Visualized temperature sensitive area identification method and system for numerical control machine tool
By combining peak hysteresis correlation and adaptive inflection point time series detection algorithms with hierarchical clustering-wavelet correlation analysis, temperature-sensitive regions are identified and visualized, solving the problem of unintuitive sensor placement and achieving efficient and economical temperature sensor installation and thermal error modeling.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI JIAOTONG UNIV
- Filing Date
- 2025-04-15
- Publication Date
- 2026-07-31
AI Technical Summary
In existing technologies, the placement of temperature sensors is not intuitive under variable operating conditions and changes in external conditions, resulting in insufficient accuracy and robustness in thermal error modeling, as well as high sensor quantity and installation costs.
The peak hysteresis correlation algorithm and the adaptive inflection point time series detection algorithm are used to identify the high-frequency and low-frequency component changes of the temperature signal. Combined with hierarchical clustering-wavelet correlation analysis, the temperature-sensitive region is determined and the sensor location is visualized.
Without measuring machine tool thermal error, the sensor location can be determined in a time-saving and cost-effective manner, improving the correlation coefficient and modeling accuracy of the sensors, adapting to changes in working conditions, reducing the number of sensors, and reducing installation complexity.
Smart Images
Figure CN120382379B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of temperature-sensitive area selection technology, specifically, it relates to a method and system for visually identifying temperature-sensitive areas of CNC machine tools, and in particular, an identification strategy for visually identifying temperature-sensitive areas of thermally induced errors in CNC machine tools. Background Technology
[0002] Changes in machine tool operating conditions or external environment can lead to a decrease in machining accuracy due to time-varying thermal effects. Thermal errors caused by thermal effects can account for 40%-70% of machining errors and can also interfere with the measurement of other machine tool error items. To reduce the impact of thermal errors, error prevention and error compensation methods are generally used to improve machining accuracy.
[0003] Error prevention methods avoid thermal errors by improving machine tool materials or optimizing structural design. However, they have a boundary effect; once accuracy reaches a certain threshold, further improvements become more costly than beneficial. Error compensation methods, on the other hand, reduce thermal errors by modeling and predicting them, and then compensating for these predictions. Comparatively, error compensation is more economical, more adaptable to changes in the external environment, and can be applied to machine tools already in service. The key to thermal error compensation lies in accurately modeling and predicting machine tool thermal errors. Temperature changes over time are the fundamental cause of thermal deformation, and because they are easily measurable in a workshop environment, they are commonly used as the primary physical quantity for thermal error modeling. Currently, establishing a mapping relationship between measurable temperatures during machining and thermal errors is the most widely used modeling method.
[0004] Accurate construction of the machine tool temperature field is a crucial prerequisite for predicting and compensating for thermal deformation. The optimal approach is to place temperature sensors in the temperature-sensitive areas of the machine tool. Traditional redundant placement strategies are costly, and if sensitive locations are not covered, the optimized sensor mapping will still result in suboptimal thermal errors. Appropriate temperature sensors can improve the accuracy and robustness of thermal error modeling and save computational resources. To reduce the number of temperature sensors and time costs, some systems primarily deploy them at the heat sources of the machine tool before selecting temperature-sensitive points. However, the selection of the number and location of sensors is often not intuitive, and determining the optimal location is time-consuming. While temperature-sensitive point selection of already deployed temperature sensors has proven to improve modeling accuracy and robustness, the combined influence of changing internal and external factors makes the thermal effects of the machine tool more complex. Studies show that temperature-sensitive points are variable when operating conditions and external conditions change. Furthermore, changes in external conditions, such as the periodic activation of active cooling, can introduce periodic variations in thermal errors, reducing the robustness of temperature-sensitive points. Therefore, even if the most representative temperature sensor is selected from the already deployed temperature sensors, it may still not be able to accurately map changes in thermal error, and there is still no clear research on the placement of temperature sensors.
[0005] The patent document "Method and System for Selecting Temperature Sensitive Points for Thermal Error Based on Comprehensive Temperature Information" (CN112307579A) discloses a method for selecting temperature sensitive points using comprehensive temperature information, determining the optimal number of clusters using multiple clustering effectiveness indicators, and improving the accuracy of the thermal error model by using the number of temperature sensitive points selected based on comprehensive temperature information. This method reduces the number of temperature sensitive points and improves model performance. However, due to the combined influence of internal and external factors under variable operating conditions and changing external conditions, the temperature sensitive points are variable. Even if the most representative measuring points are selected from the deployed temperature sensors, they may still not accurately reflect the changing trend of thermal error, and problems such as the accidental deletion of highly correlated points and the misselection of low-sensitivity points may still occur.
[0006] Based on the above reasons and needs, it is necessary to conduct further research on the initial installation location of temperature sensors under variable operating conditions and changing external conditions, and propose a visual temperature-sensitive area selection strategy to guide the installation of temperature sensors. Summary of the Invention
[0007] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for visually identifying temperature-sensitive areas in CNC machine tools.
[0008] The method for visualizing temperature-sensitive areas of CNC machine tools provided by the present invention includes: The high-frequency variation step uses a peak-hysteresis correlation algorithm to obtain the temperature gradient change caused by the high-frequency components of the temperature signal; The low-frequency change step involves using an adaptive inflection point timing detection algorithm to obtain the temperature gradient change caused by the low-frequency component of the temperature signal. The calculation steps include: measuring and acquiring temperature field data to obtain temperature signals and adjusting the data structure; and executing high-frequency change steps and low-frequency change steps. The output steps include using hierarchical clustering-wavelet correlation analysis to verify and visualize the temperature-sensitive region, and then outputting the visualization results.
[0009] Preferably, the peak hysteresis correlation algorithm includes: Step S1.1: Based on the temperature field data, obtain the cooling start time according to the cooling signal. ; in, Indicates the number of times the cooling signal has been activated; Represents a non-negative integer; CS indicates the cooling signal; CST indicates the cooling start time.
[0010] Step S1.2: Calculate the local minimum value LMinT of the interval between two adjacent cooling start times based on the cooling start time. ; in, This represents temperature data that changes over time t.
[0011] Step S1.3: Calculate local maxima between local minima. , .
[0012] Step S1.4: Calculate the cooling lag time based on the cooling start time and the local maximum. , ; Among them, temperature difference ; This indicates the set temperature change threshold, which is 10% of the maximum temperature drop from active cooling.
[0013] Preferably, the adaptive inflection point timing detection algorithm includes: Step S2.1: Based on temperature field data, calculate the timestamp of operating condition changes according to the operating condition signal. , Points where the second-order difference exceeds an adaptively determined gradient threshold are set as inflection points or points of significant acceleration change.
[0014] in, This represents a time series composed of operating conditions; Indicates the number of times the operating condition signal has been activated; Represents a non-negative integer; This represents temperature data that changes over time t. This indicates the window size used to calculate the cumulative gradient before the operating conditions change, and it is less than half the duration of active cooling.
[0015] Step S2.2, Exclude active cooling during startup Changes in operating conditions Calculate the cumulative gradient change after the operating conditions change. , ; in, This indicates the window size, which is equal to the duration of the active cooling. "end" indicates the end of the cooling signal; This function calculates the sum of each element in an array and returns a vector of the same length.
[0016] Step S2.3: Find the index points where the gradient changes significantly. ; Here, Scidx represents the index where a significant gradient change occurs for the first time after a change in operating conditions.
[0017] Step S2.4: Update the lag time of the operating condition based on the calculated index point. ; Among them, temperature difference ; This indicates the set temperature change threshold.
[0018] Preferably, adjusting the data structure includes expanding the data dimension to three dimensions [rows, cols, timesseries], and initializing the lag time to... ; in, Represents the set of real numbers; rows, cols, and timeseries represent row pixels, column pixels, and time dimension, respectively. Num represents time.
[0019] The calculation steps include performing a high-frequency change step, outputting the result and averaging it over time, performing a low-frequency change step, outputting the result and averaging it over time, and selecting a temperature-sensitive region.
[0020] The temperature-sensitive region includes the region reflecting the peak temperature rise and the region reflecting the significant temperature gradient.
[0021] Preferably, the hierarchical clustering-wavelet correlation analysis includes: Step S4.1: Perform hierarchical clustering on the temperature time series data in the temperature field data. Observe the cluster division results through the clustering tree, and determine the number of better clusters by combining the elbow rule. Use the sum of squared errors (SSE) as the cost function of the elbow rule. ; in, Indicates the first Cluster Data points in Its center of mass The square of the Euclidean distance; K represents the total number of clusters.
[0022] Step S4.2: Apply Analytic Morlet wavelet transform to the temperature and thermal error data in the temperature field data. , ; in, It is a time signal In scale and location Wavelet coefficients below; Represents the complex conjugate of the wavelet function; Indicated by frequency Oscillating complex time series signals; It represents the standard deviation.
[0023] Step S4.3: Apply Spearman correlation analysis to the time-frequency domain information after wavelet transform to obtain the Spearman correlation coefficient between temperature data and thermal error. ; in, express The i-th observation value is Ranking in; This indicates the rank of the i-th observation in Y; n represents the number of observations in the dataset.
[0024] The present invention provides a visual temperature-sensitive area identification system for CNC machine tools, comprising: The high-frequency variation module uses a peak-hysteresis correlation algorithm to obtain the temperature gradient change caused by the high-frequency components of the temperature signal; The low-frequency change module uses an adaptive inflection point timing detection algorithm to obtain the temperature gradient change caused by the low-frequency component of the temperature signal; The calculation module acquires temperature field data, obtains temperature signals, adjusts the data structure, and triggers the high-frequency change module and the low-frequency change module. The output module uses hierarchical clustering-wavelet correlation analysis to verify and visualize the temperature-sensitive region and output the visualization results.
[0025] Preferably, the peak hysteresis correlation algorithm includes: Module M1.1, based on temperature field data, obtains the cooling start time according to the cooling signal. ; in, Indicates the number of times the cooling signal has been activated; Represents a non-negative integer; CS indicates the cooling signal; CST indicates the cooling start time.
[0026] Module M1.2 calculates the local minimum LMinT of the interval between two adjacent cooling start times based on the cooling start time. ; in, This represents temperature data that changes over time t.
[0027] Module M1.3: Calculating Local Maximums Between Local Minimums , .
[0028] Module M1.4 calculates the cooling lag time based on the cooling start time and local maxima. , ; Among them, temperature difference ; This indicates the set temperature change threshold, which is 10% of the maximum temperature drop from active cooling.
[0029] Preferably, the adaptive inflection point timing detection algorithm includes: Module M2.1 calculates the timestamps of operating condition changes based on temperature field data and operating condition signals. , Points where the second-order difference exceeds an adaptively determined gradient threshold are set as inflection points or points of significant acceleration change.
[0030] in, This represents a time series composed of operating conditions; Indicates the number of times the operating condition signal has been activated; Represents a non-negative integer; This represents temperature data that changes over time t. This indicates the window size used to calculate the cumulative gradient before the operating conditions change, and it is less than half the duration of active cooling.
[0031] Module M2.2, Excluding Active Cooling During Startup Changes in operating conditions Calculate the cumulative gradient change after the operating conditions change. , ; in, This indicates the window size, which is equal to the duration of the active cooling. "end" indicates the end of the cooling signal; This function calculates the sum of each element in an array and returns a vector of the same length.
[0032] Module M2.3: Find the index points where the gradient changes significantly. ; Here, Scidx represents the index where a significant gradient change occurs for the first time after a change in operating conditions.
[0033] Module M2.4 updates the lag time of the operating condition based on the calculated index point. ; Among them, temperature difference ; This indicates the set temperature change threshold.
[0034] Preferably, adjusting the data structure includes expanding the data dimension to three dimensions [rows, cols, timesseries], and initializing the lag time to... ; in, Represents the set of real numbers; rows, cols, and timeseries represent row pixels, column pixels, and time dimension, respectively. Num represents time.
[0035] The calculation module triggers a high-frequency change module, outputs the result and averages it over time, triggers a low-frequency change module, outputs the result and averages it over time, and selects a temperature-sensitive region.
[0036] The temperature-sensitive region includes the region reflecting the peak temperature rise and the region reflecting the significant temperature gradient.
[0037] Preferably, the hierarchical clustering-wavelet correlation analysis includes: Module M4.1 performs hierarchical clustering on the temperature time-series data in the temperature field data. It observes the clustering results through a clustering tree and uses the elbow rule to determine the number of better clusters. The sum of squared errors is used as the cost function of the elbow rule. ; in, Indicates the first Cluster Data points in Its center of mass The square of the Euclidean distance; K represents the total number of clusters.
[0038] Module M4.2 applies Analytic Morlet wavelet transform to the temperature and thermal error data in the temperature field data. , ; in, It is a time signal In scale and location Wavelet coefficients below; Represents the complex conjugate of the wavelet function; Indicated by frequency Oscillating complex time series signals; It represents the standard deviation.
[0039] Module M4.3 applies Spearman correlation analysis to the time-frequency domain information after wavelet transform to obtain the Spearman correlation coefficient between temperature data and thermal error. ; in, express The i-th observation value is Ranking in; This indicates the rank of the i-th observation in Y; n represents the number of observations in the dataset.
[0040] Compared with the prior art, the present invention has the following beneficial effects: 1. Compared with the traditional cluster correlation analysis method for selecting temperature sensitive points, this invention can determine the placement of temperature sensors without measuring the thermal error of the machine tool, saving time and instrument usage costs, while the placed sensors have a high correlation coefficient.
[0041] 2. This invention proposes a criterion for determining temperature-sensitive areas, identifies temperature-sensitive areas under the influence of internal and external factors of the machine tool, and establishes a thermal error mapping model based on the variable operating conditions of the machine tool under changing working conditions, so as to perform thermal error mapping under different operating conditions.
[0042] 3. This invention visualizes the lag time from "change in operating conditions" to "change in temperature," making the placement of temperature sensors visible. This avoids the complex installation and debugging of multiple sensors and eliminates the need to rely on expert experience, providing visual guidance for the installation of temperature sensors. Attached Figure Description
[0043] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A schematic diagram of the measurement and acquisition equipment for a visual temperature-sensitive area identification system for CNC machine tools; Figure 2 This is a schematic diagram of the spindle speed spectrum for thermal testing; Figure 3 Schematic diagram of thermal imaging results for milling spindle and turning spindle; Figure 4 This is a schematic diagram illustrating the changes in spindle thermal error under the influence of operating conditions and active cooling signals. Figure 5 A schematic diagram of the temperature-sensitive region under the influence of high-frequency temperature components; Figure 6 A schematic diagram of the temperature-sensitive region under the influence of low-frequency temperature components; Figure 7(a) is a schematic diagram of the clustering tree diagram analysis results of the milling spindle.
[0044] Figure 7(b) is a schematic diagram of the results of the pipe bending method for milling spindle.
[0045] Figure 7(c) is a schematic diagram of the clustering tree diagram analysis results of the milling spindle.
[0046] Figure 7(d) shows the clustering results and correlation coefficients of the turning spindle.
[0047] Figure 7(e) is a schematic diagram of the results of the pipe bending method for turning spindle.
[0048] Figure 7(f) shows the clustering results and correlation coefficients of the turning spindle. Detailed Implementation
[0049] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.
[0050] The installation of temperature sensors and the selection of temperature sensitive points are among the most important steps before thermal error modeling and compensation. Traditional temperature sensor installation is random and depends on expert experience. Selecting the optimal sensitive point requires not only installing and screening a large number of temperature sensors, but also is highly dependent on the random location of the sensors.
[0051] To study the temperature change characteristics of machine tools and find the optimal temperature measurement area, this invention provides a method for visualizing the temperature-sensitive area of a CNC machine tool, used to guide the installation of temperature sensors. Visualizing the temperature-sensitive area allows temperature sensors placed within it to better reflect changes in thermal error when operating conditions and external conditions change. The method includes the following steps: Step 1: Measure the temperature field data, propose and use the peak hysteresis correlation algorithm to identify the temperature gradient changes caused by the high-frequency components of the temperature signal.
[0052] Specifically, for the high-frequency components of the temperature signal, a peak hysteresis correlation algorithm is applied to the temperature gradient changes caused by active cooling. The peak hysteresis correlation algorithm also includes calculation steps: Step S1.1: Based on the temperature field data obtained from the temperature field measurement, according to the cooling signal ( The principle behind identifying the timestamp when cooling begins, i.e., the Cooling Start Time (CST), is as follows:
[0053] in, Indicates the number of times the cooling signal is activated. Represents a non-negative integer.
[0054] Step S1.2: Based on the cooling start time obtained above ( Calculate the timestamp of the local minimum (LMinT) between two consecutive cooling start intervals:
[0055] in, This represents temperature data that changes over time t.
[0056] Step S1.3: Based on the calculation results of the above formula, at the beginning of cooling and the adjacent local minimum ( ) calculate local maxima between Timestamp:
[0057] Step S1.4: Based on the calculation results of the above two formulas, that is, the lag time is calculated using the cooling start time and the local maximum timestamp. ):
[0058] Among them, temperature difference Due to the influence of ambient temperature and noise signals on the sensor, a temperature change threshold is set ( To filter out the aforementioned interference, the lag time is set to [value] when the difference between two adjacent local maxima and minima does not exceed this threshold. The threshold is set to 10% of the maximum temperature drop from active cooling.
[0059] In more preferred embodiments, the dual-spindle design of a milling and turning machine tool with two spindles of different structural types is taken as the research object. The temperature field and thermal error of the dual-spindle under different working conditions are measured using a thermal imager and an eddy current displacement sensor. The equipment used includes... Figure 1 As shown. When determining the thermal effects of the rotating spindle, the spindle speed was designed as a variable speed spectrum. Each speed was maintained for 15 minutes to simulate typical machining conditions. Following the thermal testing guidelines recommended by ISO 230-3, the preheating test lasted for 4 hours, and the variable speed spectrum changed over time as shown. Figure 2 As shown.
[0060] Step 2: Propose and use an adaptive inflection point timing detection algorithm to identify temperature gradient changes caused by low-frequency components of the temperature signal.
[0061] Specifically, for the low-frequency component of the temperature signal, an adaptive inflection point timing detection algorithm is applied to the temperature gradient changes caused by operating conditions. The adaptive inflection point timing detection algorithm includes the following calculation steps: Step S2.1: Based on the original measured temperature data, identify the timestamps of changes in operating conditions according to the operating condition signals. ):
[0062] in, It is a time series composed of operating conditions (speed). After active cooling control is activated, the temperature of the cooled components drops rapidly, and high-frequency information introduces interference. Therefore, second-order difference is used to identify the inflection point of temperature or its acceleration change. When the second-order difference exceeds the adaptively determined gradient threshold ( A threshold is considered to be inflection point or significant acceleration change when a threshold is reached. The threshold is set as follows:
[0063] in, The window size used to calculate the cumulative gradient before the operating conditions change is set to be less than half the duration of active cooling.
[0064] Step S2.2: Since temperature changes in long-cycle mode can be affected by the active cooling process, additional rules are added to avoid potential interference. Specifically, during the active cooling startup period ( Changes in operating conditions are no longer included in the calculation:
[0065] in, This represents the window size, which is equal to the active cooling duration. After satisfying the above rules, the cumulative gradient change after the change in operating conditions is calculated using the results obtained from the above formula. ):
[0066] in, The function calculates the sum of each element in the array and returns a vector of the same length.
[0067] Step S2.3: Subsequently, based on the calculation results of the above two equations, find the index point where the gradient changes significantly:
[0068] Here, Scidx represents the index where a significant gradient change occurs for the first time after a change in operating conditions.
[0069] Step S2.4: Finally, update the lag time based on the calculated index point:
[0070] Among them, temperature difference .
[0071] In further preferred embodiments, experiments were conducted on both turning and milling spindles. The maximum rotational speed of the turning spindle was 2500 RPM, and the maximum speed of the milling spindle was 8000 RPM. During the experiments, thermal imagers were used to continuously photograph the turning and milling spindles. A five-point eddy current sensor was used to measure the thermal displacement during the preheating cycle in real time.
[0072] Thermal images of the turning spindle and the milling spindle were captured separately during the thermal test, such as Figure 3 As shown. Changes in machine tool operating conditions and active cooling activation data were also recorded during the complete thermal test. Under the combined effect of active cooling control and changes in operating conditions, the thermal errors of the turning and milling spindles along the X, Y, and Z directions were recorded. (See figure) Figure 4As shown, after the active cooling signal is activated, the thermal error fluctuates within a short period. Although the changes in temperature and thermal error amplitude are small, the high-frequency component affects the surface finish quality. In contrast, changes in operating conditions lead to larger changes in the amplitude of temperature and thermal error, but low frequency and long period affect the overall machining accuracy. The above experimental results indicate that it is necessary to consider the influence of two temperature frequency components in the selection strategy for temperature-sensitive areas.
[0073] The lag time identification algorithm based on the high-frequency and low-frequency components of temperature signals can be used to identify temperature-sensitive areas under the influence of internal and external factors of machine tools.
[0074] Step 3: Application and visualization of parallel computing-based algorithms to acquire machine tool temperature field data and adjust the data structure for algorithm recognition; Specifically, an experiment was conducted using a milling and turning machine as a preferred example, and the algorithm was applied to identify and visualize temperature-sensitive areas.
[0075] To visualize temperature-sensitive regions, the data structure was adjusted to include three dimensions: rows, columns, and timeseries, representing row pixels, column pixels, and a time dimension. The initial lag time was set as follows:
[0076] in, Let represent the set of real numbers. Then, all the algorithms from steps one and two are applied to find the region where the temperature gradient first begins to respond after the thermal equilibrium is disrupted.
[0077] To address the characteristics of component temperature changes over time, the cluster-correlation analysis screening process is transformed into two methods for identifying temperature-sensitive regions based on minimum temperature lag time. An algorithm based on the lag time of high / low frequency components of temperature signal changes is used to identify temperature-sensitive regions under the combined influence of internal and external factors of the machine tool.
[0078] In more preferred examples, the thermal imaging temperature data is formatted as (rows, columns, time), i.e., [480×640×14400]. The visualization result analysis process for the temperature-sensitive area of the high-frequency component of temperature change is as follows: The time series of the start-up and shutdown of the active cooling system of the milling spindle was recorded and analyzed 51 times. The set cooling amplitude of the active cooling device is 1°C, so the threshold of the algorithm is set to 0.1°C. After executing step one, the final output result has a dimension of [480×640×51]. Then, the time dimension is averaged to obtain the visualization result as follows. Figure 5As shown in the left figure, the visual distribution area of the lag time can be directly observed; the darker areas better reflect high-frequency temperature fluctuations. Similarly, during the thermal testing of the turning spindle, the active cooling system was started and stopped 54 times. That is, the final output dimension after executing step one is [480×640×54]. The final visualization is as follows. Figure 5 As shown in the right figure, the lag time from changes in operating conditions / external conditions to temperature changes is visually displayed, enabling visual identification of the temperature sensor placement location.
[0079] The visualization analysis process for the temperature-sensitive region of the low-frequency component of temperature change is as follows: During a 4-hour thermal test of the milling spindle, 15 changes in operating conditions occurred. The active cooling system activation signal was searched before and after the timestamps of these changes. After excluding the influence of active cooling control, a total of 10 changes in operating conditions were involved in the calculation. The final output dimension after step two is [480×640×10]. The final visualization after averaging the time dimension is as follows. Figure 6 As shown in the left figure. Figure 5 Left image and Figure 6 The comparison in the left figure shows that the temperature change of the milling spindle's front bearing better reflects the low-frequency components. This is because the milling spindle motor is covered by the cooling system wiring, so changes in operating conditions are not better reflected near the motor. During the thermal testing of the turning spindle, after excluding the influence of active cooling control, a total of 12 operating conditions were involved in the calculation. After performing step two, the final output dimension is [480×640×12]. The final visualization is as follows. Figure 6 As shown in the figure on the right. Figure 6 The darker area in the right-hand image shows the hydraulic oil cooling fins. The visualization demonstrates that the cooling fins better reflect changes in operating conditions. In contrast, the turning spindle's motor and bearings are covered by protective sleeves, failing to adequately reflect temperature changes caused by varying operating conditions.
[0080] This paper proposes criteria for identifying temperature-sensitive areas, including areas that reflect peak temperature rise and areas that reflect significant temperature gradients. By visualizing the lag time from "operating condition change" to "temperature change," the identification results of temperature-sensitive areas and the placement of temperature sensors are visualized, providing visual guidance for temperature sensor installation. This avoids installing and screening a large number of temperature sensors on machine tools, saving time and avoiding sensor waste. It also eliminates the need to rely on expert experience, making it more suitable for implementation in industrial (workshop) environments, and is particularly suitable for temperature point placement on large machine tools.
[0081] Step 4: Use hierarchical clustering-wavelet correlation analysis to verify the correctness of the visualized temperature-sensitive region. This step is used to demonstrate and illustrate the visualization effect. Specifically, to verify the effectiveness of the algorithm, traditional hierarchical clustering-wavelet correlation analysis was used to screen temperature-sensitive points, and the results were compared with the identified temperature-sensitive regions.
[0082] Mechanism analysis shows that the fundamental cause of machine tool thermal deformation is the change in temperature over time. A hierarchical clustering-elbow rule-wavelet correlation analysis algorithm (i.e., hierarchical clustering-wavelet correlation analysis) was employed to verify the placement of temperature sensors within the visualized temperature-sensitive area. Step S4.1: First, perform hierarchical clustering on the temperature time-series data acquired from the temperature field. Hierarchical clustering does not require pre-specifying the number of clusters. The clustering results can be observed through the clustering tree. The elbow rule is then used to determine the number of better clusters. The sum of squared errors (SSE), as the cost function of the elbow rule, is calculated as follows:
[0083] in, Indicates the first Cluster Data points in Its center of mass The square of the Euclidean distance.
[0084] Step S4.2: Subsequently, wavelet transform is performed on the temperature and thermal error data. Analytic Morlet wavelet transform is used, and its form is as follows:
[0085]
[0086] in, It is a time signal In scale and location The wavelet coefficients below. This represents the complex conjugate of the wavelet function. Indicated by frequency Oscillating complex time series signals, It represents the standard deviation.
[0087] Step S4.3: Using the variable velocity spectrum, temperature and thermal error exhibit a highly nonlinear relationship. Spearman correlation analysis is applied to the time-frequency domain information after wavelet transform. Spearman correlation coefficient. The calculation method is as follows:
[0088] in, express The i-th observation value is The ranking in the middle. This represents the rank of the i-th observation in Y. n represents the number of observations in the dataset, i.e., the sample size. The correlation coefficient between temperature data and thermal error is obtained using the above method.
[0089] Comparative experiments verified the effectiveness and advantages of the proposed algorithm.
[0090] In further preferred embodiments, to verify the effectiveness and correctness of the proposed visualization strategy for selecting temperature-sensitive regions, it is necessary to conduct comparative verification with traditional strategies. The equipment used for the verification experiments retains the same configuration, with the addition of a contact temperature sensor for acquiring temperature information. A four-wire PT-100 RTD is used as the temperature sensor, with a measurement accuracy of ±0.1°C. The temperature sensor is used in conjunction with the NI-9216 integrated temperature acquisition module. LabVIEW is used as the acquisition software, and the acquisition frequency is set to 5Hz. The collected temperature data is then averaged and filtered.
[0091] Temperature sensors for turning and milling spindles were screened separately. The number of clusters and clustering results were determined by hierarchical clustering and the elbow method. Temperature-sensitive points were then obtained through wavelet transform and correlation analysis. Figure 7 shows the clustering dendrograms, elbow method results, and correlation coefficient calculation results for the milling and turning spindles. The two left-hand plots of Figure 7 show the clustering dendrograms for the milling and turning spindles, respectively. The clustering dendrogram is viewed by drawing a horizontal line from the bottom; the number of vertical lines intersecting indicates the number of clusters. As the horizontal line continues to rise, the number of vertical lines crossing it decreases, i.e., the number of clusters decreases. The purpose of clustering is to reduce collinearity of temperature data and map thermal error data with the fewest possible variables. The optimal number of clusters is to find a balance between the sum of squares within and between clusters. The two middle plots of Figure 7 show the elbow method results for the milling and turning spindles, showing the elbow inflection point when the number of clusters is 5, indicating that the optimal number of clusters is 5. Similarly, the optimal number of clusters for the turning spindle is 6. Based on the clustering results and correlation coefficients in the two right-hand figures of Figure 7, Table 1 shows the screening results for temperature-sensitive points. The temperature points for the milling spindle are T2, T4, T10, and T12, with the group having the lowest correlation coefficient being discarded. The temperature points for the turning spindle are T2, T3, T5, T7, and T9, with the group having the lowest correlation coefficient again being discarded.
[0092] Table 1. Matching of Sensitive Points and Visualized Areas:
[0093] Compared to traditional clustering correlation analysis strategies, visualization methods can identify temperature-sensitive regions influenced by both internal and external factors of the machine tool. The selection of these temperature-sensitive regions is based on varying machine tool operating conditions, thus enabling the modeling of temperature sensors placed within these selected regions and the establishment of thermal error mapping models under changing operating conditions. This is highly helpful for optimizing sensor placement and modeling target thermal errors. Because the milling spindle motor is surrounded by a cooling water jacket, the temperature sensors on the motor cannot fully capture changes in operating conditions. However, the temperature sensors can effectively detect temperature fluctuations caused by active cooling control. Temperature sensors T5 to T7 placed on the spindle correspond to... Figure 5 The right-hand image shows a fairly small rear bearing cooling area. Although the distance between sensors T5 and T6 is less than 10 cm, their temperature trends and correlation coefficients are very different. However, the method proposed in this invention can determine the location of temperature sensors without measuring machine tool thermal errors. All selected sensor locations are within the visualized sensitive area, and temperature sensors arranged within this area exhibit high correlation coefficients. By observing gradient color regions, key locations are well identified, and the temperature-sensitive areas of different components are visualized.
[0094] In more preferred embodiments, data collected by a thermal imager is used to visualize a strategy for selecting temperature-sensitive areas.
[0095] The present invention also provides a visual temperature-sensitive area identification system for CNC machine tools. The visual temperature-sensitive area identification system for CNC machine tools can be implemented by executing the process steps of the visual temperature-sensitive area identification method for CNC machine tools. That is, those skilled in the art can understand the visual temperature-sensitive area identification method for CNC machine tools as a preferred embodiment of the visual temperature-sensitive area identification system for CNC machine tools.
[0096] The present invention provides a visual temperature-sensitive area identification system for CNC machine tools, comprising: The high-frequency variation module uses a peak-hysteresis correlation algorithm to obtain the temperature gradient change caused by the high-frequency components of the temperature signal; The low-frequency change module uses an adaptive inflection point timing detection algorithm to obtain the temperature gradient change caused by the low-frequency component of the temperature signal; The calculation module acquires temperature field data, obtains temperature signals, adjusts the data structure, and triggers the high-frequency change module and the low-frequency change module. The output module uses hierarchical clustering-wavelet correlation analysis to verify and visualize the temperature-sensitive region and output the visualization results.
[0097] In more preferred embodiments, the peak hysteresis correlation algorithm includes: Module M1.1, based on temperature field data, obtains the cooling start time according to the cooling signal. ; in, Indicates the number of times the cooling signal has been activated; Represents a non-negative integer; CS indicates the cooling signal; CST indicates the cooling start time.
[0098] Module M1.2 calculates the local minimum LMinT of the interval between two adjacent cooling start times based on the cooling start time. ; in, This represents temperature data that changes over time t.
[0099] Module M1.3: Calculating Local Maximums Between Local Minimums , .
[0100] Module M1.4 calculates the cooling lag time based on the cooling start time and local maxima. , ; Among them, temperature difference ; This indicates the set temperature change threshold, which is 10% of the maximum temperature drop from active cooling.
[0101] In more preferred embodiments, the adaptive inflection point timing detection algorithm includes: Module M2.1 calculates the timestamps of operating condition changes based on temperature field data and operating condition signals. , Points where the second-order difference exceeds an adaptively determined gradient threshold are set as inflection points or points of significant acceleration change.
[0102] in, This represents a time series composed of operating conditions; Indicates the number of times the operating condition signal has been activated; Represents a non-negative integer; This represents temperature data that changes over time t. This indicates the window size used to calculate the cumulative gradient before the operating conditions change, and it is less than half the duration of active cooling.
[0103] Module M2.2, Excluding Active Cooling During Startup Changes in operating conditions Calculate the cumulative gradient change after the operating conditions change. , ; in, This indicates the window size, which is equal to the duration of the active cooling. "end" indicates the end of the cooling signal; This function calculates the sum of each element in an array and returns a vector of the same length.
[0104] Module M2.3: Find the index points where the gradient changes significantly. ; Here, Scidx represents the index where a significant gradient change occurs for the first time after a change in operating conditions.
[0105] Module M2.4 updates the lag time of the operating condition based on the calculated index point. ; Among them, temperature difference ; This indicates the set temperature change threshold.
[0106] In more preferred embodiments, adjusting the data structure includes expanding the data dimensions to three dimensions [rows, cols, timeseries], with an initial lag time of... ; in, Represents the set of real numbers; rows, cols, and timeseries represent row pixels, column pixels, and time dimension, respectively. Num represents time.
[0107] The calculation module triggers a high-frequency change module, outputs the result and averages it over time, triggers a low-frequency change module, outputs the result and averages it over time, and selects a temperature-sensitive region.
[0108] The temperature-sensitive region includes the region reflecting the peak temperature rise and the region reflecting the significant temperature gradient.
[0109] In more preferred embodiments, the hierarchical clustering-wavelet correlation analysis includes: Module M4.1 performs hierarchical clustering on the temperature time-series data in the temperature field data. It observes the clustering results through a clustering tree and uses the elbow rule to determine the number of better clusters. The sum of squared errors is used as the cost function of the elbow rule. ; in, Indicates the first Cluster Data points in Its center of mass The square of the Euclidean distance; K represents the total number of clusters.
[0110] Module M4.2 applies Analytic Morlet wavelet transform to the temperature and thermal error data in the temperature field data. , ; in, It is a time signal In scale and location Wavelet coefficients below; Represents the complex conjugate of the wavelet function; Indicated by frequency Oscillating complex time series signals; It represents the standard deviation.
[0111] Module M4.3 applies Spearman correlation analysis to the time-frequency domain information after wavelet transform to obtain the Spearman correlation coefficient between temperature data and thermal error. ; in, express The i-th observation value is Ranking in; This indicates the rank of the i-th observation in Y; n represents the number of observations in the dataset.
[0112] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function as logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.
[0113] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.
Claims
1. A method for visualizing temperature sensitive area identification of a numerically controlled machine tool, characterized in that, include: The high-frequency variation step uses a peak-hysteresis correlation algorithm to obtain the temperature gradient change caused by the high-frequency components of the temperature signal; The peak hysteresis correlation algorithm includes: Step S1.1, based on the temperature field data, obtaining a cooling start time from the cooling signal, ; Step S1.2, calculating a local minimum LMinT of the interval between two successive cooling starts according to the cooling start time, ; Step S1.3, computing local maxima between local minima , ; Step S1.4, calculating the lag time of cooling from the cooling start time and the local maximum , ; in, Indicates the number of times the cooling signal is activated. Representing a non-negative integer, CS represents the cooling signal, and CST represents the cooling start time. This represents temperature data that changes over time t, specifically the temperature difference. , This indicates the set temperature change threshold, which is 10% of the maximum temperature drop from active cooling. The low-frequency change step involves using an adaptive inflection point timing detection algorithm to obtain the temperature gradient change caused by the low-frequency component of the temperature signal. The adaptive inflection point timing detection algorithm includes: Step S2.1: Based on temperature field data, calculate the timestamp of operating condition changes according to the operating condition signal. , Points where the second-order difference exceeds an adaptively determined gradient threshold are set as inflection points or points of significant acceleration change. Step S2.2, Exclude active cooling during startup Changes in operating conditions Calculate the cumulative gradient change after the operating conditions change. , ; Step S2.3, finding index points where gradient changes significantly, ; Step S2.4, updating the hysteresis time of the operating condition according to the calculated index point ; in, This represents a time series composed of operating conditions. Indicates the number of times the operating condition signal has been activated. Represents a non-negative integer. This represents temperature data that changes over time t. This indicates the window size used to calculate the cumulative gradient before the operating conditions change; it is less than half the duration of active cooling. This indicates the window size, which is the duration of the active cooling process. "end" indicates the end of the cooling signal. This function calculates the sum of each element in an array and returns a vector of the same length. `Scidx` represents the index of the first significant gradient change after a change in operating conditions. (Temperature difference) , This indicates the set temperature change threshold; The calculation steps include: measuring and acquiring temperature field data to obtain temperature signals and adjusting the data structure; and executing high-frequency change steps and low-frequency change steps. The data structure adjustment includes expanding the data dimensions to three dimensions [rows, cols, timeseries], and initializing the lag time to... ; wherein, represents a set of real numbers, rows, cols, timeseries represent row pixels, column pixels, time dimension, respectively, and Num represents time; The calculation steps include performing a high-frequency change step, outputting the result and averaging it over time, performing a low-frequency change step, outputting the result and averaging it over time, and selecting a temperature-sensitive region. The temperature-sensitive region includes the region that reflects the peak temperature rise and the region that reflects the significant temperature gradient. The output steps include using hierarchical clustering-wavelet correlation analysis to verify and visualize the temperature-sensitive region, and then outputting the visualization results.
2. The method of visualizing temperature sensitive areas of a CNC machine according to claim 1, wherein, The hierarchical clustering-wavelet correlation analysis includes: Step S4.1: Perform hierarchical clustering on the temperature time series data in the temperature field data. Observe the cluster division results through the clustering tree, and determine the number of better clusters by combining the elbow rule. Use the sum of squared errors (SSE) as the cost function of the elbow rule. ; in, Indicates the first Cluster Data points in Its center of mass The squared Euclidean distance, where K represents the total number of clusters; Step S4.2, Analytic Morlet wavelet transform is applied to the temperature and thermal error data in the temperature field data, ; in, It is a time signal In scale and location The wavelet coefficients below, Denotes the complex conjugate of the wavelet function. Indicated by frequency Oscillating complex time series signals, Indicates standard deviation; Step S4.3: Apply Spearman correlation analysis to the time-frequency domain information after wavelet transform to obtain the Spearman correlation coefficient between temperature data and thermal error. ; wherein, represents the rank of the ith observation in represents the rank of the ith observation in Y, n represents the number of observations in the dataset. 3. A visualized temperature sensitive area identification system for a numerically controlled machine tool, characterized in that include: The high-frequency variation module uses a peak-hysteresis correlation algorithm to obtain the temperature gradient change caused by the high-frequency components of the temperature signal; The peak hysteresis correlation algorithm includes: - the module M1.1, based on the temperature field data, obtains a cooling start time from the cooling signal, ; - module M1.2, calculating a local minimum LMinT of the interval between two successive cooling starts, as a function of the cooling start time, ; Module M1.3: Calculating Local Maximums Between Local Minimums , ; Module M1.4, calculating the lag time of cooling from the cooling start time and the local maximum , ; in, Indicates the number of times the cooling signal is activated. Representing a non-negative integer, CS represents the cooling signal, and CST represents the cooling start time. This represents temperature data that changes over time t, specifically the temperature difference. , This indicates the set temperature change threshold, which is 10% of the maximum temperature drop from active cooling. The low-frequency change module uses an adaptive inflection point timing detection algorithm to obtain the temperature gradient change caused by the low-frequency component of the temperature signal; The adaptive inflection point timing detection algorithm includes: Module M2.1, based on temperature field data, calculates a working condition change timestamp from the working condition signal , , sets points where the second-order difference exceeds an adaptively determined gradient threshold as points of inflection or significant acceleration change, the threshold Module M2.2, excluding active cooling during startup of the operating condition, calculating the accumulated gradient change after the operating condition change , ; Module M2.3, finding the index points where the gradient significantly changes, ; Module M2.4 updates the lag time of the operating condition based on the calculated index point. ; in, This represents a time series composed of operating conditions; Indicates the number of times the operating condition signal has been activated. represents a non-negative integer, temp Temperature data of t change This indicates the window size used to calculate the cumulative gradient before the operating conditions change; it is less than half the duration of active cooling. This indicates the window size, which is the duration of the active cooling process. "end" indicates the end of the cooling signal. This function calculates the sum of each element in an array and returns a vector of the same length. `Scidx` represents the index of the first significant gradient change after a change in operating conditions. (Temperature difference) , This indicates the set temperature change threshold; The calculation module acquires temperature field data, obtains temperature signals, adjusts the data structure, and triggers the high-frequency change module and the low-frequency change module. The adjustment data structure includes extending the dimension of data to three dimensions [rows, cols, timeseries], initializing the lag time as ; in, This represents the set of real numbers, where rows, cols, and timeseries represent row pixels, column pixels, and time dimension, respectively, and Num represents time. The calculation module triggers a high-frequency change module, outputs the result and averages it over time, triggers a low-frequency change module, outputs the result and averages it over time, and selects a temperature-sensitive region. The temperature-sensitive region includes the region that reflects the peak temperature rise and the region that reflects the significant temperature gradient. The output module uses hierarchical clustering-wavelet correlation analysis to verify and visualize the temperature-sensitive region and output the visualization results.
4. The visualized temperature sensitive area identification system of a CNC machine tool according to claim 3, characterized in that, The hierarchical clustering-wavelet correlation analysis includes: Module M4.1 performs hierarchical clustering on the temperature time-series data in the temperature field data. It observes the clustering results through a clustering tree and uses the elbow rule to determine the number of better clusters. The sum of squared errors is used as the cost function of the elbow rule. ; in, Indicates the first Cluster Data points in Its center of mass The squared Euclidean distance, where K represents the total number of clusters; Module M4.2, applying an Analytic Morlet wavelet transform to temperature and thermal error data in the temperature field data, , ; in, It is a time signal In scale and location The wavelet coefficients below, Denotes the complex conjugate of the wavelet function. Indicated by frequency Oscillating complex time series signals, Indicates standard deviation; The module M4.3, the Spearman correlation analysis is used on the time-frequency domain information after the wavelet transform, the Spearman correlation coefficient between the temperature data and the thermal error is obtained, and the Spearman correlation coefficient ; wherein, represents the rank of the ith observation in represents the rank of the ith observation in Y, n represents the number of observations in the dataset.