Heavy truck transmission system motor shell fine machining method, equipment and device

Abstract

The application relates to the technical field of fine machining of motor housings, in particular to a fine machining method, equipment and device for a heavy truck transmission system motor housing, which comprises the following steps: obtaining a plurality of modal inherent frequency bands of a tool tip point of a machine tool; in the process of fine machining of the motor housing, real-time acquisition of a sound pressure signal and a rotating speed signal of a machine tool main shaft; presetting an update cycle of a rotating speed disturbance parameter; for each cycle, decomposing the sound pressure signal into signal components, screening out each vibration candidate signal component suspected of containing vibration information, identifying background noise components in each vibration candidate signal component and filtering out the background noise components, reconstructing each vibration candidate signal component, and recombining to obtain a target vibration frequency band signal; evaluating the vibration incubation in the current cycle, and updating the rotating speed disturbance parameter in the next cycle. The application aims to improve the vibration suppression effect of fine machining of the motor housing, and effectively guarantee the fine machining precision of the motor housing.

Description

Technical Field

[0001] This application relates to the field of precision machining technology for motor housings, specifically to precision machining methods, equipment, and apparatus for motor housings in heavy-duty truck transmission systems. Background Technology

[0002] The transmission system of heavy-duty trucks is a core component of new energy heavy-duty trucks, and the processing quality of its parts directly affects the vehicle's power transmission efficiency, reliability, and service life. As a key component of the heavy-duty truck transmission system, the processing quality of the motor housing directly affects the motor's heat dissipation performance and the mechanical support of the transmission system.

[0003] However, motor housings are typically thin-walled components with low rigidity, making them prone to chatter during finishing machining, which severely affects machining accuracy. Existing methods mostly focus on monitoring and suppressing chatter during its incubation stage.

[0004] Existing technologies determine the target chatter frequency band through wavelet packet decomposition and the first-order natural frequency of the tool system, and realize chatter incubation characteristic early warning based on the wavelet packet energy of the target chatter frequency band. However, the target chatter frequency band obtained through the first-order natural frequency of the tool system is usually single and fixed, while the main chatter frequency of the chatter frequency band in the chatter monitoring signal is not unique under certain turning conditions. Furthermore, during the chatter incubation stage, the weak chatter information in the chatter monitoring signal is more easily interfered with by background noise. This results in not only information loss in the wavelet packet energy of the target chatter frequency band, but also the introduction of excessive background noise signal energy, leading to missed alarms or false warnings. Consequently, it affects the chatter suppression effect in the precision turning of the motor housing of the heavy truck transmission system, making it difficult to effectively guarantee the precision turning accuracy of the motor housing. Summary of the Invention

[0005] In view of the above, it is necessary to provide a method, equipment, and apparatus for precision machining of motor housings in heavy-duty truck transmission systems. Compared with traditional precision machining methods for motor housings in heavy-duty truck transmission systems, this method effectively ensures the precision machining accuracy of the motor housing by improving the chatter suppression effect during precision machining. In a first aspect, embodiments of this application provide a method for precision machining of a motor housing in a heavy-duty truck transmission system, the method comprising the following steps: The inherent frequency bands of multiple modes of the machine tool tip are obtained, and sound pressure signals and machine tool spindle speed signals are collected in real time during the precision turning of the motor housing. The update period for the rotational speed disturbance parameters for flutter suppression is preset. For each period, a mode decomposition algorithm is used to decompose the sound pressure signal into signal components. By analyzing the overlap between the signal components and the inherent frequency bands of the modes, candidate flutter signal components suspected of containing flutter information are selected. Wavelet packet decomposition algorithm is used to decompose each candidate flutter signal component. The signal is reconstructed using the wavelet packet coefficients of each node. The harmonics of the main frequency of the rotational speed signal are compared with the main frequency of the reconstructed signal of each node to obtain the signal synchronization deviation of each node. The signal energy uniformity of each node is obtained by analyzing the data dispersion in the wavelet packet coefficients of each node. Based on the signal synchronization deviation and the signal energy uniformity, a pre-trained background noise detection model is used to identify and filter out the background noise components in each candidate flutter signal component, thereby reconstructing each candidate flutter signal component and then recombining them to obtain the target flutter frequency band signal. The flutter incubation status in the current cycle is assessed by using the energy kurtosis index of the target flutter frequency band signal, and then the rotational speed perturbation parameters in the next cycle are updated.

[0006] In one embodiment, the screening process for the flutter candidate signal components is as follows: Extract the half-power bandwidth of the maximum spectral peak in the spectrum of each signal component; If there is an intersection between the half-power bandwidth of each signal component and the inherent frequency band of any mode, then each signal component is regarded as a candidate flutter signal component.

[0007] In one embodiment, the process of obtaining the signal synchronization deviation is as follows: The minimum difference between the main frequency of the reconstructed signal of each node and the harmonic of the main frequency of the rotational speed signal; The signal synchronization deviation is directly proportional to the minimum gap and inversely proportional to the main frequency of the rotation speed signal.

[0008] In one embodiment, the signal synchronization deviation is the ratio of the minimum difference to the main frequency of the rotational speed signal.

[0009] In one embodiment, the signal energy uniformity is the reciprocal of the data dispersion in the wavelet packet coefficients of each node.

[0010] In one embodiment, the step of employing a pre-trained background noise detection model to identify background noise components in each flutter candidate signal component includes: The training process of the background noise detection model is as follows: acquire historical sound pressure signals containing flutter gestation characteristics, combine the signal synchronization deviation and signal energy uniformity of each node to form the background noise detection vector of each node, extract the signal-to-noise ratio of the reconstructed signal of each node of the historical sound pressure signal, set the category labels of the background noise detection vectors of each node with a signal-to-noise ratio lower than or greater than or equal to the preset threshold value to the first preset value and the second preset value respectively, and train the background noise detection model using the background noise detection vectors and category labels of all nodes of all historical sound pressure signals. Based on the background noise detection vector of each node of each flutter candidate signal component in the current period, the background noise detection model is used to obtain the category label of the background noise detection vector of each node, and the reconstructed signal of the node where the background noise detection vector with the category label of the first preset value is located is taken as the background noise component.

[0011] In one embodiment, the method for reconstructing each flutter candidate signal component is as follows: All data in the wavelet packet coefficients of the nodes where the background noise detection vectors with the first preset value are located are assigned to 0. The signal is then reconstructed based on all reassigned wavelet packet coefficients and unassigned wavelet packet coefficients of each flutter candidate signal component.

[0012] In one embodiment, the assessment process for the flutter gestation is as follows: If the energy kurtosis index of the target flutter frequency band signal is greater than the preset threshold, it is determined that flutter has begun to develop in the current cycle.

[0013] Secondly, embodiments of this application also provide precision machining equipment for the motor housing of a heavy-duty truck transmission system, the equipment comprising: The signal acquisition module is used to acquire multiple modal inherent frequency bands of the machine tool cutting tool tip. During the precision turning of the motor housing, it acquires sound pressure signals and machine tool spindle speed signals in real time. The background noise filtering module is used to preset the update period of the rotational speed disturbance parameters for flutter suppression. For each period, a mode decomposition algorithm is used to decompose the sound pressure signal into signal components. By analyzing the overlap between the signal components and the inherent frequency bands of the modes, candidate flutter signal components suspected of containing flutter information are selected. A wavelet packet decomposition algorithm is used to decompose each candidate flutter signal component. The signal is reconstructed using the wavelet packet coefficients of each node. The harmonics of the main frequency of the rotational speed signal are compared with the main frequency of the reconstructed signal of each node to obtain the signal synchronization deviation of each node. The signal energy uniformity of each node is obtained by analyzing the data dispersion in the wavelet packet coefficients of each node. Based on the signal synchronization deviation and the signal energy uniformity, a pre-trained background noise detection model is used to identify and filter the background noise components in each candidate flutter signal component, and then reconstruct each candidate flutter signal component to obtain the target flutter frequency band signal. The flutter incubation monitoring and flutter suppression module is used to assess the flutter incubation status in the current cycle by using the energy kurtosis index of the target flutter frequency band signal, and then update the rotational speed disturbance parameters in the next cycle.

[0014] Thirdly, embodiments of this application also provide a precision machining apparatus for a motor housing of a heavy-duty truck transmission system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described precision machining methods for the motor housing of a heavy-duty truck transmission system.

[0015] This application has at least the following beneficial effects: This application acquires multiple modal inherent frequency bands of the machine tool tip and filters chatter candidate signal components based on the overlap between signal components and modal inherent frequency bands. This enables comprehensive capture of chatter information distributed across multiple modes, significantly improving the completeness and accuracy of chatter monitoring and effectively avoiding information loss due to incomplete frequency band coverage. Furthermore, by calculating the signal synchronization deviation and signal energy uniformity, the essential differences between background noise and flutter signals in synchronization characteristics and energy distribution uniformity are fully utilized to achieve accurate identification of background noise components; combined with a pre-trained background noise detection model, background noise components in flutter candidate signal components are filtered out, significantly reducing the interference of background noise on the monitoring of flutter gestation characteristics. Furthermore, the target flutter frequency band signal is obtained by reconstructing the candidate flutter signal components after filtering out background noise. The flutter incubation status is evaluated by the energy kurtosis index of the target flutter frequency band signal and the speed disturbance parameters are updated. This achieves timely and accurate detection and effective suppression of flutter, thereby improving the flutter suppression effect of precision machining of the motor housing and effectively ensuring the precision machining accuracy of the motor housing. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the steps of a precision machining method for a motor housing of a heavy-duty truck transmission system according to an embodiment of this application; Figure 2 This is a schematic diagram of the process for acquiring signals in the target flutter frequency band. Detailed Implementation

[0017] The following description, in conjunction with the accompanying drawings, details the specific scheme of the precision machining method, equipment, and apparatus for the motor housing of the heavy-duty truck transmission system provided in this application.

[0018] Please see Figure 1 The diagram illustrates a process flow of a precision machining method for a motor housing of a heavy-duty truck transmission system according to an embodiment of this application. The method includes the following steps: Step 1: Obtain the inherent frequency bands of multiple modes at the tip of the machine tool. During the precision turning of the motor housing, collect the sound pressure signal and the rotational speed signal of the machine tool spindle in real time.

[0019] Before precision turning, a modal impact test is performed on the tip of the machine tool to obtain all the modal natural frequencies of the tool tip. Then, the modal natural frequency band is obtained with each obtained modal natural frequency as the frequency band center, requiring that there is no overlap between the modal natural frequency bands. The modal impact test is a known technique and will not be described in detail in this application.

[0020] In this embodiment, the bandwidth of the modal intrinsic frequency band is set to 100Hz. The bandwidth of the modal intrinsic frequency band is preset by the user and can be set by the implementer according to the actual situation. This application does not impose any special restrictions.

[0021] During the precision turning of the motor housing workpiece, a sound pressure sensor installed near the machine tool tool post is used to collect sound pressure signals, and a speed sensor is used to collect the speed signals of the machine tool spindle. Furthermore, in order to eliminate the influence of dimensions on the collected signals, the amplitude of the collected sound pressure signal and the amplitude of the speed signal are normalized respectively.

[0022] In this embodiment, the straight-line distance between the sound pressure sensor and the tip of the machine tool is 50mm. Here, 50mm is only one embodiment of this application, and implementers can set it according to the actual situation. This application does not impose any special restrictions.

[0023] In this embodiment, the acquisition frequency of both the sound pressure sensor and the rotation speed sensor is 10kHz. The acquisition frequency value is preset by the user and can be set by the implementer according to the actual situation. This application does not impose any special restrictions.

[0024] In this embodiment, the Min-Max normalization method is used to normalize the amplitude of the sound pressure signal and the amplitude of the rotation speed signal. The Min-Max normalization method is a well-known technology and will not be described in detail here.

[0025] Step 2: Preset the update cycle of the speed disturbance parameter for flutter suppression. For each cycle, screen out each flutter candidate signal component that is suspected to contain flutter information; identify and filter out the background noise component in each flutter candidate signal component, then reconstruct each flutter candidate signal component, and then reassemble to obtain the target flutter frequency band signal.

[0026] Step 2.1: Preset the update period of the rotational speed disturbance parameter for flutter suppression. For each period, use the mode decomposition algorithm to decompose the sound pressure signal into each signal component. By the overlap between the signal component and the mode's inherent frequency band, select each flutter candidate signal component that is suspected to contain flutter information.

[0027] Generally, when chatter occurs during turning, the chatter frequency is usually distributed near a certain modal natural frequency of the machine tool tip. Therefore, in order to effectively utilize all the chatter information in the collected sound pressure signal, to more accurately monitor whether chatter is about to occur during the precision turning of the motor housing, and then update the speed disturbance parameters used to suppress chatter based on the chatter incubation status, the following processing is performed.

[0028] The update cycle of the preset chatter suppression speed disturbance parameters must be longer than the time required for the machine tool spindle to rotate one revolution.

[0029] In this embodiment, the length of the period is 1 second. The length of the period is preset by the user and can be set by the implementer according to the actual situation. This application does not impose any special restrictions.

[0030] Taking the i-th cycle as an example, the sound pressure signal is decomposed into signal components using a mode decomposition algorithm. Based on the overlap between the signal components and the inherent frequency bands of the modes, candidate flutter signal components suspected of containing flutter information are selected. Specifically: The sound pressure signal is decomposed into signal components using a mode decomposition algorithm to characterize the different signal components in the sound pressure signal within the i-th period. The number of signal components needs to be greater than the number of intrinsic frequency bands of all modes obtained. The half-power bandwidth of the maximum spectral peak in the spectrum of each signal component is extracted and used as the main frequency range of each signal component to characterize the frequency band where the signal energy is most concentrated in each signal component. If there is an intersection between the main frequency range of each signal component and the inherent frequency band of any mode, each signal component is used as a candidate flutter signal component to characterize the signal component that is suspected to contain flutter information.

[0031] In this embodiment, a variational mode decomposition algorithm is used to decompose the sound pressure signal into individual intrinsic mode function components. Each intrinsic mode function component is a signal component, and the number of signal components is twice the number of intrinsic frequency bands of all modes. The number of signal components can be selected by the implementer based on the decomposition effect. The variational mode decomposition algorithm is a well-known technology and will not be described in detail in this application. As other implementation methods, based on the ability to decompose the sound pressure signal into individual signal components, the implementer can use other existing feasible technologies, such as empirical mode decomposition algorithms, eigenmode decomposition algorithms, etc. This application does not impose any special restrictions.

[0032] In this embodiment, the Fast Fourier Transform is used to obtain the spectrum of each signal component. The Fast Fourier Transform is a well-known technology and will not be described in detail here. As other implementation methods, based on the ability to obtain the spectrum of the signal components, the implementer may use other existing feasible technologies. This application does not impose any special restrictions.

[0033] Step 2.2: Decompose each flutter candidate signal component using the wavelet packet decomposition algorithm. Reconstruct the signal using the wavelet packet coefficients of each node obtained by decomposition. Compare the harmonic of the main frequency of the rotation speed signal with the main frequency of the reconstructed signal of each node to obtain the signal synchronization deviation of each node. Obtain the signal energy uniformity of each node by using the data dispersion in the wavelet packet coefficients of each node.

[0034] When monitoring chatter initiation during turning operations—specifically, the critical state between stable cutting and full chatter—the background noise in the acquired sound pressure level (SPL) signal mainly originates from electronic and thermal noise generated by the sensor itself, as well as mechanical noise generated by the machine tool operating under stable cutting conditions. Since the chatter signal characteristics are relatively weak during the chatter initiation stage, the acquired SPL signal is easily interfered with by background noise. Therefore, to reduce the interference of background noise components in the acquired SPL signal on subsequent chatter initiation characteristic monitoring, the following processing is performed.

[0035] Under normal circumstances, during turning operations, when the machine tool is in a stable cutting state and there is no chatter, the main frequency components of the sound pressure signal are highly consistent with the spindle speed frequency and its integer multiples, exhibiting significant synchronization characteristics. When chatter occurs, a high-amplitude spike frequency component independent of the spindle speed frequency will appear in the spectrum of the sound pressure signal. At this time, the sound pressure signal is no longer synchronized with the spindle speed.

[0036] Based on the above analysis, taking flutter candidate signal component a as an example, wavelet packet decomposition algorithm is used to decompose flutter candidate signal component a. The wavelet packet coefficients of each node obtained by decomposition are used to reconstruct the signal, which is used to characterize the signal component corresponding to each node in flutter candidate signal component a. By comparing the harmonic of the main frequency of the rotation speed signal with the main frequency of the reconstructed signal of each node, the signal synchronization deviation of each node is obtained, specifically: The minimum difference between the main frequency of the reconstructed signal of each node and the harmonic of the main frequency of the rotational speed signal is used to evaluate the degree of asynchrony between the signal component corresponding to each node in the chatter candidate signal component a and the rotational speed signal of the spindle. The ratio of the minimum difference to the main frequency of the rotational speed signal is used as the signal synchronization deviation of each node. The extraction of the main frequency is a well-known technique and will not be elaborated upon in this application.

[0037] In this embodiment, the wavelet packet decomposition algorithm is used to perform three-level wavelet packet decomposition on the flutter candidate signal component a to obtain the wavelet packet coefficients of the eight nodes of the flutter candidate signal component a. Each wavelet packet coefficient is a one-dimensional vector containing multiple data points, which is used to evaluate the energy distribution of the eight different frequency sub-bands in the flutter candidate signal component a. The wavelet packet decomposition algorithm is a well-known technology and will not be described in detail in this application.

[0038] In this embodiment, the expression for the signal synchronization deviation of each node is: In the formula, R represents the signal synchronization deviation of a single node; , These represent the maximum and minimum values ​​of the reconstructed signal's main frequency and the main frequency of the rotational speed signal for a single node, respectively; n represents... and The ratio is rounded to the nearest integer; f represents the main frequency of the rotational speed signal; This indicates the absolute value operation.

[0039] Furthermore, sensor noise includes electronic noise and thermal noise, and its generation mechanism is random, resulting in the sensor noise in the acquired sound pressure signal being randomly distributed throughout the signal. Therefore, after wavelet packet decomposition of the chatter candidate signal components, the noise energy will be evenly distributed to each wavelet packet coefficient, thus exhibiting the characteristic of relatively uniform energy distribution in each frequency sub-band. However, in turning, when machine tool chatter begins to develop, the energy of the machine tool vibration will begin to concentrate in a specific narrow frequency band, resulting in a significant reduction in the uniformity of energy distribution in each frequency sub-band.

[0040] Based on the above analysis, the signal energy uniformity of each node is obtained through the data dispersion in the wavelet packet coefficients of each node. This uniformity is used to evaluate the degree of uniformity of signal energy distribution in the signal frequency sub-band of the corresponding signal component in the flutter candidate signal component a of each node. Specifically: The reciprocal of the data dispersion in the wavelet packet coefficients of each node is used as the signal energy uniformity of each node.

[0041] It should be added that: in the process of calculating the uniformity of signal energy, if there is a case where the data dispersion is 0, the data dispersion should be mapped to a positive number first, and then subsequent calculations should be performed. There are many methods to map the data dispersion to a positive number, and the implementer can choose an existing feasible method according to the actual situation. In this embodiment, the purpose of mapping the data dispersion to a positive number is achieved by calculating the sum of the data dispersion and a preset value greater than 0. In this embodiment, the preset value greater than 0 is 0.01. The preset value greater than 0 can be flexibly set by the implementer according to the actual situation, and this application does not limit it.

[0042] In this embodiment, data dispersion refers to the coefficient of variation of the data. As other implementation methods, based on the ability to measure the unevenness of data distribution, implementers may adopt other existing feasible technologies, such as information entropy, etc. This application does not impose any special restrictions.

[0043] It should be noted that the greater the calculated signal energy uniformity, the more uniform the signal energy distribution is in the signal frequency subband of the corresponding signal component in the flutter candidate signal component a of each node.

[0044] Step 2.3: Based on the signal synchronization deviation and the signal energy uniformity, a pre-trained background noise detection model is used to identify and filter out the background noise components in each flutter candidate signal component, and then reconstruct each flutter candidate signal component to obtain the target flutter frequency band signal.

[0045] Furthermore, based on the signal synchronization deviation and the signal energy uniformity, a pre-trained background noise detection model is used to identify the background noise components in each flutter candidate signal component, specifically: The signal synchronization deviation and signal energy uniformity of each node are combined to form the background noise detection vector of each node. This vector is used to evaluate whether the signal component corresponding to each node in the flutter candidate signal component a is the background noise component generated during the precision turning of the motor housing workpiece to be processed. This effectively filters out the background noise component in all signal components containing flutter information in the sound pressure signal. Multiple historical sound pressure signals with the same format and flutter incubation characteristics as those in the i-th period are obtained from the historical database of precision machining of motor housing workpieces. These signals have the same length and dimensionless processing method. Using the same acquisition method as the background noise detection vector for each node in the i-th period, the background noise detection vectors for each node of each historical sound pressure signal are obtained. The signal-to-noise ratio (SNR) of the reconstructed signal for each node of the historical sound pressure signal is extracted. The category label of the background noise detection vector for each node with an SNR lower than a preset threshold is set to a first preset value, and the category label of the background noise detection vector for the remaining nodes is set to a second preset value. A machine learning classification model is trained using the background noise detection vectors and category labels of all nodes of all historical sound pressure signals to obtain the background noise detection model. Specifically, the SNR extraction process is as follows: while acquiring historical sound pressure signals, the cutting force signal from the machine tool force gauge is simultaneously acquired. The true flutter dominant frequency is calibrated based on the cutting force signal. The energy of the true flutter dominant frequency band in the reconstructed signal of each node of the historical sound pressure signal is used as the signal, and the energy of the remaining frequency bands is used as noise. The actual chatter frequency band is a 4Hz band centered on the actual chatter frequency. 4Hz is merely one embodiment of this application; implementers can set it according to their actual circumstances, and this application does not impose any special limitations. Determining the actual chatter frequency based on the cutting force signal is a well-known technique, and this application will not elaborate further.

[0046] In this embodiment, the number of historical sound pressure signals is 300, and the preset threshold value is 20dB. The number of historical sound pressure signals and the preset threshold value are both preset by humans. The implementer can set them according to the actual situation. This application does not impose any special restrictions.

[0047] In this embodiment, the machine learning classification model is specifically a support vector machine model. The first preset value and the second preset value are 1 and 0, respectively. The accuracy is used as the evaluation index of the support vector machine model. The training of the support vector machine model is a well-known technique, and the specific process will not be described in detail.

[0048] Taking flutter candidate signal component a as an example, the background noise detection vector of each node of flutter candidate signal component a is used as the input of the trained background noise detection model, and the category label of the background noise detection vector of each node is output. If the category label of the background noise detection vector of each node is the first preset value, then the reconstructed signal of each node is used as the background noise component in flutter candidate signal component a.

[0049] All data in the wavelet packet coefficients of the nodes where the background noise detection vectors of all categories labeled with the first preset value of the flutter candidate signal component a are assigned to 0. The signal is reconstructed based on all reassigned wavelet packet coefficients and unassigned wavelet packet coefficients of the flutter candidate signal component a, thereby filtering out the background noise information in the flutter candidate signal component a. Wavelet packet signal reconstruction is a well-known technique and will not be described in detail in this application.

[0050] The target flutter frequency band signal for the i-th period is obtained by reconstructing all reconstructed flutter candidate signal components of the sound pressure signal within the i-th period. A schematic diagram of the process for obtaining the target flutter frequency band signal is shown below. Figure 2 As shown.

[0051] Step 3: Assess the flutter incubation status in the current cycle using the energy kurtosis index of the target flutter frequency band signal, and then update the rotational speed perturbation parameters for the next cycle.

[0052] The wavelet packet energy kurtosis index of the target flutter frequency band signal is calculated. When the wavelet packet energy kurtosis index is greater than a preset threshold, it is determined that flutter has begun to develop during the precision turning of the motor housing in the i-th period and is about to occur; otherwise, it is determined that flutter has not begun to develop during the precision turning of the motor housing in the i-th period. The calculation of the wavelet packet energy kurtosis index is a known technique and will not be elaborated upon in this application.

[0053] In this embodiment, the method for obtaining the preset threshold is as follows: the mean and standard deviation of the wavelet packet energy kurtosis index of the target chatter frequency band signal of the computer tool under the historical cutting stable state are calculated respectively, and the sum of the mean and three times the standard deviation is used as the preset threshold.

[0054] Furthermore, the spindle speed perturbation method is used to suppress chatter during the precision turning of the motor housing. This method involves periodically varying the machine tool spindle speed around its nominal speed at a certain frequency and amplitude, thereby disrupting the formation mechanism of regenerative chatter and thus suppressing it. Specifically: The frequency estimation algorithm is used to obtain the chatter frequency of the target chatter frequency band signal. Based on the chatter frequency, the speed disturbance parameters in the spindle speed disturbance method are calculated, including the disturbance frequency and the disturbance amplitude. The dSPACE controller is used to control the machine tool spindle speed to continuously and periodically change according to the disturbance frequency and disturbance amplitude in the calculated speed disturbance parameters. The machine tool with the changed spindle speed is used to continue the precision turning of the motor housing to be machined, thereby realizing chatter suppression in the chatter incubation stage of the precision turning of the motor housing to be machined, and reducing the impact of chatter on the precision turning accuracy of the motor housing to be machined. The frequency estimation algorithm and the calculation of the speed disturbance parameters are well known technologies and will not be described in detail in this application.

[0055] Based on the same inventive concept as the above method, this application also provides precision machining equipment for the motor housing of a heavy truck transmission system, including: The signal acquisition module is used to acquire multiple modal inherent frequency bands of the machine tool cutting tool tip. During the precision turning of the motor housing, it acquires sound pressure signals and machine tool spindle speed signals in real time. The background noise filtering module is used to preset the update period of the rotational speed disturbance parameters for flutter suppression. For each period, a mode decomposition algorithm is used to decompose the sound pressure signal into signal components. By analyzing the overlap between the signal components and the inherent frequency bands of the modes, candidate flutter signal components suspected of containing flutter information are selected. A wavelet packet decomposition algorithm is used to decompose each candidate flutter signal component. The signal is reconstructed using the wavelet packet coefficients of each node. The harmonics of the main frequency of the rotational speed signal are compared with the main frequency of the reconstructed signal of each node to obtain the signal synchronization deviation of each node. The signal energy uniformity of each node is obtained by analyzing the data dispersion in the wavelet packet coefficients of each node. Based on the signal synchronization deviation and the signal energy uniformity, a pre-trained background noise detection model is used to identify and filter the background noise components in each candidate flutter signal component, and then reconstruct each candidate flutter signal component to obtain the target flutter frequency band signal. The flutter incubation monitoring and flutter suppression module is used to assess the flutter incubation status in the current cycle by using the energy kurtosis index of the target flutter frequency band signal, and then update the rotational speed disturbance parameters in the next cycle.

[0056] Based on the same inventive concept as the above methods, this application also provides a precision machining apparatus for a motor housing of a heavy-duty truck transmission system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described precision machining methods for a motor housing of a heavy-duty truck transmission system.

[0057] In summary, this application acquires multiple modal inherent frequency bands of the machine tool tip and filters chatter candidate signal components based on the overlap between signal components and modal inherent frequency bands. This enables comprehensive capture of chatter information distributed across multiple modes, significantly improving the completeness and accuracy of chatter monitoring and effectively avoiding information loss due to incomplete frequency band coverage. Furthermore, by calculating the signal synchronization deviation and signal energy uniformity, the essential differences between background noise and flutter signals in synchronization characteristics and energy distribution uniformity are fully utilized to achieve accurate identification of background noise components; combined with a pre-trained background noise detection model, background noise components in flutter candidate signal components are filtered out, significantly reducing the interference of background noise on the monitoring of flutter gestation characteristics. Furthermore, the target flutter frequency band signal is obtained by reconstructing the candidate flutter signal components after filtering out background noise. The flutter incubation status is evaluated by the energy kurtosis index of the target flutter frequency band signal and the speed disturbance parameters are updated. This achieves timely and accurate detection and effective suppression of flutter, thereby improving the flutter suppression effect of precision machining of the motor housing and effectively ensuring the precision machining accuracy of the motor housing.

Claims

1. A method for precision machining of a motor housing in a heavy-duty truck transmission system, characterized in that, The method includes the following steps: The inherent frequency bands of multiple modes of the machine tool tip are obtained, and sound pressure signals and machine tool spindle speed signals are collected in real time during the precision turning of the motor housing. The update period for the rotational speed disturbance parameters for flutter suppression is preset. For each period, a mode decomposition algorithm is used to decompose the sound pressure signal into signal components. By analyzing the overlap between the signal components and the inherent frequency bands of the modes, candidate flutter signal components suspected of containing flutter information are selected. Wavelet packet decomposition algorithm is used to decompose each candidate flutter signal component. The signal is reconstructed using the wavelet packet coefficients of each node. The harmonics of the main frequency of the rotational speed signal are compared with the main frequency of the reconstructed signal of each node to obtain the signal synchronization deviation of each node. The signal energy uniformity of each node is obtained by analyzing the data dispersion in the wavelet packet coefficients of each node. Based on the signal synchronization deviation and the signal energy uniformity, a pre-trained background noise detection model is used to identify and filter out the background noise components in each candidate flutter signal component, thereby reconstructing each candidate flutter signal component and then recombining them to obtain the target flutter frequency band signal. The flutter incubation status in the current cycle is assessed by the energy kurtosis index of the target flutter frequency band signal, and then the rotational speed perturbation parameters in the next cycle are updated. The pre-trained background noise detection model identifies background noise components in each flutter candidate signal component, including: The training process of the background noise detection model is as follows: acquire historical sound pressure signals containing flutter gestation characteristics, combine the signal synchronization deviation and signal energy uniformity of each node to form the background noise detection vector of each node, extract the signal-to-noise ratio of the reconstructed signal of each node of the historical sound pressure signal, set the category labels of the background noise detection vectors of each node with a signal-to-noise ratio lower than or greater than or equal to the preset threshold value to the first preset value and the second preset value respectively, and train the background noise detection model using the background noise detection vectors and category labels of all nodes of all historical sound pressure signals. Based on the background noise detection vector of each node of each flutter candidate signal component in the current period, the background noise detection model is used to obtain the category label of the background noise detection vector of each node, and the reconstructed signal of the node where the background noise detection vector with the category label of the first preset value is located is taken as the background noise component.

2. The precision machining method for the motor housing of a heavy-duty truck transmission system as described in claim 1, characterized in that, The screening process for the flutter candidate signal components is as follows: Extract the half-power bandwidth of the maximum spectral peak in the spectrum of each signal component; If there is an intersection between the half-power bandwidth of each signal component and the inherent frequency band of any mode, then each signal component is regarded as a candidate flutter signal component.

3. The precision machining method for the motor housing of a heavy-duty truck transmission system as described in claim 1, characterized in that, The process of obtaining the signal synchronization deviation is as follows: The minimum difference between the main frequency of the reconstructed signal of each node and the harmonic of the main frequency of the rotational speed signal; The signal synchronization deviation is directly proportional to the minimum gap and inversely proportional to the main frequency of the rotation speed signal.

4. The precision machining method for the motor housing of a heavy-duty truck transmission system as described in claim 3, characterized in that, The signal synchronization deviation is the ratio of the minimum difference to the main frequency of the rotational speed signal.

5. The precision machining method for the motor housing of a heavy-duty truck transmission system as described in claim 1, characterized in that, The signal energy uniformity is the reciprocal of the data dispersion in the wavelet packet coefficients of each node.

6. The precision machining method for the motor housing of a heavy-duty truck transmission system as described in claim 1, characterized in that, The method for reconstructing each flutter candidate signal component is as follows: All data in the wavelet packet coefficients of the nodes containing all background noise detection vectors with the first preset value are assigned to 0. The signal is then reconstructed based on all reassigned wavelet packet coefficients and unassigned wavelet packet coefficients of each flutter candidate signal component.

7. The precision machining method for the motor housing of a heavy-duty truck transmission system as described in claim 1, characterized in that, The assessment process for the flutter gestation is as follows: If the energy kurtosis index of the target flutter frequency band signal is greater than the preset threshold, it is determined that flutter has begun to develop in the current cycle.

8. A precision machining equipment for a motor housing of a heavy-duty truck transmission system, applied to the precision machining method for the motor housing of a heavy-duty truck transmission system as described in claims 1-7, characterized in that, The device includes: The signal acquisition module is used to acquire multiple modal inherent frequency bands of the machine tool cutting tool tip. During the precision turning of the motor housing, it acquires sound pressure signals and machine tool spindle speed signals in real time. The background noise filtering module is used to preset the update period of the rotational speed disturbance parameters for flutter suppression. For each period, a mode decomposition algorithm is used to decompose the sound pressure signal into signal components. By analyzing the overlap between the signal components and the inherent frequency bands of the modes, candidate flutter signal components suspected of containing flutter information are selected. A wavelet packet decomposition algorithm is used to decompose each candidate flutter signal component. The signal is reconstructed using the wavelet packet coefficients of each node. The harmonics of the main frequency of the rotational speed signal are compared with the main frequency of the reconstructed signal of each node to obtain the signal synchronization deviation of each node. The signal energy uniformity of each node is obtained by analyzing the data dispersion in the wavelet packet coefficients of each node. Based on the signal synchronization deviation and the signal energy uniformity, a pre-trained background noise detection model is used to identify and filter the background noise components in each candidate flutter signal component, and then reconstruct each candidate flutter signal component to obtain the target flutter frequency band signal. The flutter incubation monitoring and flutter suppression module is used to assess the flutter incubation status in the current cycle by using the energy kurtosis index of the target flutter frequency band signal, and then update the rotational speed disturbance parameters in the next cycle.

9. A precision machining apparatus for a motor housing of a heavy-duty truck transmission system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the precision machining method for the motor housing of the heavy truck transmission system as described in any one of claims 1-7.

Images