Method for improving the surface quality of milling workpiece based on high-density water-based foam cutting fluid
By using high-density water-based foam cutting fluid during the milling process, the milling force and vibration are reduced and the surface quality of the workpiece is improved, and the impact of milling force and vibration on surface quality during the milling process is solved, and higher surface quality is achieved.
Patent Information
- Application Number
- CN202310612136.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-26
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2043-05-26
AI Technical Summary
Milling force and milling vibration during the milling process have a negative impact on the surface quality of the workpiece, and the prior art has failed to effectively utilize high-density water-based foam cutting fluid to reduce these effects.
A method based on high-density water-based foam cutting fluid was designed. By building a dry and wet milling test platform containing milling force and milling vibration signal acquisition, milling tests under different cutting conditions of TC4 titanium alloy were carried out, data were collected and grayed out. Correlation degree and principal component analysis, establish a generalized regression neural network model for particle swarm optimization, and predict the surface roughness of the workpiece.
By spraying high-density water-based foam cutting fluid, the milling force and milling vibration can be reduced, and the surface quality of the workpiece is significantly improved. The surface roughness of the workpiece is better than dry milling and traditional cutting fluid wet milling.
Smart Images

Figure CN116643020B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a method for improving the surface quality of a milling workpiece based on a high-density water-based foam cutting fluid, and belongs to the technical field of mechanical engineering. Background Art
[0002] Milling is widely used in mechanical processing, but the process planning of milling is uncertain, which will affect the milling performance of the workpiece. Improving the cutting performance of titanium alloy has become the main research direction of titanium alloy processing. In the actual milling process, the tool will generate milling force when milling the workpiece. The workpiece is often metal and fixed on the machine tool fixture. The milling force, the clamping force of the fixture and the metal bond atomic force of the workpiece collide with each other, causing the milling system to vibrate. The milling vibration will produce vibration marks on the surface of the workpiece, affecting the surface processing quality of the workpiece. Therefore, the milling force and milling vibration can affect the surface quality of the workpiece to a certain extent. Many scholars have studied the processing state of the cutting system and the surface quality of the workpiece, and analyzed the influence of cutting conditions and cutting parameters on the surface quality of the workpiece after milling through various methods, but few scholars have considered the influence of high-density water-based foam cutting fluid on the surface quality of the workpiece after milling. In theory, the foam burst absorbs energy, which can reduce the milling force and milling vibration to a certain extent, thereby improving the surface quality of the workpiece after milling. Therefore, it is necessary to design a method based on high-density water-based foam cutting fluid to improve the surface quality of milling workpieces from an energy perspective. Summary of the invention
[0003] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a method for improving the surface quality of milled workpieces based on high-density water-based foam cutting fluid, which can reduce the milling force and milling vibration generated during the milling process, thereby improving the surface quality of the workpiece after milling.
[0004] The present invention provides a method for improving the surface quality of a milling workpiece based on a high-density water-based foam cutting fluid, comprising the following steps:
[0005] S1. Formulate a milling test plan based on the three elements of milling, and build a dry and wet milling test platform including milling force and milling vibration signal acquisition;
[0006] S2. The milling test of TC4 titanium alloy under different cutting conditions is completed through the test scheme and the dry and wet milling test platform designed in step S1. During the test, the milling force, milling noise and milling vibration data are collected synchronously, and after the test, the surface roughness of the test workpiece is measured to obtain the surface roughness of the workpiece;
[0007] S3, extracting the test data collected in step S2, and performing comparative analysis on the extracted data by using grey correlation;
[0008] S4. Perform principal component analysis on the milling force and milling vibration to extract their eigenvalues. Then, based on the experimental data and particle swarm optimization generalized regression neural network, a multidimensional feature prediction model is established with one-dimensional and two-dimensional eigenvalues as input parameters and roughness as output results. Finally, the three combined prediction models are compared and analyzed.
[0009] The present invention conducts milling tests with different process parameters under different cutting conditions on a milling workpiece, and respectively conducts dry milling, wet milling and high-density water-based foam wet milling. Noise signals, three-dimensional force signals and vibration signals are collected synchronously in real time. After the test, the surface roughness of the test workpiece is measured to obtain the surface roughness of the workpiece. The three groups of control group data after treatment are compared and analyzed by grey correlation. The milling force and milling vibration are subjected to principal component analysis to extract their eigenvalues. Based on the test data and the particle swarm optimization generalized regression neural network, a multidimensional feature prediction model is established with one-dimensional features and two-dimensional eigenvalues as input parameters and roughness as output results. By performing grey correlation analysis on the test data and comparing and analyzing the combined prediction models, a method for improving the surface quality of milling workpieces based on high-density water-based foam cutting fluid is obtained.
[0010] The technical solution further optimized as the present invention is as follows:
[0011] Furthermore, the dry and wet milling test platform includes four parts: a dry and wet milling system, a milling force measurement system, a milling noise and vibration measurement system, and a roughness acquisition system; the dry and wet milling system is composed of a test piece, a tool, a CNC milling machine, and a liquid spray system; the milling force measurement system adopts a piezoelectric three-axis dynamometer; the milling noise and vibration measurement system is composed of a piezoelectric three-axis acceleration sensor, a charge amplifier, a data acquisition instrument, and a computer installed with acoustic and vibration analysis software; the roughness acquisition system includes a contact roughness meter.
[0012] In step S1, a piezoelectric three-axis dynamometer is used to perform synchronous real-time measurement of the milling force signal in three directions during the milling process; after the milling noise and vibration measurement system is connected via a data cable, the milling noise is measured in real time during the milling process, and the milling vibration signal is measured in three directions in real time.
[0013] Furthermore, the liquid spray system is an external cutting fluid device or a foam generating device, and the traditional cutting fluid system includes a cutting fluid container, a liquid outlet pipe and a nozzle, the cutting fluid container is connected to the nozzle through the liquid outlet pipe, and the nozzle is arranged toward the cutting area where the tool and the workpiece are located; the foam generating device includes a compressed air storage tank, a foam liquid storage tank, a water storage tank and a foaming nozzle, the foam liquid storage tank and the water storage tank are connected to a mixing pipeline, the mixing pipeline and the compressed air storage tank are connected to the foaming nozzle, and the foaming nozzle is arranged toward the cutting area where the tool and the workpiece are located.
[0014] In step S2, the milling test under different cutting conditions refers to a milling test using different cooling methods under the same processing equipment and processing environment conditions, including a dry milling test, a wet milling test based on a traditional cutting fluid, and a milling test based on a high-density water-based foam cutting fluid; the dry milling test does not use any medium with cooling and lubrication functions, the wet milling test based on the traditional cutting fluid uses a cutting fluid pouring device to pour the cutting fluid into the tool and workpiece cutting area; the milling test based on the high-density water-based foam cutting fluid sprays the foam into the cutting area through a foam generating device.
[0015] In step 3, the comparative analysis of the processed test data using grey correlation analysis is based on the test data and the grey correlation analysis method, and the correlation of force, vibration and noise with respect to processing parameters in dry milling, wet milling and foam milling are analyzed respectively.
[0016] In step S4, the three-dimensional milling force root mean square and the three-dimensional milling vibration acceleration root mean square are respectively reduced in dimension using the principal component analysis method, and the specific operations are as follows:
[0017] First, the data of the three-dimensional milling force root mean square or the three-dimensional milling vibration acceleration root mean square are standardized, and the formula is as follows:
[0018]
[0019] In the formula, Z ij is the normalized data of the original milling force or milling vibration signal data, x ij is the unidirectional milling force root mean square or the unidirectional milling vibration acceleration root mean square, n is the number of trials, m is the number of principal component variables;
[0020] Secondly, the correlation coefficient matrix is calculated based on the standardized data, and the Jacobi method is used to solve the correlation coefficient matrix to obtain the eigenvalues and corresponding eigenvectors of the correlation coefficient matrix;
[0021] Then, the eigenvalues of the correlation coefficient matrix are arranged from large to small, and the principal component contribution rate and cumulative contribution rate corresponding to each eigenvalue are solved according to formula (2):
[0022]
[0023] In the formula, e i is the principal component contribution rate, is the cumulative contribution rate, λ i is the principal component variable with respect to the eigenvalue;
[0024] The root mean square of three-dimensional milling force after dimensionality reduction F dRMS Or the root mean square of the three-dimensional milling vibration acceleration after dimensionality reduction a dRMS The eigenvector corresponding to the eigenvalue with a cumulative contribution value greater than 90% is multiplied by the root mean square of the three-dimensional milling force or the root mean square of the three-dimensional milling vibration acceleration, that is:
[0025] F dRMS =α*F RMS-x +β*F RMS-y +γ*F RMS-z (3)
[0026] a dRMS =α1*a RMS-x +β1*a RMS-y +γ1*a RMS-z (4)
[0027] Where α, β, and γ are the eigenvectors corresponding to the principal component eigenvalues when the cumulative contribution value of the root mean square of the three-dimensional milling force is greater than 90%, and F RMS-x 、F RMS-y 、F RMS-z are the principal component eigenvalues when the cumulative contribution of the root mean square of the milling force in the x, y, and z directions is greater than 90%, respectively. dRMS is the root mean square of the three-dimensional milling force after dimensionality reduction after principal component processing; α1, β1, and γ1 are the eigenvectors corresponding to the principal component eigenvalues when the cumulative contribution value of the root mean square of the three-dimensional milling vibration acceleration is greater than 90%, respectively. RMS-x 、a RMS-y 、a RMS-z They are the principal component eigenvalues when the cumulative contribution of the root mean square of the milling vibration acceleration in the x, y, and z directions is greater than 90%, and a dRMS is the root mean square of the three-dimensional milling vibration acceleration after dimensionality reduction after principal component processing;
[0028] Select the root mean square value of the sound pressure level L from the synchronously collected milling noise data p As the milling noise characteristic value.
[0029] In step S4, the specific method of establishing a multidimensional prediction model using a generalized regression neural network is as follows:
[0030] (1) Establishing a generalized regression neural network
[0031] Suppose the random variables are x and y, and the observed value of x is known to be X, then the regression function y relative to X, the joint probability density function is expressed as f(X,y), and its mathematical expectation is It is calculated by the following formula:
[0032]
[0033] Where, E(y / X) is the prediction formula;
[0034] Usually the joint probability density function f(X,y) can be estimated approximately by the training sample set, and the Gaussian kernel function is used for estimation.
[0035]
[0036] Among them, X i is the sample observation value of the random variable x, Y i is the sample observation value of the random variable y, o is the sample size, p is the dimension of the random variable x, and σ is the smoothing factor;
[0037] use Substituting f(X,y) into formula (13), we can get the basic formula of generalized regression neural network:
[0038]
[0039] In the formula, is the predicted value;
[0040] (2) Optimize the parameter smoothing factor σ in the generalized regression neural network using the particle swarm algorithm
[0041] Assume that in a D-dimensional target search space, there are N particles forming a colony, where the i-th particle is represented by a D-dimensional vector, namely X i =(x i1 ,x i2 ,…,x iD ), the flight speed of the ith particle is also a D-dimensional vector, denoted by V i =(v i1 ,v i2 ,…,v iD ), the optimal position searched by the i-th particle so far is called the individual extreme value, denoted by p best =(p i1 ,p i2 ,…,p iD ), the optimal position searched by the entire particle swarm so far is the global extreme value, denoted as g best =(p g1 ,p g2 ,…,p gD);
[0042] When the above two optimal values are found, the particle updates its speed and position according to the following formulas (8) and (9):
[0043]
[0044]
[0045]
[0046] in, is to update the location, is the update speed, is the speed before update, is the position before the update, is the optimal position searched by the i-th particle at time t, is the optimal position searched by the entire particle swarm at time t, c1 and c2 are both learning factors, is the inertia weight, r1 and r2 are uniform random numbers in the range of [0, 1]; f is the real-time objective function value of the particle, f min and f avg are the minimum target value and average target value of all particles at present, ω max and ω min are the maximum and minimum values of inertia weight respectively;
[0047] (3) Particle Swarm Optimization of Generalized Regression Neural Network
[0048] ① Call the generalized regression neural network in step (2) and import the data sample;
[0049] ② Initialize the parameters of the particle swarm: particle population number N, number of iterations T;
[0050] ③ Divide the training set and the test set according to the set ratio; the input parameters are spindle speed, milling depth, reduced-dimensionality milling force root mean square, reduced-dimensionality vibration acceleration root mean square and sound pressure level root mean square value, and the output parameter is roughness. Import the data in the training set into the generalized regression neural network model for training, and use the smoothness factor as the particle population. The difference between the roughness predicted during the smoothness factor training and the actual measured roughness is used as the fitness value Fit[i];
[0051] ④ Find the minimum fitness value Fit[i] among the particles, recorded as the individual extreme value p best ;
[0052] ⑤ After the set number of iterations is completed, the optimal parameters of the generalized regression neural network model are obtained, and the generalized regression neural network with the optimal parameters is used for model training to obtain the optimal generalized regression neural network model;
[0053] ⑥ Import the test set data into the optimal generalized regression neural network model for prediction to obtain the roughness prediction result, and compare the roughness prediction result with the actual measured roughness value to verify the prediction accuracy of the prediction model.
[0054] Furthermore, the high-density water-based foam cutting fluid is prepared by mixing a foam generator and water in a ratio of 1:100 to 120.
[0055] The present invention obtains a method for improving the surface quality of milled workpieces by comparing dry milling tests, wet milling tests based on traditional cutting fluids, and milling tests based on high-density water-based foam cutting fluids. By comparing and analyzing the data of three groups of control tests of dry milling, wet milling based on traditional cutting fluids, and milling based on high-density water-based foam cutting fluids, it is found that the surface quality of the workpiece measured by the test group based on high-density water-based foam cutting fluids is higher than that of the other two. Spraying high-density water-based foam on the milling area during the milling process can reduce the milling force and milling vibration generated by the milling process, thereby improving the surface quality of the workpiece after milling. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] Figure 1 It is a schematic diagram of the present invention.
[0057] Figure 2 This is a schematic diagram of the dry and wet milling test platform in the present invention.
[0058] Figure 3 It is a comparison curve diagram of the measured value and the fitted value of the roughness of the roughness combination prediction model under the dry milling test group in the present invention.
[0059] Figure 4 It is a comparison curve diagram of the measured value and the fitted value of the roughness of the roughness combination prediction model under the wet milling test group of pouring Castrol synthetic cutting fluid in the present invention.
[0060] Figure 5 It is a comparison curve diagram of the measured value and the fitted value of the roughness of the roughness combination prediction model in the wet milling test group of the present invention by spraying high-density water-based foam cutting fluid.
[0061] Figure 6 This is a real picture of the dry milling test group in the present invention.
[0062] Figure 7 This is a real picture of the wet milling test group poured with Castrol synthetic cutting fluid in the present invention.
[0063] Figure 8 This is a real picture of the wet milling test group sprayed with high-density water-based foam cutting fluid in the present invention. DETAILED DESCRIPTION
[0064] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings: This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation method and specific operation process are given, but the protection authority of the present invention is not limited to the following embodiments.
[0065] Example 1
[0066] like Figure 1 As shown in the figure, a method for improving the surface quality of milling workpieces based on high-density water-based foam cutting fluid is specifically provided, which includes: building a titanium alloy dry and wet milling test platform integrating milling force, milling noise, milling vibration and workpiece roughness measurement, formulating a test plan, and completing the dry and wet milling comparison test of the workpiece (TC4 titanium alloy) under three cutting conditions: dry milling, Castrol synthetic cutting fluid pouring milling and high-density foam cutting fluid jet milling. During the test, the milling force, milling noise and milling vibration data are collected synchronously, and the milling noise characteristic value is extracted from them. After the test, the roughness of the surface of the test workpiece is measured to obtain the surface roughness of the workpiece. The three groups of control group data after treatment are compared and analyzed by grey correlation. The principal component analysis of milling force and milling vibration is performed to extract their characteristic values. Based on the test data and the particle swarm optimization generalized regression neural network, a multidimensional feature prediction model with one-dimensional features and two-dimensional features as input parameters and roughness as output results is established. By performing grey correlation analysis on the test data and comparing the combined prediction models, a method for improving the surface quality of milling workpieces based on high-density water-based foam cutting fluid was obtained.
[0067] 1. Test equipment
[0068] like Figure 2As shown in the figure, the test platform consists of four parts: dry and wet milling system, milling force measurement system, milling noise and vibration measurement system, and roughness acquisition system. The dry and wet milling system consists of a 100*40*100mm TC4 titanium alloy block specimen, a cemented carbide GM-4E-D10.0 four-edge end mill, an XKA714 vertical CNC milling machine, an external cutting fluid device and a foam generating device. The milling force measurement system uses a piezoelectric three-axis dynamometer. The milling noise and vibration measurement system consists of an HS6020 sound calibrator, an HS5661 precision sound level meter, a YD-21 piezoelectric three-axis acceleration sensor, a YE5852 charge amplifier, a high-speed data acquisition instrument, and a WS-AV acoustic vibration measurement and analysis system. The roughness acquisition system mainly includes a contact roughness measuring instrument. The external cutting fluid generating device includes a cutting fluid container, a liquid outlet pipe, and a nozzle. The cutting fluid container is connected to the nozzle through the liquid outlet pipe, and the nozzle is arranged toward the cutting area where the tool and the workpiece are located. The foam generating device comprises a compressed air storage tank, a foam liquid storage tank, a water storage tank and a foaming nozzle. The foam liquid storage tank and the water storage tank are connected with a mixing pipeline. The mixing pipeline and the compressed air storage tank are connected with the foaming nozzle. The foaming nozzle is arranged toward the cutting area where the tool and the workpiece are located.
[0069] 2. Experimental plan
[0070] The test was completed under three cutting conditions: dry machining, Castrol synthetic cutting fluid pouring, and high-density water-based foam cutting fluid (the high-density water-based foam cutting fluid is made of Jumu brand high-foam concentrated foam generator produced by Wuhan Jumu United Investment Co., Ltd. and water in a ratio of 1:100 to 120). According to the three elements of milling, the feed speed v of this test is f There are 1 level, spindle speed n and cutting depth a p There are four levels respectively. The test parameters and levels are shown in Table 1, and the specific scheme is shown in Table 2.
[0071] Table 1 Test parameters and levels
[0072]
[0073] Table 2 Experimental plan
[0074]
[0075] 3. Experimental steps
[0076] (1) Build a titanium alloy dry and wet milling system: The specimen is clamped on the milling machine fixture; the milling cutter is fixed to the spindle of the milling machine through a tool holder; Castrol cutting fluid is poured into the cutting area of the tool and the workpiece through a pouring device, and the high-density water-based foam cutting fluid is squeezed and foamed through a foam generating device and sprayed into the cutting area;
[0077] (2) Connecting and testing the milling force measurement system: Connect the piezoelectric three-axis dynamometer to the milling machine through a transmission cable, start the dynamometer, open the data acquisition software, collect the milling force data during the milling process, and obtain the three-axis milling force root mean square value through data processing by an industrial computer;
[0078] (3) Connect and test the milling noise and vibration measurement system: First, calibrate the HS5661 precision sound level meter with the HS6020 sound calibrator, and then place the precision sound level meter next to the milling workpiece to collect noise signals. Fix the piezoelectric three-dimensional acceleration sensor on the machine tool fixture through a strong magnet, and then connect it to the charge amplifier through a transmission cable. The amplified signal is transmitted to the high-speed data acquisition instrument for aggregation and input into the data processing computer through the USB data cable. The computer completes the milling vibration data collection and obtains the milling noise and three-dimensional milling vibration acceleration root mean square value through the built-in vibration measurement and analysis system;
[0079] (4) Completion of the milling test: According to the milling test plan in Table 2, a program is compiled into the CNC milling machine to complete the collection and processing of the milling force and milling vibration signals under three different cutting conditions. The test data is classified and summarized in the computer, and the power is turned off after the data is saved. The test is completed.
[0080] (5) After the test, a contact roughness measuring instrument is used to measure the surface quality of the workpiece to obtain the roughness value.
[0081] 4. Test results and analysis
[0082] In order to avoid data errors caused by operational errors and milling cutter breakage during the test, according to the above experimental steps, each group of tests was completed twice, a total of 96 groups of data, and 48 groups of data that were not affected were selected for analysis. Among them, 16 groups of test data under dry milling conditions are shown in Table 3, 16 groups of test data under Castrol cutting fluid pouring conditions are shown in Table 4, and 16 groups of test data under high-density water-based foam cutting fluid spraying conditions are shown in Table 5.
[0083] Table 3 16 groups of test data under dry milling conditions
[0084]
[0085]
[0086] Table 4 16 groups of test data under Castrol cutting fluid pouring conditions
[0087]
[0088] Table 5 16 groups of test data under high-density water-based foam cutting fluid spraying conditions
[0089]
[0090] 4.1 Grey relational analysis
[0091] Table 6 Grey correlation analysis of roughness and milling force
[0092]
[0093]
[0094] Table 7 Grey correlation analysis of roughness and milling vibration
[0095]
[0096] Table 8 Grey correlation analysis of roughness and milling noise
[0097]
[0098]
[0099] By combining the data in Table 3, Table 4 and Table 5 and the grey correlation analysis software, the grey absolute correlation, grey relative correlation value and grey comprehensive correlation between the roughness and the milling force of the dry milling test group, the wet milling test group with cast Castrol synthetic cutting fluid and the wet milling test group with high-density water-based foam cutting fluid were obtained, as shown in Table 6. It can be found that in the wet milling test group with high-density water-based foam cutting fluid, the grey correlation between the roughness and the milling force is greater than that of the other two groups, which also shows that the connection between the roughness and the milling force in the wet milling test group with high-density water-based foam cutting fluid is stronger than that in the other two groups.
[0100] By combining the data in Table 3, Table 4 and Table 5 and the grey correlation analysis software, the grey absolute correlation, grey relative correlation value and grey comprehensive correlation between roughness and milling vibration of the dry milling test group, the wet milling test group with cast Castrol synthetic cutting fluid and the wet milling test group with high-density water-based foam cutting fluid were obtained, as shown in Table 7. It can be found that in the wet milling test group with high-density water-based foam cutting fluid, the grey correlation between roughness and milling vibration is greater than that of the other two groups, which also shows that the connection between roughness and milling vibration in the wet milling test group with high-density water-based foam cutting fluid is stronger than that in the other two groups.
[0101] By combining the data in Table 3, Table 4 and Table 5 and the grey correlation analysis software, the grey absolute correlation, grey relative correlation value and grey comprehensive correlation between roughness and milling noise in the dry milling test group, the wet milling test group with cast Castrol synthetic cutting fluid and the wet milling test group with high-density water-based foam cutting fluid were obtained, as shown in Table 8. It can be found that in the wet milling test group with high-density water-based foam cutting fluid, the grey correlation between roughness and milling noise is greater than that of the other two groups, which also shows that the connection between roughness and milling noise in the wet milling test group with high-density water-based foam cutting fluid is stronger than that in the other two groups.
[0102] In summary, based on the experimental data and grey correlation analysis method, the correlations between roughness and milling force, vibration and noise in the dry milling test group, the wet milling test group with Castrol synthetic cutting fluid and the wet milling test group with high-density water-based foam cutting fluid were analyzed respectively. The grey correlation between roughness and milling force, vibration and noise in the wet milling test group with high-density water-based foam cutting fluid was better than that in the other two groups, indicating that high-density water-based foam cutting fluid has a certain optimization effect on improving the surface quality of milling workpieces.
[0103] 4.2 Principal component analysis of milling force and milling vibration
[0104] The principal component analysis of the milling force and milling vibration of the dry milling test group, the wet milling test group with Castrol synthetic cutting fluid and the wet milling test group with high-density water-based foam cutting fluid was performed, as shown in Table 9.
[0105] Table 9 Principal component eigenvalues and variance contribution rates of the milling force of the dry milling test group in principal component analysis
[0106]
[0107] The principal component eigenvalues and variance contribution rates of the milling force of the Castrol test group
[0108]
[0109] Principal component eigenvalues and variance contribution rates of milling force in the foam test group
[0110]
[0111]
[0112] Main component eigenvalues and variance contribution rates of milling vibration in dry milling test group
[0113]
[0114] The principal component eigenvalues and variance contribution rates of milling vibration in the Castrol test group
[0115]
[0116] Principal component eigenvalues and variance contribution rates of milling vibration in the foam test group
[0117]
[0118] The root mean square F of the three control groups after dimensionality reduction after principal component analysis dRMS and the root mean square of vibration acceleration after dimension reduction a dRMS As shown in Table 10.
[0119] Table 10 RMS force and RMS vibration acceleration after dimensionality reduction
[0120]
[0121]
[0122] 4.3 Roughness prediction model analysis
[0123] The test data is imported into the particle swarm prediction model to predict the roughness, and the correlation analysis is performed on the roughness prediction value and the measured value. The prediction accuracy of the particle swarm combination prediction model is verified by the correlation coefficient. The correlation coefficient R and error MES are shown in Table 11.
[0124] Table 11 Optimization parameters and prediction accuracy of multidimensional feature prediction model Dry milling test group
[0125]
[0126] Castrol Milling Test Group
[0127]
[0128] Foam milling test group
[0129]
[0130] The roughness prediction values obtained by the three groups of roughness prediction models are shown in Table 12.
[0131] Table 12 Three groups of roughness prediction values
[0132]
[0133]
[0134] The fitting curve is as follows Figure 3 , Figure 4 and Figure 5 shown.
[0135] Through the roughness combination prediction model, it can be seen that the correlation coefficient R of the roughness prediction model of the wet milling test group sprayed with high-density water-based foam cutting fluid is 0.9542, which is much larger than the dry milling test group and the wet milling test group poured with Castrol synthetic cutting fluid. The roughness prediction model error of the wet milling test group sprayed with high-density water-based foam cutting fluid is maintained at 0.003μm, which is better than the other two test groups. It can be seen that high-density water-based foam cutting fluid has a strong effect on improving the surface quality of milling workpieces.
[0136] The above description is only a specific implementation of the present invention, but the protection scope of the present invention is not limited thereto. Any person familiar with the technology can understand and think of any changes or substitutions within the technical scope disclosed by the present invention, which should be included in the scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A method for improving the surface quality of a milling workpiece based on a high-density water-based foam cutting fluid, characterized in that: The following steps are involved: S1. Based on the three elements of milling, a milling test plan is formulated, and a dry and wet milling test platform including milling force and milling vibration signal acquisition is built; the dry and wet milling test platform includes four parts: dry and wet milling system, milling force measurement system, milling noise and vibration measurement system, and roughness acquisition system; the dry and wet milling system consists of a test piece, a tool, a CNC milling machine, and a liquid spray system; the milling force measurement system uses a piezoelectric three-axis dynamometer; the milling noise and vibration measurement system consists of a piezoelectric three-axis acceleration sensor, a charge amplifier, a data acquisition instrument, and a computer installed with acoustic and vibration analysis software ; The roughness acquisition system includes a contact roughness measuring instrument; the liquid spraying system is an external cutting fluid device or a foam generating device. The traditional cutting fluid system includes a cutting fluid container, a liquid outlet pipe and a nozzle. The cutting fluid container is connected to the nozzle through the liquid outlet pipe, and the nozzle is arranged toward the cutting area where the tool and the workpiece are located; the foam generating device includes a compressed air storage tank, a foam liquid storage tank, a water storage tank and a foaming nozzle. The foam liquid storage tank and the water storage tank are connected to a mixing pipeline. The mixing pipeline and the compressed air storage tank are connected to the foaming nozzle. The foaming nozzle is arranged toward the cutting area where the tool and the workpiece are located; S2. The milling test of TC4 titanium alloy under different cutting conditions is completed through the test scheme and dry and wet milling test platform designed in step S1. During the test, the milling force, milling noise and milling vibration data are collected synchronously, and the surface roughness of the test workpiece is measured after the test to obtain the surface roughness of the workpiece; the milling test under different cutting conditions refers to the milling test using different cooling methods under the same processing equipment and processing environment conditions, including dry milling test, wet milling test based on traditional cutting fluid and milling test based on high-density water-based foam cutting fluid; the dry milling test does not use any medium with cooling and lubrication function, and the wet milling test based on traditional cutting fluid uses a cutting fluid pouring device to pour the cutting fluid into the cutting area of the tool and the workpiece; the milling test based on high-density water-based foam cutting fluid sprays the foam into the cutting area through a foam generating device; the high-density water-based foam cutting fluid is prepared by mixing a foam generator and water in a ratio of 1:100 to 120; the foam generator is the Jumu brand high-foam concentrated foam generator of Wuhan Jumu United Investment Co., Ltd. S3, extracting characteristic values from the test data collected in step S2, and performing comparative analysis on the extracted data using grey correlation; S4. Perform principal component analysis on the milling force and milling vibration to extract their eigenvalues. Then, based on the experimental data and particle swarm optimization generalized regression neural network, a multidimensional feature prediction model is established with one-dimensional and two-dimensional eigenvalues as input parameters and roughness as output results. Finally, the three combined prediction models are compared and analyzed.
2. The method for improving the surface quality of a milling workpiece based on a high-density water-based foam cutting fluid according to claim 1, characterized in that: In the step S1, a piezoelectric three-axis dynamometer is used to perform synchronous real-time measurement of the milling force signal in three directions during the milling process; after the milling noise and vibration measurement system is connected via a data cable, the milling noise is measured in real time during the milling process, and the milling vibration signal is measured in three directions in real time.
3. The method for improving the surface quality of a milling workpiece based on a high-density water-based foam cutting fluid according to claim 1, characterized in that: In the step 3, the processed test data are compared and analyzed using grey correlation, and the correlations of force, vibration and noise with respect to processing parameters in dry milling, wet milling and foam milling are analyzed respectively.
4. The method for improving the surface quality of a milling workpiece based on a high-density water-based foam cutting fluid according to claim 1, characterized in that: In step S4, the three-dimensional milling force root mean square and the three-dimensional milling vibration acceleration root mean square are respectively reduced in dimension using the principal component analysis method, and the specific operations are as follows: First, the data of the three-dimensional milling force root mean square or the three-dimensional milling vibration acceleration root mean square are standardized, and the formula is as follows: In the formula, Z ij is the normalized data of the original milling force or milling vibration signal data, x ij is the unidirectional milling force root mean square or the unidirectional milling vibration acceleration root mean square, n is the number of trials, m is the number of principal component variables; Secondly, the correlation coefficient matrix is calculated based on the standardized data, and the Jacobi method is used to solve the correlation coefficient matrix to obtain the eigenvalues and corresponding eigenvectors of the correlation coefficient matrix; Then, the eigenvalues of the correlation coefficient matrix are arranged from large to small, and the principal component contribution rate and cumulative contribution rate corresponding to each eigenvalue are solved according to formula (2): In the formula, e i is the principal component contribution rate, is the cumulative contribution rate, λ i is the principal component variable with respect to the eigenvalue; The root mean square of three-dimensional milling force after dimensionality reduction F dRMS Or the root mean square of the three-dimensional milling vibration acceleration after dimensionality reduction a dRMS The eigenvector corresponding to the eigenvalue with a cumulative contribution value greater than 90% is multiplied by the root mean square of the three-dimensional milling force or the root mean square of the three-dimensional milling vibration acceleration, that is: F dRMS =α*F RMS-x +β*F RMS-y +γ*F RMS-z (3) a dRMS =α1*a RMS-x +β1*a RMS-y +γ1*a RMS-z (4) Where α, β, and γ are the eigenvectors corresponding to the principal component eigenvalues when the cumulative contribution value of the root mean square of the three-dimensional milling force is greater than 90%, and F RMS-x 、F RMS-y 、F RMS-z are the principal component eigenvalues when the cumulative contribution of the root mean square of the milling force in the x, y, and z directions is greater than 90%, respectively. dRMS is the root mean square of the three-dimensional milling force after dimensionality reduction after principal component processing; α1, β1, and γ1 are the eigenvectors corresponding to the principal component eigenvalues when the cumulative contribution value of the root mean square of the three-dimensional milling vibration acceleration is greater than 90%, respectively. RMS-x 、a RMS-y 、a RMS-z They are the principal component eigenvalues when the cumulative contribution of the root mean square of the milling vibration acceleration in the x, y, and z directions is greater than 90%, and a dRMS is the root mean square of the three-dimensional milling vibration acceleration after dimensionality reduction after principal component processing; Select the root mean square value of the sound pressure level L from the synchronously collected milling noise data p As the milling noise characteristic value.
5. The method for improving the surface quality of a milling workpiece based on a high-density water-based foam cutting fluid according to claim 4, characterized in that: In step S4, the specific method of establishing a multidimensional prediction model using a generalized regression neural network is as follows: (1) Establish a generalized regression neural network Suppose the random variables are x and y, and the observed value of x is known to be X, then the regression function y relative to X, the joint probability density function is expressed as f(X,y), and its mathematical expectation is It is calculated by the following formula: Where, E(y / X) is the prediction formula; Usually the joint probability density function f(X,y) can be estimated approximately by the training sample set, and the Gaussian kernel function is used for estimation. Among them, X i is the sample observation value of the random variable x, Y i is the sample observation value of the random variable y, o is the sample size, p is the dimension of the random variable x, and σ is the smoothing factor; use Substituting f(X,y) into formula (13), we can get the basic formula of generalized regression neural network: In the formula, is the predicted value; (2) Optimize the parameter smoothing factor σ in the generalized regression neural network using the particle swarm algorithm Assume that in a D-dimensional target search space, there are N particles forming a colony, where the i-th particle is represented by a D-dimensional vector, namely X i =(x i1 ,x i2 ,…,x iD ), the flight speed of the ith particle is also a D-dimensional vector, denoted by V i =(v i1 ,v i2 ,…,v iD ), the optimal position searched by the i-th particle so far is called the individual extreme value, denoted by p best =(p i1 ,p i2 ,…,p iD ), the optimal position searched by the entire particle swarm so far is the global extreme value, denoted as g best =(p g1 ,p g2 ,…,p gD ); When the above two optimal values are found, the particle updates its speed and position according to the following formulas (8) and (9): in, is to update the location, is the update speed, is the speed before update, is the position before the update, is the optimal position searched by the i-th particle at time t, is the optimal position searched by the entire particle swarm at time t, c1 and c2 are both learning factors, is the inertia weight, r1 and r2 are uniform random numbers in the range of [0, 1]; f is the real-time objective function value of the particle, f min and f avg are the minimum target value and average target value of all particles at present, ω max and ω min are the maximum and minimum values of inertia weight respectively; (3) Particle Swarm Optimization of Generalized Regression Neural Network ① Call the generalized regression neural network in step (2) and import the data sample; ② Initialize the parameters of the particle swarm: particle population number N, number of iterations T; ③ Divide the training set and the test set according to the set ratio; the input parameters are spindle speed, milling depth, reduced-dimensionality milling force root mean square, reduced-dimensionality vibration acceleration root mean square and sound pressure level root mean square value, and the output parameter is roughness. Import the data in the training set into the generalized regression neural network model for training, and use the smoothness factor as the particle population. The difference between the roughness predicted during the smoothness factor training and the actual measured roughness is used as the fitness value Fit[i]; ④ Find the minimum fitness value Fit[i] among the particles, recorded as the individual extreme value p best ; ⑤ After the set number of iterations is completed, the optimal parameters of the generalized regression neural network model are obtained, and the generalized regression neural network with the optimal parameters is used for model training to obtain the optimal generalized regression neural network model; ⑥ Import the test set data into the optimal generalized regression neural network model for prediction to obtain the roughness prediction result, and compare the roughness prediction result with the actual measured roughness value to verify the prediction accuracy of the prediction model.
