Method for predicting thermal error of spindle of computer numerical control machine tool on basis of twin feature transfer between virtual sample machine tool and physical sample machine tool
Through the twin feature migration method of virtual and real prototypes and combined with the gray system theory, a CNC machine tool spindle thermal error prediction model is established, which solves the problems of limited space and difficult sensor layout, and realizes high-precision spindle thermal error prediction, which improves machining accuracy.
Patent Information
- Application Number
- PCT/CN2023/139877
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-06-01
- Filing Date
- 2023-12-19
- Publication Date
- 2024-12-05
AI Technical Summary
The existing CNC machine tool spindle thermal error prediction methods cannot guarantee prediction accuracy when space is limited and sensor arrangement is difficult, especially because multiple temperature sensors cannot be arranged near the spindle and thermal error cannot be measured by displacement sensors, which affects the machining accuracy.
Using the method of transferring twin features of virtual and real prototypes, the temperature sensitive points are screened by building a physical prototype experimental bench, and the relationship between the autoregressive distribution of thermal error and the temperature hysteresis point is established. The virtual prototype model is constructed, and the twin features are migrated to determine the temperature sensitive points of virtual prototypes. Combined with the theory of gray system, a spindle thermal error prediction model is established, and the real value is measured by the displacement sensor to optimize the prediction accuracy.
Without the need to arrange multiple temperature sensors and displacement sensors, the accuracy of thermal error prediction of CNC machine tool spindle is improved, and it is suitable for machine tool environments with limited space, improving prediction accuracy, and solving the accuracy problem of traditional methods.
Smart Images

Figure CN2023139877_05122024_PF_FP_ABST
Abstract
Description
Thermal error prediction method for CNC machine tool spindle based on twin feature transfer of virtual and real prototypes
[0001] This application claims priority to the Chinese patent application filed with the Patent Office of China on June 1, 2023, with application number 202310643624.4 and invention name “CNC machine tool spindle thermal error prediction method based on virtual and real prototype twin feature migration”, the entire contents of which are incorporated into this application by reference. Technical Field
[0002] The present invention belongs to the technical field of precision machining machine tools, and relates to a method for predicting thermal errors of a CNC machine tool spindle, and in particular to a method for predicting thermal errors of a CNC machine tool spindle based on twin feature migration of virtual and real prototypes. Background Art
[0003] The rapid development of the manufacturing industry and the rapid growth in demand for precision parts production are placing higher demands on the machining accuracy of CNC machine tools. During the actual machining process of CNC machine tools, errors such as geometric errors, thermal errors, and errors caused by cutting forces affect the machining accuracy of machine tools. Among the many error factors, thermal errors caused by factors such as heat generated by bearing friction and ambient temperature account for 40%-70% of the total machine tool error and are the most important factor affecting machine tool machining accuracy. The spindle is the core component that determines the machining accuracy of machine tools. The large amount of heat generated when the machine tool is running at high speed can cause the spindle's center of rotation to drift, increase the turning radius, and other problems. Predicting the thermal error of the machine tool spindle can provide data reference for the subsequent error compensation stage and is of great significance.
[0004] The prediction of thermal error of machine tool spindle mainly includes three steps: selecting temperature sensitive points, establishing thermal error prediction model, and calculating model prediction performance. Existing research selects several temperature sensitive points from the temperature sensors arranged outside the spindle; uses the selected temperature sensitive points to model the thermal error of the spindle; and compares the model prediction value with the actual value to calculate the accuracy of the prediction method. However, the existing machine tool spindle thermal error prediction method has the following problems: (1) The space near the machine tool spindle is limited, and it is not possible to arrange a large number of temperature sensors for screening temperature sensitive points; (2) The spindle end of the machine tool is tightly wrapped, and it is impossible to measure the spindle thermal error for thermal error modeling by displacement sensor. Under the condition that it is difficult to arrange the spindle sensor of CNC machine tools, the prediction accuracy of the existing spindle thermal error prediction method cannot be guaranteed.
[0005] Summary of the Invention
[0006] To address the problems presented in the prior art, the present invention proposes a CNC machine tool spindle thermal error prediction method based on twin feature transfer between virtual and real prototypes. This method improves the accuracy of CNC machine tool spindle thermal error prediction and eliminates the need for displacement sensors on the machine tool. This method overcomes the shortcomings of existing methods and improves the accuracy of spindle thermal error prediction, even when placement of sensors on CNC machine tool spindles is difficult.
[0007] To achieve the above object, the technical solution of the present invention is as follows:
[0008] S1: Build a spindle physical prototype test bench, evenly set multiple measuring points outside the physical prototype spindle, screen out temperature-sensitive points outside the physical prototype spindle from the multiple measuring points, and establish a thermal error autoregressive distributed hysteresis model;
[0009] S2: Determine the temperature stagnation point of the physical prototype spindle at the temperature sensitive point outside the physical prototype spindle, establish a temperature relationship between the temperature sensitive point outside the physical prototype spindle and the temperature stagnation point on the physical prototype spindle based on the grey system theory and record it as a first temperature relationship, and construct a physical prototype spindle thermal error analysis model based on the first temperature relationship, using the temperature of the temperature stagnation point on the physical prototype spindle as a thermal error feature;
[0010] S3: Establish a virtual prototype model of the machine tool. Use the temperature stagnation point on the physical prototype spindle as a twin feature and migrate it to the virtual prototype spindle of the machine tool to obtain the temperature migration point of the virtual prototype spindle. Then, determine the position of each temperature-sensitive point outside the virtual prototype spindle. Then, based on the grey system theory, establish the temperature relationship between the temperature migration point on the virtual prototype spindle and each temperature-sensitive point outside the virtual prototype spindle and record it as the second temperature relationship. Then, combine the twin coupling relationship between the physical prototype spindle and the virtual prototype spindle, the physical prototype spindle thermal error analysis model, and the second temperature relationship to construct a CNC machine tool spindle thermal error prediction model based on the virtual and real prototype twin feature migration.
[0011] S4: Adjust the parameters of the physical prototype test bench so that it always operates in the same working conditions as the machine tool. The thermal error of the physical prototype spindle measured by the displacement sensor is integrated with the twin coupling relationship in S3 to obtain the true value of the thermal error of the machine tool spindle. Then, based on the temperature-sensitive points outside the virtual prototype spindle, temperature sensors are arranged at corresponding positions outside the machine tool spindle. The collected temperature data is input into the CNC machine tool spindle thermal error prediction model in real time to obtain the predicted value of the machine tool spindle thermal error. Finally, the prediction accuracy is calculated based on the true value and predicted value of the machine tool spindle thermal error. The CNC machine tool spindle thermal error prediction model is optimized based on the prediction accuracy to obtain a more accurate predicted value of the machine tool spindle thermal error.
[0012] Compared with the prior art, the present invention has the following beneficial effects:
[0013] 1) The present invention arranges a relatively small number of temperature sensors on a machine tool, and is suitable for arranging temperature sensors on a machine tool with limited space near the spindle;
[0014] 2) Based on the twin features of virtual and real prototype migration, the machine tool spindle thermal error prediction model established in this invention is not affected by the different structures between the physical prototype test bench and the machine tool, as well as the layout of the temperature sensors, and has high prediction accuracy;
[0015] 3) The present invention does not require the placement of displacement sensors on the machine tool when calculating the prediction accuracy of the thermal error model of the machine tool spindle, thereby solving the accuracy calculation problem of the traditional machine tool spindle thermal error prediction method.
[0016] Figures in the specification
[0017] FIG1 is a diagram illustrating key parameters of thermal error prediction according to the present invention.
[0018] FIG2 is a flow chart of the method of the present invention.
[0019] Figure 3 is a flow chart for selecting temperature sensitive points outside the spindle of the physical prototype.
[0020] Figure 4 is a flow chart for establishing a thermal error analysis model for the physical prototype spindle.
[0021] Figure 5 is a performance calculation flow chart of the CNC machine tool spindle thermal error prediction method. DETAILED DESCRIPTION
[0022] The present invention will be further described below with reference to the accompanying drawings and specific examples:
[0023] As shown in Figures 1 and 2, the following steps are included:
[0024] S1: Build a spindle physical prototype test bench, evenly set multiple measuring points outside the physical prototype spindle, screen out temperature-sensitive points outside the physical prototype spindle from the multiple measuring points, and establish a thermal error autoregressive distributed hysteresis model.
[0025] S1 is specifically:
[0026] S11: Build a spindle physical prototype test bench, evenly set 10-20 measuring points outside the physical prototype spindle, and arrange corresponding temperature sensors at each measuring point to record the temperature changes at each measuring point.
[0027] S12: Establish a finite element model of the physical prototype test bench, and use the simulation of the working condition mutation process to preliminarily screen out the advanced measuring points outside the physical prototype spindle that have a faster temperature mutation response than a thermal error mutation response when the working condition mutation occurs based on the finite element model of the physical prototype test bench.
[0028] S13: A thermal hysteresis clustering algorithm is used to select several temperature-sensitive points outside the physical prototype spindle from the advanced measurement points outside the physical prototype spindle. The thermal hysteresis effect and the self-memory of the thermal error are combined to establish a thermal error autoregressive distributed hysteresis model.
[0029] As shown in FIG3 , a thermal hysteresis clustering algorithm is used to select several temperature sensitive points outside the spindle of the physical prototype from the advanced measurement points outside the spindle of the physical prototype, specifically:
[0030] First, calculate the average response time Δt of each leading measurement point under each sudden change of working condition. r , r=1,2,…s, s represents the number of leading measurement points, and the calculation formula is as follows:
[0031] Where Δt rq It represents the response time of the rth leading measurement point under the qth operating condition, q = 1, 2, ... N, where N represents the number of sudden changes in the operating condition.
[0032] Then, each advanced measurement point is divided into k categories according to the average response time, k∈[k min ,k max ], k min represents the minimum number of clusters, k max Indicates the maximum number of clusters, which is determined based on experience. max and minimum average response time Δt min The categories of the leading measuring points are determined according to the following formula:
[0033] Then the average of the average response time of all the advanced measurement points in each type of advanced measurement point is taken as the cluster center c of this type of advanced measurement point. l , l = 1, 2, ..., k, and then calculate the sum of squares of errors under the current number of clusters according to the cluster centers of various types of advanced measurement points; change the number of clusters k, and calculate the sum of squares of errors under different numbers of clusters, that is, the sum of squares of errors of the response time of all measurement points under different k values. In the specific implementation, first let k = k min , calculate and divide the measurement points into k min The sum of square errors of the response time of all measurement points in the class. Increase the value of k until k = k max, sequentially calculating the sum of squared errors in the response time of all measurement points for different values of k. A graph is then plotted against the number of clusters k and the corresponding sum of squared errors. The number of clusters at which the sum of squared errors decreases sharply in the graph is used as the target number of clusters. Each advanced measurement point is clustered according to the target number of clusters to obtain target clusters. The advanced measurement point with the highest correlation within each target cluster is then designated as a temperature-sensitive point outside the physical prototype spindle. In specific implementations, the correlation refers to the correlation between the temperature of each advanced measurement point and the true value of the thermal error of the physical prototype spindle.
[0034] S2: Determine the temperature stagnation point of the temperature sensitive point outside the physical prototype spindle on the physical prototype spindle, establish the temperature relationship between the temperature sensitive point outside the physical prototype spindle and the temperature stagnation point on the physical prototype spindle based on the grey system theory and record it as the first temperature relationship, and construct a physical prototype spindle thermal error analysis model based on the first temperature relationship, using the temperature of the temperature stagnation point on the physical prototype spindle as the thermal error feature.
[0035] As shown in Figure 4, S2 is specifically:
[0036] S21: Use the finite element method to simulate the sudden change of working conditions and find the temperature stagnation points of each temperature-sensitive point outside the physical prototype spindle on the physical prototype spindle. The temperature stagnation points refer to the two points where the ratio of the average response time is within the allowable range (usually, δ = 0.05) in all sub-working condition sudden changes. Determine the temperature stagnation points of the temperature-sensitive points outside the physical prototype spindle on the physical prototype spindle according to the following formula:
[0037] Among them, Δt1 and Δt2 are the average response times of the temperature sensitive point outside the physical prototype spindle and the temperature stagnation point on the physical prototype spindle, respectively.
[0038] S22: intercepting a multi-condition finite temperature sequence of each temperature-sensitive point outside the physical prototype spindle and the temperature stagnation point on the physical prototype spindle, and based on the multi-condition finite temperature sequence, using grey system theory to determine the temperature relationship between the temperature-sensitive point outside the physical prototype spindle and the temperature stagnation point on the physical prototype spindle and record it as the first temperature relationship.
[0039] S23: The first temperature relationship is integrated with the existing thermal error autoregressive distributed hysteresis model to construct a physical prototype spindle thermal error analysis model that uses the temperature on the physical prototype spindle and the temperature of the stagnation point as the thermal error feature. The formula is as follows:
[0040] Among them, Y g represents the predicted value of the thermal error of the machine tool spindle at time g, a0 represents the adjustment coefficient, a irepresents the weight of the i-th order lag variable of the thermal error, m represents the autoregressive order, p represents the number of exogenous variables, n represents the distribution lag order, ε g represents white noise, β ji represents the weight of the i-th order lagged variable of temperature under the j-th exogenous variable, represents the i-th order lag variable of temperature, represents the 1-AGO sequence of system temperature characteristic data under the jth exogenous variable, a j represents the grey system development coefficient, b j2 represents the driving coefficient, Represents the 1-AGO sequence of the relevant factor data under the j-th exogenous variable.
[0041] S3: A virtual prototype model of the machine tool is established. Finite element methods are used to exclude areas with no significant temperature changes. Based on this, the possible placement area for the twin machine tool's external spindle temperature sensors is delineated based on the actual machining environment. Temperature hysteresis points on the physical prototype spindle are used as twin features and are mapped one by one to the virtual prototype spindle. The temperature migration points of the virtual prototype spindle are obtained, and the locations of the temperature-sensitive points outside the virtual prototype spindle are determined. Specifically, the virtual prototype finite element model of the machine tool undergoes a simulation of the same operating condition mutation process as in S12. A binary search method is used to identify the temperature-sensitive points outside the virtual prototype spindle that exhibit the same thermal hysteresis effect as the temperature migration points on the virtual prototype spindle. Based on grey system theory, the temperature relationships between the temperature migration points on the virtual prototype spindle and the temperature-sensitive points outside the virtual prototype spindle are established and recorded as the second temperature relationship. Finally, a twin coupling relationship is constructed between the temperatures of the corresponding points on the physical and virtual prototype spindles, as well as the twin coupling relationship between the spindle thermal errors. In this example, the physical prototype spindle and the machine tool spindle are identical models. In S12 and S3, identical sudden operating condition simulation experiments were performed on the physical prototype finite element model and the machine tool's virtual prototype, respectively. This indicates that the temperature variations at corresponding locations on the two spindles are consistent. Analysis from a thermal mechanism perspective indicates that spindle thermal error is closely related to temperature variations at various locations on the spindle. Therefore, the thermal errors of the physical and virtual prototype spindles are consistently equal.
[0042] Finally, by integrating the twin coupling relationship between the physical prototype spindle and the virtual prototype spindle, the thermal error analysis model of the physical prototype spindle, and the second temperature relationship, a CNC machine tool spindle thermal error prediction model based on the twin feature migration of virtual and real prototypes is constructed.
[0043] S4: As shown in Figure 5, S4: Adjust the parameters of the physical prototype test bench so that it always operates under the same working conditions as the machine tool. The physical prototype spindle thermal error measured by the displacement sensor is integrated with the twin coupling relationship in S3 to obtain the true value of the machine tool spindle thermal error. Then, based on the temperature sensitive points outside the virtual prototype spindle, temperature sensors are arranged at corresponding positions outside the machine tool spindle. The collected temperature data is input into the CNC machine tool spindle thermal error prediction model in real time to obtain the predicted value of the machine tool spindle thermal error. Finally, the prediction accuracy is calculated based on the true value and predicted value of the machine tool spindle thermal error. The CNC machine tool spindle thermal error prediction model is optimized based on the prediction accuracy to obtain a more accurate predicted value of the machine tool spindle thermal error. The accuracy of the machine tool spindle thermal error prediction method is calculated based on indicators such as the root mean square error. The present invention calculates the accuracy of the machine tool spindle thermal error prediction method by a method that does not require the placement of displacement sensors on the machine tool.
[0044] Based on the twin coupling relationship established in S3, the thermal error of the physical prototype spindle measured by the displacement sensor in this embodiment is the true value of the thermal error of the machine tool spindle.
[0045] The above embodiments are used to illustrate the present invention rather than to limit the present invention. Any modifications and changes made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.
Claims
1. A method for predicting thermal errors of CNC machine tool spindles based on twin feature migration of virtual and real prototypes, characterized in that: The following steps are involved: S1: Build a spindle physical prototype test bench, evenly set multiple measuring points outside the physical prototype spindle, select and obtain the temperature sensitive points outside the physical prototype spindle from the multiple measuring points, and establish a thermal error autoregressive distributed hysteresis model; S2: Determine the temperature stagnation point of the temperature sensitive point outside the physical prototype spindle on the physical prototype spindle, establish the temperature relationship between the temperature sensitive point outside the physical prototype spindle and the temperature stagnation point on the physical prototype spindle according to the grey system theory and record it as the first temperature relationship, and construct a physical prototype spindle thermal error analysis model based on the first temperature relationship, which takes the temperature of the temperature stagnation point on the physical prototype spindle as the thermal error feature; S3: Establish a virtual prototype model of the machine tool, use the temperature isostasis point on the physical prototype spindle as the twin feature and migrate it to the virtual prototype spindle of the machine tool, obtain the temperature migration point of the virtual prototype spindle, and then determine the position of each temperature sensitive point outside the virtual prototype spindle. Then, based on the grey system theory, establish the temperature relationship between the temperature migration point on the virtual prototype spindle and each temperature sensitive point outside the virtual prototype spindle and record it as the second temperature relationship; then, integrate the twin coupling relationship between the physical prototype spindle and the virtual prototype spindle, the thermal error analysis model of the physical prototype spindle, and the second temperature relationship, and construct a CNC machine tool spindle thermal error prediction model based on the twin feature migration of virtual and real prototypes; S4: Adjust the parameters of the physical prototype test bench so that it always runs in the same working condition as the machine tool. The thermal error of the physical prototype spindle measured by the displacement sensor is integrated with the twin coupling relationship in S3 to obtain the true value of the thermal error of the machine tool spindle. Then, according to the temperature sensitive points outside the virtual prototype spindle, the temperature sensor is arranged at the corresponding position outside the machine tool spindle, and the collected temperature data is input into the CNC machine tool spindle thermal error prediction model in real time to obtain the predicted value of the machine tool spindle thermal error; Finally, according to the machine tool spindle The prediction accuracy is calculated by the real value and predicted value of the thermal error. The CNC machine tool spindle thermal error prediction model is optimized according to the prediction accuracy to obtain a more accurate prediction value of the machine tool spindle thermal error.
2. The method for predicting thermal error of a CNC machine tool spindle by twin feature migration of virtual and real prototypes according to claim 1 is characterized in that: The S1 is specifically: S11: Build a spindle physical prototype test bench, evenly set 10-20 measuring points outside the physical prototype spindle, and arrange corresponding temperature sensors at each measuring point; S12: Establish a finite element model of the physical prototype test bench, and use the simulated working condition mutation process to preliminarily screen out the leading measuring points outside the main axis of the physical prototype whose temperature mutation response is faster than the thermal error mutation response when the working condition mutation occurs based on the finite element model of the physical prototype test bench; S13: A thermal hysteresis clustering algorithm is used to select several temperature sensitive points outside the physical prototype spindle from the advanced measurement points outside the physical prototype spindle, and a thermal error autoregressive distributed hysteresis model is established by combining the thermal hysteresis effect and the self-memory of the thermal error.
3. The method for predicting thermal error of a CNC machine tool spindle by twin feature migration of virtual and real prototypes according to claim 2 is characterized in that: In S13, a number of temperature sensitive points outside the main axis of the physical prototype are selected from the advanced measurement points outside the main axis of the physical prototype using a thermal hysteresis clustering algorithm, specifically: First, the average response time of each advance measuring point under each sudden change of working condition is calculated in turn; then, each advance measuring point is divided into k categories according to the average response time, and the average value of the average response time of all advance measuring points in each category of advance measuring points is used as the cluster center of this category of advance measuring points, and then the sum of square errors under the current number of clusters is calculated according to the cluster centers of each category of advance measuring points; the number of clusters k is changed, and the sum of square errors under different numbers of clusters is calculated, and then a relationship diagram between the number of clusters k and the corresponding sum of square errors is drawn, and the number of clusters when the decrease of the sum of square errors in the relationship diagram suddenly decreases is used as the target number of clusters; each advance measuring point is clustered according to the target number of clusters to obtain each target cluster cluster, The advanced measurement points with the greatest correlation in each target cluster are taken as temperature sensitive points outside the main axis of the physical prototype.
4. The method for predicting thermal error of a CNC machine tool spindle by twin feature migration of virtual and real prototypes according to claim 3 is characterized in that: The correlation is specifically the correlation between the temperature of each advanced measurement point and the true value of the thermal error of the spindle of the physical prototype.
5. The method for predicting thermal error of a CNC machine tool spindle by twin feature migration of virtual and real prototypes according to claim 1 is characterized in that: The S2 is specifically: S21: Use the finite element method to simulate the sudden change of working conditions, and find the temperature stagnation points of each temperature sensitive point outside the physical prototype spindle on the physical prototype spindle; S22: intercepting a multi-condition finite temperature sequence of each temperature sensitive point outside the physical prototype spindle and a temperature stagnation point on the physical prototype spindle, and based on the multi-condition finite temperature sequence, using the grey system theory to determine the temperature relationship between the temperature sensitive point outside the physical prototype spindle and the temperature stagnation point on the physical prototype spindle and record it as the first temperature relationship; S23: The first temperature relationship is integrated with the existing thermal error autoregressive distribution hysteresis model to construct a physical prototype spindle thermal error analysis model that uses the temperature on the physical prototype spindle and the temperature of the hysteresis point as the thermal error feature.
6. The method for predicting thermal error of a CNC machine tool spindle by twin feature migration of virtual and real prototypes according to claim 1 is characterized in that: In S3, a temperature sensitive point outside the virtual prototype spindle having the same thermal hysteresis effect as the temperature migration point on the virtual prototype spindle is determined by a binary search method.
Citation Information
Patent Citations
Heavy-duty tool thermal error prediction method taking environmental temperature into account
CN103823991A
Digital controlled lathe thermal error measuring compensation system and method
CN106736848A
Grey-theory sectional weighted heat error modeling method with lag data processing
CN110262393A
Machine tool thermal error self-learning prediction model modeling method and machine tool thermal error control method based on digital twinning
CN113051831A
Numerical control machine tool spindle thermal error segmentation modeling method considering thermal hysteresis
CN114896771A
Cited By
Cross-working-condition migration machine tool thermal error step-by-step prediction method and system
CN120706286A
