Indirect prediction method of milling force considering tool wear based on power signal

Abstract

The present invention provides an indirect prediction method for milling force based on power signal considering tool wear. By establishing a theoretical power consumption model and considering the influence of tool wear on cutting force, the influence relationship from tool wear to cutting force and then to machine tool power is established from the mechanism level, and a variable working condition milling force prediction model based on machine tool power consumption is established, thereby realizing accurate prediction of tool wear and cutting force under any working condition. The method provided by the present invention overcomes the shortcomings of other indirect monitoring signals in terms of versatility and extensibility. At the same time, the present invention is relatively simple to install and maintain, and has higher feasibility in actual production line deployment.

Description

Technical Field

[0001] The invention belongs to the technical field of state monitoring of cutting process of numerical control machine tools, and in particular relates to an indirect prediction method of milling force considering tool wear. Background Art

[0002] With the intelligent development of manufacturing industry and the increasing complexity of processed products, online / offline monitoring of the processing process has become increasingly important. As one of the key physical parameters in the milling process, cutting force plays an important role in determining the machining dimensional accuracy, surface integrity, cutting temperature, tool-workpiece vibration, etc. Therefore, accurate and reliable prediction of cutting force is indispensable for monitoring, optimizing and improving the processing quality.

[0003] The traditional cutting force monitoring method is achieved by a flat dynamometer fixed to the workbench or a rotating dynamometer installed on the tool shaft. However, this method and device usually have high restrictions on the size, shape and operation requirements of the workpiece, and the non-negligible weight of this device will affect the dynamic characteristics of the machine tool or the workpiece, resulting in instability during the processing.

[0004] Existing research shows that the use of indirect methods to predict cutting forces has good prospects. Indirect methods are usually divided into empirical formula methods based on a large number of experiments and time-frequency domain analysis methods based on signal calculation processing. Empirical methods, such as summarizing the relationship between the different energy development stages of acoustic emission signals and the size of cutting forces, derive relevant cutting force empirical models. Signal analysis usually constructs a disturbance transfer function between the cutting force and the drive current or spindle acceleration, as well as Kalman filter compensation to achieve the purpose of predicting the cutting force through the spindle current signal or spindle vibration signal. By using the above indirect signals to achieve cutting force prediction, the cost of cutting force measurement can be greatly reduced.

[0005] Regarding the indirect prediction of milling force, since the tool state is constantly changing during the machining process, the existing indirect prediction models such as current, acoustic emission, and spindle acceleration can usually only directly reflect the relationship between the cutting force and the measurement signal, and it is difficult to establish a complete relationship between the measurement signal, tool wear, and cutting force from a mechanistic point of view. This has greatly restricted the development of these monitoring methods in terms of versatility and extensibility. Taking into account the mature development of the mechanical model of power-cutting force and the easy acquisition of power signals, the present invention provides an indirect prediction method for milling force based on machine tool power signals and considering tool wear and variable working conditions, and can provide a basis for subsequent research and promotion to industrial applications. Summary of the invention

[0006] The purpose of the present invention is to avoid the shortcomings of the prior art and provide an indirect prediction method for milling force based on power signals and considering tool wear by establishing a milling force prediction model, which can realize the prediction of instantaneous cutting force under any cutting condition through real-time power information of the machine tool.

[0007] To achieve the above object, the technical solution adopted by the present invention is: an indirect prediction method of milling force considering tool wear based on power signal, comprising the following steps:

[0008] Step 1: Obtain cutting force data, machine tool power data and tool wear value during variable working condition milling processing, and establish data samples required for milling force prediction;

[0009] Step 2: extracting the time domain features of the variable working condition cutting force signal and the machine tool power signal from the data sample to establish a basic cutting state data set; then, according to the working condition corresponding to the basic cutting state data set, dividing the basic cutting state data set into a calibration set, a calculation set and a verification set;

[0010] Step 3: Based on the analysis of the basic cutting state data set, a cutting force mechanical model considering tool wear is established, and then a machine tool power consumption model based on instantaneous cutting force is established according to the calibration set divided by the basic cutting state data set;

[0011] Step 4: Use simulated annealing algorithm to solve the parameters in the machine tool power consumption model, and establish a variable working condition milling force prediction model based on the machine tool power consumption;

[0012] Step 5: Use the established variable working condition milling force prediction model to realize the variable working condition instantaneous milling force prediction.

[0013] Furthermore, the cutting force data in step 1 is selected from the cutting force signal data obtained by a desktop dynamometer within the time period from 1 second after the tool advances to 1 second before the tool retracts, so as to ensure the stability of the cutting force data;

[0014] The machine tool power data is obtained in real time through a power sensor installed in the external power supply circuit of the machine tool. The method for selecting the extraction range of the machine tool power data is the same as that for the cutting force data.

[0015] The tool wear value is the average value of the maximum wear width of the back face of each tooth of the tool participating in the cutting area.

[0016] Furthermore, the process of extracting the time domain feature of the variable working condition cutting force signal in the step 2 is as follows: selecting the last milling stroke before each offline measurement of tool wear, taking the average cutting force at each sampling point with the tooth angle as the interval within the tool rotation cycle from the middle 40% to 60% sample length of the cutting force signal from the tool cutting in to the cutting out as the cutting force, thereby obtaining the average cutting force on the cutter teeth of one cycle as the time domain feature of the cutting force signal, which is used to eliminate the influence of tool runout and instability factors in the machining process;

[0017] The process of extracting the time domain characteristics of the machine tool power signal is as follows: the last milling stroke before each offline measurement of tool wear is selected, and the middle 40% to 60% sample length of the machine tool power signal from the tool cutting in to the cutting out is used as a stable signal for time domain analysis to avoid the influence of the instability factors of the cutting in and out process, and the average value of the machine tool power signal is used as the time domain characteristics of the machine tool power signal;

[0018] The time domain characteristics of the cutting force signal and the machine tool power signal and the tool wear value are matched and combined with the process parameter conditions of different working conditions according to the timing logic of data acquisition, so as to establish the basic cutting state data set; wherein the process parameter conditions include spindle speed, feed speed and cutting depth;

[0019] Then, the basic cutting state dataset was randomly divided into calibration set, calculation set and validation set in a ratio of 1:3:5, and each dataset was not selected repeatedly.

[0020] Furthermore, in step three, the step of establishing the cutting force mechanical model considering tool wear is as follows:

[0021] Considering the contact mode between the tool flank and the workpiece, based on the cutting mechanism analysis, the instantaneous cutting force expression is obtained as follows:

[0022] ,

[0023] Where, dF j,t , dF j,r Respectively represent the instantaneous cutting force of the tangential and radial micro-element cutting edge; dF j,tc , dF j,rc Respectively represent the tangential and radial micro-element cutting edge shear forces; dF j,te , dF j,re They represent the tangential and radial micro-element cutting edge forces respectively; VB is the tool wear value; φ represents the position angle of the micro-element cutting edge; z represents the axial position of the micro-element cutting edge, and the integral range is from the cutting depth a p Decide;

[0024] Micro-element shear force dF j,qc and micro-edge force dFj,qe The specific calculation expression is:

[0025] ,

[0026] ,

[0027] In the formula, K qc They represent the tangential and radial shear force coefficients, i.e., q=t, r, and the directions of other physical quantities are defined in the same way as above; f z is the feed rate per tooth of the tool, which is determined by the spindle speed n and the feed speed f; F q (VB) represents the edge force per unit length of the flank surface;

[0028] Then the F q The calculation expression of (VB) is:

[0029] ,

[0030] Where RC is the length of the actual contact area of ​​the tool flank, which is determined by the tool wear value VB and the elastic contact length s; τ(x) and σ(x) are the normal stress and shear stress generated by the contact between the workpiece and the tool flank, respectively. The magnitudes of the normal stress and shear stress are determined by the maximum values ​​of the normal stress and shear stress τ0(x), σ0(x) and the action form of the microelement flank on the workpiece;

[0031] The expressions of normal stress τ(x) and shear stress σ(x) are:

[0032] ,

[0033] ,

[0034] In the formula, VB p is the width of the plastic contact area; when RC is less than or equal to the upper limit of the elastic contact area width VB*, only elastic contact occurs, that is, VB p The value is 0; when RC is greater than the upper limit of the elastic contact area width VB*, the elastic contact area no longer increases, the plastic contact area begins to appear and continues to increase with the wear value, VB p =RC-VB*;

[0035] That is, the establishment of the mechanical model of cutting force considering tool wear is completed;

[0036] Furthermore, the cutting power P cutting It is composed of two components: spindle rotation and feed. The expression of cutting power on the micro-element cutting edge is:

[0037] ,

[0038] Where, dP n is the spindle rotation power, dP f Is the spindle feed power;

[0039] Spindle rotation power dP n and spindle feed power dP f The calculation expression is:

[0040] ,

[0041] Since the machine power P total From the cutting power P cutting and machine idling power P idle The energy conversion efficiency η of the cutting process is calibrated by the calibration set divided from the basic cutting state data set in step 2; the energy consumption P of the machine tool when idling idle The fitting is realized through the power data of the machine tool idling operation; finally, the machine tool power is calculated, that is, the expression of the machine tool power consumption model based on the instantaneous cutting force is obtained:

[0042] .

[0043] Furthermore, the simulated annealing algorithm in step 4 solves the parameters in the machine tool power consumption model, specifically including the following steps:

[0044] The first step is to determine the decision variables as the parameters in the machine power consumption model, including the tangential shear force coefficient K tc , radial shear force coefficient K rc , maximum normal stress σ0, maximum shear stress τ0, upper limit of elastic contact area width VB * ;

[0045] The second step is to determine the constraints:

[0046] Determine the tangential shear force coefficient K tc , radial shear force coefficient K rc , the range of maximum normal stress σ0, maximum shear stress τ0; the upper limit of the width of the elastic contact area VB * The range is determined to be within the wear value range of normal tool use;

[0047] The third step is to determine the objective function:

[0048] The minimum value of the sum of squares of the error between the measured machine power and the power value calculated based on the machine power consumption model is:

[0049] ,

[0050] Where P measureFor measuring power, i refers to the i-th basic cutting state of the basic cutting state data set;

[0051] The fourth step is to optimize and solve the objective function:

[0052] First, set the initial temperature and select a temperature T0 higher than 100° to ensure that the solution space can be fully explored during the operation; within the constraints described in the second step, randomly select an initial solution x0 as the starting point of the algorithm and calculate its corresponding objective function value E(x0);

[0053] Secondly, set the solution set update strategy as: for the current solution x current , given the current solution x current Based on the change of 10% to 30% in the range width of the constraint condition, the current solution is updated to obtain the new solution x new , where, for the initial solution set update, x current =x0;

[0054] Then, the acceptance criteria are designed to calculate the objective function value E(x new ), and compared with the objective function value E(x current ) for comparison; According to the Metropolis criterion, if E(x new ) <E(x current ), then directly accept x new As the new current solution, that is, x current =x new ; otherwise, with probability exp{[E(x current )-E(x new )] / T}Accept x new ,This step allows the algorithm system to jump out of the local optimum with a certain probability;

[0055] Then, according to the preset cooling strategy, the current temperature value is updated to reduce the probability of the system accepting a poor solution and gradually approach the global optimal solution;

[0056] Finally, the above steps are repeated until the preset minimum temperature or maximum number of iterations is reached, and the algorithm stops iterating; at this time, the current solution x current That is, the optimal solution of the machine tool power consumption model parameters, and its corresponding objective function value E(x current ) shows the accuracy of the solution parameters; the solved machine tool power consumption model parameters are back-substituted to obtain the fully solved variable working condition milling force prediction model based on machine tool power consumption.

[0057] Furthermore, the step five specifically includes the following steps:

[0058] Firstly, the real-time machine power signal is obtained from the machine power circuit, and the real-time cutting power is calculated through the variable working condition milling force prediction model based on power consumption, and then the tool wear value is calculated;

[0059] Secondly, the real-time spindle speed, feed speed and cutting depth information are read from the CNC system of the machine tool, and according to the tool wear value, the real-time cutting force is finally calculated based on the variable working condition milling force prediction model of the machine tool power consumption.

[0060] The beneficial effects of the present invention are as follows: the present invention provides an indirect prediction method of milling force based on power signal considering tool wear. The method establishes a theoretical power consumption model and considers the influence of tool wear on cutting force. The influence relationship from tool wear to cutting force and then to machine tool power is established from the mechanism level. The method overcomes the shortcomings of other indirect monitoring signals in terms of universality and expansibility. The method of installing a power sensor in the external power supply circuit of the machine tool is adopted to avoid the limitation of the machining process by the dynamometer or other sensing devices and the interference of the machining environment.

[0061] At the same time, the present invention is relatively simple to install and maintain, and has higher feasibility in actual production line deployment. The method realizes accurate prediction of tool wear and cutting force under any working conditions by decoupling the power consumption components of the machine tool and using a simulated annealing algorithm to realize the correlation of cutting force characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is a flow chart of the present invention;

[0063] Figure 2 It is a schematic diagram of cutting mechanism analysis of cutting force mechanical modeling considering tool wear of the present invention;

[0064] Figure 3 It is a logic block diagram of a simulated annealing algorithm used in solving the parameters in the power consumption model of a machine tool according to the present invention;

[0065] Figure 4 It is a schematic diagram of the implementation effect of the method of the present invention;

[0066] Figure 4 (1) and (2) are schematic diagrams of the predicted effects of the initial and late wear of the C6 tool in the experimental group, respectively. Figure 4 (3) and (4) are schematic diagrams of the predicted effects of tool wear in the early and late stages of C7 wear in the experimental group, respectively. DETAILED DESCRIPTION

[0067] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0068] In order to achieve the above object, the present invention provides the following specific implementation methods:

[0069] Example 1: Figure 1 As shown, an indirect prediction method of milling force considering tool wear based on power signal includes the following steps:

[0070] S01. Obtain cutting force data, machine tool power data and tool wear value during variable working condition milling processing, and establish data samples required for milling force prediction;

[0071] Among them, the cutting force data is the cutting force signal data obtained by the desktop dynamometer during the period from 1 second after the tool enters the tool to 1 second before the tool retracts the tool, which is used to ensure the stability of the cutting force data;

[0072] The machine tool power data is obtained in real time through a power sensor installed in the external power circuit of the machine tool. The extraction range selection method of the machine tool power data is the same as that of the cutting force data.

[0073] The tool wear value is the average value of the maximum wear width of the back face of each tooth of the tool participating in the cutting area.

[0074] S02. Extract the time domain features of the cutting force signal under variable working conditions: select the last milling stroke before each offline measurement of tool wear, and take the average cutting force at each sampling point with the tooth angle as the interval within the tool rotation cycle from the middle 40% to 60% sample length of the cutting force signal from the tool cutting in to the cutting out as the cutting force. In this way, the average cutting force on the cutter teeth of one cycle is obtained as the time domain feature of the cutting force signal, which is used to eliminate the influence of tool runout and instability factors in the machining process;

[0075] Extract the time domain features of the machine tool power signal: select the last milling stroke before each offline measurement of tool wear, and use the middle 40% to 60% sample length of the machine tool power signal from tool entry to cutout as a stable signal for time domain analysis to avoid the influence of instability factors in the cut-in and cut-out process, and use the average value of the machine tool power signal as the time domain feature of the machine tool power signal;

[0076] The time domain characteristics of the cutting force signal and the machine tool power signal and the tool wear value are matched and combined with the process parameter conditions of different working conditions according to the timing logic of data acquisition, so as to establish the basic cutting state data set; wherein the process parameter conditions include spindle speed, feed speed and cutting depth;

[0077] Thus, a basic cutting state data set is established;

[0078] S03. According to the working conditions corresponding to the basic cutting state data set, the basic cutting state data set is randomly divided into a calibration set, a calculation set and a verification set in a ratio of 1:3:5, and each data set is not selected repeatedly.

[0079] S04. Based on the analysis of the basic cutting state data set, a cutting force mechanical model considering tool wear is established, specifically:

[0080] Considering the contact mode between the tool flank and the workpiece, based on the cutting mechanism analysis, the instantaneous cutting force expression is obtained as follows:

[0081] ,

[0082] Where, dF j,t , dF j,r Respectively represent the instantaneous cutting force of the tangential and radial micro-element cutting edge; dF j,tc , dF j,rc Respectively represent the tangential and radial micro-element cutting edge shear forces; dF j,te , dF j,re They represent the tangential and radial micro-element cutting edge forces respectively; VB is the tool wear value; φ represents the position angle of the micro-element cutting edge; z represents the axial position of the micro-element cutting edge, and the integral range is from the cutting depth a p Decide;

[0083] Micro-element shear force dF j,qc and micro-edge force dF j,qe The specific calculation expression is:

[0084] ,

[0085] ,

[0086] In the formula, K qc They represent the tangential and radial shear force coefficients, i.e., q=t, r, and the directions of other physical quantities are defined in the same way as above; f z is the feed rate per tooth of the tool, which is determined by the spindle speed n and the feed speed f; F q (VB) represents the edge force per unit length of the flank surface;

[0087] Then the F q The calculation expression of (VB) is:

[0088] ,

[0089] Where RC is the length of the actual contact area of ​​the tool flank, which is determined by the tool wear value VB and the elastic contact length s; τ(x) and σ(x) are the normal stress and shear stress generated by the contact between the workpiece and the tool flank, respectively. The magnitudes of the normal stress and shear stress are determined by the maximum values ​​of the normal stress and shear stress τ0(x), σ0(x) and the action form of the microelement flank on the workpiece;

[0090] The expressions of normal stress τ(x) and shear stress σ(x) are:

[0091] ,

[0092] ,

[0093] In the formula, VB p is the width of the plastic contact area; when RC is less than or equal to the upper limit of the elastic contact area width VB*, only elastic contact occurs, that is, VB p The value is 0; when RC is greater than the upper limit of the elastic contact area width VB*, the elastic contact area no longer increases, the plastic contact area begins to appear and continues to increase with the wear value, VB p =RC-VB*;

[0094] That is to say, the establishment of the cutting force mechanical model considering tool wear is completed.

[0095] S05. Based on the calibration set divided by the basic cutting state data set, a machine tool power consumption model based on instantaneous cutting force is established, specifically:

[0096] Cutting power P cutting It is composed of two components: spindle rotation and feed. The expression of cutting power on the micro-element cutting edge is:

[0097] ,

[0098] Where, dP n is the spindle rotation power, dP f Is the spindle feed power;

[0099] Spindle rotation power dP n and spindle feed power dP f The calculation expression is:

[0100] ,

[0101] Since the machine power P total From the cutting power P cutting and machine idling power P idleThe energy conversion efficiency η of the cutting process is calibrated by the calibration set divided from the basic cutting state data set in step 2; the energy consumption P of the machine tool when idling idle The fitting is realized through the power data of the machine tool idling operation; finally, the machine tool power is calculated, that is, the expression of the machine tool power consumption model based on the instantaneous cutting force is obtained:

[0102] .

[0103] S06. Based on the simulated annealing algorithm, the actual decision variables of the algorithm are set as the parameters in the machine tool power consumption model, including the tangential shear force coefficient K tc , radial shear force coefficient K rc , maximum normal stress σ0, maximum shear stress τ0, upper limit of elastic contact area width VB * .

[0104] S07. Determine the constraints:

[0105] Determine the tangential shear force coefficient K tc , radial shear force coefficient K rc , the range of maximum normal stress σ0, maximum shear stress τ0; the upper limit of the width of the elastic contact area VB * The range is determined within the wear value range of normal tool use.

[0106] S08. Determine the objective function:

[0107] The minimum value of the sum of squares of the error between the measured machine power and the power value calculated based on the machine power consumption model is:

[0108] ,

[0109] Where P measure For measuring power, i refers to the i-th basic cutting state of the basic cutting state data set.

[0110] S09. Set the initial temperature and select a temperature T0 higher than 100° to ensure that the solution space can be fully explored during the operation; within the constraints described in the second step, randomly select an initial solution x0 as the starting point of the algorithm and calculate the corresponding objective function value E(x0).

[0111] S10, set the solution set update strategy to: for the current solution x current , given the current solution x current Based on the change of 10% to 30% in the range width of the constraint condition, the current solution is updated to obtain the new solution x new , where, for the initial solution set update, x current =x0.

[0112] S11. Design acceptance criteria and calculate the objective function value E(x new ), and compared with the objective function value E(x current ) for comparison; According to the Metropolis criterion, if E(x new ) <E(x current ), then directly accept x new As the new current solution, that is, x current =x new ; otherwise, with probability exp{[E(x current )-E(x new )] / T}Accept x new , this step allows the algorithm system to jump out of the local optimum with a certain probability.

[0113] S12. Update the current temperature value according to the preset cooling strategy to reduce the probability of the system accepting a poor solution and gradually approach the global optimal solution.

[0114] S13, repeat the above steps until the preset minimum temperature or maximum number of iterations is reached, and the algorithm stops iterating; at this time, the current solution x current That is, the optimal solution of the machine tool power consumption model parameters, and its corresponding objective function value E(x current ) shows the accuracy of the solution parameters; the solved machine tool power consumption model parameters are back-substituted to obtain the fully solved variable working condition milling force prediction model based on machine tool power consumption.

[0115] S14, obtaining a real-time machine tool power signal from a machine tool power circuit, calculating the real-time cutting power through a variable working condition milling force prediction model based on power consumption, and then calculating the tool wear value.

[0116] S15, reading the real-time spindle speed, feed speed and cutting depth information from the numerical control system of the machine tool, and finally calculating the real-time cutting force size based on the variable working condition milling force prediction model of the machine tool power consumption according to the tool wear value.

[0117] The following is a verification explanation combined with a specific application example. The example given is a VMC-850 three-axis slot milling experiment of an end mill, and the workpiece material is Ti6Al4V. The cutting force data is measured in real time by a Kistler9119A installed on the machine tool table, and the machine tool power consumption data is obtained by a power sensor YOKOGAWA WT1803 connected to the machine tool's external power supply circuit. The tool wear value is measured by regularly stopping the machine to unload the tool and using an optical microscope Alicona IF-EdgeMaster.

[0118] Step 1: Create a data sample:

[0119] The process parameters involved in the variable working condition milling process include spindle speed n, cutting depth a p As well as the feed speed f, the tool wear value is the average value of the maximum wear width of the back face of each tooth of the tool participating in the cutting area at the current moment.

[0120] Milling experiments were conducted under different working conditions, and the tool wear was measured over the entire tool life (0-300 μm) to ensure that a wide range of working conditions were covered, as shown in Table 1 below:

[0121] Table 1 Cutting parameter settings

[0122]

[0123] Step 2: Dataset division:

[0124] To extract the time domain features of the cutting force signal under variable working conditions, the last milling stroke before each offline measurement of tool wear is selected. The average cutting force at each sampling point with the tooth angle as the interval within the tool rotation cycle is taken from the middle 40% to 60% sample length of the cutting force signal from the tool cutting in to the tool cutting out as the cutting force. The average cutting force on the cutter teeth in one cycle is obtained as the time domain feature of the cutting force signal, which is used to eliminate the influence of tool runout and instability factors in the machining process.

[0125] To extract the time domain features of the machine tool power signal, the last milling stroke before each offline measurement of tool wear is selected. The middle 40% to 60% sample length of the machine tool power signal from the time when the tool cuts in to the time when the tool cuts out is used as a stable signal for time domain analysis to avoid the influence of unstable factors in the process of cutting in and out, and the average value of the machine tool power signal is used as the time domain feature of the machine tool power signal.

[0126] The time domain characteristics of the cutting force signal and the machine tool power signal and the tool wear value are matched and combined with the process parameter conditions of different working conditions according to the timing logic of data acquisition, so as to establish the basic cutting state data set; considering the size of the process parameter domain and the tool wear data, C5 is selected as the calibration set; among the remaining 8 groups of data, 3 groups (C1, C4, and C9 in this embodiment) are randomly selected as calculation groups for training the model to determine the unknown parameters in the prediction model. The remaining 5 groups of data are used as verification sets to verify the solution effect (C2, C6, and C8 in this embodiment) and the prediction of tool wear and instantaneous cutting force (C6 and C7 in this embodiment), which are used to evaluate the prediction performance of the training model on new samples.

[0127] Step 3: Establish a power consumption model considering tool wear:

[0128] The first part is to establish an instantaneous cutting force model that takes into account tool wear. Considering the contact mode between the tool back face and the workpiece, based on the cutting mechanism analysis, the instantaneous cutting force expression is obtained as follows:

[0129] ,

[0130] Where, dF j,t , dF j,r Respectively represent the instantaneous cutting force of the tangential and radial micro-element cutting edge; dF j,tc , dF j,rc Respectively represent the tangential and radial micro-element cutting edge shear forces; dF j,te , dF j,re They represent the tangential and radial micro-element cutting edge forces respectively; VB is the tool wear value; φ represents the position angle of the micro-element cutting edge; z represents the axial position of the micro-element cutting edge, and the integral range is from the cutting depth a p Decide;

[0131] Micro-element shear force dF j,qc and micro-edge force dF j,qe The specific calculation expression is:

[0132] ,

[0133] ,

[0134] In the formula, K qc They represent the tangential and radial shear force coefficients, i.e., q=t, r, and the directions of other physical quantities are defined in the same way as above; f z is the feed rate per tooth of the tool, which is determined by the spindle speed n and the feed speed f; F q (VB) represents the edge force per unit length of the flank surface;

[0135] The F q The calculation expression of (VB) is:

[0136] ,

[0137] Among them, considering that there is a certain rebound area between the tool and the workpiece, which is determined by the tool wear value VB and the elastic contact length s, the actual contact length between the two is represented by RC for integration operation, and the expression is:

[0138] ,

[0139] Where k1=34 is the empirical model coefficient, H=349GPa is the Vickers hardness of the workpiece material, and E=165000ksi is the elastic modulus of the workpiece material. Therefore, in this example, the calculation relationship between RC and VB is:

[0140] ,

[0141] τ(x) and σ(x) are the normal stress and shear stress generated by the contact between the workpiece and the tool flank, respectively. The normal stress and shear stress are determined by the maximum values ​​of the normal stress and shear stress, τ0(x) and σ0(x), respectively, and the action form of the micro-element flank on the workpiece.

[0142] The expressions of normal stress τ(x) and shear stress σ(x) are:

[0143] ,

[0144] ,

[0145] In the formula, VB p is the width of the plastic contact area, such as Figure 2 As shown in area A; in addition, Figure 2 The middle area B represents the elastic contact area, I represents the chip, II represents the tool rake face, and III represents the tool flank face; when the tool wear value exceeds the upper limit of the elastic contact area width VB*, the elastic contact area no longer increases, and the plastic contact area begins to appear and continues to increase with the increase of the wear value. Therefore, VB p It can be expressed as:

[0146] ,

[0147] Based on the above equations, a cutting force mechanical model considering tool wear was established.

[0148] The second part of step 3 is to establish a machine power consumption model based on the instantaneous cutting force, and then the cutting power P cutting It is composed of two components: spindle rotation and feed. The expression of cutting power on the micro-element cutting edge is:

[0149] ,

[0150] Where, dP n is the spindle rotation power, dP f Is the spindle feed power;

[0151] Spindle rotation power dP n and spindle feed power dP f The calculation expression is:

[0152] ,

[0153] ,

[0154] Total power consumption of machine tool Ptotal From the cutting power P cutting and machine idling power P idle It consists of two parts, among which P idle Regarding the fitting of the machine tool idling speed, the results are as follows:

[0155] ,

[0156] Considering that there is a certain loss in energy conversion during the cutting process, the actual cutting power P acting on the cutting process is cutting The cutting power P allocated to the cutting part of the total power consumption of the machine tool additional There is a certain deviation. Therefore, this method calibrates the loss deviation by setting a conversion efficiency coefficient η, and fits the calibration set divided from the basic cutting state data set in step 2 to obtain η=0.81, which is expressed as:

[0157] ,

[0158] Finally, by summing the two parts of power, the total power consumption of the machine tool is obtained, and the expression is:

[0159] ,

[0160] That is to realize the machine tool power consumption modeling based on instantaneous cutting force.

[0161] Step 4: Use the simulated annealing algorithm to solve the parameters in the machine tool power consumption model and establish a variable working condition milling force prediction model based on the machine tool power consumption, which specifically includes the following processes:

[0162] The first step is to determine the decision variables as the parameters in the machine power consumption model, including the tangential shear force coefficient K tc , radial shear force coefficient K rc , maximum normal stress σ0, maximum shear stress τ0, upper limit of elastic contact area width VB * ;

[0163] The second step is to determine the constraints, which are:

[0164] ,

[0165] The third step is to determine the objective function:

[0166] Minimum value of the sum of squares of the errors between the measured power and the power calculated according to the model:

[0167] ,

[0168] The fourth step is to optimize and solve the objective function:

[0169] First, set the initial temperature and select a temperature T0 higher than 100° to ensure that the solution space can be fully explored during the operation; within the constraints described in the second step, randomly select an initial solution x0 as the starting point of the algorithm and calculate its corresponding objective function value E(x0);

[0170] Secondly, set the solution set update strategy as: for the current solution x current , given the current solution x current Based on the change of 30% of the range width of the constraint condition, the current solution is updated to obtain the new solution x new , where, for the initial solution set update, x current =x0;

[0171] Design acceptance criteria and calculate the objective function value E(x new ), and compared with the objective function value E(x current ) for comparison. According to the Metropolis criterion, if E(x new ) <E(x current ), then directly accept x new As the new current solution, that is, x current =x new Otherwise, with probability exp{[E(x current )- E(x new )] / T}Accept x new This step allows the algorithm system to jump out of the local optimum with a certain probability. In the same temperature level, m=50 repeated searches are set to ensure sufficient search at each temperature.

[0172] The algorithm's selection decision during the search process is directly affected by the current temperature value T. Differences in cooling rate and mode may directly affect the convergence speed, accuracy and stability of the algorithm. After comparing exponential cooling, linear cooling and logarithmic cooling strategies, this method selected the exponential cooling strategy, and the function is as follows:

[0173] ,

[0174] According to the preset cooling strategy, the current temperature value is updated to reduce the probability of the system accepting a poor solution and gradually approach the global optimal solution. Finally, the above steps are repeated until the preset maximum number of iterations N=800 is reached. When the preset termination condition is reached, the algorithm stops iterating. At this time, the current solution x current This is the optimal solution found by the algorithm, and its corresponding objective function value E(x current ) indicates the accuracy of the solution parameters.

[0175] In this embodiment, the above algorithm is used to solve the following K based on the cutting force sample data of the calculation group:tc =1872.1 N∙mm -2 , K tc =722.5 N∙mm -2 , σ0=0.3697 GPa, τ0=0.8530 GPa, VB * =131.1 μm. Substituting the above model parameters back, we can get the complete solution of the cutting force prediction model. The prediction effect is as follows Figure 4 shown.

[0176] The comparison between the prediction results and the measured data shows that the method proposed in this application provides an economical and efficient solution for predicting cutting forces and monitoring tool wear. In addition, the proposed monitoring method of replacing the dynamometer with a power sensor shows significant potential and important value in industrial production applications. This method provides strong technical support for the intelligent and efficient transformation of the machining process.

[0177] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An indirect prediction method of milling force considering tool wear based on power signal, characterized in that: The following steps are involved: Step 1: Obtain cutting force data, machine tool power data and tool wear value during variable working condition milling processing, and establish data samples required for milling force prediction; Step 2: extracting the time domain features of the variable working condition cutting force signal and the machine tool power signal from the data sample to establish a basic cutting state data set; then, according to the working condition corresponding to the basic cutting state data set, dividing the basic cutting state data set into a calibration set, a calculation set and a verification set; Step 3: Based on the analysis of the basic cutting state data set, a cutting force mechanical model considering tool wear is established, and then according to the calibration set divided by the basic cutting state data set, a machine tool power consumption model based on instantaneous cutting force is established; The steps of establishing the cutting force mechanical model considering tool wear are as follows: Considering the contact mode between the tool flank and the workpiece, based on the cutting mechanism analysis, the instantaneous cutting force expression is obtained as follows: ， Where, dF j,t , dF j,r Respectively represent the instantaneous cutting force of the tangential and radial micro-element cutting edge; dF j,tc , dF j,rc Respectively represent the tangential and radial micro-element cutting edge shear forces; dF j,te , dF j,re They represent the tangential and radial micro-element cutting edge forces respectively; VB is the tool wear value; φ represents the position angle of the micro-element cutting edge; z represents the axial position of the micro-element cutting edge, and the integral range is from the cutting depth a p Decide; Micro-element shear force dF j,qc and micro-edge force dF j,qe The specific calculation expression is: ， ， In the formula, K qc They represent the tangential and radial shear force coefficients, i.e., q=t, r, and the directions of other physical quantities are defined in the same way as above; f z is the feed rate per tooth of the tool, which is determined by the spindle speed n and the feed speed f; F q (VB) represents the edge force per unit length of the flank surface; Then the F q The calculation expression of (VB) is: ， Where RC is the length of the actual contact area of ​​the tool flank, which is determined by the tool wear value VB and the elastic contact length s; τ(x) and σ(x) are the normal stress and shear stress generated by the contact between the workpiece and the tool flank, respectively. The magnitudes of the normal stress and shear stress are determined by the maximum values ​​of the normal stress and shear stress τ0(x), σ0(x) and the action form of the microelement flank on the workpiece; The expressions of normal stress τ(x) and shear stress σ(x) are: ， ， In the formula, VB p is the width of the plastic contact area; when RC is less than or equal to the upper limit of the elastic contact area width VB*, only elastic contact occurs, that is, VB p The value is 0; when RC is greater than the upper limit of the elastic contact area width VB*, the elastic contact area no longer increases, the plastic contact area begins to appear and continues to increase with the wear value, VB p =RC-VB*; That is, the establishment of the mechanical model of cutting force considering tool wear is completed; Furthermore, the cutting power P cutting It is composed of two components: spindle rotation and feed. The expression of cutting power on the micro-element cutting edge is: ， Where, dP n is the spindle rotation power, dP f Is the spindle feed power; Spindle rotation power dP n and spindle feed power dP f The calculation expression is: ， Since the machine power P total From the cutting power P cutting and machine idling power P idle The energy conversion efficiency η of the cutting process is calibrated by the calibration set divided from the basic cutting state data set in step 2; the energy consumption P of the machine tool when idling idle The fitting is realized through the power data of the machine tool idling operation; finally, the machine tool power is calculated, that is, the expression of the machine tool power consumption model based on the instantaneous cutting force is obtained: ； Step 4: Use simulated annealing algorithm to solve the parameters in the machine tool power consumption model, and establish a variable working condition milling force prediction model based on the machine tool power consumption; The simulated annealing algorithm described in step 4 solves the parameters in the machine tool power consumption model, specifically including the following steps: The first step is to determine the decision variables as the parameters in the machine power consumption model, including the tangential shear force coefficient K tc , radial shear force coefficient K rc , maximum normal stress σ0, maximum shear stress τ0, upper limit of elastic contact area width VB * ; The second step is to determine the constraints: Determine the tangential shear force coefficient K tc , radial shear force coefficient K rc , the range of maximum normal stress σ0, maximum shear stress τ0; the upper limit of the width of the elastic contact area VB * The range is determined to be within the wear value range of normal tool use; The third step is to determine the objective function: The minimum value of the sum of squares of the error between the measured machine power and the power value calculated based on the machine power consumption model is: ， Where P measure For measuring power, i refers to the i-th basic cutting state of the basic cutting state data set; The fourth step is to optimize and solve the objective function: First, set the initial temperature and select a temperature T0 higher than 100° to ensure that the solution space can be fully explored during the operation; within the constraints described in the second step, randomly select an initial solution x0 as the starting point of the algorithm and calculate its corresponding objective function value E(x0); Secondly, set the solution set update strategy as: for the current solution x current , given the current solution x current Based on the change of 10% to 30% in the range width of the constraint condition, the current solution is updated to obtain the new solution x new , where, for the initial solution set update, x current =x0; Then, the acceptance criteria are designed to calculate the objective function value E(x new ), and compared with the objective function value E(x current ) for comparison; According to the Metropolis criterion, if E(x new ) <E(x current ), then directly accept x new As the new current solution, that is, x current =x new ; otherwise, with probability exp{[E(x current )-E(x new )] / T}Accept x new ,This step allows the algorithm system to jump out of the local optimum with a certain probability; Then, according to the preset cooling strategy, the current temperature value is updated to reduce the probability of the system accepting a poor solution and gradually approach the global optimal solution; Finally, the above steps are repeated until the preset minimum temperature or maximum number of iterations is reached, and the algorithm stops iterating; at this time, the current solution x current That is, the optimal solution of the machine tool power consumption model parameters, and its corresponding objective function value E(x current ) shows the accuracy of the solution parameters; the solved machine power consumption model parameters are substituted back to obtain the fully solved variable working condition milling force prediction model based on machine power consumption; Step 5: Use the established variable working condition milling force prediction model to realize the variable working condition instantaneous milling force prediction.

2. The indirect prediction method of milling force considering tool wear based on power signal according to claim 1, characterized in that: The cutting force data in step 1 is selected from the cutting force signal data obtained by a desktop dynamometer during the period from 1 second after the tool advances to 1 second before the tool retracts, so as to ensure the stability of the cutting force data; The machine tool power data is obtained in real time through a power sensor installed in the external power supply circuit of the machine tool. The method for selecting the extraction range of the machine tool power data is the same as that for the cutting force data. The tool wear value is the average value of the maximum wear width of the back face of each tooth of the tool participating in the cutting area.

3. The indirect prediction method of milling force considering tool wear based on power signal as claimed in claim 1, characterized in that: The process of extracting the time domain feature of the variable working condition cutting force signal in step 2 is as follows: selecting the last milling process stroke before each offline measurement of tool wear, taking the average cutting force at each sampling point with the tooth angle as the interval within the tool rotation cycle from the middle 40% to 60% sample length of the cutting force signal from the tool cutting in to the cutting out as the cutting force, thereby obtaining the average cutting force on the tool teeth of one cycle as the time domain feature of the cutting force signal, which is used to eliminate the influence of tool runout and instability factors in the machining process; The process of extracting the time domain characteristics of the machine tool power signal is as follows: the last milling stroke before each offline measurement of tool wear is selected, and the middle 40% to 60% sample length of the machine tool power signal from the tool cutting in to the cutting out is used as a stable signal for time domain analysis to avoid the influence of the instability factors of the cutting in and out process, and the average value of the machine tool power signal is used as the time domain characteristics of the machine tool power signal; The time domain characteristics of the cutting force signal and the machine tool power signal and the tool wear value are matched and combined with the process parameter conditions of different working conditions according to the timing logic of data acquisition, so as to establish the basic cutting state data set; wherein the process parameter conditions include spindle speed, feed speed and cutting depth; Then, the basic cutting state dataset was randomly divided into calibration set, calculation set and validation set in a ratio of 1:3:5, and each dataset was not selected repeatedly.

4. The indirect prediction method of milling force based on power signal considering tool wear according to any one of claims 1 to 3, characterized in that: The step five specifically includes the following steps: Firstly, the real-time machine power signal is obtained from the machine power circuit, and the real-time cutting power is calculated through the variable working condition milling force prediction model based on power consumption, and then the tool wear value is calculated; Secondly, the real-time spindle speed, feed speed and cutting depth information are read from the CNC system of the machine tool, and according to the tool wear value, the real-time cutting force is finally calculated based on the variable working condition milling force prediction model of the machine tool power consumption.

Images