CN118210273A - Numerical control center anti-collision processing state monitoring method based on data driving - Google Patents
Numerical control center anti-collision processing state monitoring method based on data driving Download PDFInfo
- Publication number
- CN118210273A CN118210273A CN202410628428.4A CN202410628428A CN118210273A CN 118210273 A CN118210273 A CN 118210273A CN 202410628428 A CN202410628428 A CN 202410628428A CN 118210273 A CN118210273 A CN 118210273A
- Authority
- CN
- China
- Prior art keywords
- sequence
- moment
- collision
- cutter
- normal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 69
- 238000012544 monitoring process Methods 0.000 title claims abstract description 34
- 230000002159 abnormal effect Effects 0.000 claims abstract description 16
- 238000003754 machining Methods 0.000 claims abstract description 14
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 10
- 108010076504 Protein Sorting Signals Proteins 0.000 claims description 47
- 230000001932 seasonal effect Effects 0.000 claims description 35
- 230000008569 process Effects 0.000 claims description 21
- 230000001133 acceleration Effects 0.000 claims description 15
- 239000013598 vector Substances 0.000 claims description 11
- 230000008859 change Effects 0.000 claims description 10
- 230000033001 locomotion Effects 0.000 claims description 8
- 230000005856 abnormality Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000006185 dispersion Substances 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 abstract description 7
- 238000005516 engineering process Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000001052 transient effect Effects 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Landscapes
- Numerical Control (AREA)
Abstract
The application relates to the technical field of anti-collision processing state monitoring, in particular to a method for monitoring an anti-collision processing state of a numerical control center based on data driving, which comprises the following steps: collecting related data in the machining process of the numerical control center; determining a real-time dynamic safety distance coefficient of the tool at any moment based on the related data; determining a normal current sequence at any moment and each sub-sequence after dividing the current sequence, and determining a collision early warning index of the tool at any moment by adopting a sequence decomposition algorithm and a real-time dynamic safety distance coefficient; based on collision early warning indexes of normal data samples, constructing an isolated forest model by adopting an isolated forest; and inputting the related data at the moment to be monitored into an isolated forest model to obtain an abnormal score, and judging the possibility of collision risk. The application aims to timely and accurately monitor potential collision risks and ensure the safe production and efficient operation of the numerical control center.
Description
Technical Field
The application relates to the technical field of anti-collision processing state monitoring, in particular to a numerical control center anti-collision processing state monitoring method based on data driving.
Background
With the rapid development of intelligent manufacturing technology, the core position of the numerical control machine tool in the field of machining is increasingly prominent, and particularly in the manufacturing process of high-precision complex parts, the numerical control machining center plays a vital role. However, in the highly automated numerical control machining process, due to factors such as equipment failure, misoperation, inaccurate process parameters and the like, collision accidents between a cutter and a workpiece or a machine tool are extremely easy to occur, so that expensive production equipment is damaged, and serious production stagnation and safety risks are possibly brought. Therefore, the anti-collision processing state monitoring technology of the numerical control center based on data driving is explored and developed to predict and avoid potential collision risks accurately in real time, and the method has great practical significance for promoting the development and safety guarantee of the intelligent manufacturing field.
The existing anti-collision monitoring measures mainly comprise hardware sensor monitoring alarm, fixed program limiting protection and a software early warning system based in part on simple rule logic. Although these methods are capable of preventing the occurrence of a collision event to some extent, there are significant limitations in practical use. First, the hardware sensor is relatively expensive to install and is responsive only to specific locations and types of collisions, making it difficult to cover all of the various complex collision possibilities; and secondly, a software early warning system based on a preset rule cannot adapt to changeable processing environment and process condition changes, the sensitivity and accuracy of the software early warning system are limited by a preset static threshold value, and the software early warning system is not suitable for dynamic changes and uncertain factors in the processing process. In addition, in the face of the big data age, the traditional monitoring means do not fully utilize massive real-time operation data generated by a modern numerical control machine tool, and potential collision risk modes cannot be effectively mined, so that the overall efficiency of a collision early warning system and the capability of early intervention are affected. Therefore, an intelligent analysis method based on data driving is needed, and through deep analysis of multi-element data such as speed, acceleration, current, voltage and the like in the processing process, possible collision risks are accurately captured and predicted, so that the reliability and predictability of a numerical control center anti-collision monitoring system are improved.
Disclosure of Invention
In order to solve the technical problems, the application provides a numerical control center anti-collision processing state monitoring method based on data driving so as to solve the existing problems.
The application discloses a method for monitoring an anti-collision processing state of a numerical control center based on data driving, which adopts the following technical scheme:
The embodiment of the application provides a method for monitoring the anti-collision processing state of a numerical control center based on data driving, which comprises the following steps:
collecting related data in the machining process of the numerical control center;
Determining a real-time dynamic safety distance coefficient of the cutter at any moment based on the minimum safety distance, the target position distance and the advancing speed and the acceleration of the cutter at any moment in the related data;
Determining a normal current sequence and a current sequence at any moment and sub-sequences after dividing the current sequence, and respectively determining a trend sequence and a seasonal sequence of each sub-sequence by adopting a sequence decomposition algorithm; determining collision early warning indexes of the cutters at any moment based on each subsequence and corresponding trend sequences, seasonal sequences and real-time dynamic safety distance coefficients thereof;
Based on collision early warning indexes of normal data samples, constructing an isolated forest model by adopting an isolated forest; and inputting the related data at the moment to be monitored into an isolated forest model to obtain an abnormal score, and judging the possibility of collision risk.
Preferably, the collecting the related data in the machining process of the numerical control center includes:
collecting the minimum safety distance and the target position distance of a lower cutter at any collecting time in the processing process of the numerical control center, and the advancing speed, the acceleration, the output current of a servo motor and the output voltage of the servo motor;
the minimum safety distance is the minimum value of the relative distances between the cutter and surrounding devices; the target position distance is the distance between the cutter and the end point; the output current and the output voltage of the servo motor are current and voltage data of the servo motor for controlling the movement of the cutter.
Preferably, the method for determining the real-time dynamic safety distance coefficient comprises the following steps:
obtaining a predicted minimum safety distance and a predicted target position distance of a cutter;
Taking the product of the predicted minimum safety distance and the predicted target position distance as the input of a threshold function to obtain the output result of the threshold function;
And taking the product of the output result and the predicted target position distance as a real-time dynamic safety distance coefficient of the cutter at any moment.
Preferably, the expression of the predicted minimum safe distance and the predicted target position distance is:
In the method, in the process of the invention, 、/>Are respectively/>Minimum safety distance of moment cutter and target position distance,/>、/>Are respectively/>The predicted minimum safety distance and the predicted target position distance of the moment cutter, t is the acquisition time interval,/>Is/>Travel speed of the time tool,/>Is/>Acceleration of tool travel at time,/>Is/>The deviation angle of the tool travel path at the moment passes/>Time of day relative to/>The travel distance of the cutter at the moment and the minimum An Quanju dispersion of the travel distance are calculated, and the travel distance is/is calculated、Are respectively/>Sine and cosine values of the deviation angle of the tool travelling path at the moment;
The acquisition method of n comprises the following steps: selecting a predicted target position distance And predicting minimum safe distance/>The time difference between the first time point of 0 or less and the time point b is denoted as n.
Preferably, the threshold function is: when the input of the threshold function is less than or equal to 0, the output result of the threshold function is-1; otherwise, the output result of the threshold function is 1.
Preferably, the determining the current sequence at any time and the sub-sequences after division of the current sequence at any time respectively determines a trend sequence and a seasonal sequence of each sub-sequence by adopting a sequence decomposition algorithm, and the method comprises the following steps:
The output currents of the servo motors in the related data at all times in the acquisition period under the normal condition are formed into a normal current sequence; the output currents of the servo motors in the related data at all moments before any moment in the acquisition period are formed into a current sequence at any moment;
dividing a normal current sequence and a current sequence at any moment into subsequences according to a preset processing stage;
taking each subsequence as input of Hannan-Rissanen algorithm, and outputting seasonal period of each subsequence; and taking each subsequence and the seasonal period thereof as input of a sequence decomposition algorithm, and outputting the decomposed trend sequence and seasonal sequence.
Preferably, the method for determining the collision early warning index includes:
the voltage sequence and the current sequence are collectively referred to as an electrical signal sequence;
Determining the stability coefficient of the electric signal sequence of the cutter at any moment based on each subsequence in the electric signal sequence at any moment and the normal electric signal sequence and the corresponding trend sequence;
determining a regular change coefficient of the electrical signal sequence of the tool at any moment based on each subsequence in the normal electrical signal sequence at any moment and the seasonal sequence corresponding to the subsequence;
Calculating the product of the stability coefficient and the regular change coefficient of the electric signal sequence, calculating the product of the voltage sequence and the current sequence, and recording the product as a first product; calculating a difference between a number 1 and the first product; calculating the sum of the autocorrelation coefficient of the current sequence at any moment and a preset parameter adjusting coefficient; taking the ratio of the difference value to the sum value as an index of an exponential function based on a natural constant;
Calculating the product of the real-time dynamic safety distance coefficient of the cutter at any moment and the calculation result of the exponential function, and recording the product as a second product; and taking the difference value of the second product and the absolute value of the real-time dynamic safety distance coefficient of the cutter at any moment as a collision early warning index of the cutter at any moment.
Preferably, the stability coefficient is expressed as:
In the method, in the process of the invention, Is/>Electric signal sequence of time cutter/>Is the number of processing stages, p/(stability coefficient)Is an exponential function based on natural constant,/>、/>Are respectively/>Time of day, normal electrical signal sequence/>In/>Slope of fitted line of trend sequence of subsequence of each processing stage,/>Is the first/>, in the normal electrical signal sequence iSubsequence of individual processing phases,/>Is/>Time-of-day electrical signal sequence/>Middle/>Subsequence of individual processing phases,/>、/>The first/>, in the electrical signal sequence i at the normal and b times respectivelyTrend intensity of trend sequence of subsequences of individual processing stages,/>Is a variance function,/>Is a preset parameter adjusting coefficient.
Preferably, the expression of the regular change coefficient is:
In the method, in the process of the invention, Is/>Electric signal sequence of time cutter/>Regular coefficient of variation,/>Is a correlation coefficient function,/>、Are respectively/>Sequence formed by fitting straight line slopes of trend sequences of all subsequences in time and normal electric signal sequence i according to processing sequence,/>Is the minimum value,/>、/>The first/>, in the electrical signal sequence i at the normal and b times respectivelySeasonal intensity of seasonal sequence of sub-sequences of individual processing phases,/>、/>Respectively is normal,/>Time-of-day electrical signal sequence iThe seasonal sequence of sub-sequences of individual processing phases, p being the number of processing phases.
Preferably, the collision early warning index based on the normal data sample adopts an isolated forest to construct an isolated forest model; inputting relevant data at the moment to be monitored into an isolated forest model to obtain an abnormal score, and judging the possibility of collision risk, wherein the method comprises the following steps:
collecting the related data of a normal sample at each moment in the normal condition; acquiring collision early warning indexes of normal samples at each moment;
The relevant data of the normal samples at each moment and the collision early warning index form a characteristic vector at each moment; taking the feature vectors of all normal samples as the input of an isolated forest algorithm, and outputting an isolated forest model;
inputting the feature vector of the moment to be monitored into an isolated forest model, and outputting an abnormality score of the moment to be monitored;
when the abnormal score is larger than a preset value, judging that collision risk exists at the moment to be monitored; otherwise, judging that the moment to be monitored has no collision risk.
The application has at least the following beneficial effects:
According to the method, by monitoring the feed path in the machining process of the numerical control center and combining the target position distance of the cutter, the minimum safety distance between the cutter and surrounding devices, the speed and acceleration data, the real-time dynamic distance safety coefficient is calculated to judge whether collision risk is caused by abnormal feed path, and the risk of cutter collision is more accurately excavated from the actual motion angle of the cutter; then, the current and voltage data of the servo motor are combined, the current and voltage data are compared with the current and voltage data under normal conditions, the collision early warning index is calculated, and the fine monitoring of potential collision risks in the machining process of the numerical control center is realized by integrating various sensor data and utilizing a strategy combining the real-time dynamic safety distance coefficient and the collision early warning index; finally, predicting collision risk through an isolated forest model, and effectively filling the defects of the existing scheme when dealing with complex and sudden collision risk on the basis of keeping the original hard protection by means of an isolated forest algorithm, thereby greatly improving the intelligence, flexibility and reliability of the numerical control center anti-collision monitoring system. By the method, potential collision risks can be monitored more timely and accurately, and safe production and efficient operation of the numerical control center are powerfully ensured.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the application, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for monitoring the anti-collision processing state of a numerical control center based on data driving;
FIG. 2 is a schematic view of the minimum safety distance of the tool at time b;
fig. 3 is a flowchart of index construction of collision warning index.
Detailed Description
In order to further describe the technical means and effects adopted by the application to achieve the preset aim, the following detailed description is given below of the data-driven numerical control center anti-collision processing state monitoring method according to the application, which is based on the specific implementation, structure, characteristics and effects thereof, with reference to the attached drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
The following specifically describes a specific scheme of the data-driven numerical control center anti-collision processing state monitoring method provided by the application with reference to the accompanying drawings.
The application provides a data-driven-based numerical control center anti-collision processing state monitoring method.
Specifically, the following method for monitoring the anti-collision processing state of a numerical control center based on data driving is provided, please refer to fig. 1, and the method comprises the following steps:
and S001, acquiring related data in the machining process of the numerical control center by using a sensor, and preprocessing the data.
The method comprises the steps that data in the machining process of a numerical control center are collected through sensors, laser ranging sensors are installed at the front end of a cutter, a working table and main structural accessories of a machine tool, the relative distance between the cutter and surrounding devices and the distance between the cutter and an end point are recorded in real time, the surrounding devices refer to other fixed or movable parts (such as a clamp, a working table, a tailstock and a protection device) in the machine tool around the cutter, and the end point refers to the machining end point of a preset cutter path; the travelling speed and the acceleration of the cutter are recorded in real time by installing a speed and acceleration sensor on the main shaft; the current and voltage sensors are arranged on an output interface of the servo motor for controlling the movement of the cutter, and the current fluctuation and the voltage fluctuation of the servo motor are monitored in real time.
For a certain moment, acquiring the relative distance between the cutter and surrounding devices, and taking the shortest distance as the minimum safe distance of the momentThe relative distance of the cutter relative to surrounding devices is measured by laser ranging sensors mounted on the working table and at the main structural accessories of the machine tool; the distance of the tool relative to the end point is measured as the target position distance/>, at that moment, by a laser ranging sensor mounted at the front end of the tool. The travelling speed/>, of the tool at each moment is acquired through a speed sensor, an acceleration sensor and a current and voltage sensor which are respectively arranged on a main shaft and an output interface of a servo motorAcceleration/>And servo motor output current/>, of a servo motor controlling the movement of the toolOutput voltage of servo motor/>. In order to realize real-time monitoring, data are acquired every 10 seconds, the acquisition period is N seconds, the value of N in the embodiment is 1000, and an operator can set the value according to actual conditions.
The minimum safe distance, the target position distance, the advancing speed and the acceleration of the cutter at each moment, the output current of the servo motor for controlling the movement of the cutter and the output voltage of the servo motor can be obtained, and the output current and the output voltage of the servo motor are recorded as related data in the processing process of the numerical control center.
Step S002, calculating a real-time dynamic distance safety coefficient according to the minimum safety distance of the cutter and the target position distance and combining the advancing speed and the acceleration; and then, calculating a collision early warning index by combining the current and voltage fluctuation conditions of a servo motor for controlling the movement of the cutter.
When the numerical control center works, the abnormal feeding path or part wear can cause the increase of collision risk, namely the too small distance between the cutter and surrounding devices is caused by programming errors, systematic errors, mechanical part wear and other reasons, so that the possibility of collision is increased; meanwhile, collision risk can also cause the cutter to encounter suddenly increased resistance in the advancing process, so that the current and the voltage of the servo motor correspondingly fluctuate.
When the collision risk caused by abnormal feeding path or part abrasion is increased, the travelling path of the cutter can deviate, and the distance between the cutter and surrounding devices is always kept unchanged or within a safe distance in the travelling process of the cutter under normal conditions, so that the collision risk can not be generated until the end point; however, when the travel path of the tool is shifted, the minimum distance between the tool and surrounding devices becomes smaller, and if the distance between the target position is small enough, the shift will not generate a large risk of collision, and if the distance between the target position is large enough, the risk of collision will be greater. Therefore, the embodiment calculates the real-time dynamic safety distance coefficient of the cutter according to the following formula toThe time is as follows:
In the method, in the process of the invention, Is/>Real-time dynamic safety distance coefficient of moment cutter,/>、/>Are respectively/>Minimum safety distance of moment cutter and target position distance,/>、/>Are respectively/>The predicted minimum safe distance and the predicted target position distance of the cutter at the moment are selected, and the predicted target position distance/>And predicting minimum safe distance/>The time difference between the first time less than or equal to 0 and the time b is denoted as n, t is the acquisition time interval, and the embodiment is set to 10s,/>Is/>Travel speed of the time tool,/>Is/>Acceleration of tool travel at time,/>Is/>The deviation angle of the tool travel path at the moment passes/>Time of day relative to/>The travel distance of the cutter at the moment and the minimum An Quanju dispersion of the travel distance are calculated, and the travel distance is/is calculated、/>Are respectively/>Sine and cosine values of the tool path offset angle at the moment,Is a threshold function, and when the value input by the threshold function in brackets is less than or equal to 0, the output result of the threshold function is/>When the value input by the threshold function in the brackets is greater than 0, the output result of the threshold function is/>. The minimum safety distance of the cutter at the time b is shown in fig. 2.
When collision risk is caused by abnormal feeding path or part wear, the travel path of the cutter is deviated and collision is causedMinimum safe distance of time/>At this time if the target location has not been reached, i.e./>Then/>For/>Thereby/>Real-time dynamic safety distance coefficient/>Is a non-positive number, and/>The smaller the collision risk, the greater; if at/>Arrival within the time span of the moment, then/>Then/>Is-1 or 1, and when/>When it is-1,/>=0; When (when)When 1,/><0, Thus/>Real-time dynamic safety distance coefficient/>Is a non-positive number. Therefore, when collision risk is caused by abnormal feed path or part wear, the real-time dynamic safety distance coefficient is a non-positive number, andThe smaller the collision risk, the greater; and under normal conditions if/><0, If there is no collision,/>The real-time dynamic safety distance coefficient results in a positive number, and the larger the positive number is, the safer the tool travelling path is, and the less possibility of collision is.
Although the real-time dynamic safety distance coefficient assumes uniform acceleration, the assumption can be adapted to most situations because the target distance of the numerical control center is limited. However, the situation of increased collision risk cannot be completely captured through detection of the feed path, for example, dynamic response errors of a machine tool system, such as mechanical gaps, low accuracy of a transmission mechanism, control system delay, or sudden and transient collision risk, such as the situation of rapid deviation from an original planned path caused by sudden component breakage or system misalignment, may not be early-warned in time by calculating the real-time dynamic safety distance coefficient alone.
In this case, further consideration is required in combination with the electrical signal of the servo motor controlling the movement of the tool, when the collision risk or sudden and transient collision risk is caused by the dynamic response error, the feeding path may still be forward according to the normal path, but in this case, the tool encounters suddenly increased resistance during the traveling process, so that the voltage and the current change.
Under normal machining conditions, the voltage and current of the servo motor should be maintained to be relatively stable or change according to a certain rule, for example, rough machining is possible just at the beginning, as the longer the tool is, the higher the machining precision requirement is, the higher the frequency of speed adjustment is in a short time, so that the voltage and current increase and the fluctuation frequency is higher, but the fluctuation is regular. In case of collision risk, a sudden increase of resistance, such as abnormal friction or jamming between the tool and the workpiece, is caused, and the servo motor often needs to provide a larger driving force in order to overcome such additional resistance, which leads to a significant increase of the motor current.
Firstly, the embodiment calculates voltage and current data in normal condition, i.e. in complete period without collision, and takes the current data as an example, and forms a normal current sequence according to time sequenceThe stages of processing in a normal complete cycle, e.g. just before coming into contact with the processed article, roughing stage, finishing stage, etc., are then counted, assuming a common/>A plurality of processing stages, wherein the normal current sequence/>, is determined according to the processing stagesDivided into/>The subsequence, in this embodiment, the value m is 3, which can be set by the practitioner according to the actual situation.
Each subsequence is taken as input, and a Hannan-Rissanen algorithm is adopted to obtain a seasonal period; and then respectively taking each subsequence and the corresponding seasonal period as input, adopting STL decomposition, and outputting the decomposed trend sequence and seasonal sequence. The trend sequence acquisition process comprises the following steps: and carrying out smoothing treatment on the original sequence to obtain a smoothed sequence, and subtracting the smoothed sequence from the original sequence to obtain a trend sequence. The seasonal sequence is obtained by the following steps: subtracting the trend sequence from the original sequence to obtain a residual fluctuation sequence, carrying out seasonal decomposition on the residual sequence, carrying out pattern matching through a seasonal period, and carrying out differential operation on the residual sequence and the matched seasonal pattern to obtain the seasonal sequence. And then taking the trend sequence as input, adopting linear fitting, and outputting the slope of a fitting straight line of the trend sequence. And then, respectively calculating the trend intensity and the seasonal intensity of the trend sequence and the seasonal sequence, and adopting the same processing mode for the normal voltage data of the normal condition. The Hannan-Rissanen algorithm, STL decomposition, pattern matching, linear fitting, and calculation of trend intensity and seasonal intensity are all known techniques, and detailed descriptions thereof are omitted.
To be used forTime of day is an example, assume/>Moment is at/>Stage by stage, the present embodiment will/>The current and voltage data before the moment are formed in time sequence/>The current sequence and the voltage sequence at the moment are divided according to the processing stage, and the fitting straight line slope, the trend intensity and the seasonal intensity of the trend sequence are obtained by adopting the same processing mode for each sub-sequence after the division. Calculating/>, according to the following formula, based on the slope, trend intensity and seasonal intensity of a fitted straight line of a trend sequence corresponding to the current and voltage sequences at the time of normal and time bCollision early warning index of the cutter at the moment:
In the method, in the process of the invention, Is/>Collision early warning index of moment cutter,/>Is/>Real-time dynamic safety distance coefficient of moment cutter,/>Is an exponential function based on a natural constant e,/>、/>Are respectively/>Electric signal sequence of time cutter/>Stability coefficient, regular change coefficient, electrical signal sequence/>The category of (a) includes voltage sequence and current sequence,/>The autocorrelation coefficient of the current sequence at time b, and the hysteresis value is 3,/>Is a parameter adjusting coefficient, avoids the denominator to be 0 and takes the value of 0.01; the autocorrelation coefficients are known techniques and will not be described in detail;
p is the number of processing stages and, 、/>Are respectively/>Time of day, normal electrical signal sequence/>In/>Slope of fitted line of trend sequence of subsequence of each processing stage,/>Is the first/>, in the normal electrical signal sequence iSubsequence of individual processing phases,/>Is/>Time-of-day electrical signal sequence/>Middle/>Subsequence of individual processing phases,/>、/>The first/>, in the electrical signal sequence i at the normal and b times respectivelyTrend intensity of trend sequence of subsequences of individual processing stages,/>Is a variance function;
Is a correlation coefficient function,/> 、/>Are respectively/>Sequence formed by fitting straight line slopes of trend sequences of all subsequences in time and normal electric signal sequence i according to processing sequence,/>Is the minimum value,/>、/>The first/>, in the electrical signal sequence i at the normal and b times respectivelySeasonal intensity of seasonal sequence of sub-sequences of individual processing phases,/>、/>Respectively is normal,/>Time-of-day electrical signal sequence iSeasonal sequence of sub-sequences of individual processing stages. The index construction flow chart of the collision early warning index is shown in fig. 3.
When a collision risk occurs, an abnormality in the feed path may be detected, and an abnormal fluctuation in the current or voltage of the servo motor may be detected. If the abnormality of the feed path is directly detected to have collision risk, the real-time dynamic safety distance coefficient is a non-positive number, namelyIf resistance is increased, namely abnormal fluctuation of current and voltage of the servo motor is detected, the stability of the current and the voltage is poor compared with that of the normal condition, namely the trend of the electric signal sequence is changed compared with that of the normal condition, so that the slope of a fitting straight line of the trend sequence is changed, and the stability is poor as the slope difference is larger; the trend intensity measures the trend of the sequence, the stability of the electrical signal is better under normal conditions, the trend is more obvious, the stability is poorer when the current and voltage abnormally fluctuate, the trend is poorer, and in addition, the stability is poorer, so that/>When the electric signal sequence at the moment is inserted into the normal electric signal sequence, the stability of elements in the sequence is poorer, and the variance is increased, so that the electric signal sequence/>Stability coefficient/>; The sudden increase of resistance can lead to the remarkable increase of current, break the regularity of elements in a current sequence and lead to the autocorrelation coefficient/>The current sequence is closer to 0, and compared with the seasonal sequence of the current sequence in the normal condition, the current sequence has inconsistent regularity, so that the correlation coefficient of the current sequence and the seasonal sequence is smaller, the change of trend slopes is inconsistent, the correlation coefficient between the slope sequences of the fitting straight lines of the trend sequence is smaller, the seasonal intensity measures the regularity of the sequence, the regularity of the electric signal sequence in the normal condition is obvious, and the regularity is worse when the current and the voltage abnormally fluctuate, so that the electric signal sequence/>Regular coefficient of variation/>，/>Collision early warning index/>, time of dayThe larger the exponential function portion is greater than 1, and the larger the exponential function portion is, the more serious the fluctuation of the electrical signal is. And then when/>Collision early warning index/>, when there is a risk of collision at the momentAnd the greater the probability of collision risk, the greater the probability of collisionThe smaller.
If no resistance increase occurs, the stability of the current and voltage of the servo motor is similar to that of the normal state, so thatCollision early warning index/>, time of dayThe middle exponential function should be close to 1, so that the collision early warning index. In summary, when a collision risk occurs, the collision early warning index/>And/>The smaller the probability of collision risk is, the greater.
Real-time dynamic safety distance coefficient when there is no collision riskAnd the stability of the current and the voltage of the servo motor is similar to that of the normal state, thereby/>Collision early warning index/>, time of dayThe middle exponential function should be close to 1, and then the collision early warning index/>。
And S003, calculating a collision early warning index of the normal data sample, and constructing an isolated forest model by utilizing an isolated forest algorithm. And inputting the related data at the moment needing to be monitored into a model, and predicting the possibility of collision risk.
And counting a large amount of normal data under the normal condition, acquiring the relevant data of each normal sample at the corresponding time according to the data acquisition mode in the step one, in the embodiment, the normal samples at M times and the relevant data corresponding to the normal samples are taken together, and the collision early warning index of the normal samples at each time is calculated. In this embodiment, the value of M is 3600, and the practitioner can set the value according to the actual situation.
And (3) the related data of the normal samples acquired at each moment are: speed of speedAcceleration/>Current/>Voltage/>And normalizing each datum by using a maximum value normalization method, and forming the normalized datum into a feature vector at each moment.
And taking the feature vectors of M normal samples as input, adopting an isolated forest algorithm, wherein the number of samples extracted each time is 256, and the number of isolated trees in the forest is 100, so as to construct an isolated forest model. The isolated forest algorithm is a known technology and will not be described in detail.
And then taking the feature vector at the moment to be monitored as input of an isolated forest model, calculating an abnormality score of the feature vector by using the trained isolated forest model, and considering that the collision risk exists at the moment to be monitored corresponding to the feature vector when the abnormality score is smaller than an abnormality threshold value of 0.5. The abnormal threshold can be set by the practitioner according to the actual situation, and the isolated forest is not described in detail as a known technology.
Therefore, the monitoring of the anti-collision processing state of the numerical control center is realized.
It should be noted that: the sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the foregoing description has been directed to specific embodiments of this specification. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and the same or similar parts of each embodiment are referred to each other, and each embodiment mainly describes differences from other embodiments.
The above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; the technical solutions described in the foregoing embodiments are modified or some of the technical features are replaced equivalently, so that the essence of the corresponding technical solutions does not deviate from the scope of the technical solutions of the embodiments of the present application, and all the technical solutions are included in the protection scope of the present application.
Claims (10)
1. The method for monitoring the anti-collision processing state of the numerical control center based on data driving is characterized by comprising the following steps of:
collecting related data in the machining process of the numerical control center;
Determining a real-time dynamic safety distance coefficient of the cutter at any moment based on the minimum safety distance, the target position distance and the advancing speed and the acceleration of the cutter at any moment in the related data;
Determining a normal current sequence and a current sequence at any moment and sub-sequences after dividing the current sequence, and respectively determining a trend sequence and a seasonal sequence of each sub-sequence by adopting a sequence decomposition algorithm; determining collision early warning indexes of the cutters at any moment based on each subsequence and corresponding trend sequences, seasonal sequences and real-time dynamic safety distance coefficients thereof;
Based on collision early warning indexes of normal data samples, constructing an isolated forest model by adopting an isolated forest; and inputting the related data at the moment to be monitored into an isolated forest model to obtain an abnormal score, and judging the possibility of collision risk.
2. The method for monitoring the anti-collision processing state of the numerical control center based on data driving according to claim 1, wherein the step of collecting the relevant data in the processing process of the numerical control center comprises the following steps:
collecting the minimum safety distance and the target position distance of a lower cutter at any collecting time in the processing process of the numerical control center, and the advancing speed, the acceleration, the output current of a servo motor and the output voltage of the servo motor;
the minimum safety distance is the minimum value of the relative distances between the cutter and surrounding devices; the target position distance is the distance between the cutter and the end point; the output current and the output voltage of the servo motor are current and voltage data of the servo motor for controlling the movement of the cutter.
3. The method for monitoring the anti-collision processing state of the numerical control center based on data driving according to claim 2, wherein the method for determining the real-time dynamic safety distance coefficient comprises the following steps:
obtaining a predicted minimum safety distance and a predicted target position distance of a cutter;
Taking the product of the predicted minimum safety distance and the predicted target position distance as the input of a threshold function to obtain the output result of the threshold function;
And taking the product of the output result and the predicted target position distance as a real-time dynamic safety distance coefficient of the cutter at any moment.
4. The method for monitoring the anti-collision processing state of the numerical control center based on data driving according to claim 3, wherein the expression for predicting the minimum safe distance and the predicted target position distance is as follows:
In the method, in the process of the invention, 、/>Are respectively/>Minimum safety distance of moment cutter and target position distance,/>、/>Are respectively/>The predicted minimum safety distance and the predicted target position distance of the moment cutter, t is the acquisition time interval,/>Is/>Travel speed of the time tool,/>Is/>Acceleration of tool travel at time,/>Is/>The deviation angle of the tool travel path at the moment passes/>Time of day relative to/>The travel distance of the cutter at the moment and the minimum An Quanju dispersion of the travel distance are calculated, and the travel distance is/is calculated、/>Are respectively/>Sine and cosine values of the deviation angle of the tool travelling path at the moment;
The acquisition method of n comprises the following steps: selecting a predicted target position distance And predicting minimum safe distance/>The time difference between the first time point of 0 or less and the time point b is denoted as n.
5. The data-driven numerical control center anti-collision processing state monitoring method according to claim 4, wherein the threshold function is: when the input of the threshold function is less than or equal to 0, the output result of the threshold function is-1; otherwise, the output result of the threshold function is 1.
6. The method for monitoring the anti-collision processing state of the numerical control center based on data driving according to claim 1, wherein the determining the current sequence at any time and the sub-sequences after dividing the current sequence at any time respectively determines the trend sequence and the seasonal sequence of each sub-sequence by adopting a sequence decomposition algorithm, and the method comprises the following steps:
The output currents of the servo motors in the related data at all times in the acquisition period under the normal condition are formed into a normal current sequence; the output currents of the servo motors in the related data at all moments before any moment in the acquisition period are formed into a current sequence at any moment;
dividing a normal current sequence and a current sequence at any moment into subsequences according to a preset processing stage;
taking each subsequence as input of Hannan-Rissanen algorithm, and outputting seasonal period of each subsequence; and taking each subsequence and the seasonal period thereof as input of a sequence decomposition algorithm, and outputting the decomposed trend sequence and seasonal sequence.
7. The method for monitoring the collision avoidance processing state of a numerically controlled center based on data driving according to claim 1, wherein the method for determining the collision warning index comprises:
the voltage sequence and the current sequence are collectively referred to as an electrical signal sequence;
Determining the stability coefficient of the electric signal sequence of the cutter at any moment based on each subsequence in the electric signal sequence at any moment and the normal electric signal sequence and the corresponding trend sequence;
determining a regular change coefficient of the electrical signal sequence of the tool at any moment based on each subsequence in the normal electrical signal sequence at any moment and the seasonal sequence corresponding to the subsequence;
Calculating the product of the stability coefficient and the regular change coefficient of the electric signal sequence, calculating the product of the voltage sequence and the current sequence, and recording the product as a first product; calculating a difference between a number 1 and the first product; calculating the sum of the autocorrelation coefficient of the current sequence at any moment and a preset parameter adjusting coefficient; taking the ratio of the difference value to the sum value as an index of an exponential function based on a natural constant;
Calculating the product of the real-time dynamic safety distance coefficient of the cutter at any moment and the calculation result of the exponential function, and recording the product as a second product; and taking the difference value of the second product and the absolute value of the real-time dynamic safety distance coefficient of the cutter at any moment as a collision early warning index of the cutter at any moment.
8. The data-driven numerical control center anti-collision processing state monitoring method according to claim 7, wherein the expression of the stability coefficient is:
In the method, in the process of the invention, Is/>Electric signal sequence of time cutter/>Is the number of processing stages, p/(stability coefficient)Is an exponential function based on natural constant,/>、/>Are respectively/>Time of day, normal electrical signal sequence/>In/>Slope of fitted line of trend sequence of subsequence of each processing stage,/>Is the first/>, in the normal electrical signal sequence iSubsequence of individual processing phases,/>Is/>Time-of-day electrical signal sequence/>Middle/>Subsequence of individual processing phases,/>、/>The first/>, in the electrical signal sequence i at the normal and b times respectivelyTrend intensity of trend sequence of subsequences of individual processing stages,/>Is a variance function,/>Is a preset parameter adjusting coefficient.
9. The method for monitoring the anti-collision processing state of the numerical control center based on data driving according to claim 7, wherein the expression of the regular change coefficient is:
In the method, in the process of the invention, Is/>Electric signal sequence of time cutter/>Regular coefficient of variation,/>Is a correlation coefficient function,/>、/>Are respectively/>Sequence formed by fitting straight line slopes of trend sequences of all subsequences in time and normal electric signal sequence i according to processing sequence,/>Is the minimum value,/>、/>The first/>, in the electrical signal sequence i at the normal and b times respectivelySeasonal intensity of seasonal sequence of sub-sequences of individual processing phases,/>、/>Respectively is normal,/>Time-of-day electrical signal sequence iThe seasonal sequence of sub-sequences of individual processing phases, p being the number of processing phases.
10. The method for monitoring the collision avoidance processing state of the numerically controlled center based on data driving according to claim 1, wherein the collision early warning index based on the normal data sample is used for constructing an isolated forest model by adopting an isolated forest; inputting relevant data at the moment to be monitored into an isolated forest model to obtain an abnormal score, and judging the possibility of collision risk, wherein the method comprises the following steps:
collecting the related data of a normal sample at each moment in the normal condition; acquiring collision early warning indexes of normal samples at each moment;
The relevant data of the normal samples at each moment and the collision early warning index form a characteristic vector at each moment; taking the feature vectors of all normal samples as the input of an isolated forest algorithm, and outputting an isolated forest model;
inputting the feature vector of the moment to be monitored into an isolated forest model, and outputting an abnormality score of the moment to be monitored;
when the abnormal score is larger than a preset value, judging that collision risk exists at the moment to be monitored; otherwise, judging that the moment to be monitored has no collision risk.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410628428.4A CN118210273B (en) | 2024-05-21 | Numerical control center anti-collision processing state monitoring method based on data driving |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410628428.4A CN118210273B (en) | 2024-05-21 | Numerical control center anti-collision processing state monitoring method based on data driving |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118210273A true CN118210273A (en) | 2024-06-18 |
| CN118210273B CN118210273B (en) | 2024-07-30 |
Family
ID=
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080012519A1 (en) * | 2006-06-30 | 2008-01-17 | Star Micronics Co., Ltd. | Moving member movement control apparatus, moving member movement control method and machine tool movement control apparatus |
| CN105005266A (en) * | 2015-05-29 | 2015-10-28 | 华中科技大学 | Method for quickly identifying collision of milling cutter based on current of motor |
| US20160082562A1 (en) * | 2014-09-19 | 2016-03-24 | Mikron Agie Charmilles SA | High voltage crash prevention device for machine tools |
| US20200324413A1 (en) * | 2019-04-10 | 2020-10-15 | Industrial Technology Research Institute | Method and system for predicting collision of machining path |
| CN116627087A (en) * | 2023-04-18 | 2023-08-22 | 大族激光科技产业集团股份有限公司 | Collision protection method, processing device, equipment and storage medium |
| CN116810493A (en) * | 2023-08-31 | 2023-09-29 | 山东惠硕重工机械有限公司 | Anti-collision detection method and system for numerical control machine tool based on data driving |
| CN117311266A (en) * | 2023-10-23 | 2023-12-29 | 成都悦蓉智诚科技有限公司 | Protection system of numerical control machining process full-state multi-signal collision machine |
| CN117421687A (en) * | 2023-12-18 | 2024-01-19 | 圣道天德电气(山东)有限公司 | Method for monitoring running state of digital power ring main unit |
| CN117788568A (en) * | 2024-02-23 | 2024-03-29 | 天津港太平洋国际集装箱码头有限公司 | Ship shellfish position optimizing and positioning method based on digital twin |
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080012519A1 (en) * | 2006-06-30 | 2008-01-17 | Star Micronics Co., Ltd. | Moving member movement control apparatus, moving member movement control method and machine tool movement control apparatus |
| US20160082562A1 (en) * | 2014-09-19 | 2016-03-24 | Mikron Agie Charmilles SA | High voltage crash prevention device for machine tools |
| CN105005266A (en) * | 2015-05-29 | 2015-10-28 | 华中科技大学 | Method for quickly identifying collision of milling cutter based on current of motor |
| US20200324413A1 (en) * | 2019-04-10 | 2020-10-15 | Industrial Technology Research Institute | Method and system for predicting collision of machining path |
| CN116627087A (en) * | 2023-04-18 | 2023-08-22 | 大族激光科技产业集团股份有限公司 | Collision protection method, processing device, equipment and storage medium |
| CN116810493A (en) * | 2023-08-31 | 2023-09-29 | 山东惠硕重工机械有限公司 | Anti-collision detection method and system for numerical control machine tool based on data driving |
| CN117311266A (en) * | 2023-10-23 | 2023-12-29 | 成都悦蓉智诚科技有限公司 | Protection system of numerical control machining process full-state multi-signal collision machine |
| CN117421687A (en) * | 2023-12-18 | 2024-01-19 | 圣道天德电气(山东)有限公司 | Method for monitoring running state of digital power ring main unit |
| CN117788568A (en) * | 2024-02-23 | 2024-03-29 | 天津港太平洋国际集装箱码头有限公司 | Ship shellfish position optimizing and positioning method based on digital twin |
Non-Patent Citations (1)
| Title |
|---|
| 段裕刚;刘志兵;王西彬;肖三强;: "基于可视锥理论的五轴加工刀具碰撞检验方法", 计算机集成制造***, no. 11, 15 November 2014 (2014-11-15) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11614728B2 (en) | Machine tool management system that obtains a next maintenance period from a maintenance period model and a refinement algorithm | |
| CN110008565B (en) | Industrial process abnormal working condition prediction method based on operation parameter correlation analysis | |
| CN102929210B (en) | Control and optimization system for feature-based numerical control machining process and control and optimization method therefor | |
| CN104808585B (en) | A kind of quick inspection method of lathe health status | |
| US20190005413A1 (en) | Control device and machine learning device | |
| JP6832327B2 (en) | Data-driven method for automatic detection of anomalous workpieces during the production process | |
| CN101477351A (en) | Intelligent numerical control method with three-stage process self-optimization function | |
| CN111222247A (en) | Early fault early warning method for rotary machine | |
| CN111948976B (en) | Cutter state monitoring method, non-volatile readable storage medium and electronic device | |
| CN108445838A (en) | A kind of numerically-controlled machine tool processing quality analysis method, grader and equipment | |
| Mou et al. | A study of tool tipping monitoring for titanium milling based on cutting vibration | |
| CN116893643A (en) | Intelligent robot driving track safety control system based on data analysis | |
| CN116880356A (en) | Method and device for monitoring machining state of numerical control machine tool | |
| CN117331344A (en) | Multi-signal quality monitoring and controlling system in numerical control machining process | |
| CN112262027B (en) | Abnormality determination device and abnormality determination method | |
| CN102467113B (en) | Controller of mechanical device with parameter learning and parameter learning method thereof | |
| Schwenzer et al. | Machine learning for tool wear classification in milling based on force and current sensors | |
| CN118210273B (en) | Numerical control center anti-collision processing state monitoring method based on data driving | |
| EP3951528B1 (en) | Abnormality detection apparatus, and abnormality detection method | |
| CN118210273A (en) | Numerical control center anti-collision processing state monitoring method based on data driving | |
| CN116690313B (en) | Failure monitoring method for machining cutter of web plate of aircraft structural member | |
| CN117420792A (en) | Central control system for composite processing machine tool | |
| CN111052016B (en) | Method for monitoring a machine tool | |
| Jiang et al. | Quality control method of used product remanufacturing process based on dynamic and non-normal EWMA control charts | |
| Park et al. | A study on the predictive maintenance of 5 axis CNC machine tools for cutting of large aircraft parts |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |