinding wheel spindle power signals are synchronously collected in real time; online sensing and adaptive correction are carried out based on the vibration acceleration signals, the grinding wheel spindle speed signals, the acoustic emission signals and the grinding wheel spindle power signals; a comprehensive stability index is constructed; the working state is judged based on the comprehensive stability index; when the working state meets the triggering condition, the adjustment range of the process parameters is restricted; the minimum process parameter adjustment amplitude is taken as the target, and the optimal adjustment instruction is solved based on the preset process parameter adjustment priority; and the optimal adjustment instruction is executed through a machine tool communication interface. Advantages: the method can adaptively control the gear forming grinding time-varying chatter stability according to real-time signal data, and ensures that the gear forming grinding process can be processed with high quality.

Description

Technical Field

[0001] This application relates to the field of gear precision machining technology, and in particular to an adaptive control method for time-varying chatter stability during gear forming grinding. Background Technology

[0002] Form grinding is a key final machining process for obtaining high-precision gear tooth profiles with high surface integrity. However, this process is essentially a strongly coupled, nonlinear dynamic system. The extremely high energy density in the grinding zone leads to significant temperature rise and even phase transition on the workpiece surface; the wear, breakage, and adhesion of abrasive grains on the grinding wheel cause continuous changes in their sharpness and morphology; furthermore, the dynamic characteristics of the "machine tool-fixture-grinding wheel-workpiece" system under actual machining load differ from those under no-load conditions. These factors collectively cause key process parameters (such as grinding force coefficient, process damping, and contact stiffness) and system modal parameters (frequency, damping ratio) to drift with machining time / progress, thus making the stability boundary based on linear regenerative chatter theory no longer fixed and exhibiting obvious time-varying characteristics. In practice, the "stable" parameters selected in the stability lobe diagram (SLD) drawn offline based on initial no-load parameters often trigger chatter in the later stages of machining due to boundary contraction, resulting in tooth surface chatter marks, burns, and out-of-tolerance geometric accuracy, severely restricting the consistency of machining quality and efficiency.

[0003] Current technologies suffer from two main shortcomings: First, offline stability prediction methods based on no-load experimental modal analysis and constant process parameters cannot track time-varying parameters under operating conditions, resulting in limited prediction accuracy and a susceptibility to misjudgments. Second, passive alarm and suppression strategies based on vibration amplitude exceeding thresholds typically intervene only after flutter has occurred and damage has taken place, representing a "post-hoc remedy" lacking foresight. Therefore, developing a closed-loop intelligent control method capable of online sensing of system status, rolling correction of the prediction model, and forward-looking parameter adjustment based on predictions is of great significance for ensuring stable, efficient, and high-quality processing throughout the gear forming grinding process. Summary of the Invention

[0004] Therefore, it is necessary to provide an adaptive control method for time-varying chatter stability in gear forming grinding, including:

[0005] S1: During the gear forming grinding process, the vibration acceleration signal, grinding wheel spindle speed signal, acoustic emission signal, and grinding wheel spindle power signal are collected synchronously in real time.

[0006] S2: Based on vibration acceleration signal, grinding wheel spindle speed signal, acoustic emission signal, and grinding wheel spindle power signal, online sensing and adaptive correction are performed to construct a comprehensive stability index;

[0007] S3: Based on the comprehensive stability index, the working state is determined. When the working state meets the triggering conditions, the adjustment range of the process parameters is constrained. The goal is to minimize the adjustment range of the process parameters. Based on the preset process parameter adjustment priority, the optimal adjustment command is solved. The optimal adjustment command is then sent and executed through the machine tool communication interface.

[0008] Beneficial effects: This method can adaptively control the stability of time-varying chatter during gear forming grinding based on real-time signal data, ensuring stable, efficient, and high-quality processing throughout the entire gear forming grinding process. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart of the adaptive control method for time-varying chatter stability during gear forming grinding in the embodiments of this application.

[0011] Figure 2 This is a schematic diagram of the multi-source sensor arrangement and data link in an embodiment of this application.

[0012] Figure 3 This is a logic diagram of the calculation of the comprehensive stability index and the state triggering in the embodiments of this application.

[0013] Figure 4 This is a data flow diagram that is updated on a rolling basis in the embodiments of this application.

[0014] Figure 5 This is a schematic diagram of the time-varying stability boundary and stability probability field solved by the fully discrete method in the embodiments of this application.

[0015] Figure 6 This is a flowchart of the parameter transition optimization solution based on high confidence stability region constraints in the embodiments of this application. Detailed Implementation

[0016] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application are described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0017] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0018] like Figure 1 As shown, this embodiment provides an adaptive control method for time-varying chatter stability during gear forming grinding, including:

[0019] S1: During the gear forming grinding process, the vibration acceleration signal, grinding wheel spindle speed signal, acoustic emission signal, and grinding wheel spindle power signal are collected synchronously in real time.

[0020] This embodiment is implemented on a five-axis CNC gear grinding machine. To achieve synchronous acquisition of multi-source signals, such as... Figure 2 As shown, a triaxial ICP-type accelerometer (sensitivity 100mV / g) is mounted vertically (Z-axis) above the grinding wheel spindle housing. This position is sensitive to grinding force excitation and is used to collect vibration acceleration signals. ;An acoustic emission sensor can be optionally placed near the grinding zone to collect data. Grinding wheel spindle speed signal The grinding wheel spindle power signal is read in real time by the CNC system via a fieldbus (such as EtherCAT). or current signal Feedback is obtained from the servo drive unit. All signals are synchronously acquired via a data acquisition card at a sampling frequency of 25.6kHz, and synchronized with a unified timestamp in the data acquisition card or edge control computer.

[0021] S2: Based on vibration acceleration signal, grinding wheel spindle speed signal, acoustic emission signal, and grinding wheel spindle power signal, online sensing and adaptive correction are performed to construct a comprehensive stability index.

[0022] Specifically, such as Figure 3 As shown, this step includes:

[0023] S2.1: Preprocess the vibration acceleration signal and divide it into segments according to the sliding time window to obtain the vibration response signal corresponding to each time window.

[0024] In this embodiment, the vibration acceleration signal is preprocessed and windowed, and within each time window, the working mode parameters are identified, and the vibration intensity index (VSI) and comprehensive stability index (SI) are calculated and graded triggering is determined. Specifically, this includes online processing of the acquired continuous vibration signal. The sliding time window length is set to 0.5 seconds, and the overlap rate is 50%.

[0025] For the vibration acceleration signal within the t-th time window :

[0026] 1. Calculate the root mean square value in the time domain: , This represents the root mean square value in the time domain for the t-th time window. This represents the mean function.

[0027] 2. Perform a Fast Fourier Transform (FFT) on the windowed signal to obtain its one-sided amplitude spectrum. .

[0028] 3. Operating Mode Identification: The Covariance-Driven Stochastic Subspace Identification (SSI-COV) algorithm is employed. A Hankel matrix is ​​constructed from the signals within the window, and QR decomposition and SVD decomposition are performed to obtain the system state matrix, thereby calculating the system poles. By analyzing the stability diagram, the significant pole that remains stable across adjacent model orders (frequency change <1%, damping ratio change <5%) and has the smallest damping ratio is selected; its frequency is the center frequency of the dominant flutter mode in the current window. Damping ratio is For example, identifying It fluctuates in the range of 4920-4980Hz.

[0029] S2.2: The working modal analysis method is used to identify each vibration response signal online, and the center frequency and damping ratio of the dominant flutter mode in each time window are output. For any time window, an adaptive frequency band is constructed based on the center frequency, and the amplitude characteristics of the adaptive frequency band are extracted. The root mean square value of the vibration response signal in the time domain is calculated. The vibration intensity index is constructed based on the amplitude characteristics and the root mean square value in the time domain.

[0030] Furthermore, the formula for calculating the amplitude characteristic may be:

[0031] ;

[0032] Or:

[0033] ;

[0034] Or:

[0035] ;

[0036] in, This represents the amplitude characteristics of the t-th time window. This represents the center frequency of the t-th time window. The frequency domain representation of the vibration response signal within the t-th time window. Amplitude, This indicates the half-width of the frequency band (e.g., 50Hz).

[0037] Furthermore, a vibration intensity index is constructed based on amplitude characteristics and time-domain root mean square value, including:

[0038] ;

[0039] in, This represents the vibration intensity index at the t-th time window. The vibration intensity index can adaptively track the drift of the flutter mode frequency and comprehensively reflect the vibration energy level. This represents the root mean square value in the time domain for the t-th time window. This represents the amplitude characteristics of the t-th time window. The fusion function can be a product type, a weighted sum type, or a normalized product type. In this embodiment, the normalized product type is preferred.

[0040] In this embodiment, the working mode analysis method includes a random subspace identification algorithm, a recursive SSI algorithm, or an equivalent algorithm; the dominant flutter mode is the mode most relevant to the regenerative flutter mechanism and has a relatively low damping ratio.

[0041] S2.3: Based on the mapping of the grinding wheel spindle power signal and process parameters, the thermal load characterization quantity of the grinding zone is obtained, and the grinding wheel wear state is estimated based on the acoustic emission signal, grinding wheel spindle power signal and / or vibration acceleration signal; based on the center frequency, damping ratio, thermal load characterization quantity and grinding wheel wear state of the same time window, the state vector of the corresponding time window is constructed, and the time-varying model parameters are updated in a rolling manner based on the state vector.

[0042] Furthermore, based on the mapping of the grinding wheel spindle power signal and process parameters, the thermal load characterization parameters of the grinding zone are derived, including:

[0043] The grinding energy is derived from the grinding wheel spindle power signal and the material removal rate, and the thermal load characterization of the grinding zone is obtained through a preset mapping relationship.

[0044] Alternatively, query a pre-established database or empirical model of process parameters and thermal load characterization quantities to obtain the thermal load characterization quantity of the grinding zone;

[0045] Alternatively, the grinding wheel spindle power signal and process parameters can be input into the online proxy model of the thermo-mechanical-wear coupled multiphysics simulation model to obtain the thermal load characterization of the grinding zone.

[0046] To achieve online tracking of time-varying drift of the stability boundary, this embodiment, based on thermal load characterization and wear state estimation, performs thermo-mechanical-wear coupled rolling correction on key process parameters and updates the time-varying regenerative flutter model. For example... Figure 4As shown, the center frequency obtained online Damping ratio and thermal load characterization Grinding wheel wear condition Common driving grinding force coefficient Process damping coefficient With contact stiffness The direction of the parameters is kept updated, forming a time-varying model parameter set that updates over time. .

[0047] Furthermore, the time-varying model parameters include grinding force coefficient and process damping coefficient;

[0048] The updating of the grinding force coefficient follows the physical direction of "thermal softening reduces the force coefficient, and wear and passivation increase the force coefficient." The updating method for the grinding force coefficient is as follows:

[0049] ;

[0050] ;

[0051] in, This represents the grinding force coefficient for the t-th time window. Indicates the tangential reference force coefficient, and indicates the temperature rise. Indicates the thermal softening coefficient. Indicates the wear passivation coefficient. This represents the state vector for the t-th time window. This represents the thermal load characteristic quantity for the t-th time window. Indicates the initial temperature;

[0052] The updating of the process damping coefficient follows the mechanism that "increased energy dissipation leads to increased damping, while accelerated wear leads to decreased damping." The updating method for the process damping coefficient is as follows:

[0053] ;

[0054] in, This represents the process damping coefficient at time window t. Indicates the damping of the reference process. This represents the specific grinding energy at the t-th time window. Indicates the reference specific grinding energy. Indicates the first index, This indicates the second index.

[0055] Taking the grinding of TC11 titanium alloy gears as an example:

[0056] 1. Thermal load characterization quantity estimate:

[0057] Real-time reading of instantaneous power fed back by the grinding wheel spindle servo unit (Unit: W). Calculate the current material removal rate: The equivalent grinding width For a given tooth groove that is constant (e.g., 5 mm), then the specific grinding energy... (Unit: J / mm3).

[0058] Based on preliminary dry / wet grinding experiments and infrared temperature measurement calibration, a quadratic mapping model for TC11 was established: (Unit: °C). Real-time... Substitute and you will get Alternatively, an online surrogate model trained using thermo-mechanical-wear (TFW) multiphysics simulation can be directly output. To improve online generalization capabilities under different operating conditions.

[0059] 2. Grinding wheel wear condition Fusion estimation:

[0060] Synchronously acquire acoustic emission signals and calculate within the window Define three characteristics:

[0061] The root mean square value of the acoustic emission signal: (Assuming 2.0V is the severe wear threshold)

[0062] Normalized drift of grinding wheel spindle power relative to the sharp state: , For sharp grinding wheels in the same The average power under [condition].

[0063] The normalized drift of the vibration intensity index relative to the sharp state: , This is a reference for the corresponding sharpness condition.

[0064] Linear fusion is employed.

[0065] .

[0066] The weights are based on the sensitivity to wear determined in the experiment. It is restricted to [0,1].

[0067] 3. Rolling correction of time-varying model parameters (period 0.25 seconds):

[0068] Stiffness: Assume initial... , This is a reference value. Therefore:

[0069] .

[0070] Tangential grinding force coefficient : Calibration , , .but:

[0071] .

[0072] Process damping coefficient ( ): Calibration , Given γ=0.3 and η=0.5, then:

[0073] ( ).

[0074] S2.4: Substitute the updated time-varying model parameters into the time-varying regenerative flutter model and solve the time-varying stability boundary using the fully discrete method; perform Monte Carlo sampling simulation on the center frequency and damping ratio and construct a stable probability field based on the time-varying stability boundary.

[0075] Furthermore, Monte Carlo sampling simulations were performed on the center frequency and damping ratio, and a stable probability field was constructed based on the time-varying stability boundary, including:

[0076] A random perturbation consistent with the statistical characteristics of its identification error is applied to the center frequency and damping ratio, and multiple Monte Carlo samplings are performed. The percentage of stable times of the process parameters in multiple simulations is statistically determined to obtain the stable probability field.

[0077] Perform steps S2.3-S2.4 on the corresponding process parameters to calculate the corresponding time-varying stability boundary and the spectral radius of the time-varying stability boundary; the stability criterion is that the spectral radius corresponding to the process parameter is less than 1.

[0078] Specifically, such as Figure 5 As shown, the time-varying model parameters obtained from the above rolling update are substituted into the three-degree-of-freedom time-varying regenerative chatter model, and the time-varying stability boundary at the current moment is numerically solved using the fully discrete method. Specifically, the grinding wheel spindle speed is... Perform discrete scanning within a preset range (e.g., 6000 rpm to 25000 rpm, with a step size of 100 rpm), and solve for the corresponding critical grinding depth at each discrete speed point. Thus, the stability boundary curve is obtained. Based on this, in order to quantify the prediction uncertainty introduced by the modal parameter identification error, the center frequency obtained online is... With damping ratio Apply a random perturbation consistent with its error statistics, perform Monte Carlo sampling, and statistically analyze each process parameter point. The proportion of samples that meet the stability criterion under multiple samplings forms a stable probability field. Among them, when (For example When the condition is met, the corresponding region is defined as a high-confidence stability region to provide a constraint basis for subsequent process point migration control.

[0079] To further characterize the statistical fluctuation range of the stability boundary curve itself, it can be measured at each discrete rotational speed point. Statistical analysis is performed on the critical grinding depth samples obtained from Monte Carlo sampling. Suppose N Monte Carlo samplings are performed, and the critical grinding depth sample at the i-th rotational speed point is obtained using the fully discrete method under the corresponding perturbation parameters. Then the mean boundary at that rotational speed point is: The standard deviation is: Construct the confidence band for the mean boundary: ;in This is the confidence coefficient; when a 95% confidence band is required, it is preferred. The confidence band The statistical fluctuation range used to intuitively express the "boundary itself" is a visual representation of the uncertainty at the "boundary line level," and is distinct from the method used to express the "probability of process point stability." P stable The stable probability fields are complementary, together forming a complete characterization of the prediction results and uncertainties of time-varying stability.

[0080] S2.5: Weighted summation of vibration intensity index, damping ratio, frequency drift, and stability probability field yields the comprehensive stability index; the comprehensive stability index... Normalized to the interval [0, 1], the larger the value, the higher the stability.

[0081] In this embodiment, the frequency drift is used to characterize the deviation of the center frequency obtained online relative to the reference frequency, and its expression is:

[0082] ;

[0083] in, This represents the frequency shift at time window t. This represents the center frequency of the t-th time window. This represents the reference frequency for the t-th time window.

[0084] The reference frequency f ref ( t The value can be a constant or a time-varying value, preferably determined in one of the following ways:

[0085] Baseline reference (constant): taken at the initial stage of machining or the initial stable stage after wheel dressing. f d ( t The mean of () is used as the reference frequency, i.e. .

[0086] Sliding Reference (Time-Varying, Online Recommendation): Takes data from a past period of time. f d ( t The moving average of ) is used as the reference frequency, i.e. L represents the total number of time window slides. Indicates the index of the number of times the time window has slid. Indicates the size of the time window.

[0087] Theoretical / nominal reference (constant): The system's natural frequency obtained from no-load modal experiments or simulations. f d,0 As a reference frequency, i.e. .

[0088] In the implementation of a discrete sliding time window, the above equation is equivalent to:

[0089]

[0090] Where k is the time window number.

[0091] Alternatively, to enhance sensitivity to short-term mutations, N-window drift can also be used:

[0092]

[0093] Where N is a positive integer.

[0094] Furthermore, the formula for calculating the comprehensive stability index is:

[0095] ;

[0096] ;

[0097] ;

[0098] ;

[0099] ;

[0100] ;

[0101] in, This represents the overall stability index for the t-th time window. This represents the clamp function, which restricts the variable x to the interval [0,1]. , , , These represent the weights of the vibration intensity penalty, the system damping characteristic penalty, the frequency drift penalty, and the stability probability field penalty, respectively. , , , Let represent the vibration intensity penalty, system damping characteristic penalty, frequency drift penalty, and stability probability field penalty for the t-th time window, respectively. Indicates the first vibration intensity threshold. This indicates the second vibration intensity threshold. Indicates the first damping threshold. This indicates the second damping threshold. This represents the first frequency drift threshold. This indicates the threshold for the second frequency drift. Representing process parameter points The stable probability field, This represents the grinding wheel spindle speed during the t-th time window. The critical grinding depth is represented by the t-th time window; the vibration intensity penalty term is constrained by the fact that the larger the vibration intensity index, the more unstable the system; the system damping characteristic penalty term is constrained by the fact that the smaller the damping ratio, the more unstable the system; the frequency drift penalty term is constrained by the fact that the larger the frequency drift, the more unstable the system; and the stability probability field penalty term is constrained by the fact that the lower the stability probability, the more unstable the system.

[0102] In this embodiment, it also includes a vibration intensity index. Comprehensive stability index Stability probability In addition to the material removal mechanism, a mapping relationship or rule base is established to identify the dominant removal mode of different workpiece materials under different process conditions online, and output the corresponding damage risk level and differentiated process adjustment suggestions.

[0103] S3: Based on the comprehensive stability index, the working state is determined. When the working state meets the triggering conditions, the adjustment range of the process parameters is constrained. The goal is to minimize the adjustment range of the process parameters. Based on the preset process parameter adjustment priority, the optimal adjustment command is solved. The optimal adjustment command is then sent and executed through the machine tool communication interface.

[0104] Specifically, the steps include:

[0105] The comprehensive stability index is compared with a preset threshold to determine the operating status as stable, warning, or unstable; a first index threshold is set. Second exponential threshold The working state is divided into stable ( ), early warning ( ) and instability ).

[0106] When the working state meets the triggering condition, the adjustment range of the constrained process parameters is within the high-confidence stability region. The triggering condition is that the working state is in a warning or unstable state for M consecutive time windows. The high-confidence stability region is the stable probability field corresponding to the process parameter that is greater than the stability threshold. With the goal of minimizing the adjustment range of the process parameters, and based on the preset process parameter adjustment priority, the optimal adjustment command is solved. The optimal adjustment command is then sent and executed through the machine tool communication interface.

[0107] In this embodiment, the expression for the objective function is:

[0108] ;

[0109] in, This indicates the adjusted linear velocity of the grinding wheel. This indicates the adjusted grinding depth. This indicates the adjusted feed rate. This indicates the amount of adjustment for the grinding wheel's linear speed. Indicates the grinding depth adjustment amount. Indicates the amount of feed rate adjustment. Indicates the linear velocity of the grinding wheel. Indicates the depth of grinding. Indicates the feed rate.

[0110] In this embodiment, M is a positive integer that can be adaptively adjusted according to the on-site noise level or processing stage to reduce the false trigger rate; for example, if the comprehensive stability index of three consecutive time windows is less than the first index threshold, the working state is considered to be continuously warning; if the comprehensive stability index of two consecutive time windows is less than the second index threshold, the working state is considered to be continuously unstable.

[0111] In this embodiment, the process parameters include grinding wheel linear speed, grinding depth, and feed rate. Increasing the grinding wheel linear speed has the highest adjustment priority, decreasing the grinding depth has a medium adjustment priority, and decreasing the feed rate has the lowest adjustment priority. Specifically: First, try increasing the grinding wheel linear speed to expand the stability region; if optimization fails or the effect is insufficient, then decrease the grinding depth; if the stability region constraint still cannot be met, then decrease the feed rate; when the estimated grinding wheel wear state... If the safety threshold is exceeded, or the cost of reverting to parameter adjustments is too high, a grinding wheel dressing or safety shutdown command will be triggered.

[0112] like Figure 6 As shown, the parameter transition optimization solution process based on high-confidence stability region constraints includes:

[0113] When the comprehensive stability index When the warning / instability trigger condition is met, the migration optimization control is initiated. The constraint is "the adjusted process point falls into the high-confidence stability region". A constrained optimization problem is constructed with the objective of minimizing the parameter adjustment range, and the grinding wheel linear velocity is adjusted sequentially according to a preset control priority. Grinding depth With feed Adjustments are made; when wear conditions are present. If the limit is exceeded or optimization fails, a grinding wheel dressing or safe shutdown suggestion will be triggered.

[0114] Assuming the current parameters are: , , The system has been in the warning zone for three consecutive windows. This triggers migration optimization.

[0115] Construct the following constrained optimization problem:

[0116] Objective function: .

[0117] Constraints:

[0118] ;in, , Indicates the outer diameter of the grinding wheel;

[0119] ;

[0120] ;

[0121] .

[0122] Optimize variables: , , .

[0123] An online solution is provided using a Sequential Quadratic Programming (SQP) solver. Following a priority system, the solver will attempt to improve performance first. Assume the optimal solution is: , , The recommended adjustment is to increase the linear speed to 32 m / s and slightly reduce the grinding depth to 0.018 mm. This instruction is written to the CNC system via the EtherCAT bus, and the system automatically updates the corresponding S-instruction (spindle speed) and axial grinding depth in the NC code, completing the online adjustment. After adjustment, the system detected that the SI value rose back to 0.82, returning to the stable range.

[0124] A knowledge base can be established. For example, if the material is a nickel-based superalloy and the VSI consistently exceeds 2.5, a message can be displayed indicating "high risk of brittle fracture; it is recommended to check for a white layer on the surface." If the material is a titanium alloy and the VSI gradually increases... A significant increase indicates "increased adhesive wear, pay attention to the depth of the α-case layer." This information can assist process engineers in conducting in-depth diagnosis and decision-making.

[0125] In summary, this invention constructs a complete perception-decision-control closed loop through the above-described process sequence. It utilizes online vibration signals to capture the system's dynamic characteristics and process state changes in real time, continuously updates a high-fidelity prediction model by fusing thermal, force, and wear information, and performs forward-looking intelligent control based on a probabilistic stability domain. Ultimately, it achieves effective suppression of time-varying chatter in gear forming grinding and autonomous assurance of stability throughout the entire process, significantly improving the reliability and quality consistency of the machining.

[0126] The adaptive control method for time-varying chatter stability in gear forming grinding provided in this embodiment has the following beneficial effects:

[0127] 1. High prediction accuracy and close fit to working conditions: The system dynamic parameters under actual load are obtained online through working modal analysis, and the process parameters are rolled over by combining the thermo-mechanical-wear state, so that the stability prediction model can dynamically track the time-varying evolution in actual processing, which significantly improves the prediction accuracy and reliability.

[0128] 2. The evaluation mechanism is sensitive and highly interpretable: The proposed VSI and SI indicators have clear physical meanings and are easy to calculate. They can not only sensitively capture changes in process status, but also facilitate engineers to understand and set thresholds, forming an effective online "health" evaluation and hierarchical early warning system.

[0129] 3. Visualized and forward-looking decision-making basis: The constructed stable probability field and confidence stability domain transform the abstract stability boundary into an intuitive "risk map", providing a clear and quantitative decision-making basis for prediction-based feedforward decision control, realizing the transformation from "passive response" to "active defense".

[0130] 4. Intelligent control strategy with minimal disturbance: Based on the minimum adjustment backtracking strategy of optimization solution, the system minimizes disturbance to the predetermined process plan while ensuring stability, thus maintaining processing efficiency. At the same time, the preset priority conforms to the process mechanism, ensuring the engineering practicality of the control.

[0131] 5. Strong System Integration and Easy Integration: This invention provides a complete technical closed loop from perception, analysis, prediction to execution. Each module has a clear function and can be embedded into existing CNC systems or standalone industrial PCs via software, facilitating integration and application in industrial settings. Furthermore, the system can consist of an industrial PC / edge control computer, a data acquisition card, and a CNC interface. Each functional module is implemented in software and interacts with data via shared memory or message queues. The corresponding computer-readable storage medium stores the program instructions, model parameters, and threshold configuration files used to implement the above steps, supporting rapid deployment and version updates.

[0132] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0133] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An adaptive control method for time-varying chatter stability in gear forming grinding, characterized in that, include: S1: During the gear forming grinding process, the vibration acceleration signal, grinding wheel spindle speed signal, acoustic emission signal, and grinding wheel spindle power signal are collected synchronously in real time. S2: Based on vibration acceleration signals, grinding wheel spindle speed signals, acoustic emission signals, and grinding wheel spindle power signals, online sensing and adaptive correction are performed to construct a comprehensive stability index, including: S2.1: Preprocess the vibration acceleration signal and divide it into segments according to the sliding time window to obtain the vibration response signal corresponding to each time window; S2.2: The working modal analysis method is used to identify each vibration response signal online, and the center frequency and damping ratio of the dominant flutter mode in each time window are output. For any time window, an adaptive frequency band is constructed based on the center frequency, and the amplitude characteristics of the adaptive frequency band are extracted. The root mean square value of the vibration response signal in the time domain is calculated. The vibration intensity index is constructed based on the amplitude characteristics and the root mean square value in the time domain. S2.3: Based on the mapping of the grinding wheel spindle power signal and process parameters, the thermal load characterization quantity of the grinding zone is obtained, and the grinding wheel wear state is estimated based on the acoustic emission signal, grinding wheel spindle power signal and / or vibration acceleration signal; based on the center frequency, damping ratio, thermal load characterization quantity and grinding wheel wear state of the same time window, the state vector of the corresponding time window is constructed, and the time-varying model parameters are updated in a rolling manner based on the state vector; S2.4: Substitute the updated time-varying model parameters into the time-varying regenerative flutter model and solve the time-varying stability boundary using the fully discrete method; perform Monte Carlo sampling simulation on the center frequency and damping ratio and construct a stability probability field based on the time-varying stability boundary; S2.5: Weighted summation of vibration intensity index, damping ratio, frequency drift, and stability probability field yields the comprehensive stability index; S3: Based on the comprehensive stability index, the working state is determined. When the working state meets the triggering conditions, the adjustment range of the process parameters is constrained. The goal is to minimize the adjustment range of the process parameters. Based on the preset process parameter adjustment priority, the optimal adjustment command is solved. The optimal adjustment command is then sent and executed through the machine tool communication interface.

2. The adaptive control method for time-varying chatter stability in gear forming grinding according to claim 1, characterized in that, The formula for calculating the amplitude characteristic may be: ； Or: ； Or: ； in, This represents the amplitude characteristics of the t-th time window. This represents the center frequency of the t-th time window. The frequency domain representation of the vibration response signal within the t-th time window. Amplitude, It indicates the half-width of the frequency band.

3. The adaptive control method for time-varying chatter stability in gear forming grinding according to claim 2, characterized in that, Vibration intensity indices are constructed based on amplitude characteristics and root mean square values ​​in the time domain, including: ； in, This represents the vibration intensity index for the t-th time window. This represents the root mean square value in the time domain for the t-th time window. This represents the amplitude characteristics of the t-th time window. This represents the fusion function.

4. The adaptive control method for time-varying chatter stability in gear forming grinding according to claim 2, characterized in that, Working mode analysis methods include random subspace identification algorithms or recursive SSI algorithms.

5. The adaptive control method for time-varying chatter stability in gear forming grinding according to claim 2, characterized in that, The thermal load characterization parameters of the grinding zone are mapped from the grinding wheel spindle power signal and process parameters, including: The grinding energy is derived from the grinding wheel spindle power signal and the material removal rate, and the thermal load characterization of the grinding zone is obtained through a preset mapping relationship. Alternatively, query a pre-established database or empirical model of process parameters and thermal load characterization quantities to obtain the thermal load characterization quantity of the grinding zone; Alternatively, the grinding wheel spindle power signal and process parameters can be input into the online proxy model of the thermo-mechanical-wear coupled multiphysics simulation model to obtain the thermal load characterization of the grinding zone.

6. The adaptive control method for time-varying chatter stability in gear forming grinding according to claim 2, characterized in that, The time-varying model parameters include grinding force coefficient and process damping coefficient; The grinding force coefficient is updated as follows: ； ； in, This represents the grinding force coefficient at time window t. Indicates the tangential reference force coefficient, and indicates the temperature rise. Indicates the thermal softening coefficient. Indicates the wear passivation coefficient. This represents the state vector for the t-th time window. This represents the thermal load characteristic quantity for the t-th time window. Indicates the initial temperature; The update method for the process damping coefficient is as follows: ； in, This represents the process damping coefficient at time window t. Indicates the damping of the reference process. This represents the specific grinding energy at the t-th time window. Indicates the reference specific grinding energy. Indicates the first index, This indicates the second index.

7. The adaptive control method for time-varying chatter stability in gear forming grinding according to claim 2, characterized in that, Monte Carlo sampling simulations were performed on the center frequency and damping ratio, and a stable probability field was constructed based on the time-varying stability boundary, including: Random perturbations are applied to the center frequency and damping ratio, and multiple Monte Carlo samplings are performed. The percentage of stable times of the process parameters in multiple simulations is statistically analyzed to obtain the stable probability field. Perform steps S2.3-S2.4 on the corresponding process parameters to calculate the corresponding time-varying stability boundary and the spectral radius of the time-varying stability boundary; the stability criterion is that the spectral radius corresponding to the process parameter is less than 1.

8. The adaptive control method for time-varying chatter stability in gear forming grinding according to claim 2, characterized in that, S3 include: The comprehensive stability index is compared with a preset threshold to determine whether the working status is stable, under warning, or unstable. When the working state meets the triggering condition, the adjustment range of the constrained process parameters is within the high confidence stability region. The triggering condition is that the working state is continuously in a warning or unstable state. The high confidence stability region is the stable probability field corresponding to the process parameter that is greater than the stability threshold. With the goal of minimizing the adjustment range of the process parameters, and based on the preset process parameter adjustment priority, the optimal adjustment command is solved. The optimal adjustment command is then sent and executed through the machine tool communication interface.

9. The adaptive control method for time-varying chatter stability in gear forming grinding according to claim 1, characterized in that, The process parameters include grinding wheel linear speed, grinding depth, and feed rate; increasing the grinding wheel linear speed has the highest adjustment priority, decreasing the grinding depth has a medium adjustment priority, and decreasing the feed rate has the lowest adjustment priority.

Images