A tool wear self-adaptive monitoring method and system based on power signal

Abstract

The application discloses a tool wear and breakage self-adaptive alarm method and system based on a power signal, utilizes a real-time power signal acquisition system, reflects the change of a power signal in a cutting process of a machine tool on an amplitude with tool wear and breakage, and self-adaptively monitors the tool wear and breakage state; an adaptive alarm system is utilized to establish a tool wear and breakage dynamic alarm line, the difference between a real-time signal and the alarm line is compared, and the tool wear and breakage self-adaptive monitoring is realized. The application realizes the self-adaptive monitoring of tool wear and breakage in a tool dynamic cutting process, simultaneously establishes an evaluation standard of the adaptive alarm algorithm, can realize the real-time monitoring of the tool wear and breakage state, has higher convenience and obvious economy, and meanwhile, the signal is stable, the operation is convenient, and the application is safe and reliable.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of intelligent monitoring, and particularly relates to a tool wear and breakage adaptive monitoring method and system based on power signals. BACKGROUND

[0002] As a tool commonly used in mechanical manufacturing, a tool plays an important role in precision machining. In the numerical control milling process, tool wear and breakage is one of the common faults of numerical control machine tools. Tool wear and breakage has a great influence on the energy consumption of the machine tool spindle and is directly related to the machining quality of the part. Relevant research results show that tool wear is the root cause of tool failure, and the maintenance cost caused by tool failure accounts for 15% to 40% of the cost of goods produced, and the downtime caused by tool failure accounts for about 20% of the total downtime of the tool. Therefore, real-time monitoring of the tool state is very critical.

[0003] Tool wear and breakage monitoring has always been a very critical technology in automated processing. It is an important means to realize production process automation, unmanned operation, ensure product quality, improve production efficiency, and reduce equipment failure. However, due to the harsh environment in the actual cutting process, the diversity of tools and workpieces, and the large amount of discrete data collected, actual monitoring is very difficult.

[0004] There are many tool wear and breakage monitoring methods. According to the different principles of tool wear detection, tool wear monitoring is mainly divided into direct monitoring and indirect monitoring. Direct monitoring refers to judging the wear state by identifying the geometric shape of the cutting edge, the surface quality or measuring the change of the tool cutting edge parameters, such as optical measurement method, resistance measurement method, computer image processing method, etc. Indirect monitoring method monitors not the tool itself, but the signals related to the tool itself, such as cutting force, acoustic emission, vibration, current and power signal, etc. According to the change of the signal in the cutting process, the tool state information is obtained. It is worth noting that although the indirect monitoring method is difficult to construct a model and has lower accuracy than the direct method, the sensor used is easy to install, can be monitored in real time and online, and has lower detection cost. Therefore, the indirect method has been widely used in tool wear monitoring in the past few years. Among them, the power signal monitoring sensor is easy to install and generally does not affect the machine tool processing, and has high signal-to-noise ratio, so the power monitoring method becomes the main method selected in this paper. SUMMARY

[0005] The purpose of the present application is to overcome the disadvantages of high monitoring cost and inconvenience of real-time monitoring of direct monitoring method, and to propose a tool wear and breakage adaptive monitoring method and system based on power signals. The method constructs an adaptive alarm algorithm, which can realize adaptive monitoring of tool wear and breakage, and reduces the requirement for the user's own experience.

[0006] Before the tool wear self-adaptive monitoring method based on power signal is proposed, the self-adaptive alarm system is first proposed, which is described in detail as follows.

[0007] 1) In the cutting state of the tool, the current signal and voltage signal of the tool driving motor are sampled at equal time intervals (power grid frequency 50 Hz, period 20 ms, sampling frequency M*50, M can be 32, 64, 128, 256).

[0008] 2) M voltage values U t ,U t+1 ,...,U M-1 and corresponding current signal values I t ,I t+1 ,...,I M-1 collected in step 1) are sequentially sampled, and the average power P t at time t is calculated.

[0009]

[0010] 3) The position of time t is slid by the value of M divided by 10, and step 2) is repeated.

[0011] 4) According to K (K≥60) average power signals obtained in step 3) in sequence, the training data set T = {X1,X2,…,X K} of power is preprocessed, and the first-order difference feature of the training data set T is extracted, that is, the feature training set S = {Y1,Y2,…,Y K} of power is obtained.

[0012] 5) The probability density f(Y) of the training set S is derived as:

[0013]

[0014] Under the assumption that its distribution is Gaussian distribution, the mean m and standard deviation h of the training set S are:

[0015]

[0016] According to the Lydard criterion, set the initial threshold P1 of the training set S, and the parameter λ controls the upper and lower limits, and the upper and lower limits are P1_up and P1_low respectively, which are:

[0017] P1_up = m + λ*h

[0018] P1_low = m - λ*h

[0019] 6) All data of the training data set T are sorted in ascending order to obtain the data set Z = {Z1,Z2,…,Z MThe first quantile Q1 and the third quantile Q3 of the data set Z are calculated and expressed as:

[0020]

[0021] As known from the above, the interquartile range IQR of the training data set Z is derived as:

[0022]

[0023] The initial threshold P2, alpha of the training data set T is determined according to the multiple interquartile range principle, and the upper and lower limits thereof are represented as P2_up and P2_low, respectively, and the determination manner is as follows:

[0024] P2_up = Q3 + alpha * IQR

[0025] P2_low = Q1 - alpha * IQR

[0026] 7) With the lapse of running time, whether the K+1th X k+1 and Y k+1 is out of the limit is judged by P2, and P1, respectively;

[0027] The working method of the adaptive alarm system is as follows: when Y k+1 is within the threshold P1_up and P 1- low, the system judges that the state is normal; when X k+1 is within the threshold P2_up and P2_low, the system judges that the state is normal; when Y k+1 is out of the threshold P1_up and P 1- low, and X k+1 is within the threshold P2_up and P2_low, the system judges that the state is normal; when Y k+1 is out of the threshold P1_up and P 1- low, and X k+1 is out of the threshold P2_up and P2_low, the system judges that the state is an alarm state.

[0028] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0029] A tool wear and breakage adaptive monitoring method based on a power signal, which utilizes a real-time power signal acquisition system to reflect the change of the power signal in the amplitude in the cutting process of a machine tool with tool wear and breakage, and dynamically identifies the tool wear and breakage condition; an adaptive alarm system is utilized to establish a dynamic alarm line of the power signal, and the difference between the real-time power signal and the alarm line is compared to realize adaptive monitoring of tool wear and breakage.

[0030] Preferably, the adaptive alarm system comprises the following steps:

[0031] 1) In the cutting state, the current signal and voltage signal of the tool driving motor are sampled at equal time intervals (power grid frequency 50 Hz, period 20 ms, sampling frequency M*50, M can be 32, 64, 128, 256).

[0032] 2) The M voltage values U t ,U t+1 ,...,U M-1 and corresponding current signal values I t ,I t+1 ,...,I M-1 successively collected in step 1) are used to calculate the average power P t at time t.

[0033]

[0034] 3) The position of time t is slid by a value of M divided by 10, and step 2) is repeated.

[0035] 4) According to the K (K≥60) average power signals successively obtained in step 3), the training data set T = {X1,X2,…,X K} of power is preprocessed, and the first-order difference feature of the training data set T is extracted, that is, the feature training set S = {Y1,Y2,…,Y K} of power is obtained.

[0036] 5) The probability density f(Y) of the training set S is derived as:

[0037]

[0038] Under the assumption that its distribution is Gaussian distribution, the mean m and standard deviation h of the training set S are:

[0039]

[0040] According to the Lydian criterion, the initial threshold P1 of the training set S is set, and λ is a parameter for controlling the upper and lower limits, and the upper and lower limits are P1_up and P1_low, respectively, which are:

[0041] P1_up = m + λ*h

[0042] P1_low = m - λ*h

[0043] 6) All data of the training data set T are sorted in ascending order to obtain the data set Z = {Z1,Z2,…,Z M}, and the first quantile Q1 and the third quantile Q3 of the data set Z are calculated, which are:

[0044]

[0045] From the above, the interquartile range IQR of the training data set Z is derived as:

[0046]

[0047] The initial threshold P2,α of the training data set T is determined according to the multiple interquartile range principle, which is an empirical parameter artificially selected, and the upper and lower limits thereof are represented as P2_up and P2_low respectively, and the determination method is as follows:

[0048] P2_up = Q3 + α * IQR

[0049] P2_low = Q1 - α * IQR

[0050] 7) With the passage of time, P2, and P1 are used to determine whether the K+1th X k+1 and Y k+1 is out of the limit;

[0051] Repeat steps 4) to 7), draw a curve according to the change of time, and analyze the tool wear and damage trend.

[0052] The working method of the adaptive alarm system includes: when Y k+1 is within the threshold P1_up and P 1- low, the system determines that it is in a normal state; when X k+1 is within the threshold P2_up and P2_low, the system determines that it is in a normal state; when Y k+1 is outside the threshold P1_up and P 1- low, and X k+1 is within the threshold P2_up and P2_low, the system determines that it is in a normal state; when Y k+1 is outside the threshold P1_up and P 1- low, and X k+1 is outside the threshold P2_up and P2_low, the system determines that it is in an alarm state.

[0053] Preferably, the real-time power signal acquisition system obtains an average power signal by calculating the acquired three-phase voltage and current signals; specifically including the following steps:

[0054] 1) The hardware system containing the data acquisition device is connected in series between the tool driving motor and the power supply, the voltage and current signals are acquired by using the hardware system, and the real-time power signal is calculated;

[0055] 2) The standard power time sequence is saved into the hardware system as a training data set;

[0056] 3) System begins to collect and store power signal periodically throughout the cutting process.

[0057] A tool wear and breakage adaptive monitoring system based on power signal, comprising an adaptive alarm system and a real-time power signal acquisition system, which realizes adaptive monitoring of tool wear and breakage by comparing the difference between real-time signal and alarm line.

[0058] Compared with the prior art, the present application has the following advantages:

[0059] The present application realizes adaptive monitoring of tool wear and breakage in the case of tool cutting, and establishes an evaluation standard based on adaptive alarm algorithm, which can monitor the wear and breakage state of the tool in real time, and fills the gap of adaptive monitoring method of tool wear and breakage based on power signal. Compared with the existing tool wear and breakage monitoring, this method has high convenience and significant economy, and the signal is stable, the operation is convenient, safe and reliable. BRIEF DESCRIPTION OF DRAWINGS

[0060] Fig. 1 A flow chart of tool wear and breakage adaptive monitoring is provided for the present application.

[0061] Fig. 2 A three-phase alternating current real-time power calculation program block diagram is provided for the present application. DETAILED DESCRIPTION

[0062] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0063] Please refer to Figs. 1-2 In the embodiments of the present application, a tool wear and breakage adaptive monitoring method based on power signal uses a real-time power signal acquisition system to reflect the change of power signal in the amplitude reflected by tool wear and breakage in the cutting process of the machine tool, and to adaptively monitor the tool wear and breakage condition; an adaptive alarm system is used to establish a dynamic alarm line for tool wear and breakage, and to compare the difference between real-time signal and alarm line, so as to realize adaptive monitoring of tool wear and breakage.

[0064] The adaptive alarm system specifically comprises the following steps:

[0065] 1) In the cutting state of the tool, the current signal and voltage signal of the tool driving motor are sampled at equal time intervals (power grid frequency 50Hz, period 20ms, sampling frequency Mx50, M can be 32, 64, 128, 256);

[0066] 2) M voltage values U t ,U t+1 ,...,U M-1and the corresponding current signal value I t , t+1 , M-1 , calculate the average power P at time t t ;

[0067]

[0068] 3) slide the position of time t with the value of M divided by 10, repeat step 2);

[0069] 4) according to the K (K≥60) average power signals obtained in sequence in step 3), pre-process the training data set T = {X1, X2, …, X K} of power, extract the first-order difference features of the training data set T, that is, obtain the feature training set S = {Y1, Y2, …, Y K} of power;

[0070] 5) the probability density f(Y) of the training set S is derived as:

[0071]

[0072] Under the assumption that its distribution is Gaussian distribution, the mean m and the standard deviation h of the training set S are:

[0073]

[0074]

[0075] According to the Lydian criterion, set the initial threshold P1 of the training set S, and the parameter λ is the control upper and lower limit, and the upper and lower limits are P1_up and P1_low respectively:

[0076] P1_up = m + λ * h

[0077] P1_low = m - λ * h

[0078] 6) sort all data of the training data set T in ascending order to obtain the data set Z = {Z1, Z2, …, Z M}, calculate the first quantile Q1 and the third quantile Q3 of the data set Z, which are:

[0079]

[0080] As can be seen from the above, the interquartile range IQR of the training data set Z is derived as:

[0081]

[0082] The initial threshold P2,α of the training data set T is an empirical parameter artificially selected according to the principle of multiple quantile distance, and the upper and lower limits thereof are represented as P2_up and P2_low respectively, and the determination manner thereof is as follows:

[0083] P2_up=Q3+α*IQR

[0084] P2_low=Q1-α*IQR

[0085] 7) With the running time, P2, and P1 are used to determine whether the K+1th X k+1 and Y k+1 is out of the limit;

[0086] Repeat steps 4) to 7), draw a curve according to the change of time, and analyze the tool wear and damage trend.

[0087] The working method of the adaptive alarm system: when Y k+1 is within the threshold P1_up and P1_low, the system determines that it is in a normal state; when X k+1 is within the threshold P2_up and P2_low, the system determines that it is in a normal state; when Y k+1 is outside the threshold P1_up and P1_low, and X k+1 is within the threshold P2_up and P2_low, the system determines that it is in a normal state; when Y k+1 is outside the threshold P1_up and P1_low, and X k+1 is outside the threshold P2_up and P2_low, the system determines that it is in an alarm state.

[0088] The real-time power signal acquisition system obtains the average power signal through the calculation of the acquired three-phase voltage and current signals; and specifically includes the following steps:

[0089] 1) The hardware system containing the data acquisition device is connected in series between the machine tool and the power supply, the voltage and current signals are acquired by using the hardware system, and the real-time power signal is calculated;

[0090] 2) The standard power time sequence is saved into the hardware system as the training data set;

[0091] 3) The system starts to periodically acquire and store the power signals of the entire cutting process.

[0092] A tool wear and damage adaptive monitoring system based on a power signal comprises an adaptive alarm system and a real-time power signal acquisition system, and realizes adaptive monitoring of tool wear and damage by comparing the difference between the real-time signal and the alarm line.

[0093] Although the present application has been described in detail with reference to the foregoing embodiments, the technical solutions recorded in the foregoing embodiments can be modified, or some of the technical features can be replaced by equivalent features, by those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for adaptive monitoring of tool wear and breakage based on power signals, characterized in that, Comprising the following steps: 1) In the cutting state, the current signal and voltage signal of the tool driving motor are sampled at equal time intervals, wherein the power grid frequency is 50 Hz, the period is 20 ms, and the sampling frequency is Hz, 32, 64, 128, 256 can be taken. 2) M voltage values sequentially collected in step 1) and corresponding current signal values , calculate the average power at time t ; 3) slide with values that are multiples of 10 modulo M repeating step 2) at the time position. 4) sequentially obtained according to step 3) a training data set of power obtained by pre-processing the average power signals , wherein ; and extracting first-order differential features from the training data set T to obtain a feature training set of power ; 5) Establishing the first alarm threshold : Based on the feature training set , applying the Lyapunov criterion to determine its upper and lower limits and ; 6) Establishing a second alarm threshold : Based on the training data set , applying the principle of multiple quantile range to determine its upper and lower limits and ; 7) making a cooperative judgment: in real-time monitoring, the real-time power value at the moment and its corresponding first-order differential characteristics are obtained , when the first alarm threshold and are exceeded, and the second alarm threshold and are exceeded, the system determines that it is in an alarm state.

2. A tool wear adaptive monitoring method based on power signal according to claim 1, characterized in that, establishing the first alarm threshold in the step 5) Specifically comprises: Assuming the distribution of the feature training set is Gaussian, compute its mean and standard deviation ; set the upper and lower limits of the first alarm threshold as: wherein are control parameters chosen according to the Lyapunov criterion.

3. A tool wear adaptive monitoring method based on power signal according to claim 1, characterized in that, establishing a second alarm threshold in the step 6) Specifically comprises: All data of the training data set are sorted in ascending order to obtain a data set ; Computing a dataset of first quantiles and third quantiles ; Computing the interquartile range ; The upper and lower limits of the second alarm threshold P2 are set as: Wherein, α is an artificial experience parameter selected according to the principle of multiple quantile distance.

4. A method of adaptive tool wear and breakage monitoring based on power signal according to claim 1, characterized in that, In the step 7), the cooperative judgment specifically comprises: If When the threshold And Within the system is determined to be normal state; If The system determines as normal state when the threshold And Within, If At a threshold And P Otherwise, and At a threshold And The system is determined to be in a normal state; If At a threshold And Otherwise, and At a threshold And The system is determined to be in an alarm state.

5. A power signal based tool wear adaptive monitoring method according to claim 1, characterized in that, Further comprising the steps: Repeat steps 4) to 7), draw a curve according to the change of time, and analyze the tool wear and damage trend.

6. A method of adaptive tool wear and breakage monitoring based on power signal according to claim 1, characterized in that, The method obtains the average power signal in several cycles through the calculation of the collected three-phase voltage and current signals; specifically comprising the following steps: 1) The hardware system containing the data collector is connected in series between the tool driving motor and the power supply, the voltage and current signals are collected by using the hardware system, and the real-time power signal is calculated; 2) Save the standard power time sequence into the hardware system as the training data set; 3) The system starts to collect and store the power signal of the entire cutting process periodically.

7. A power signal based adaptive tool wear out monitoring system for performing the method of any one of claims 1 to 6, characterized by It comprises an adaptive alarm system and a real-time power signal acquisition system, and realizes adaptive monitoring of tool wear and damage by comparing the difference between the real-time signal and the alarm line.

Images