High-efficiency machining method and device for reducer cycloid wheel based on machining machine tool

Abstract

The present application relates to the technical field of cycloidal wheel processing control, and specifically to an efficient processing method and device for a cycloidal wheel of a reducer based on a processing machine tool. The method comprises: obtaining processing data during the cycloidal wheel processing process and performing preprocessing; preliminarily determining whether chatter occurs by analyzing the periodic changes in cutting depth and the periodic fluctuations in cutting force under different temperature changes; and combining vibration, current, and cutting force data to assess the likelihood of chatter occurrence; utilizing the likelihood of chatter occurrence, classifying the processing data through cluster analysis to screen out high-quality processing data and defective processing data; and employing a dragonfly optimization algorithm to optimize the initial position distribution of the algorithm using the high-quality processing data and defective processing data, and solving for the optimal cutting speed as the cutting speed during the current cycloidal wheel processing process. The present application aims to comprehensively analyze chatter phenomena through multi-dimensional processing data to improve the overall processing efficiency during the current cycloidal wheel processing process.

Description

Technical Field

[0001] The present application relates to the technical field of cycloid gear processing control, and in particular to an efficient processing method and device for a reducer cycloid gear based on a processing machine tool. Background Art

[0002] RV reducers are high-precision reduction gears. Their compact structure, high torque, and high transmission efficiency make them widely used in industrial machinery such as machine tools, robotics, and automated equipment. The cycloid wheel, a core component of the RV reducer, has a crucial impact on its transmission efficiency, load-bearing capacity, and service life due to its tooth profile accuracy (such as tooth shape and meshing clearance). While traditional machining methods (such as gear shaping and grinding) can meet certain precision requirements, they lack efficiency and process stability in modern manufacturing, which demands high-precision, high-volume, and low-cost production. Therefore, optimizing the machining method for the cycloid wheel will help precisely control its dimensional accuracy, form and position accuracy, and surface roughness, thereby ensuring the stability and reliability of the RV reducer.

[0003] Since cycloidal wheels are precision-machined parts, existing research has proposed adaptive machining control technologies for cycloidal wheels to improve their machining accuracy. These technologies can adaptively adjust machining parameters based on data such as temperature, vibration, and current. However, existing adaptive machining control technologies for cycloidal wheels often focus on optimizing machining parameters using the mean of machining data or a single indicator to achieve improved machining accuracy. These technologies do not fully consider the fact that in cycloidal wheel machining processes where cutting depths are constantly changing, changes in machining parameters can easily lead to chatter, resulting in reduced overall machining efficiency. Summary of the Invention

[0004] In order to solve the above technical problems, the present application provides an efficient processing method and device for a cycloid gear of a reducer based on a processing machine tool. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides an efficient machining method for a cycloid gear of a reducer based on a machining machine, the method comprising the following steps:

[0006] Acquire and pre-process processing data during the cycloid wheel processing process; the processing data includes vibration data, cutting temperature, cutting force, cutting speed, cutting depth, and feed motor current data, and construct a sequence in time;

[0007] The cutting force is divided into multiple subsequences using the periodicity of cutting depth. The degree of discreteness of the similarity between any subsequence and the rest of the subsequences, as well as the periodic fluctuations of the subsequences under different temperature changes, are analyzed to construct the cutting turbulence coefficient of the cutting force sequence to preliminarily determine whether chatter occurs.

[0008] Vibration, current, and cutting force data are recorded as target processing data. Based on the difference between the peak value and the respective average peak value in the target processing data and the local fluctuation characteristics of the difference, the significance index of each peak is calculated. All peak values ​​are clustered based on this index to extract the detection peak value. The chatter confidence level is constructed by combining the cutting disorder coefficient of the cutting force sequence, the mean of the range of the internal element means of all clusters in all target processing data, and the mean distance between the detection peak position sequences of all target processing data to assess the possibility of chatter occurrence.

[0009] Using the possibility of chatter occurrence, the processing data is classified through cluster analysis to screen out high-quality processing data and defective processing data;

[0010] The dragonfly optimization algorithm is adopted to optimize the initial position distribution of the algorithm with high-quality processing data and defective processing data, and the optimal cutting speed is solved as the cutting speed in the current cycloid wheel processing process.

[0011] Preferably, the method for preliminarily determining whether chattering occurs is:

[0012] Get the cycle length of the cutting depth sequence;

[0013] The cutting force is divided into multiple cutting force subsequences using the cycle length, and the discrete degree of similarity between the first cutting force subsequence and all other cutting force subsequences is calculated to determine the first discrete degree of the cutting force sequence;

[0014] The cutting temperature is divided into two temperature subsequences by using a sequence segmentation algorithm, and recorded as a rising temperature subsequence and a stable temperature subsequence respectively according to the order of position.

[0015] Calculate the rank mean of each cutting force subsequence in the cutting force sequence, and record the cutting force subsequence whose rank mean falls within the rank range of the rising temperature subsequence as the temperature-rising cutting force subsequence, and record the remaining cutting force subsequences as the temperature-normal cutting force subsequence;

[0016] Sort the temperature-rise cutting force subsequences in ascending order according to their positional mean values, calculate the difference between the cutting force means of the two temperature-rise cutting force subsequences, count the number of negative elements between all the differences, and record the ratio between the number of negative elements and the number of all differences as the first ratio of the cutting force sequence.

[0017] Obtain the degree of dispersion between the cutting force means of all warm-level cutting force subsequences, which is recorded as the second degree of dispersion of the cutting force sequence;

[0018] The cutting turbulence coefficient of the cutting force sequence is constructed by using the first and second discrete degrees and the first ratio of the cutting force sequence to preliminarily determine whether chatter occurs.

[0019] Preferably, the cycle length is determined by a hysteresis corresponding to a first peak value among a plurality of autocorrelation coefficients from the cutting depth.

[0020] Preferably, the method for assessing the possibility of chatter occurrence is:

[0021] For various target processing data, all peaks and valleys in the target processing data are obtained, and the mean between all peaks is obtained; the difference between each peak and the peak mean is calculated, and recorded as the first difference of each peak;

[0022] Taking each peak as the center, calculate the degree of dispersion between all elements within the range of the nearest neighbor valley value before and the nearest neighbor valley value after each peak, and record it as the local fluctuation value of each peak;

[0023] Calculate the product of the first difference of each peak value and the local fluctuation value, and record it as the significance index of each peak value; cluster the significance indexes of all peak values ​​as the input of the clustering algorithm;

[0024] Calculate the mean of the internal elements of each cluster and find the range of all means, which is recorded as the first range of the target processing data; at the same time, record the peak value corresponding to the internal element of the cluster with the largest mean value among all clusters as the detection peak value;

[0025] Obtain the bit sequence of all detected peaks in the target processed data, and construct a bit sequence of the target processed data according to the order of bit sequence from small to large;

[0026] The chatter confidence is constructed using the cutting turbulence coefficient of the cutting force sequence, the mean of the first range of all target processing data, and the mean of the distance between the bit sequence of all target processing data to evaluate the possibility of chatter occurrence.

[0027] Preferably, the method of classifying the processed data by cluster analysis to screen out high-quality processed data is:

[0028] The attribute vector of the machining data in each machining process is constructed using the chatter confidence, machining time and machining error in each machining process.

[0029] Cluster the attribute vectors of the processing data collected in all processing steps, record the cluster with the smallest mean value of the internal element modulus as the high-quality cluster, and record the cluster with the smallest mean value of the internal element modulus as the defective cluster;

[0030] The processing data corresponding to all elements in the defective cluster are defective processing data, and the processing data corresponding to all elements in the high-quality cluster are high-quality processing data.

[0031] Preferably, the dragonfly optimization algorithm is used to optimize the initial position distribution of the algorithm using high-quality processing data and defective processing data, and the method for solving the optimal cutting speed is:

[0032] The cutting speed in high-quality processing data is used as the food position in the dragonfly optimization algorithm, and the cutting speed in defective processing data is used as the enemy position;

[0033] During initialization, dragonflies were first placed at the food location, and the remaining dragonflies were randomly distributed at non-natural enemy locations;

[0034] Construct the global objective function: ;Where Y is the global objective function value; is the minimum value function; S is the modulus of the attribute vector corresponding to the processed data;

[0035] The number of dragonflies, initial positions, preset related parameters, global objective function, and cutting speed of the processing data in the current processing process are used as inputs of the dragonfly optimization algorithm. The output of the dragonfly optimization algorithm is the optimal cutting speed that minimizes the value of the global objective function.

[0036] Among them, the number of dragonflies is determined by the number of processing data obtained from the historical database.

[0037] Preferably, the cutting turbulence coefficient is calculated as follows:

[0038] The ratio of the sum of the first discrete degree and the second discrete degree of the cutting force sequence to the first ratio of the cutting force sequence is taken as the cutting turbulence coefficient of the cutting force sequence.

[0039] Preferably, the calculation formula for the chatter confidence is:

[0040] The chatter confidence of the u-th processing data is recorded as , ;

[0041] Where, is the cutting turbulence coefficient of the cutting force sequence in the u-th machining process; is the first range mean of all target processing data for the uth time; is the mean distance between the bit sequence of all target processing data in the uth time.

[0042] In a second aspect, another embodiment of the present application provides an efficient processing device for a cycloid gear of a reducer based on a processing machine tool, which implements the efficient processing method for a cycloid gear of a reducer based on a processing machine tool described above, and the device includes:

[0043] Data acquisition and preprocessing module, used to collect vibration, temperature, cutting force, cutting speed, cutting depth and current data during the cycloid wheel processing process and perform preprocessing;

[0044] Chatter detection module, used to analyze the periodic changes in cutting depth and periodic fluctuations in cutting force to detect chatter phenomena;

[0045] Flutter confidence analysis module, used to evaluate the possibility of flutter occurrence by combining multi-dimensional data;

[0046] Attribute clustering module, used to classify processed data;

[0047] The optimization solution module is used to optimize the parameters of the dragonfly optimization algorithm using the classification results, and solve the optimal cutting speed through the dragonfly optimization algorithm to improve the processing efficiency of the current processing and avoid chatter.

[0048] Preferably, the data acquisition step in the data acquisition and preprocessing module includes:

[0049] Arrange vibration sensors on the spindle of the processing machine tool to collect vibration data in real time during the processing;

[0050] The cutting temperature, cutting force, cutting speed, cutting depth and current data of the feed motor during the processing are obtained in real time through the central control system of the processing machine tool.

[0051] This application has at least the following beneficial effects:

[0052] This application can reflect whether the cutting force remains stable and periodic under the periodic cutting depth changes of the cycloidal wheel by preliminarily judging whether the chatter phenomenon occurs, thereby judging the stability of the machining process; this application can reflect the coupling and synchronization between multi-dimensional machining data in the machining process by evaluating the possibility of chatter occurrence, thereby accurately identifying whether chatter phenomenon has occurred; this application can classify the machining data by using the possibility of chatter occurrence to distinguish the machining quality, thereby optimizing the relevant parameters of the optimal solution algorithm, thereby improving computational efficiency. This application addresses the problem that the existing technology does not fully consider the changes in the dynamic cutting depth of the cycloidal wheel, which may lead to chatter phenomenon and thus reduce machining efficiency. This application evaluates the possibility of chatter occurrence and optimizes the initial position distribution in the algorithm through high-quality machining data, thereby being able to more efficiently find the optimal cutting speed, thereby maximizing the overall machining efficiency in the current cycloidal wheel machining process while ensuring machining accuracy and avoiding chatter. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A flowchart of the steps of an efficient machining method for a cycloidal gear of a reducer based on a machining machine provided in one embodiment of the present application;

[0054] Figure 2 A schematic diagram of a module of an efficient processing device for a cycloidal wheel of a reducer based on a processing machine tool provided in one embodiment of the present application. DETAILED DESCRIPTION

[0055] Example 1

[0056] This embodiment provides a flow chart of the steps of an efficient processing method for a cycloid gear of a reducer based on a processing machine tool, as shown in the attached figure. Figure 1 As shown, the method includes the following steps:

[0057] Step 1: Obtain the processing data during the cycloid wheel processing and perform preprocessing.

[0058] This application collects and preprocesses data through a data collection and preprocessing module built into a high-efficiency processing device for a cycloid gear of a reducer based on a processing machine tool, specifically as follows:

[0059] A vibration sensor is arranged on the spindle of the processing machine tool to collect the machine tool vibration data in real time during the processing; the cutting temperature, cutting force, cutting speed, cutting depth and current data of the feed motor during the cycloidal wheel processing are obtained in real time through the central control system of the processing machine tool.

[0060] All data is collected synchronously and in real time at a frequency of 10 Hz, from the start of the cycloid wheel machining process to the end of the process. After each machining process, the machining accuracy (percentage data) of the cycloid wheel is collected using a three-dimensional point-by-point scanning method.

[0061] According to the chronological order of data acquisition, the vibration sequence, temperature sequence, cutting force sequence, cutting speed sequence, cutting depth sequence, and current sequence are constructed respectively and recorded as one-time processing data. The processing data of the cycloid wheel for N times of the cycloid wheel is obtained from the historical database of the processing machine tool. In this embodiment, N is set to 100.

[0062] In order to eliminate the dimensional influence between the data, all the data are normalized. The normalization methods include Z-score normalization, sigmoid normalization, maximum and minimum value normalization, etc. This embodiment adopts maximum and minimum value normalization.

[0063] Step 2: Preliminarily determine whether chattering occurs.

[0064] Normally, faster cutting speeds result in higher cutting efficiency and shorter machining times. However, during high-speed cutting operations, even small changes in machining parameters can cause significant fluctuations in the stress between the tool and the workpiece, which can easily lead to chatter. This necessitates further adjustment of machining parameters to maintain a stable machining rhythm, which in turn reduces machining efficiency.

[0065] During the machining process of the cycloid wheel, since the tooth profile of the cycloid wheel is multi-petal meshing and the cutting depth changes periodically with the tooth phase, the cycloid wheel is more prone to vibration than conventional machining parts, thereby affecting the machining accuracy and efficiency.

[0066] This application uses a built-in vibration detection module in a high-efficiency machining device for a cycloid gear of a reducer based on a machining machine to analyze machining data to preliminarily detect whether vibration has occurred, as follows:

[0067] During the cutting process, when the cutting speed is reasonable and chatter is not present, the cutting force should also exhibit periodic fluctuations due to the cyclical variation in cutting depth. However, rising temperature causes the workpiece to soften, which in turn reduces the mean cutting force, but the cyclical variation trend of the cutting force remains unchanged. As the temperature reaches a steady state, the mean cutting force also begins to stabilize. If the cutting speed is too fast and chatter occurs, the cutting force variation loses its cyclical characteristics, and the synchronization between the cutting force and temperature is also disrupted.

[0068] Based on this, the present application obtains the cycle length of the cutting depth; uses the cycle length to divide the cutting force into multiple cutting force subsequences, and calculates the discrete degree of similarity between the first cutting force subsequence and all other cutting force subsequences to determine the first discrete degree of the cutting force sequence.

[0069] Preferably, as a preferred embodiment, the cycle length is determined by the hysteresis corresponding to the first peak value among several autocorrelation coefficients of the cutting depth. In other embodiments, the main frequency cycle of the cutting depth in the frequency domain can be obtained by fast Fourier transform to determine its cycle length.

[0070] This embodiment takes the processing data of the u-th cycloid wheel as an example for analysis:

[0071] The cutting depth series is used as the input of the autocorrelation coefficient, and the lag [1- ] range, output the autocorrelation coefficient corresponding to each hysteresis, and record the hysteresis corresponding to the first peak in all autocorrelation coefficients as the period length of the cutting depth sequence. Where N is the data length of the cutting depth sequence; t is 5 in this embodiment, indicating that the cycloid wheel has at least five gears. Implementers can adjust the value according to actual conditions.

[0072] The cutting depth sequence is divided into multiple depth subsequences based on the period length. Multiple cutting force subsequences are extracted from the cutting force sequence based on the position range of each depth subsequence within the cutting depth sequence. The similarity between the first cutting force subsequence and all cutting force subsequences is calculated, and the degree of dispersion between all similarities is calculated, which is recorded as the first dispersion degree of the cutting force sequence.

[0073] The similarity calculation method is not limited to the Pearson correlation coefficient, Spearman correlation coefficient, or cosine similarity; this embodiment uses the Pearson correlation coefficient. The dispersion calculation method is not limited to information entropy, variance, or coefficient of variation; this embodiment uses information entropy. The first dispersion can reflect whether the data change trends between the first cutting force subsequence and the remaining cutting force subsequences are relatively consistent, thereby reflecting whether the change trends in the cutting force sequence conform to periodicity.

[0074] Furthermore, it is analyzed whether the change of cutting force is synchronized with the temperature.

[0075] During the cutting process, the cutting temperature will gradually increase and then begin to stabilize after reaching a certain temperature, showing two obvious data trends.

[0076] Therefore, the present application uses a sequence segmentation algorithm to segment the cutting temperature into two temperature subsequences, and records them in order of position as a rising temperature subsequence and a steady temperature subsequence. The sequence segmentation algorithm is not limited to the BG segmentation algorithm or the MK segmentation algorithm. This embodiment adopts the MK sequence segmentation algorithm.

[0077] The rank mean of each cutting force subsequence in the cutting force sequence is calculated, and the cutting force subsequence whose rank mean belongs to the rank range of the rising temperature subsequence is recorded as the temperature-rising cutting force subsequence, and the remaining cutting force subsequences are recorded as the temperature-normal cutting force subsequence.

[0078] The temperature-rise cutting force subsequences are sorted in ascending order according to their rank mean values, and the differences between the cutting force means of the two temperature-rise cutting force subsequences are calculated in sequence. The number of negative elements between all differences is counted, and the ratio between the number of negative elements and the number of all differences is recorded as the first ratio of the cutting force sequence.

[0079] Obtain the degree of dispersion between the cutting force means of all warm-level cutting force subsequences, which is recorded as the second degree of dispersion.

[0080] Based on this, the present application utilizes the first and second discrete degrees of the cutting force sequence and the first ratio of the cutting force sequence to construct a cutting turbulence coefficient of the cutting force sequence to preliminarily determine whether chatter occurs.

[0081] In this embodiment, the cutting disorder coefficient of the cutting force sequence in the u-th processing process is constructed The calculation expression is:

[0082]

[0083] Where, is the first discrete degree of the cutting force sequence during the u-th machining process; is the second discrete degree of the cutting force sequence in the u-th machining process; is the first ratio of the cutting force sequence in the u-th machining process.

[0084] It should be understood that the first discrete degree It can reflect whether the cutting force changes of the cycloid wheel are in line with the periodic trend when the cutting depth has periodic variation characteristics and the current cutting speed is used for cutting. The larger the value, the worse the periodic trend is. It can reflect the fluctuation difference of cutting force in different cutting cycles after the cutting temperature is stable. The larger the value, the greater the violent fluctuation of cutting force. It can reflect whether the average cutting force has a continuous downward trend during the temperature rise stage, and thus reflect whether the cutting change conforms to the physical law of material softening. It can comprehensively reflect the overall stability of the cutting force when the cycloid wheel is processed at the current cutting speed. The larger the value, the more likely the current cutting speed has exceeded the most appropriate speed range and the more likely chatter has occurred.

[0085] Step 3: Assess the likelihood of chatter occurring.

[0086] Furthermore, when chatter is not occurring, machine tool vibration is primarily concentrated at the tooth-pass frequency and its harmonics, resulting in small vibration data amplitudes and relatively stable fluctuations. However, if chatter occurs due to an inappropriate cutting speed, the vibration data will exhibit abnormal, sudden peaks and significantly increase its variance. Furthermore, when chatter occurs, the fluctuation variance of the cutting force and the current in the feed motor will also mutate. Therefore, by combining vibration, current, and cutting force data during machining, we can further analyze whether chatter is occurring and assess its likelihood.

[0087] Based on this, the present application records the vibration, current and cutting force data as target processing data; for various target processing data, obtain all peaks and valleys in the target processing data, and obtain the mean between all peaks; calculate the difference between each peak and the peak mean, and record it as the first difference of each peak; with each peak as the center, calculate the degree of dispersion between all elements within the range of the previous nearest neighbor valley value and the next nearest neighbor valley value of each peak, and record it as the local fluctuation value of each peak; calculate the product of the first difference of each peak and the local fluctuation value, and record it as the significance index of each peak; the significance index of all peaks is calculated as Cluster the input of the clustering algorithm; calculate the mean of the internal elements of each cluster, and find the range of all means, which are recorded as the first range of the target processing data; at the same time, record the peak corresponding to the internal element of the cluster with the largest mean among all clusters as the detection peak; obtain the position sequence of all detection peaks in the target processing data, and construct the position sequence of the target processing data according to the position sequence from small to large; use the cutting disorder coefficient of the cutting force sequence, the first range mean of all target processing data and the mean distance between the position sequence of all target processing data to construct the vibration confidence to evaluate the possibility of vibration occurrence.

[0088] In this embodiment, the following analysis is performed using vibration data as an example:

[0089] Using a peak-to-valley detection algorithm, we obtain all peaks and valleys in the vibration sequence and the mean of all peaks. We then calculate the difference between each peak and the mean, recording this as the first difference for each peak. Note that this first difference can be positive or negative.

[0090] With each peak as the center, the degree of dispersion between the nearest neighbor valley value preceding each peak and all elements within the range of the nearest neighbor valley value following each peak is calculated, and this is recorded as the local fluctuation value of each peak. The local fluctuation value can reflect the degree of sudden change of each peak within a local range. It should be noted that when a peak is near the end of the sequence, resulting in no valley value on either side, the degree of dispersion between the peak and all elements within the range of the first or last element of the sequence is calculated.

[0091] The product of the first difference of each peak and the local fluctuation value is calculated and recorded as the significance index of each peak. The significance index of all peaks is used as the input of the clustering algorithm for clustering. Clustering algorithms are not limited to the k-means algorithm, the DPC algorithm, and the DBSCAN algorithm. This embodiment uses the K-means algorithm for clustering, and uses the elbow rule to obtain the optimal number of clusters, and outputs multiple clusters.

[0092] Calculate the mean of the internal elements of each cluster and find the range of all means, which is recorded as the first range of the vibration sequence; at the same time, record the peak value corresponding to the internal element of the cluster with the largest mean value among all clusters as the detection peak value.

[0093] It should be understood that the significance index can reflect the degree of deviation of each vibration peak value compared to the vibration mean and whether the local fluctuation is severe; the clustering algorithm can cluster all peak values, and extract a group of vibration peak values ​​with the largest deviation from the cluster mean and the most severe local fluctuation by comparing the cluster means, thereby improving subsequent detection accuracy and efficiency.

[0094] Obtain the position order of all detected peaks in the vibration sequence, and construct the position order sequence of the vibration sequence according to the order from small to large. In the same way, obtain the position order sequence and the first range difference of the current sequence and cutting force sequence respectively.

[0095] In this embodiment, the chatter confidence of the u-th processing data is constructed The calculation expression is:

[0096]

[0097] Where, is the cutting turbulence coefficient of the cutting force sequence in the u-th machining process; is the first range mean of all target processing data for the uth time; is the mean distance between the bit sequence sequences of all target processing data for the uth time, expressed as the mean of the DTW distances between the three bit sequence sequences of vibration sequence, current sequence, and cutting force sequence. It should be noted that the DTW distance is used here to take into account the possible inconsistency in the data length of the bit sequence.

[0098] It should be understood that the first range mean It can reflect whether there are abnormal data fluctuations in the vibration data, current data and cutting force data in the target processing data of the uth time; the distance from the mean It can reflect whether the abnormal peak occurrence time of the three target processing data of vibration, current and cutting force is relatively consistent. It can combine multi-dimensional data factors to comprehensively reflect whether chatter occurs during the u-th processing at the current cutting speed. If the value is smaller, it reflects that chatter does not occur, and it is still possible to consider increasing the cutting speed of the processing machine tool to improve the processing efficiency.

[0099] Step 4: Filter out high-quality processing data and defective processing data.

[0100] Furthermore, after obtaining the chatter confidence of the processing data, the cutting time and cutting accuracy required at each cutting speed can be combined to obtain a cutting speed that minimizes the processing time and provides better processing accuracy.

[0101] This application uses the built-in attribute clustering module of the efficient processing device of the reducer cycloid wheel based on the processing machine tool to distinguish the processing data, as follows:

[0102] Obtain the chatter confidence of each processing data; at the same time, the data length of the cutting force sequence corresponding to each processing data is used as the processing time of each processing data, and the difference between 1 and the processing accuracy (percentage) is recorded as the processing error of each processing data.

[0103] To eliminate the dimensional influence of the data, the chatter confidence, machining time, and machining error collected from all machining processes were normalized. An attribute vector of each machining data was constructed based on the normalized data, namely [chatter confidence, machining time, machining error]. The attribute vectors of the machining data collected from all machining processes were used as the input of the clustering algorithm, and multiple machining clusters were output.

[0104] Calculate the mean modulus of the internal elements of each processing cluster. The processing cluster with the largest mean is designated as the defective cluster, and the processing cluster with the smallest mean is designated as the high-quality cluster. Accordingly, the processing data corresponding to all elements in the defective cluster is designated as defective processing data, and the processing data corresponding to all elements in the high-quality cluster is designated as high-quality processing data.

[0105] Through clustering algorithms, different processing qualities can be distinguished, making it easier to find the optimal cutting speed.

[0106] Step 5: Calculate the optimal cutting speed.

[0107] This application obtains the optimal solution through the built-in optimization solution module of the high-efficiency processing device of the reducer cycloid gear based on the processing machine tool, as follows:

[0108] In order to further optimize the cutting speed, this application uses the Dragonfly Optimization Algorithm for optimal solution. In the Dragonfly Optimization Algorithm, the distribution of initial positions has a significant impact on the convergence speed and calculation accuracy of the algorithm. A good initial distribution can accelerate the search for the global optimal solution.

[0109] To this end, this application uses the cutting speed in high-quality processing data as the food location in the dragonfly optimization algorithm, i.e., the local optimal solution, and the cutting speed in defective processing data as the natural enemy location, i.e., the local worst solution. The number of dragonflies is determined by the number of processing data N obtained from the historical database.

[0110] During initialization, dragonflies were first placed at the food location, and the remaining dragonflies were randomly distributed at non-natural enemy locations.

[0111] Construct the global objective function: Where Y is the global objective function value; is the minimum function; S is the module length of the attribute vector corresponding to the processed data.

[0112] In this embodiment, the relevant parameters of the preset dragonfly optimization algorithm are: separation weight is 0.1, alignment weight is 0.1, aggregation weight is 0.7, food factor is 1, natural enemy weight is 1, inertia weight is initially 0.9, and decreases linearly with the number of iterations to 0.4, and the number of iterations is 50.

[0113] The number of dragonflies, initial positions, preset related parameters, global objective function and cutting speed of processing data in the current processing process are taken as the input of the dragonfly optimization algorithm. The output of the dragonfly optimization algorithm is the optimal cutting speed that minimizes the value of the global objective function.

[0114] At this point, under the premise of not causing vibration and ensuring processing accuracy, the optimal cutting speed is output as the cutting speed in the current cycloid wheel processing process, thereby minimizing the processing time of the current cycloid wheel and improving processing efficiency.

[0115] Example 2

[0116] This embodiment provides a module schematic diagram of an efficient processing device for a reducer cycloid wheel based on a processing machine tool, as shown in the attached figure. Figure 2 As shown, the above-mentioned efficient processing method of the reducer cycloid wheel based on the processing machine tool is realized. The device has multiple built-in modules, which are as follows:

[0117] Data acquisition and preprocessing module, used to collect vibration, temperature, cutting force, cutting speed, cutting depth and current data during the cycloid wheel processing process and perform preprocessing;

[0118] Chatter detection module, used to analyze the periodic changes in cutting depth and periodic fluctuations in cutting force to detect chatter phenomena;

[0119] Flutter confidence analysis module, used to evaluate the possibility of flutter occurrence by combining multi-dimensional data;

[0120] Attribute clustering module, used to classify processed data;

[0121] The optimization solution module is used to optimize the parameters of the dragonfly optimization algorithm using the classification results, and solve the optimal cutting speed through the dragonfly optimization algorithm to improve the processing efficiency of the current processing and avoid chatter.

[0122] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not invented herein.

[0123] It will be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.

Claims

1. An efficient machining method for a cycloid gear of a reducer based on a machining machine, characterized in that: The method includes: Acquire and pre-process processing data during the cycloid wheel processing process; the processing data includes vibration data, cutting temperature, cutting force, cutting speed, cutting depth, and feed motor current data, and construct a sequence in time; The cutting force is divided into multiple subsequences using the periodicity of cutting depth. The degree of discreteness of the similarity between any subsequence and the rest of the subsequences, as well as the periodic fluctuations of the subsequences under different temperature changes, are analyzed to construct the cutting turbulence coefficient of the cutting force sequence to preliminarily determine whether chatter occurs. Vibration, current, and cutting force data are recorded as target processing data. Based on the difference between the peak value and the respective average peak value in the target processing data and the local fluctuation characteristics of the difference, the significance index of each peak is calculated. All peak values ​​are clustered based on this index to extract the detection peak value. The chatter confidence level is constructed by combining the cutting disorder coefficient of the cutting force sequence, the mean of the range of the internal element means of all clusters in all target processing data, and the mean distance between the detection peak position sequences of all target processing data to assess the possibility of chatter occurrence. Using the possibility of chatter occurrence, the processing data is classified through cluster analysis to screen out high-quality processing data and defective processing data; The dragonfly optimization algorithm is used to optimize the initial position distribution of the algorithm with high-quality processing data and defect processing data, and the optimal cutting speed is solved as the cutting speed in the current cycloid wheel processing process; The dragonfly optimization algorithm is used to optimize the initial position distribution of the algorithm using high-quality processing data and defective processing data. The method for solving the optimal cutting speed is: the cutting speed in the high-quality processing data is used as the food position in the dragonfly optimization algorithm, and the cutting speed in the defective processing data is used as the natural enemy position; During initialization, dragonflies were first placed at the food location, and the remaining dragonflies were randomly distributed at non-natural enemy locations; Construct the global objective function: ;Where Y is the global objective function value; is the minimum value function; S is the modulus of the attribute vector corresponding to the processed data; The number of dragonflies, initial positions, preset related parameters, global objective function, and cutting speed of the processing data in the current processing process are used as inputs of the dragonfly optimization algorithm. The output of the dragonfly optimization algorithm is the optimal cutting speed that minimizes the value of the global objective function. Among them, the number of dragonflies is determined by the number of processing data obtained from the historical database.

2. The high-efficiency processing method of the cycloid gear of the reducer based on the processing machine tool according to claim 1 is characterized in that: The method for preliminarily judging whether chattering occurs is: Get the cycle length of the cutting depth sequence; The cutting force is divided into multiple cutting force subsequences using the cycle length, and the discrete degree of similarity between the first cutting force subsequence and all other cutting force subsequences is calculated to determine the first discrete degree of the cutting force sequence; The cutting temperature is divided into two temperature subsequences by using a sequence segmentation algorithm, and recorded as a rising temperature subsequence and a stable temperature subsequence respectively according to the order of position. Calculate the rank mean of each cutting force subsequence in the cutting force sequence, and record the cutting force subsequence whose rank mean falls within the rank range of the rising temperature subsequence as the temperature-rising cutting force subsequence, and record the remaining cutting force subsequences as the temperature-normal cutting force subsequence; Sort the temperature-rise cutting force subsequences in ascending order according to their positional mean values, calculate the difference between the cutting force means of the two temperature-rise cutting force subsequences, count the number of negative elements between all the differences, and record the ratio between the number of negative elements and the number of all differences as the first ratio of the cutting force sequence. Obtain the degree of dispersion between the cutting force means of all warm-level cutting force subsequences, which is recorded as the second degree of dispersion of the cutting force sequence; The cutting turbulence coefficient of the cutting force sequence is constructed by using the first and second discrete degrees and the first ratio of the cutting force sequence to preliminarily determine whether chatter occurs.

3. The high-efficiency processing method of the cycloid gear of the reducer based on the processing machine tool according to claim 2 is characterized in that: The cycle length is determined by the hysteresis corresponding to the first peak value among several autocorrelation coefficients of the cutting depth.

4. The high-efficiency processing method of the cycloid gear of the reducer based on the processing machine tool according to claim 1 is characterized in that: The method for evaluating the possibility of chatter occurrence is: For various target processing data, all peaks and valleys in the target processing data are obtained, and the mean between all peaks is obtained; the difference between each peak and the peak mean is calculated, and recorded as the first difference of each peak; Taking each peak as the center, calculate the degree of dispersion between all elements within the range of the nearest neighbor valley value before and the nearest neighbor valley value after each peak, and record it as the local fluctuation value of each peak; Calculate the product of the first difference of each peak value and the local fluctuation value, and record it as the significance index of each peak value; cluster the significance indexes of all peak values ​​as the input of the clustering algorithm; Calculate the mean of the internal elements of each cluster and find the range of all means, which is recorded as the first range of the target processing data; at the same time, record the peak value corresponding to the internal element of the cluster with the largest mean value among all clusters as the detection peak value; Obtain the bit sequence of all detected peaks in the target processed data, and construct a bit sequence of the target processed data according to the order of bit sequence from small to large; The chatter confidence is constructed using the cutting turbulence coefficient of the cutting force sequence, the mean of the first range of all target processing data, and the mean of the distance between the bit sequence of all target processing data to evaluate the possibility of chatter occurrence.

5. The high-efficiency processing method of the cycloid gear of the reducer based on the processing machine tool according to claim 3 is characterized in that: The method of classifying the processing data through cluster analysis to screen out high-quality processing data is as follows: The attribute vector of the machining data in each machining process is constructed using the chatter confidence, machining time and machining error in each machining process. Cluster the attribute vectors of the processing data collected in all processing steps, record the cluster with the smallest mean value of the internal element modulus as the high-quality cluster, and record the cluster with the smallest mean value of the internal element modulus as the defective cluster; The processing data corresponding to all elements in the defective cluster are defective processing data, and the processing data corresponding to all elements in the high-quality cluster are high-quality processing data.

6. The high-efficiency processing method of the cycloid gear of the reducer based on the processing machine tool according to claim 2, characterized in that: The calculation method of the cutting turbulence coefficient is: The ratio of the sum of the first discrete degree and the second discrete degree of the cutting force sequence to the first ratio of the cutting force sequence is taken as the cutting turbulence coefficient of the cutting force sequence.

7. The high-efficiency processing method of the cycloid gear of the reducer based on the processing machine tool according to claim 4 is characterized in that: The calculation formula of the chatter confidence is: The chatter confidence of the u-th processing data is recorded as , ; Where, is the cutting turbulence coefficient of the cutting force sequence in the u-th machining process; is the first range mean of all target processing data for the uth time; is the mean distance between the bit sequence of all target processing data in the uth time.

8. An efficient processing device for a cycloid gear of a reducer based on a processing machine tool, characterized in that: The device implements the efficient processing method of the reducer cycloid gear based on a processing machine tool according to any one of claims 1 to 7, and the device includes: Data acquisition and preprocessing module, used to collect vibration, temperature, cutting force, cutting speed, cutting depth and current data during the cycloid wheel processing process and perform preprocessing; Chatter detection module, used to analyze the periodic changes in cutting depth and periodic fluctuations in cutting force to detect chatter phenomena; Flutter confidence analysis module, used to evaluate the possibility of flutter occurrence by combining multi-dimensional data; Attribute clustering module, used to classify processed data; The optimization solution module is used to optimize the parameters of the dragonfly optimization algorithm using the classification results, and solve the optimal cutting speed through the dragonfly optimization algorithm to improve the processing efficiency of the current processing and avoid chatter.

9. The high-efficiency processing device for the cycloid gear of the reducer based on the processing machine tool according to claim 8, characterized in that: The steps of data acquisition in the data acquisition and preprocessing module include: Arrange vibration sensors on the spindle of the processing machine tool to collect vibration data in real time during the processing; The cutting temperature, cutting force, cutting speed, cutting depth and current data of the feed motor during the processing are obtained in real time through the central control system of the processing machine tool.

Images