Intelligent suppression method for signal noise in self-adaptive milling process
Through the intelligent signal noise suppression method of adaptive milling process, the root mean square analysis and dynamic threshold calculation are used to automatically screen the stable milling signal segment, which solves the problems of low tool wear detection accuracy and waste of resources in the existing technology, and achieves more efficient milling signal processing and system response.
Patent Information
- Application Number
- CN202411233354.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-11-12
AI Technical Summary
In the prior art, tool wear detection relies on the operator's experience and manual pre-processing, resulting in low accuracy, waste of resources and potential economic risks, and the inability to achieve automatic adaptation of thresholds.
The intelligent signal noise suppression method of adaptive milling process is adopted, and the stable milling signal segment is automatically filtered out by reading the milling signal segment, root mean square analysis and dynamic threshold calculation, thereby reducing manual operation of the operator to realize adaptive adjustment of the threshold.
It improves the accuracy and system adaptability of milling signal processing, reduces labor costs, reduces resource waste, and enhances the response speed and accuracy of the monitoring system.
Smart Images

Figure CN118927004A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of milling signal processing, and in particular relates to an adaptive milling process signal noise intelligent suppression method. Background Art
[0002] The machinery manufacturing industry is developing towards automation and intelligence, and has set higher standards for production efficiency, product quality and cost control. In this process, machine tools, as the core equipment of mechanical processing, play a vital role, and they continue to provide the necessary power for the manufacturing process. As a key component of machine tools, cutting tools are responsible for removing excess material from the workpiece. However, cutting tools will inevitably suffer from wear and tear during use, and may even break, which is directly related to the quality and production efficiency of processing.
[0003] At present, the detection of tool wear mainly relies on the operator's experience and the physical signals generated during the cutting process to judge the tool status, and the tool is replaced irregularly based on this. Although this practice can prevent processing defects caused by tool problems, in order to ensure product quality, it often leads to premature replacement of tools, which not only increases production costs, but also causes waste of resources, which is contrary to the concept of green manufacturing. On the contrary, if the tool is not replaced in time, it may affect product quality and bring potential economic risks. In order to solve this problem, the tool monitoring and early warning system came into being. By monitoring the tool status in real time, the system can identify and warn of abnormal conditions of the tool in advance, thereby avoiding product quality problems or production accidents caused by tool damage. Therefore, in the early stage of system development, it is particularly important to effectively preprocess the tool milling signal, which is directly related to the accuracy and response speed of the monitoring system.
[0004] However, the effective information of the tool mainly comes from the stable milling process, so the milling signal needs to be preprocessed to filter out the stable milling signal segments and remove the noise stages such as idle cutting, cutting in and cutting out. At present, the preprocessing of milling signals usually relies on manual drawing and subjective judgment of operators to identify the stable milling stage. This method not only increases the burden on operators, but also is prone to errors. In addition, operators usually use fixed thresholds when making judgments, which can only be applied to specific milling signals, resulting in low accuracy. Different milling signals require manual adjustment of thresholds, and automatic adaptation of thresholds cannot be achieved. Summary of the invention
[0005] In order to solve the deficiencies in the prior art, the present invention proposes an adaptive milling process signal noise intelligent suppression method, which can not only automatically preprocess the tool milling signal, but also improve the accuracy of signal processing by dynamically adjusting the threshold, thereby enhancing the adaptability and reliability of the entire system.
[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows: An adaptive milling process signal noise intelligent suppression method comprises the following steps: S1: Read the milling signal segment with a duration of 2T; S2: Use signal segment segmentation and root mean square (RMS) analysis to identify whether the milling signal segment in step S1 contains a valid milling signal. If no valid signal is detected, reset and return to step S1 to read a new milling signal segment; if a valid signal is detected, proceed to the next step; S3: performing adaptive threshold calculation on the 2T time-length signal segment confirmed to contain a valid milling signal in step S2; S4: Screen the signal segments using the threshold value calculated in step S3 to identify quasi-valid signal segments. Then, further verify the validity of the quasi-valid signal segments by the duration and RMS value of the signal segments. If the verification result is invalid, reset and return to step S1 to read a new milling signal segment; if the verification result is valid, mark the signal segment as a fully valid signal segment; S5: merging the fully valid signal segments identified in step S4, then resetting, and then returning to step S1 to read a new milling signal segment; S6: by looping through steps S1 to S5, the continuous reading and evaluation of the entire milling signal sequence is completed, and all identified fully valid signal segments are merged to achieve intelligent preprocessing of the milling signal; Among them, in step S1, T is defined as the cycle of a single stroke of milling processing; the reset operation in steps S2 and S4 refers to clearing the milling signal segment of 2T duration in step S1; the reset operation in step S5 refers to clearing the milling signal segment from the starting point in step S1 to a seconds after the end point of the quasi-valid signal segment in step S4.
[0007] Furthermore, the a second is set to 5 seconds, and the total duration of the 2T duration signal segment in step S1 is 100 seconds.
[0008] Furthermore, in step S2, the following sub-steps are further included: S201: time-segment the milling signal segment in step S1, and the duration of each sub-signal segment is t; S202: Calculate the root mean square value of each sub-signal segment, store the result in an array, and then sort the array from small to large; S203: Select values at positions X% and Y% from the sorted array as Z1 and Z2, where X% is set to 15%, Y% is set to 85%, Z1 represents a possible noise stage, and Z2 represents a possible stable milling stage; S204: Determine whether the milling signal segment in step S1 contains a valid milling signal by judging whether the condition Z1<0.02*Z2 is satisfied. If the condition is satisfied, there is a valid signal, and the next step is entered; if not, there is no valid signal, and a reset is performed, and step S1 is returned to read a new milling signal segment.
[0009] Furthermore, in step S3, the adaptive threshold of the 2T duration signal segment in which the effective milling signal segment exists is calculated according to the threshold formula (Z1+Z2) / 2.
[0010] Furthermore, the step S4 includes the following sub-steps: S401: Identify the start point and end point of the signal by comparing the RMS value with the threshold value to determine a quasi-valid signal segment; S402: Determine whether the duration of the quasi-valid signal segment is within the estimated stable milling duration range of ±5 seconds. If it is within the range, delete b seconds from the beginning and end of the quasi-valid signal segment, and then proceed to the next step; if it is not within the range, reset and return to step S1 to read a new milling signal segment; S403: Use the RMS value to verify whether the quasi-valid signal segment deleted in step S402 is a completely valid signal segment. If the verification is successful, proceed to the next step; if not, reset and return to step S1 to read a new milling signal segment. The initialization in step S402 and step S403 refers to deleting the milling signal segment with a duration of 2T in step S1.
[0011] Furthermore, in step S401, the following sub-steps are included: S401a: time-segment the milling signal segment in step S1, and the duration of each sub-signal segment is t; S401b: traverse each sub-signal segment and calculate its RMS value; S401c: By comparing the RMS values of the first 20 sub-signal segments and the last 20 sub-signal segments with the threshold, the starting point and the ending point of the signal are identified, thereby determining a quasi-valid signal segment.
[0012] Further, in step S401c, the judgment condition for the starting point signal segment is: at least 75% of the root mean square values of the first 20 sub-signal segments are less than the threshold, and at least 75% of the root mean square values of the last 20 sub-signal segments are greater than the threshold; the judgment condition for the ending point signal segment is: at least 75% of the root mean square values of the first 20 sub-signal segments are greater than the threshold, and at least 75% of the root mean square values of the last 20 sub-signal segments are less than the threshold.
[0013] Further, in step S402, if the duration of the quasi-effective signal segment is within the estimated stable milling duration range of 25 seconds to 35 seconds, then b seconds is set to 5 seconds.
[0014] Further, in step S403, the condition for a quasi-valid signal segment to be judged as a completely valid signal segment is that 5 times the root mean square value of the first 5% of the signal segment is less than the root mean square value of the last 5%, or 5 times the root mean square value of the last 5% is less than the root mean square value of the first 5%.
[0015] Furthermore, the RMS calculation formula is as follows: Where: represents the root mean square, Indicates the total number of data. Indicates The value of the data.
[0016] Compared with the prior art, the present invention has the following significant advantages: (1) The present invention determines whether there is a valid signal in the milling signal segment of 2T duration through automatic signal segment segmentation and root mean square (RMS) analysis, and performs dynamic threshold calculation. Using these thresholds, quasi-valid signal segments are further screened out, and their validity is verified by the duration and RMS value of the signal segments. The valid signal segments are then spliced, and this process is executed in a loop to achieve automatic reading and judgment of the entire signal sequence. The present invention automatically screens out signal segments for stable milling, and suppresses noise stages such as idle tool movement, cutting in and cutting out. This fully automated process reduces the need for manual operation by operators and reduces the workload. At the same time, through continuous initialization and loop execution, the present invention achieves adaptive adjustment of thresholds, improves the accuracy of milling signal processing, and does not require operators to manually adjust thresholds, saving labor costs.
[0017] (2) The present invention specifically sets the milling signal segment of 2T duration to 100 seconds, and in the initialization process of step S5, a second is set to 5 seconds, that is, a 5-second signal segment is deleted from the starting point of step S1 to the end point of the quasi-valid signal segment in step S4. These specific settings further improve the accuracy of milling signal processing.
[0018] (3) In step S2, the present invention divides the milling signal segment into sub-segments of length t, sorts the RMS values, selects the values at specific percentage positions, and determines whether the signal segment contains a valid milling signal by the condition Z1<0.02*Z2. This method achieves accurate identification of valid signal segments through precise signal segmentation and threshold calculation.
[0019] (4) In step S4, the present invention gradually determines the beginning and end points and duration of the quasi-valid signal segment, and deletes b seconds from the signal segment in step S402, and finally confirms the completely valid signal segment. In step S401, by setting specific judgment conditions for the 20 sub-segments before and after the signal segment, the start and end point signal segments are effectively identified.
[0020] (5) The present invention sets the estimated stable milling duration within the range of 25 seconds to 35 seconds, which is a duration selection optimized through practice and provides a more accurate time basis for signal validity judgment. At the same time, b seconds in step S402 is set to 5 seconds, which further improves the accuracy of effective milling signal division. The judgment condition in step S403 ensures the accurate judgment of the completely effective signal segment. In addition, the root mean square calculation method adopted by the present invention is simple and sensitive, and can accurately reflect the characteristics of data distribution. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 This is the overall flow chart of the method for intelligently suppressing signal noise in the adaptive milling process of the present invention.
[0022] Figure 2 The flowchart of the present invention is to determine whether a valid milling signal segment is included.
[0023] Figure 3 The present invention is a flow chart for determining and identifying quasi-effective milling signal segments.
[0024] Figure 4 The present invention is a flow chart for determining a quasi-valid signal segment based on the comparison between the root mean square and the threshold. DETAILED DESCRIPTION
[0025] The present invention is further described below in conjunction with the accompanying drawings and examples. The embodiments of the present invention include but are not limited to the following examples. Example
[0026] like Figures 1 to 4 As shown, this embodiment provides an adaptive milling process signal noise intelligent suppression method, comprising the following steps: S1: Read the milling signal segment with a duration of 2T; S2: Evaluate whether the signal segment obtained in step S1 contains a valid milling signal by segmenting the signal segment and calculating the root mean square (RMS) value. If no valid signal is detected, reset the system and return to step S1 to read a new signal segment; if a valid signal is detected, proceed to the next step; S3: For the effective milling signal segment identified in step S2, calculate the threshold value of its period 2T; S4: Using the threshold determined in step S3, filter the signal segments and identify quasi-valid signal segments. Further verify the validity of the signal segments by their duration and RMS value. If the verification fails, reset the system and return to step S1; if the verification succeeds, confirm that the signal segments are fully valid. S5: merging the fully valid signal segments identified in step S4, resetting the milling signal segments, and then returning to step S1 to read new signal segments; S6: by looping through steps S1 to S5, the continuous reading and evaluation of the entire milling signal sequence is completed, and all confirmed fully valid signal segments are merged to achieve intelligent preprocessing of the milling signal; Among them, in step S1, T is defined as the cycle of a single stroke of milling processing; "initialization" in step S2 and step S4 refers to clearing the 2T long milling signal segment in step S1; "initialization" in step S5 refers to clearing the 2T long signal segment in step S1, from the starting point to a seconds after the end point of the quasi-valid signal segment in step S4.
[0027] In this embodiment, step S1 involves reading a 100-second milling signal segment, which is equivalent to two milling cycles. In steps S2 and S4, if no valid signal is found, the 100-second signal segment is deleted. In step S5, if the signal segment is valid, the signal segment from the starting point to 5 seconds after the end point of the quasi-valid signal segment is deleted. The milling signal in this embodiment refers to the vibration, current and other signals collected by the sensor during the milling process, and its noise mainly comes from the stages of idle tool passing, cutting in and cutting out.
[0028] In this embodiment, step S2 includes the following steps: S201: Divide the signal segment into sub-segments with a duration of t; S202: Calculate the root mean square value of each sub-segment and arrange them in order from small to large; S203: Selecting the values of the X% and Y% positions of the array after arrangement as Z1 and Z2, wherein X% is set to 15%, Y% is set to 85%, Z1 represents a possible noise stage, and Z2 represents a possible stable milling stage; S204: Determine whether the condition Z1<0.02*Z2 is established. If so, it is considered to contain a valid milling signal segment. Otherwise, reset the system and return to step S1.
[0029] Among them, the duration t is set to 0.1 second. The setting of a smaller duration t can achieve a finer division of the signal segment, thereby improving the accuracy of judging the effectiveness of the milling signal segment. In this embodiment, X% is set to 15%, Y% is set to 85%, and the setting of the coefficient 0.02 in Z1<0.02*Z2 is based on the optimization parameter selection during the milling process.
[0030] In step S3 of the present embodiment, the threshold of the milling signal segment with a duration of 2T during which there is a valid milling signal segment is calculated according to the threshold formula (Z1+Z2) / 2. The calculated threshold is more accurate. At the same time, in the loop execution of steps S1-S5, the threshold calculated each time is adapted to the current milling signal segment, thereby realizing active adaptation of the threshold and overcoming the problem that a single threshold is usually used in manual operation, which is only effective for specific milling signals, has low accuracy, and requires manual change of the threshold for different milling signals.
[0031] In this embodiment, step S4 includes the following steps: S401: Identify the start and end points of the signal by comparing the RMS value and the threshold, and determine a quasi-valid signal segment; S402: If the duration of the quasi-valid signal segment is within ±5 seconds of the estimated stable milling duration, b seconds are deleted from the beginning and end of the signal segment; if not, the system is reset and returns to step S1; S403: Determine whether the deleted quasi-valid signal segment is completely valid based on the RMS value, if so, continue; otherwise, reset the system and return to step S1.
[0032] Among them, the initialization in step S402 and step S403 refers to deleting the milling signal segment with a duration of 100 seconds in step S1. This embodiment judges the head and tail endpoints of the quasi-valid signal segment in turn, and judges the length of the quasi-valid signal segment. It judges whether the quasi-valid signal segment after deleting b seconds in step S402 is a completely valid signal segment, and finally obtains an accurate completely valid signal segment.
[0033] In this embodiment, step S401 includes the following steps: S401a: Segment the milling signal segment in step S1 into signal segments, where the duration of each signal segment is t; S401b: traverse each signal segment and calculate the root mean square of each signal segment; S401c: By comparing the RMS values of the first 20 signal segments and the last 20 signal segments with the threshold, the starting point signal segment and the ending point signal segment are determined to obtain a quasi-valid signal segment.
[0034] Wherein, in step S401c, the start and end signal segments are identified by comparing the root mean square values of the first 20 sub-signal segments and the last 20 sub-signal segments with the threshold. The judgment condition of the start point signal segment is: more than 75% of the root mean square values of the first 20 sub-signal segments are less than the threshold, and more than 75% of the root mean square values of the last 20 sub-signal segments are greater than the threshold; the judgment condition of the end point signal segment is: more than 75% of the root mean square values of the first 20 sub-signal segments are greater than the threshold, and more than 75% of the root mean square values of the last 20 sub-signal segments are less than the threshold.
[0035] For the convenience of judgment, it is also possible to make a direct judgment based on the number of signal segments. If more than 15 of the first 20 signal segments have RMS values less than the threshold, and more than 15 of the last 20 signal segments have RMS values greater than the threshold, then it is the starting point signal segment; if more than 15 of the first 20 signal segments have RMS values greater than the threshold, and more than 15 of the last 20 signal segments have RMS values less than the threshold, then it is the ending point signal segment.
[0036] In this embodiment, the estimated stable milling time is set to 25 seconds to 35 seconds, which provides an optimized time basis for signal validity judgment. b seconds in step S402 is set to 5 seconds, which further improves the accuracy of effective milling signal division. The judgment condition in step S403 is: 5 times the root mean square value of the first 5% of the signal segment is less than the root mean square value of the last 5%, or 5 times the root mean square value of the last 5% is less than the root mean square value of the first 5%, effectively identifying a completely valid signal segment.
[0037] In this embodiment, the RMS calculation formula is as follows: Where: represents the root mean square, Indicates the total number of data. Indicates The value of the data.
[0038] The RMS calculation formula used in this embodiment is simple to calculate and responsive, and achieves an accurate description of the data distribution form.
[0039] The present invention implements an adaptive milling process signal noise intelligent suppression method. When the method is applied, the signal segmentation and root mean square (RMS) analysis of the milling signal segment of 100 seconds are first performed to determine the validity of the signal and calculate the corresponding threshold. Then, the signal is screened using the obtained threshold to identify the quasi-valid signal segment. Subsequently, the validity of these quasi-valid signal segments is further verified by evaluating the duration and root mean square value of the signal segment. Once the signal segment is confirmed to be valid, they will be marked as fully valid signal segments. Subsequently, these fully valid signal segments will be spliced together, and the above process will be executed in a loop to achieve continuous reading and judgment of the entire signal sequence. Finally, all the identified fully valid signal segments will be merged to complete the preprocessing of the milling signal. Through this coherent automated process, the present invention can automatically screen out the signal segments during stable milling, and effectively suppress the noise stages such as the idle pass, cut-in and cut-out stages. The entire process is automated, and there is no need for manual segmentation by the operator, thereby reducing the burden on the operator. In addition, the present invention dynamically adjusts the threshold through continuous initialization and loop execution, thereby achieving adaptive matching of the threshold. This method not only improves the accuracy of milling signal processing, but also, due to its strong generalization, is applicable to milling signals under various working conditions, and does not require professionals to manually adjust the threshold, further reducing labor costs.
[0040] This embodiment is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any application based on the design principle of the present invention, as well as non-inventive changes based on the present invention, should be deemed to belong to the scope of protection of the present invention.
Claims
1. An adaptive milling process signal noise intelligent suppression method, characterized in that: The following steps are involved: S1: Read the milling signal segment with a duration of 2T; S2: judging whether the milling signal segment in step S1 contains a valid milling signal segment by segmentation and RMS analysis of the signal segment; if not, initializing and returning to step S1 to read the next milling signal segment; if yes, proceeding to the next step; S3: performing threshold calculation on the milling signal segment with a duration of 2T confirmed to contain a valid milling signal segment in step S2; S4: Perform threshold judgment on the valid milling signal segment in step S3 to identify the quasi-valid signal segment, and then further judge whether the quasi-valid signal segment is completely valid through the duration and root mean square of the signal segment. If not, initialize and return to step S1 to read the next milling signal segment; if valid, confirm it as a completely valid signal segment; S5: splicing the completely valid signal segments obtained in step S4, then initializing, and returning to step S1 to read the next milling signal segment; S6: by looping through steps S1 to S5, the reading and judgment of the entire length of the milling signal segment is completed, and all confirmed completely valid signal segments are spliced to complete the preprocessing of the milling signal; Among them, in step S1, T represents the cycle of a single stroke of milling processing; in steps S2 and S4, initialization refers to removing the milling signal segment with a duration of 2T in step S1; in step S5, initialization refers to removing the milling signal segment from the starting point in step S1 to a seconds after the end point of the quasi-valid signal segment in step S4.
2. According to claim 1, an adaptive milling process signal noise intelligent suppression method is characterized by: The a second is 5 seconds, and the duration of the milling signal segment with a duration of 2T in step S1 is 100 seconds.
3. The method for intelligently suppressing signal noise during adaptive milling process according to claim 2, characterized in that: Step S2 includes the following steps: S201: Segment the milling signal segment in step S1, where the duration of each signal segment is t; S202: Calculate the root mean square of each signal segment and store the result in an array, and then sort the array from small to large; S203: Select the values at the X% and Y% positions from the array, which are Z1 and Z2 respectively, where X% is set to 15%, Y% is set to 85%, Z1 represents the noise stage, and Z2 represents the stable milling stage; S204: Determine whether the condition Z1<0.02*Z2 is established. If the condition is established, it is considered that the milling signal segment in step S1 contains a valid milling signal segment, and proceed to the next step; if not, it is considered that it does not contain a valid milling signal segment, and initialization is performed, and return to step S1 to read the initialized milling signal segment.
4. The method for intelligently suppressing signal noise during adaptive milling process according to claim 3, characterized in that: In step S3, the threshold of the milling signal segment with a duration of 2T and including the effective milling signal segment is calculated according to the threshold formula (Z1+Z2) / 2.
5. The method for intelligently suppressing signal noise during adaptive milling process according to claim 4, characterized in that: Step S4 includes the following steps: S401: Identify the start point signal segment and the end point signal segment by comparing the root mean square and the threshold to determine the quasi-valid signal segment; S402: Determine whether the duration of the quasi-valid signal segment is within the range of the estimated stable milling duration of ±5 seconds. If so, delete b seconds from the beginning and the end of the quasi-valid signal segment and proceed to the next step. If not, perform initialization and return to step S1 to read the initialized milling signal segment. S403: judging whether the quasi-valid signal segment after b seconds are deleted in step S402 is a completely valid signal segment according to the RMS, if it is completely valid, proceeding to the next step, otherwise, initializing and returning to step S1 to read the initialized milling signal segment; In steps S402 and S403, initialization refers to removing the milling signal segment with a duration of 2T in step S1. The method for intelligently suppressing signal noise in an adaptive milling process according to claim 5 is characterized in that, in step S401, the following steps are included: S401a: Segment the milling signal segment in step S1 into signal segments, where the duration of each signal segment is t; S401b: traverse each signal segment and calculate the root mean square of each signal segment; S401c: By comparing the root mean square of the first 20 signal segments and the last 20 signal segments with the threshold, the starting point signal segment and the ending point signal segment are identified to determine the quasi-valid signal segment.
6. The method for intelligently suppressing signal noise during adaptive milling process according to claim 6, characterized in that: In step S401c, the judgment condition for the starting point signal segment is: more than 75% of the root mean square values of the first 20 signal segments are less than the threshold, and more than 75% of the root mean square values of the last 20 signal segments are greater than the threshold; the judgment condition for the ending point signal segment is: more than 75% of the root mean square values of the first 20 signal segments are greater than the threshold, and more than 75% of the root mean square values of the last 20 signal segments are less than the threshold.
7. The method for intelligently suppressing signal noise during adaptive milling process according to claim 7, characterized in that: In step S402, if it is determined that the duration of the quasi-effective signal segment is within the estimated stable milling duration range of 25 seconds to 35 seconds, b seconds is set to 5 seconds.
8. The method for intelligently suppressing signal noise during adaptive milling process according to claim 8, characterized in that: In step S403, the condition for a quasi-valid signal segment to be judged as a completely valid signal segment is that 5 times the root mean square of the first 5% of the signal segment is less than the root mean square of the last 5%, or 5 times the root mean square of the last 5% is less than the root mean square of the first 5%.
9. The method for intelligently suppressing signal noise during adaptive milling process according to any one of claims 1 to 9, characterized in that: The RMS calculation formula is as follows: Where: represents the root mean square, Indicates the total number of data. Indicates The value of the data.
Citation Information
Patent Citations
Breakage monitoring method for milling tool
CN105312965A
Tool breakage identification method based on vibration monitoring
CN112008495A
Active adaptive milling signal noise suppression method
CN117369365A
