Method for diagnosing running states of lead screws and bearings of shafts of machine tool

Abstract

The invention discloses a method for diagnosing operation states of lead screws and bearings of all shafts of a machine tool, and relates to the technical field of machine tool state monitoring and diagnosis. Comprising the following steps: collecting current data; constructing a current change curve and dividing a zeroing operation time interval and a processing time interval; constructing an offset trend chart related to a switching-on and switching-off zeroing operation; obtaining a time interval when a lead screw assembly is in an acceleration and deceleration stage, and obtaining an energy feature vector and sensitive feature data in the energy feature vector. Constructing a wear state diagnosis model, performing diagnosis evaluation, determining a wear position and a wear trend, evaluating the wear state of the bearing, and the like. According to the method, the wear state of the lead screw assembly is diagnosed by collecting the current data, analyzing the sensitive characteristic data and constructing the model, meanwhile, the wear state of the bearing component is determined according to the periodic change of the current data deviation value, the wear state of the lead screw assembly and the wear state of the bearing are conveniently and rapidly monitored, the monitoring cost of the mode is lower, and the monitoring accuracy is improved. The method is not influenced by environmental parameters and is high in precision.

Description

Technical Field

[0001] The invention relates to the technical field of machine tool status monitoring and diagnosis, and in particular to a method for diagnosing the operating status of lead screws and bearings of various axes of a machine tool. Background Art

[0002] The operating status of ball screws and bearings on CNC machine tools has a vital impact on the performance, efficiency, accuracy and operating stability of machine tools. Ball screws and bearings are key components for machine tools to achieve precision motion. If the operating status is not good, it will directly lead to a decrease in product accuracy and abnormal noise from the machine tools. Due to the structural complexity and diverse functions of CNC machine tools, sometimes it is not easy to find and troubleshoot the wear of ball screws and bearings, resulting in downtime, long maintenance time and other production-impacting time.

[0003] Existing monitoring and diagnostic methods include monitoring the temperature data, vibration data, and acoustic emission signal data of the screw and bearing. However, since the environmental parameters of the machine tool are prone to change when it is in operation, it is easy to interfere with the relevant monitoring data. In addition, the relevant monitoring methods require the use of corresponding sensors, which have very limited installation space inside compact components and are costly. Therefore, we propose a method for diagnosing the operating status of the screws and bearings of each axis of the machine tool. Summary of the invention

[0004] The object of the present invention is to provide a method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool.

[0005] The technical problem solved by the present invention is: how to collect the current data of the spindle during the zeroing operation and the acceleration and deceleration stages during the machining process, analyze the characteristic data of the current data, and build a model to diagnose the screw assembly, judge the wear state of the screw assembly, and determine the wear state of the bearing component according to the periodic change of the current data offset value, so as to solve the problem that the existing monitoring methods are easily affected by environmental interference and have high costs.

[0006] The present invention can be implemented by the following technical solution: A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool comprises the following steps:

[0007] Step 1: Set a collector on the power motor and collect current data in real time;

[0008] Step 2: Obtain the effective current value and construct a current change curve, dividing the power-on, power-off and zeroing operation time intervals and the processing time intervals;

[0009] Step 3: Based on the current change curve segment of the zeroing operation, construct an offset trend diagram about the power-on and power-off zeroing operations;

[0010] Step 4: Obtain the time interval during which the lead screw assembly is in the acceleration and deceleration stages within the processing time interval;

[0011] Step 5: Perform wavelet packet decomposition according to the current data in the time interval of the acceleration and deceleration state to obtain the energy feature vector, and select feature data sensitive to the wear state;

[0012] Step 6: Based on the BP neural network, the wear state diagnosis model is constructed by taking the characteristic data sensitive to the wear state and the processing technology name as the model input and the wear level of the screw assembly as the output;

[0013] Step 7: Use the wear state diagnosis model to diagnose and evaluate the wear level of the screw assembly corresponding to the main shaft when the screw assembly is in a specific processing process;

[0014] Step 8: Use the power-on and power-off zero offset trend graph to determine the wear location and wear trend;

[0015] Step 9: Based on whether there is uniform periodic change and offset in the current data during the zeroing process, evaluate whether there is wear on the bearing and the degree of wear.

[0016] A further technical improvement of the present invention is that the specific steps of constructing the offset trend diagram of the zeroing operation in step 3 include:

[0017] Select the earliest opening and closing zeroing operation curve segment in the current change curve as the reference curve;

[0018] Select a curve segment of the power on / off reset operation at a certain interval and align it with the time start / end point of the corresponding curve segment;

[0019] Randomly select multiple time points, calculate the current offset value of each curve segment relative to the reference curve at that time point, and finally calculate the root mean square deviation δ of any curve relative to the reference curve m , m is a positive integer greater than 1, and its maximum value is the same as the number of selected curves;

[0020] Based on coordinates (times,δ times ) construct a coordinate system, and times = 2, 3...m max , connect the coordinate points in sequence to obtain the on / off offset trend graph.

[0021] A further technical improvement of the present invention is that the method for obtaining the period of time when the lead screw assembly is in the acceleration phase in step 4 includes:

[0022] Calculate the rate of change of current over time I'(t) during the processing period. When I'(t) is greater than the set threshold, the current value is in an increasing state at that time point, and mark the first time t in the increasing state.0 , when t 0 +Δt 1 Each acquisition point in the time period satisfies I(t)>I(t 0 ), it means that the current is in the rising state, and the starting time is t 0 ;

[0023] When there is I”(t 1 )=0, record the current value I(t 1 ), and t 0 +Δt 2 Each collection point within the time period satisfies I(t 1 )-λ<I(t)<I(t 1 )+λ, it means that the current is in a steady state after rising, and the starting time of the steady state is t 1 , where λ is the allowed fluctuation threshold;

[0024] Similarly, based on the acquisition logic of the current rising state and the steady-state start time after the rise, the current falling state and the steady-state start time after the fall are obtained, which are marked as t 2 ,t 3 ;

[0025] Thus, the time interval [t 0 ,t 3 ]; Similarly, the time interval [t' 0 ,t' 3 ].

[0026] A further technical improvement of the present invention is that the process of performing wavelet packet decomposition in step 5 includes:

[0027] Select the Symlets wavelet basis function, select the decomposition layer number as five, calculate the energy of the wavelet packet coefficient of each node to obtain the energy value of the corresponding sub-band, the calculation formula is: E a,b =∑|x a,b (k)| 2 , E a,b represents the energy of the bth node in the ath layer, x a,b (k) represents the kth wavelet packet coefficient of the corresponding node; the energy values ​​of all nodes are arranged in hierarchical order, and the energy feature vector is constructed after normalization.

[0028] The further technical improvement of the present invention is that the wear state uses the position accuracy error of the machined workpiece and the wear coefficient calculated by the total working time as the characterization parameter, and the calculation formula is used to calculate The wear coefficient of the corresponding screw assembly is calculated, where ε is the wear coefficient, Δx is the position accuracy error, T is the total working time of the screw assembly, and T L The service life of the screw assembly.

[0029] A further technical improvement of the present invention is that the characteristic data sensitive to the wear state is the characteristic data sensitive to the wear coefficient, and the selection process is: statistical analysis is performed on the characteristic values ​​in the energy characteristic vector and the wear coefficient to obtain nodes where the characteristic values ​​change significantly relative to different wear coefficients, and the sensitive characteristics are determined through expert screening, thereby selecting the corresponding characteristic data.

[0030] A further technical improvement of the present invention is that a fixed diagnostic evaluation cycle is set in step seven, and a diagnostic evaluation is performed once each time the shutdown and zeroing operation is completed.

[0031] A further technical improvement of the present invention is that the method for determining the wear position includes: calculating the time of the slide movement from the aligned time start point / time end point, and since the main shaft rotation speed is consistent at this time, the corresponding wear position is determined.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. The present invention collects current data of the spindle during the zeroing operation and the acceleration and deceleration stages during the machining process, analyzes the characteristic data of the current data, and constructs a model to diagnose the screw assembly and judge the wear state of the screw assembly. At the same time, the wear state of the bearing component is determined according to the periodic change of the current data offset value. The wear state of the screw assembly and the bearing is monitored conveniently and quickly. Compared with other monitoring methods, the monitoring cost is lower, and it is almost not affected by environmental parameters and has higher accuracy.

[0034] 2. The present invention can determine the location where wear occurs by measuring the time the slide moves during the zeroing operation and the offset value of the current data within the corresponding time. Compared with other methods, it is more conducive to inspection and maintenance by staff. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0036] Figure 1 The present invention is a schematic diagram of the method execution flow. DETAILED DESCRIPTION

[0037] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0038] See also Figure 1 As shown, a method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool specifically comprises the following steps:

[0039] Step 1: Select the feed spindle and set the collector for the power motor

[0040] The spindle direction with the largest mass of the driven slide table when performing the feed motion is selected. In this case, the load of the screw assembly is large, the wear is also faster and more obvious, and the data differentiation is high; in this embodiment, the transmission screw and support bearing in the X direction are selected as the diagnosis target, the power motor adopts a permanent magnet synchronous motor, and the Hall sensor is used to collect the current and speed of the power motor in real time, and the collected real-time data is stored in time sequence; in addition, the workpiece load and processing technology (turning, milling, drilling, boring, grinding, etc.) are recorded during each operation;

[0041] Step 2: Obtain the effective current value and divide the intervals in the constructed current change curve

[0042] The AC current is converted into an equivalent DC current by the dq coordinate axis conversion method, and the effective current value is

[0043] A curve of the change of effective current value with respect to time is drawn, and the time period of zeroing operation at each power on and off is marked in the current change curve. During this time period, the spindle slide is in a no-load state, and the spindle speed remains constant during this stage, and the whole process is relatively stable; in addition, the time period between the front and rear zeroing operation time periods is defined as the processing time period.

[0044] Step 3: Build an offset trend graph for zeroing operation

[0045] Extract the zeroing period in the current change curve, build a new current-time coordinate system, select the time starting point of the zeroing period when the power is turned on once every certain number of times, align them, and calculate the deviation of each curve segment relative to the first curve; at the same time, select the time ending point of the zeroing period when the power is turned off once every certain number of times, align them, and calculate the deviation of each curve segment relative to the first curve, and finally form a deviation trend graph;

[0046] The construction method of the offset trend graph includes the following steps, taking the zeroing operation at startup as an example:

[0047] (1) The earliest curve segment in the selected curve is used as the reference curve;

[0048] (2) The current change curve at startup is aligned with the time starting point. Since the spindle speed is constant during the zeroing process, the displacement of the position of the slide relative to the lead screw in each zeroing operation is the same within the same time. That is, when the slide moves to a certain position relative to the lead screw, the offset value between the corresponding effective current values ​​can largely reflect the wear degree of the corresponding lead screw at the corresponding position.

[0049] (3) Randomly select multiple time points, obtain the effective current value at the corresponding time point according to the selected time point, and calculate the root mean square deviation δ of the curve relative to the reference curve on a specific curve m , the calculation formula is:

[0050] Among them, t i represents the selected i-th time point and i≤N is a positive integer, y 1 (t), y m (t) represent the effective current value on the reference curve and the effective current value on the specific curve respectively, m is a positive integer greater than 1, and the maximum value is the same as the number of selected curves.

[0051] (4) Associate the RMS deviation values ​​according to the startup times of the corresponding curves to construct a key-value pair (times,δ times ), where times = 2, 3...m max , use the key-value pairs to build a coordinate system according to the order of the number of startup times, and use the key-value pairs as coordinate points to connect them in sequence to obtain the startup zero offset trend graph.

[0052] Following the above steps, a shutdown zero offset trend graph can be constructed in the same way.

[0053] Step 4: Obtain the time period when the screw assembly is in the acceleration and deceleration stage within the processing time interval

[0054] Select any processing technology, extract the curve segments in the processing time interval of the corresponding processing technology from all current change curves, and distinguish the acceleration stage and the deceleration stage in the curve segments;

[0055] Calculate the difference formula: Calculate the rate of change of current over time I'(t), and compare the rate of change I'(t) with the set threshold. When I'(t) is greater than the set threshold, it means that the current value is in an increasing state at this point. The first moment in the increasing state is marked as t 0 ; and t 0 +Δt 1 Each sampling point in the time period (generally, Δt 1 Set to 0.5s), all satisfy I(t)>I(t 0), it means that the current is in a rising state, and the starting time of the rising state is t 0 ;

[0056] Calculate the curvature of the current change curve I”(t). When there is I”(t 1 )=0, it indicates that the current rising stage ends, and the current value I(t 1 ), and at t 1 +Δt 2 Each collection point within the time period satisfies I(t 1 )-λ<I(t)<I(t 1 )+λ, it means that the current is in a steady state after rising, and the starting time of the steady state is t 1 , λ is the allowed fluctuation threshold;

[0057] Based on the logic of the current rising state, the starting time t when the current is in the falling state can be obtained by the same logic. 2 And the starting time t when the current is in a steady state after decreasing 3 , specifically:

[0058] Calculate the difference formula: Calculate the rate of change of current over time I'(t), and compare the rate of change I'(t) with the set threshold. When -I'(t) is greater than the set threshold, it means that the current value is in a decreasing state at this point. The first moment in the decreasing state is marked as t 2 ; and t 2 +Δt 1 Each sampling point in the time period satisfies I(t)<I(t 2 ), it means that the current is in a rising state, and the starting time of the rising state is t 2 ;

[0059] Calculate the curvature of the current change curve I”(t). When there is I”(t 3 )=0, it indicates that the current drop phase ends, and the current value I(t 3 ), and at t 3 +Δt 2 Each collection point within the time period satisfies I(t 3 )-λ<I(t)<I(t 3 )+λ, it means that the current is in a steady state after decreasing, and the starting time of the steady state is t 3 , λ is the allowed fluctuation threshold.

[0060] Combining the above process, we can get the time interval [t 0 ,t 3]; Similarly, the time interval [t' 0 ,t' 3 ].

[0061] Step 5: Extract features based on the current data of the acceleration and deceleration periods

[0062] When the screw assembly is in an accelerating or decelerating state, the axial load of the screw assembly will change, and the current impact and frequency change of the motor current signal caused by the wear of the screw assembly will be more obvious, which is easy to improve the sensitivity of the data and improve the monitoring accuracy;

[0063] The current data of the acceleration and deceleration states in different processing technologies are classified and sorted, and the current data in this state in each processing technology is respectively introduced into the wavelet packet decomposition algorithm, so as to extract the energy characteristics characterizing the wear state of the screw assembly, specifically:

[0064] The decomposition principle of the wavelet packet decomposition algorithm is mainly to decompose the signal of the upper layer into high-frequency components and low-frequency components from the perspective of frequency band analysis, and then decompose the low-frequency components and high-frequency components of the upper layer into high-frequency components and low-frequency components one by one, forming nodes distributed in a tree shape;

[0065] Select the Symlets wavelet basis function, select the decomposition layer number as five, calculate the energy of the wavelet packet coefficient of each node to obtain the energy value of the corresponding sub-band, the calculation formula is: E a,b =∑|x a,b (k)| 2 , E a,b represents the energy of the bth node in the ath layer, x a,b (k) represents the kth wavelet packet coefficient of the corresponding node;

[0066] The energy values ​​of all nodes are arranged in hierarchical order, and then normalized to construct an energy feature vector, which represents the energy distribution of the current signal in different frequency bands;

[0067] Frequency band data that is sensitive to the wear state is selected from the vector as feature data for subsequent diagnosis of the wear state of the lead screw assembly.

[0068] Step 6: Build a wear status diagnosis model

[0069] Based on the position accuracy of the workpiece in the monitored target spindle movement direction in the historical data and the total working time of the corresponding screw assembly as the characterization parameters of its wear state, its wear coefficient is evaluated according to the calculation formula: Where ε is the wear coefficient, Δx is the position accuracy error, T is the total working time of the screw assembly, T Lis the service life of the screw assembly; the significance of this calculation formula is that the error before running-in is not caused by the wear of the screw assembly, but the closer to the service life, the accuracy error is mainly caused by wear, so as to correct the deviation when using position accuracy error to evaluate the wear state in different time periods;

[0070] Based on the historical current data of each processing technology in the time section where the screw assembly is in acceleration and deceleration state, the energy eigenvector is solved based on the wavelet packet decomposition algorithm, and the eigenvalues ​​in the eigenvector and the wear coefficient are combined for statistical analysis (including mean and variance), and the nodes where the eigenvalues ​​change significantly relative to different wear coefficients are obtained. After expert screening, they are determined to be sensitive features. The sensitive frequency band selection method in step five is the same;

[0071] Based on the range of the wear coefficient, multiple wear state interval thresholds are set to match the corresponding wear levels, including: running-in period, normal state, early wear period, primary wear, secondary wear and tertiary wear;

[0072] Based on the BP neural network model, the characteristic values ​​of multiple sensitive features and the name of the processing technology are taken as input, and the corresponding wear level is taken as output. The training set and test set are constructed, and the wear status diagnosis model is trained.

[0073] Step 7: Use the model to evaluate the wear status of the screw assembly

[0074] In the actual production process, the data acquisition and processing of steps 4 and 5 are followed to construct the model input, so as to match the wear level of the screw assembly of the main shaft and send it to the staff;

[0075] It should be noted that the automatic diagnosis cycle is set by a timer, and a diagnosis is performed once at regular intervals during operation, and once each time the power is shut down and reset to zero.

[0076] Step 8: Use the power-on and power-off zero offset trend graph to determine the wear location and wear trend

[0077] When the diagnostic wear level enters the first level of wear, the change trend and speed of the root mean square deviation are determined based on the most recent zero offset trend graph, thereby determining whether the screw assembly is in accelerated wear;

[0078] In addition, according to the moment when the current offset value exceeds the set value during the zeroing process, the position of the screw nut is determined, and then the corresponding wear position is determined, providing a reference for the staff to determine the wear position on site.

[0079] Step 9: Determine the wear status of bearing components

[0080] Observe whether there is current data with uniform periodic changes and offset during the zeroing process. If so, it means that there is wear on the bearing. The larger the offset difference of the current data, the more serious the wear.

[0081] It should be noted that the execution order of step five and step six can be swapped.

[0082] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool, characterized in that: The steps include: Step 1: Set a collector on the power motor and collect current data in real time; Step 2: Obtain the effective current value and construct a current change curve, dividing the power-on, power-off and zeroing operation time intervals and the processing time intervals; Step 3: Based on the current change curve segment of the zeroing operation, construct an offset trend diagram about the power-on and power-off zeroing operations; Step 4: Obtain the time interval during which the lead screw assembly is in the acceleration and deceleration stages within the processing time interval; Step 5: Perform wavelet packet decomposition according to the current data in the time interval of the acceleration and deceleration state to obtain the energy feature vector, and select feature data sensitive to the wear state; Step 6: Based on the BP neural network, the wear state diagnosis model is constructed by taking the characteristic data sensitive to the wear state and the processing technology name as the model input and the wear level of the screw assembly as the output; Step 7: Use the wear state diagnosis model to diagnose and evaluate the wear level of the screw assembly corresponding to the main shaft when the screw assembly is in a specific processing process; Step 8: Use the power-on and power-off zero offset trend graph to determine the wear location and wear trend; Step 9: Based on whether there is uniform periodic change and offset in the current data during the zeroing process, evaluate whether there is wear on the bearing and the degree of wear.

2. A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool according to claim 1, characterized in that: The specific steps of constructing the offset trend graph of the zeroing operation in step 3 include: Select the earliest opening and closing zeroing operation curve segment in the current change curve as the reference curve; Select a curve segment of the power on / off reset operation at a certain interval and align it with the time start / end point of the corresponding curve segment; Randomly select multiple time points, calculate the current offset value of each curve segment relative to the reference curve at that time point, and finally calculate the root mean square deviation δ of any curve relative to the reference curve m , m is a positive integer greater than 1, and its maximum value is the same as the number of selected curves; Based on coordinates (times,δ times ) construct a coordinate system, and times = 2, 3...m max , connect the coordinate points in sequence to obtain the on / off offset trend graph.

3. A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool according to claim 1, characterized in that: The method for obtaining the period of time when the lead screw assembly is in the acceleration phase in step 4 includes: Calculate the rate of change of current over time I'(t) within the processing period. When I'(t) is greater than the set threshold, the current value is in an increasing state at this time point. Mark the first moment t0 in the increasing state. When each acquisition point in the time period t0+Δt1 satisfies I(t)>I(t0), it means that the current is in an increasing state, and the starting time is t0. When I”(t1)=0, record the current value I(t1) at this time, and each acquisition point in the time period t0+Δt2 satisfies I(t1)-λ<I(t)<I(t1)+λ, which means that the current is in a steady state after rising, and the starting time of the steady state is t1, where λ is the allowable fluctuation threshold; Similarly, based on the acquisition logic of the current rising state and the steady-state starting time after the rising state, the current falling state and the steady-state starting time after the falling state are obtained, which are marked as t2 and t3 respectively; Thus, the time interval [t0, t3] representing the acceleration state of the screw assembly in a certain processing time period of a processing technology can be obtained; similarly, the time interval [t'0, t'3] representing the deceleration state of the screw assembly in the corresponding processing time period of the processing technology can be obtained.

4. A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool according to claim 1, characterized in that: The process of wavelet packet decomposition in step 5 includes: Select the Symlets wavelet basis function, select the decomposition layer number as five, calculate the energy of the wavelet packet coefficient of each node to obtain the energy value of the corresponding sub-band, the calculation formula is: E a,b =∑|x a,b (k)| 2 , E a,b represents the energy of the bth node in the ath layer, x a,b (k) represents the kth wavelet packet coefficient of the corresponding node; the energy values ​​of all nodes are arranged in hierarchical order, and the energy feature vector is constructed after normalization.

5. A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool according to claim 1, characterized in that: The wear state uses the position accuracy error of the machined workpiece and the wear coefficient calculated from the total working time as the characterization parameter, and the calculation formula is used to calculate the wear state. The wear coefficient of the corresponding screw assembly is calculated, where ε is the wear coefficient, Δx is the position accuracy error, T is the total working time of the screw assembly, and T L The service life of the screw assembly.

6. A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool according to claim 5, characterized in that: The characteristic data sensitive to the wear state is the characteristic data sensitive to the wear coefficient. The selection process is as follows: statistical analysis is performed on the eigenvalues ​​in the energy eigenvector and the wear coefficient to obtain nodes where the eigenvalues ​​change significantly relative to different wear coefficients. The sensitive features are determined through expert screening, and the corresponding characteristic data are selected.

7. A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool according to claim 1, characterized in that: In step seven, a fixed diagnostic evaluation cycle is set, and a diagnostic evaluation is performed each time the shutdown and zeroing operation is completed.

8. A method for diagnosing the operating status of the lead screws and bearings of each axis of a machine tool according to claim 1, characterized in that: The method for determining the wear position includes: calculating the time of the slide movement from the aligned time start point / time end point, and since the main shaft rotation speed is consistent at this time, the corresponding wear position is determined.

Images