Machining precision intelligent analysis method and system for numerical control machine tool

Abstract

The invention relates to the field of electric digital data processing, in particular to an intelligent processing precision analysis method and system for a numerical control machine tool. Analyzing the consistency of the data points of each dimension according to the change of the local feature intensity of the data points of different dimensions at the same moment by calculating the local feature intensity of the data points of each dimension; the method comprises the following steps: dynamically calculating an optimal window length according to the consistency and local feature intensity of data points, smoothing processing parameters in the optimal window length by using a moving average method, and applying an LOF anomaly detection algorithm to smoothed processing data to obtain an anomaly detection result to assist implementation personnel in analyzing the processing precision of the numerical control machine tool. Through the dynamic window optimization technology, the analysis range is adjusted in a self-adaptive mode, and data smoothing and exception reservation requirements are effectively balanced; and in combination with local feature intensity and multi-dimensional consistency analysis, the accuracy of anomaly recognition is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of digital data processing. In particular, it relates to a machining precision intelligent analysis method for a numerical control machine tool. BACKGROUND

[0002] A numerical control machine tool is a high-automation and high-precision device, which is applied in the industries of mechanical manufacturing, aerospace, automobile and electronics. It can efficiently process complex parts and improve production efficiency and product quality. Machining precision is the key to measure the performance of a numerical control machine tool, which is directly related to the size, shape and surface quality of products. High-precision machining can not only ensure product performance, but also improve product reliability and durability, and enhance the market competitiveness of enterprises.

[0003] In the processing of numerical control machine tool machining precision analysis scenarios using existing algorithms, the sliding window average method is often used to smooth data and detect anomalies. In actual application process, the sliding window average method has some limitations. On the one hand, the selection of window size has important influence on the smoothing effect and the sensitivity of anomaly detection. If the selected window is too small, the smoothing effect will be unsatisfactory, which cannot effectively filter out noise and further affect the accuracy analysis of machining precision. On the other hand, if the window is too large, although it may improve the smoothing effect, it may mask the abnormal values, making the anomaly detection ineffective, which is also not conducive to the accuracy analysis of machining precision.

[0004] In the machining of numerical control machine tools, the fluctuation characteristics of data change over time, but the fixed window size of the sliding window average method is difficult to adapt to such dynamic changes. The fixed window cannot be adjusted according to the actual fluctuation characteristics of data, and once the data fluctuation characteristics change, the fixed window may not be able to discover anomalies in time, affecting the accuracy and reliability of the analysis. SUMMARY

[0005] In order to solve the problem that the fixed window of the sliding window average method is difficult to adapt to the change of data fluctuation characteristics over time in the machining of numerical control machine tools, resulting in limited analysis accuracy, the present application provides solutions in the following aspects.

[0006] In a first aspect, a machining precision intelligent analysis method for a numerical control machine tool comprises: acquiring machining data of the numerical control machine tool and preprocessing; presetting a local range for the preprocessed machining data, calculating local feature strength of each dimension data point in the machining data based on the difference of data points in the local range, reflecting the change characteristics of the data points in the local window, analyzing the consistency of each dimension data point according to the change of the local feature strength of the data points of different dimensions at the same time; dynamically calculating the optimal window length of each dimension data point at different times according to the consistency and local feature strength of the data points, to balance the needs of data smoothing and abnormality reservation; using the moving average method to smooth the machining parameters in the optimal window length at each time, applying the LOF anomaly detection algorithm to the smoothed machining data to obtain the anomaly detection result and judge the machining precision of the numerical control machine tool analyzed by the numerical control auxiliary implementer.

[0007] The effect is that: through preprocessing to ensure data integrity and quality, and then calculating local feature strength based on data difference in local range, the change characteristics of data points in local window are accurately reflected. Further, by analyzing the change of local feature strength of data points of different dimensions at the same time, the consistency of each dimension data point is evaluated, and finally the optimal window length is dynamically calculated combined with the consistency and local feature strength of the data points, which can more accurately adapt to the data fluctuation characteristics, significantly improve the accuracy and reliability of machining precision analysis; by using the moving average method to smooth the machining parameters, noise can be effectively filtered out while abnormal features are retained. On this basis, applying the LOF anomaly detection algorithm to analyze the smoothed data can timely discover potential abnormal situations. Not only the accuracy of anomaly detection is improved, but also timely alarm can be sent in the early stage of abnormality, which is convenient for operators to take measures quickly.

[0008] Preferably, the machining data includes: tool temperature, tool vibration frequency, tool cutting.

[0009] Preferably, the preset local range comprises: The size of the preset initial window is set, the initial window is centered on any data point, and a certain number of data points are expanded before and after the center data to form a sequence segment, and the sequence segment is taken as the local range of the center data.

[0010] The effect is that: presetting the size of the initial window can provide a clear range for the calculation of local feature strength, ensuring that each data point has a fixed local range for feature analysis, thereby improving the consistency and repeatability of the calculation process. By reasonably setting the size of the initial window, a balance between capturing local features of data points and filtering out noise can be achieved, enhancing the adaptability of the analysis method to different data characteristics, laying a foundation for subsequent anomaly detection and data smoothing.

[0011] Preferably, the local feature intensity comprises: Taking any data point in any dimension sequence as target data, a local window is preset, the absolute value of the difference between the target data and each data point in the local window is calculated, and the average value of the sum of the absolute values of the differences is taken as the average deviation degree of the target data; The ratio between the average deviation degree and the standard deviation of all data points in the local window is calculated, and normalized, and the normalized result is taken as the local feature intensity of the target data.

[0012] The effect is that the overall deviation level of the target point and the neighborhood data is quantified by the average deviation degree, and the data scale influence is eliminated by the standard deviation ratio, so that the feature intensity is comparable and can effectively reflect the abnormality degree of the data point in the local range, considering both the absolute deviation and the data distribution characteristics, providing reliable local feature representation for subsequent consistency analysis and dynamic window adjustment.

[0013] Preferably, the local feature intensity comprises: Taking any data point in any dimension sequence as target data, a local window is preset, the absolute value of the difference between the target data and each data point in the local window is calculated, and the average value of the sum of the absolute values of the differences is taken as the average deviation degree of the target data, and the average deviation degree is normalized, and the normalized result is taken as the local feature intensity of the target data.

[0014] The effect is that the overall deviation level of the target point and the neighborhood data is quantified by the average absolute deviation, reflecting the local abnormality, and the feature intensity value can directly represent the relative abnormality degree of the point in the window, and the larger the value, the more significant the difference between the point and the surrounding data, which can accurately capture the local data fluctuation characteristics.

[0015] Preferably, the calculation method of the consistency of each dimension data point comprises: The difference between the local feature intensity of each two dimension data points at the same time is calculated, the difference is squared, and the sum of the squares of all dimension data points is obtained to get the local difference sum of squares; The average value of the local difference sum of squares is squared to get the root mean square of the local feature difference of all dimension data points, and the root mean square is exponentially attenuated by using a negative exponential function to get the consistency of the processed data.

[0016] The effect is that the square sum is used to highlight the significant difference, the root mean square is used to integrate the overall difference level, and finally the normalized consistency analysis is obtained by negative exponential mapping. The effective evaluation of the collaborative state of complex multi-dimensional data can identify local dimension mismatch and monitor the overall abnormal trend of the system, providing an objective quantitative basis for dynamic analysis window adjustment.

[0017] Preferably, the calculation method of the consistency of each dimension data point comprises: The absolute difference value of each two dimension data points corresponding to the local feature intensity at the same time is calculated respectively, and the average value of the sum of all absolute difference values is obtained as the average difference value of each dimension data point, and the average difference value is exponentially attenuated by using a negative exponential function to obtain the consistency of the processing data.

[0018] Preferably, the optimal window length comprises: The local feature intensity and the consistency of each dimension data point are normalized by using a negative hyperbolic tangent function respectively, and the normalized local feature intensity and the data consistency are multiplied by Again for linear transformation, the window correction coefficient is obtained by multiplying the two linear transformation results, and the optimal window length of the data point of each dimension data point at each time is obtained by multiplying the window correction coefficient by the preset initial sliding average window and taking the upper integer.

[0019] The effect is that by hyperbolic tangent normalization and linear transformation, the local feature intensity and the data consistency are fused to generate a dynamic window coefficient, the intelligent adjustment of the window length is realized, the window is automatically reduced to capture abnormal details when the data fluctuates greatly and the dimensions are inconsistent, and the window is expanded to improve the smoothing effect when the data is smooth and the dimensions are coordinated.

[0020] Preferably, the preprocessing of the processing data comprises: For each dimension data point in the processing data, the missing value is filled by using the difference method, the missing data point is supplemented, and the integrity of the data is ensured.

[0021] Secondly, a machining precision intelligent analysis system for a numerical control machine tool comprises a processor and a memory, and the memory stores computer program instructions.

[0022] The present application has the following effects: 1、The present application dynamically calculates the optimal window length, which can automatically adjust the window size according to the real-time data fluctuation characteristics, effectively solving the problem that the analysis precision is limited when the traditional fixed window method faces dynamic data changes. It can accurately capture small abnormalities and potential fluctuations in the processing process, even in complex and variable processing environments, and can ensure the high precision and reliability of the analysis results. This means that in actual production, potential problems that may cause processing defects can be found earlier, so that measures can be taken in time to adjust and optimize, greatly reducing the problem of rising scrap rate and equipment damage caused by not finding abnormalities in time, and effectively ensuring the stability of processing quality.

[0023] 2、The application has extremely high sensitivity and accurate recognition ability for abnormal data through comprehensive utilization of local feature strength and multi-dimensional consistency evaluation. The calculation of local feature strength can deeply analyze the change characteristics of a single data point in its local range, and the multi-dimensional consistency evaluation measures the coordination between different dimensional data from the overall perspective. This comprehensive analysis method makes the abnormal detection no longer limited to a single dimension or simple threshold judgment, but can combine multi-dimensional information for comprehensive evaluation, effectively reducing misjudgment and omission. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 is a method flowchart of steps S1-step S4 in a machining precision intelligent analysis method for a numerical control machine tool.

[0025] Figure 2 is a structural block diagram of a server fault diagnosis system for a server group. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, not all the embodiments.

[0027] Referring to Figure 1 A machining precision intelligent analysis method for a numerical control machine tool includes steps S1-S4, as follows: S1: Obtain machining data of the numerical control machine tool and perform preprocessing.

[0028] The machining data includes but is not limited to tool temperature, tool vibration frequency, and tool cutting.

[0029] It should be noted that temperature sensors, vibration sensors, and force sensors are installed on the numerical control machine tool to collect temperature, vibration frequency, and cutting force data. The collected data is transmitted to the central processing unit for analysis through a data acquisition system or interface, and the preset acquisition frequency is 30 Hz to ensure the timeliness of the data.

[0030] However, during data acquisition, data loss may occur, which will affect the accuracy of subsequent analysis to some extent. Therefore, for each dimensional data point in the machining data, the missing values are filled by using the difference method, the missing values are estimated by calculating the difference between adjacent data points, so as to supplement the missing data points and ensure the integrity of the data.

[0031] Further analysis reveals that during high-speed cutting in CNC machine tools, normal operating conditions (such as periodic vibration) can cause data fluctuations. However, these fluctuations are usually accompanied by relatively consistent data, similar to the surrounding fluctuations. Abnormal signals, on the other hand, such as non-periodic vibrations caused by tool wear, tend to have more concentrated data distribution, and when abnormal signals occur, they differ significantly from other data. Therefore, this invention calculates the local feature intensity of all data points in a given dimension by comparing the data points with the data points within their local windows. The specific steps are as follows: S2: Preset a local range for the preprocessed data. Based on the differences of data points within the local range, calculate the local feature intensity of data points in each dimension of the data to reflect the change characteristics of data points within the local window. Analyze the consistency of data points in each dimension based on the changes in the local feature intensity of data points in different dimensions at the same time.

[0032] Preset local area, including: The initial window size is preset. The initial window takes any data point as the center data and expands to several data points before and after the center data to form a sequence segment. The sequence segment is used as the local range of the center data.

[0033] Local feature intensity includes: Using any data point in any dimension sequence as the target data, a local window is preset, and the absolute value of the difference between the target data and each data point in the local window is calculated. The average value of the sum of the absolute values ​​of the differences is taken as the average deviation of the target data. Calculate the ratio between the average deviation and the standard deviation of all data points within the local window, and normalize it. Use the normalized result as the local feature intensity of the target data.

[0034] Specifically, the intensity of local features satisfies the following relationship: ; In the formula, Indicates the first In the dimension Local feature intensity of data points at time t. Indicates the preset local window size. This indicates the ordinal number of the data points in the local window. Indicates the first In the dimension The data values ​​of the data points at each time point. Indicates the first In the dimension The first data point in the local window at time 1 The data values ​​of each data point Show the first In the dimension a standard deviation of all data points in the local window of the data point at the time, The standard normalization function is represented.

[0035] That is, by calculating the deviation of the data points in the local window from the current data point, the characteristics of the local data are captured, that is, the changes of the data points in the local area. The size of the local window determines the size of the local range considered. A smaller local window focuses on finer local features, and a larger local window reflects more macro local trends. The greater the average deviation degree value, the greater the difference, and the stronger the local feature strength of the data point. When the numerical control machine is cutting at high speed, the data fluctuates violently, and the data point has strong volatility. If the difference with all data points in the local window is considered, the local feature strength of this type of data will also be strong. Therefore, by introducing the standard deviation of the data in the local window, the credibility of the average deviation degree is measured. The smaller the standard deviation of all data points in the local window, the more concentrated the data distribution in the local window, and the stronger the local feature strength of the data point.

[0036] By introducing the standard deviation of the data points in the local range, the average deviation degree and the dispersion degree of the data points are combined. Not only can the difference between the data points and the surrounding data be reflected, but also the stability of the data fluctuation can be reflected. In numerical control machine processing, when the data fluctuation is small (that is, the data distribution is concentrated), even if the deviation of a certain data point is slightly large, it may be a significant abnormal feature. When the data fluctuation is large, the same deviation may not be prominent. By dividing by the standard deviation, the relative significance of the deviation can be more accurately measured.

[0037] In the case of concentrated data distribution, a small deviation may indicate an anomaly, and this potential anomaly can be effectively captured. For numerical control machines with extremely high processing precision, it is crucial to discover these subtle abnormal features in a timely manner, which helps to take preventive measures to avoid the accumulation and expansion of processing errors. After considering the data dispersion, the calculation of the local feature strength has stronger robustness to noise. In actual processing, sensor data may be disturbed by various disturbances, resulting in occasional large fluctuations, thereby reducing the impact of such noise on the analysis results, and making the change of the local feature strength more truly reflect the actual situation in the processing process.

[0038] In addition, another embodiment also includes: Taking any data point in any dimensional sequence as a target data, a local window is preset, the absolute value of the difference between the target data and each data point in the local window is calculated, the average of the sum of the absolute value of the difference is taken as the average deviation degree of the target data, and the average deviation degree is normalized. The normalized result is taken as the local feature strength of the target data.

[0039] Specifically, the intensity of local features satisfies the following relationship: ; In the formula, Indicates the first In the dimension Local feature intensity of data points at time t. Indicates the preset local window size. This indicates the ordinal number of the data points in the local window. Indicates the first In the dimension The data values ​​of the data points at each time point. Indicates the first In the dimension The first data point in the local window at time 1 The data values ​​of each data point This represents the standard normalization function.

[0040] By directly calculating the average deviation of data points within a local window and then normalizing it, this method offers advantages such as simple and efficient calculation, fast response, strong applicability, good stability, and high flexibility. This makes it widely applicable in various CNC machine tool machining monitoring scenarios, especially suitable for applications requiring rapid response and real-time monitoring.

[0041] Further analysis reveals that sensor data from CNC machine tools is susceptible to noise (such as electromagnetic interference causing short-term drift in temperature sensors). Relying solely on the local feature intensity of single-dimensional data may misidentify noise as an anomaly; however, calculating multi-dimensional consistency can effectively identify isolated anomalies. For example, when the local feature intensity of the vibration dimension suddenly increases but the temperature and cutting force dimensions show no significant change, the consistency value is low, indicating that the anomaly may be caused by sensor failure rather than actual operating condition anomalies. The specific steps are as follows: The methods for calculating the consistency of data across different dimensions include: Calculate the difference in local feature intensity between each pair of dimensions at different times, square the difference, and sum the squares of all dimensions to obtain the sum of squares of local differences. The square root of the average of the sum of squares of local differences is taken as the root mean square of the local feature differences in all dimensions. The root mean square is then exponentially decayed using a negative exponential function to obtain the consistency of the processed data at each time step.

[0042] Specifically, consistency satisfies the following relationship: ; In the formula, Indicates the first Consistency of data processing at all times. Indicates the total number of dimensions. 、 respectively represent the dimension ordinal, represent the local feature intensity of the data point at the time in the dimension, represent the local feature intensity of the data point at the time in the dimension, represent the exponential function with the natural constant as the base number.

[0043] That is, the dispersion degree of the data in different dimensions is reflected by calculating the sum of squares of the local feature intensity difference between all pairs of dimensions; the local feature difference is mapped to an interval close to 0 to 1 by the exponential function . When the local feature difference is small, the exponential part is negative and has a large absolute value, causing the exponential function value to tend to 0, indicating that the data has low consistency; when the local feature difference is small, the absolute value of the exponential part is small, and the exponential function value tends to 1, indicating that the data has high consistency.

[0044] By calculating the sum of squares of the local feature intensity difference and then taking the root mean square value, the dispersion degree of the data is more sensitive. In multi-dimensional data, subtle differences between different dimensions can be more accurately captured, especially in scenarios that require high-precision analysis. When there are anomalies in the data, the square operation will amplify the influence of the difference, making it easier to detect anomalies.

[0045] In addition, another embodiment also includes: The absolute difference of the local feature intensity of each two-dimensional data point at the same time is calculated, and the average value of the sum of all absolute differences is obtained, that is, the average difference value of each dimensional data point, and the average difference value is exponentially attenuated using a negative exponential function to obtain the consistency of the processed data.

[0046] Specifically, the consistency satisfies the following relationship: ; In the formula, represent the consistency of the processed data at the time, represent the total number of dimensions, 、 respectively represent the dimension ordinal, represent the local feature intensity of the data point at the time in the dimension, represent the local feature intensity of the data point at the time in the dimension, Represented by natural constant An exponential function with base 1.

[0047] Because it uses absolute differences instead of squared differences, this method is less sensitive to outliers and more robust. It can stably reflect data consistency even when there are occasional large fluctuations. It is applicable to various data distributions, providing a reliable consistency assessment even when the data is non-normally distributed or contains a small number of extreme values.

[0048] S3: Based on the consistency of data points and the strength of local features, dynamically calculate the optimal window length for data points in each dimension at different times to balance the needs of data smoothing and anomaly preservation.

[0049] Optimal window length, including: For the local feature intensity and data consistency of data points at each time point in each dimension, normalization is performed using the negative hyperbolic tangent function. The normalized local feature intensity and data consistency are then multiplied by [a specific factor]. Add 1 to perform a linear transformation. Multiply the two linear transformation results to obtain the window correction coefficient. Multiply the window correction coefficient by the preset initial moving average window and round up to obtain the optimal window length for each dimension's data points at each time point.

[0050] Specifically, the optimal window length satisfies the following relationship: ; In the formula, Indicates the first In the dimension The optimal window length for data points at time t. This indicates the preset initial moving average window. Indicates the first In the dimension Local feature intensity of data points at time t. Indicates the first Consistency of data collected at any given time Represents the hyperbolic tangent function. This represents the function for rounding up.

[0051] In other words, when A larger value indicates a greater difference between the data point and other points within the local window, meaning the local features are more pronounced; when... When the size is small, the differences are small, and local features are not obvious. The negative sign in the middle makes The larger, The closer Conversely, the closer to This indicates the intensity of local characteristics. When the value increases, the value after the hyperbolic tangent function processing will decrease, reflecting a kind of inhibition or weakening effect on the feature strength, and after the transformation of , the obtained value range is in . The smaller the value is, the greater the local feature strength is, and more weight or attention needs to be given; otherwise, The greater the value is, the smaller the local feature strength is, and it can be appropriately ignored or weakened.

[0052] When the data consistency is high, the adjustment range of the window length is small, and the optimal window length is closer to the initial value; when the data consistency is low, the adjustment range of the window length is large, and the optimal window length will decrease, so as to better capture local features or abnormalities.

[0053] Exemplarily, the preset initial sliding average window can be adjusted according to specific conditions.

[0054] S4: The sliding average method is used to smooth the machining parameters in the optimal window length at each time, and the LOF anomaly detection algorithm is applied to the smoothed machining data to obtain an anomaly detection result to assist the implementer in analyzing the machining precision of the numerical control machine tool.

[0055] In response to the anomaly detection result being greater than a preset anomaly score threshold, the data point is marked as abnormal, and otherwise, less than or equal to the preset anomaly score threshold, the data point is marked as normal.

[0056] Exemplarily, the preset anomaly score threshold , and the abnormal data is stored separately, so as to facilitate the analysis of the staff.

[0057] The application also provides a machining precision intelligent analysis system for a numerical control machine tool. As shown in Figure 2 , the system comprises a processor and a memory, and the memory stores computer program instructions, which realize the machining precision intelligent analysis method for a numerical control machine tool according to the first aspect of the application when executed by the processor. The system also comprises a communication bus and a communication interface and other components familiar to those skilled in the art, and their settings and functions are known in the art, so they will not be described here.

[0058] It should be pointed out that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the protection scope of the present application patent should be subject to the appended claims.

Claims

1. A method for intelligent analysis of machining accuracy in CNC machine tools, characterized in that, include: Acquire machining data from CNC machine tools and perform preprocessing; A local range is preset for the preprocessed data. Based on the differences of data points within the local range, the local feature intensity of data points in each dimension of the data is calculated to reflect the change characteristics of data points within the local window. The consistency of data points in each dimension is analyzed based on the changes in the local feature intensity of data points in different dimensions at the same time. Based on the consistency of data points and the strength of local features, the optimal window length for data points in each dimension at different times is dynamically calculated to balance the needs of data smoothing and anomaly preservation. The machining parameters in the optimal window length at each time point are smoothed using the moving average method. The LOF anomaly detection algorithm is then applied to the smoothed machining data to obtain anomaly detection results, which helps implementers analyze the machining accuracy of CNC machine tools.

2. The intelligent analysis method for machining accuracy of CNC machine tools according to claim 1, characterized in that, The machining data includes: tool temperature, tool vibration frequency, and tool cutting speed.

3. The intelligent analysis method for machining accuracy of CNC machine tools according to claim 1, characterized in that, The preset local range includes: The initial window size is preset. The initial window takes any data point as the center data and extends several data points before and after the center data to form a sequence segment. The sequence segment is used as the local range of the center data.

4. The intelligent analysis method for machining accuracy of CNC machine tools according to claim 1, characterized in that: The local feature intensity includes: Using any data point in any dimension sequence as the target data, a local window is preset, and the absolute value of the difference between the target data and each data point in the local window is calculated. The average value of the sum of the absolute values ​​of the differences is taken as the average deviation of the target data. Calculate the ratio between the average deviation and the standard deviation of all data points within the local window, and normalize it. Use the normalized result as the local feature intensity of the target data.

5. The intelligent analysis method for machining accuracy of CNC machine tools according to claim 1, characterized in that: The local feature intensity includes: Using any data point in any dimension sequence as the target data, a local window is preset, and the absolute value of the difference between the target data and each data point in the local window is calculated. The average value of the sum of the absolute values ​​of the differences is taken as the average deviation of the target data, and the average deviation is normalized. The normalized result is taken as the local feature intensity of the target data.

6. The intelligent analysis method for machining accuracy of CNC machine tools according to claim 1, characterized in that, The calculation methods for the consistency of the data points in each dimension include: Calculate the difference in local feature intensity between each pair of data points at the same time, square the difference, and sum the squares of all data points to obtain the sum of squares of local differences. The square root of the average of the sum of squares of local differences is taken as the root mean square of the local feature differences of all data points in all dimensions. The root mean square is then exponentially decayed using a negative exponential function to obtain the consistency of the processed data.

7. The intelligent analysis method for machining accuracy of CNC machine tools according to claim 1, characterized in that, The calculation methods for the consistency of the data points in each dimension include: Calculate the absolute difference in the local feature intensity of each pair of data points at the same time, and sum all the absolute differences to obtain the average difference of the data points in each dimension. Use a negative exponential function to exponentially decay the average difference to obtain the consistency of the processed data.

8. The intelligent analysis method for machining accuracy of CNC machine tools according to claim 1, characterized in that, The optimal window length includes: The local feature strength and consistency of data points in each dimension are normalized using a negative hyperbolic tangent function. The normalized local feature strength and data consistency are then multiplied by [a specific factor]. Add 1 to perform a linear transformation. Multiply the results of the two linear transformations to obtain the window correction coefficient. Multiply the window correction coefficient by the preset initial moving average window and round up to obtain the optimal window length for each dimension's data points at each time point.

9. The intelligent analysis method for machining accuracy of CNC machine tools according to claim 1, characterized in that, Preprocessing the processing data includes: For each dimension of the processed data, the interpolation method is used to fill in missing values, thus completing the missing data points and ensuring data integrity.

10. A machining accuracy intelligent analysis system for CNC machine tools, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a machining accuracy intelligent analysis method for CNC machine tools according to any one of claims 1-9.

Images