Method for monitoring the state of wear of a cutting tool of a machine tool

Abstract

The application discloses a kind of cutting machine tool wear state monitoring methods, comprising the following steps: S1, the vibration signal and pressure signal of acquisition are made into original signal dataset;S2, original signal dataset in step S1 is analyzed, screened, and the vibration signal dataset after dimensionality reduction processing of processing is obtained;S3, obtains tool wear state model G;S4, input vibration sensor real-time acquisition tool vibration signal v and pressure sensor real-time acquisition tool pressure signal P, and utilize tool wear state model G in step S3 processing, host computer exports tool the tool wear state result under this vibration signal v, pressure signal P of tool.The vibration signal and pressure signal of tool are monitored to tool wear state by the application, whether tool is securely judged fast, accurately and the wear state of tool is obtained, improve the monitoring precision of tool wear state.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tool wear state monitoring, and particularly relates to a cutting machine tool wear state monitoring method. BACKGROUND

[0002] During cutting machining of a cutting machine, tool wear occurs due to severe tool vibration, severe friction between a workpiece machining surface and the tool, cutting heat generated during machining, etc. Tool wear will affect tool vibration and machining workpiece quality. Therefore, a cutting machine tool wear state monitoring method is needed to monitor tool wear state.

[0003] Currently, tool wear state monitoring methods mainly include direct monitoring and indirect monitoring.

[0004] 1. Direct monitoring is to monitor tool wear state by professional instruments (e.g., a profilometer, a microscope, etc.) when the cutting machine stops running.

[0005] 2. Indirect monitoring is to measure physical quantities (e.g., vibration signals, acoustic emission signals, etc.) related to tool wear.

[0006] Direct monitoring is complicated, and indirect monitoring only considers the relationship between physical quantities generated when the tool contacts the workpiece and tool wear state. Both monitoring methods do not consider the influence of external environment on tool wear, which will affect tool wear state monitoring accuracy. SUMMARY

[0007] The present application solves the technical problem of low tool wear state monitoring accuracy. The present application provides a cutting machine tool wear state monitoring method, which can monitor tool wear state from tool vibration and pressure to improve tool wear state monitoring accuracy.

[0008] The technical solution adopted by the present application to solve the technical problem is a cutting machine tool wear state monitoring method, comprising the following steps:

[0009] S1. Collecting tool vibration signals by using a vibration sensor and collecting tool pressure signals by using a pressure sensor, and forming raw signal data sets from the collected vibration signals and pressure signals.

[0010] S2. Analyzing, screening, and dimensionally reducing the raw signal data sets in step S1 to obtain vibration signal data sets after analysis, screening, and dimensionally reduction.

[0011] S3, build an HMM model, and train the HMM model using the vibration signal data set analyzed, screened and dimensionally reduced in step S2, to obtain a tool wear state model G;

[0012] S4, input the vibration signal v of the tool collected by the vibration sensor in real time and the pressure signal P of the tool collected by the pressure sensor in real time, and process them using the tool wear state model G in step S3, so that the upper computer outputs the tool wear state result of the tool under the vibration signal v and the pressure signal P.

[0013] Therefore, the tool wear state is monitored through the vibration signal and the pressure signal of the tool, the tool wear state can be quickly and accurately judged, the monitoring precision of the tool wear state is improved, and the economic loss caused by excessive tool wear or tool loosening is reduced.

[0014] Further, the step S2 comprises the following steps:

[0015] S2-1, according to the typical wear rate change, the tool wear degree is divided into an initial wear stage, a normal wear stage and an acute wear stage;

[0016] S2-2, the vibration signals v1, v2 and v3 of the tool collected by the vibration sensor on the X-axis, the Y-axis and the Z-axis are obtained, and the vibration signals v1, v2 and v3 are analyzed in the time domain and the frequency domain to obtain characteristic parameters of the vibration signals v1, v2 and v3;

[0017] S2-3, the pressure signal P of the tool collected by the pressure sensor is obtained;

[0018] S2-4, the characteristic parameters in step S2-2 and the pressure signal P in step S2-3 are combined to form an original sample D, and the original sample D is subjected to characteristic parameter screening by using a ReliefF algorithm, and the weight W (A i ) of each characteristic parameter:

[0019] ;

[0020] S2-5, the characteristics with W (A i )>0.1 are selected to form an overall data sample E:

[0021] ;

[0022] S2-6, the overall data sample E in step S2-5 is subjected to dimension reduction by using principal component analysis to obtain a covariance matrix R:

[0023] ;

[0024] S2-7, calculate the p eigenvalues of the covariance matrix R in step S2-6, and arrange them from small to large to obtain m1, m2, …, mp and the corresponding p eigenvectors L1, L2, …, Lp;

[0025] S2-8, the i-th principal component is , select several principal components to form a training sample to replace the data sample E, so that the cumulative contribution rate ψ ≥ 99%,

[0026] The calculation formula of the cumulative contribution rate ψ is:

[0027] ;

[0028] In step S2-4, R is a sample data randomly extracted from the feature set, H j (j=1, 2, …, k) are the k most nearest neighbors H j (j=1, 2, …, k) are the k most nearest neighbors M j (C) (j=1, 2, …, k), P(C) is the proportion of the class, and P(Class(R)) is the proportion of the sample class randomly selected;

[0029] Wherein: m is the sample number, is the threshold value of the feature statistical index, k is the number of nearest neighbors, and T is the output of each characteristic feature statistical index T. Thus, the feature parameters with poor correlation and poor sensitivity to tool wear can be removed, the data dimension is reduced, and the operation speed and accuracy of the monitoring method are improved; dimension reduction can eliminate redundant information.

[0030] Further, the step S3 includes the following steps:

[0031] S3-1, the initial probability distribution vector π i is represented as:

[0032] ;

[0033] S3-2, the probability a of transferring from the current state i to state j based on the observation value k of state j ij is represented as:

[0034] ,

[0035] The state transition probability matrix A is represented as:

[0036] ;

[0037] S3-3, the probability b of the occurrence of the observation value k based on the state j jk is represented as:

[0038] ,

[0039] The observation probability matrix B is represented as:

[0040] ;

[0041] The HMM model is represented as:

[0042] ;

[0043] The HMM model parameter re-estimation formula is represented as:

[0044]

[0045] The HMM model parameters in step S3-4 are re-estimated using the HMM model parameter re-estimation formula until the HMM model converges, and a tool wear state model G is obtained.

[0046] wherein q1 represents the state at initial time 1, N represents the number of Markov chain states in the HMM model, O is the observation sequence, V is the set of observable states, a is the forward variable, b is the backward variable, and L is the number of observation sequences.

[0047] Further, the step S4 comprises the following steps:

[0048] S4-1, the pressure signal P input by the tool is judged, if the pressure P is less than the set threshold value, the host computer issues an alarm, and the loose tool is tightened, if the pressure P is greater than the set threshold value, the wear state of the tool is continuously monitored;

[0049] S4-2, the input tool vibration signal v and the pressure signal P are input into the tool wear state model G, and the likelihood probabilities P1, P2 and P3 are output;

[0050] S4-3, if P1>P2 and P1>P3, it indicates that the tool is in the initial wear stage, and the wear state of the tool is continuously monitored;

[0051] S4-4, if P2>P1 and P2>P3, it indicates that the tool is in the normal wear stage, and the wear state of the tool is continuously monitored;

[0052] S4-5, if P3>P1 and P3>P2, it indicates that the tool is in the rapid wear stage, and the host computer issues an alarm, and the tool needs to be replaced. Thus, whether the tool is loose can be judged during the tool wear state monitoring, the influence of the loose tool on the tool wear state monitoring is avoided, and the accuracy of the tool wear state monitoring is further improved.

[0053] Further, in step S2-2, the characteristic parameters include:

[0054] mean value , root mean square , standard deviation , peak value , absolute value , root amplitude , kurtosis factor , margin factor , peak factor , pulse factor , waveform factor , center of gravity frequency .

[0055] Further, in step S2-2, the mean value :

[0056] ;

[0057] root mean square :

[0058] ;

[0059] standard deviation :

[0060] ;

[0061] peak value :

[0062] ;

[0063] absolute value :

[0064] ;

[0065] root amplitude :

[0066] ;

[0067] kurtosis factor :

[0068] ;

[0069] margin factor :

[0070] ;

[0071] peak factor :

[0072] ;

[0073] pulse factor :

[0074] ;

[0075] waveform factor :

[0076] ;

[0077] centroid frequency :

[0078] ;

[0079] wherein: x i is a signal point, N is the number of signal points, f i is the signal frequency, and p(f i ) is the power spectrum of the signal.

[0080] Further, in step S2-5, the overall data sample E includes:

[0081] a data sample E1 of an initial wear stage, a data sample E2 of a normal wear stage, and a data sample E3 of an abrupt wear stage.

[0082] Further, the step S2-6 includes the following steps:

[0083] S2-6-1, calculating the mean value of each feature and the standard deviation ;

[0084] S2-6-2, normalizing the features to obtain a feature matrix X:

[0085] ;

[0086] S2-6-3, calculating the covariance matrix R of the normalized feature matrix X:

[0087] .

[0088] Further, in step S2-1, the wear degree of the tool in the initial wear stage is [0mm, 0.10mm), the wear degree of the tool in the normal wear stage is [0.10mm, 0.18mm], and the wear degree of the tool in the abrupt wear stage is (0.18mm, 0.30mm].

[0089] Further, in step S4-2, the tool wear state model G includes:

[0090] Tool wear state model G1 in the initial wear stage, tool wear state model G2 in the normal wear stage, and tool wear state model G3 in the rapid wear stage.

[0091] Compared with the prior art, the beneficial effects of the present invention are:

[0092] By monitoring the vibration and pressure signals of the cutting tool, it is possible to quickly and accurately determine whether the tool is securely installed and to determine its wear condition, thereby improving the accuracy of tool wear monitoring and reducing economic losses caused by excessive tool wear or tool loosening. Attached Figure Description

[0093] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0094] Figure 1 This is a flowchart of the method for monitoring the wear condition of cutting machine tool tools according to the present invention;

[0095] Figure 2 This is a flowchart of step S2 of the present invention;

[0096] Figure 3 This is a flowchart of step S4 of the present invention;

[0097] Figure 4 This is a flowchart of steps S2-6 of the present invention;

[0098] Figure 5 This is a control block diagram of the method for monitoring the wear state of cutting machine tool tools according to the present invention.

[0099] In the diagram: 1. Vibration sensor; 2. Pressure sensor; 3. Host computer; 4. Terminal module; 5. Piezoelectric conversion module; 6. Power supply module; 7. Router module; 8. Coordinator module; 9. ZigBee network. Detailed Implementation

[0100] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0101] In the description of the present application, it needs to be understood that the orientation or positional relationship indicated by the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the features defined as "first", "second" can be explicitly or implicitly included one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0102] In the description of the present application, it needs to be understood that unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0103] As Figures 1 to 5 shown is the most preferred embodiment of the present application, the cutting tool wear state monitoring method of the cutting machine tool of the embodiment comprises the following steps:

[0104] S1, the vibration signal of the tool is collected by using the vibration sensor 1, the pressure signal of the tool is collected by using the pressure sensor 2, and the collected vibration signal and pressure signal are composed into an original signal data set;

[0105] S2, the original signal data set in step S1 is analyzed, filtered and dimensionally reduced to obtain an analyzed, filtered and dimensionally reduced vibration signal data set;

[0106] S3, an HMM model (Hidden Markov Model) is constructed, and the analyzed, filtered and dimensionally reduced vibration signal data set in step S2 is used to train the HMM model to obtain a tool wear state model G;

[0107] S4, the vibration sensor 1 collects the vibration signal v of the tool in real time and the pressure sensor 2 collects the pressure signal P of the tool in real time, and processes by using the tool wear state model G in step S3, and the upper computer 3 outputs the tool wear state result of the tool under the vibration signal v and the pressure signal P. Thus, the tool wear state is monitored by the vibration signal and the pressure signal of the tool, the tool wear state can be quickly and accurately judged, the monitoring precision of the tool wear state is improved, and the economic loss caused by excessive tool wear or tool loosening is reduced.

[0108] Specifically, the vibration sensor 1 is installed on the side surface of the tool handle, and the pressure sensor 2 is installed between the lower surface of the tool handle and the machine tool.

[0109] Specifically, the vibration sensor 1 is connected with the terminal module 4, the pressure sensor 2 is connected with the terminal module 4 through the piezoelectric conversion module 5, the power module 6 is connected with the terminal module 4, the terminal module 4, the router module 7 and the coordinator module 8 are networked through the ZigBee network 9, and the coordinator module 8 is connected with the upper computer 3 through USB to TTL.

[0110] Specifically, the power module 6 supplies power for the operation of the vibration sensor 1, the pressure sensor 2, the piezoelectric conversion module 5, the terminal module 4, the router module 7, the coordinator module 8 and the upper computer 3; the router module 7 is used for data transmission transfer to increase the transmission distance; the coordinator module 8 is used for collecting the cutting machine tool data collected by each terminal module 4 and transmitting to the upper computer 3; and the ZigBee network 9 is used for ensuring the data transmission between the terminal module 4, the router module 7 and the coordinator module 8.

[0111] Specifically, the vibration sensor 1, the pressure sensor 2, the piezoelectric conversion module 5 and the terminal module 4 are provided with multiple, which can simultaneously monitor the wear state of multiple tools.

[0112] For example, the model of the vibration sensor 1 is ADXL345, the model of the pressure sensor 2 is MD30-60, the model of the terminal module 4 is CC2530F256, the working frequency of the ZigBee network 9 is 2.4GHz, the networking mode is tree type, the transmission rate is high and the networking is convenient.

[0113] In the embodiment, step S2 includes the following steps:

[0114] S2-1, according to the typical wear rate change, the tool wear degree is divided into initial wear stage, normal wear stage and rapid wear stage;

[0115] S2-2, acquire vibration signals v1, v2, v3 of the tool in X-axis, Y-axis, and Z-axis collected by the vibration sensor 1, and perform time domain and frequency domain analysis on the vibration signals v1, v2, v3 to obtain characteristic parameters of the vibration signals v1, v2, v3;

[0116] S2-3, acquire the pressure signal P of the tool collected by the pressure sensor 2;

[0117] S2-4, form an original sample D by combining the characteristic parameters in step S2-2 and the pressure signal P in step S2-3, and perform characteristic parameter screening on the original sample D by using a ReliefF algorithm (the ReliefF algorithm is an existing algorithm), and the weight WA of each characteristic parameter i :

[0118] ;

[0119] S2-5, select the characteristics with WA i > 0.1 to form an overall data sample E:

[0120] ;

[0121] S2-6, perform dimension reduction on the overall data sample E in step S2-5 by using principal component analysis to obtain a covariance matrix R:

[0122] ;

[0123] S2-7, calculate p characteristic values of the covariance matrix R in step S2-6, and arrange them from small to large to obtain m1, m2, …, mp and p characteristic vectors L1, L2, …, Lp corresponding thereto;

[0124] S2-8, the i-th principal component is , select a plurality of principal components to form a training sample instead of the data sample E, so that the cumulative contribution rate ψ ≥ 99%,

[0125] The calculation formula of the cumulative contribution rate ψ is:

[0126] ;

[0127] In step S2-4, R is a sample data randomly extracted from the characteristic set, H j (j=1, 2, …, k) are k most nearest neighbors H j (j=1, 2, …, k) found from the characteristic set inconsistent with each category, M j (C) (j=1, 2, …, k), P(C) is the proportion of the category, and P(Class(R)) is the proportion of the randomly selected sample category.

[0128] wherein: m is the sample sampling times, is the threshold value of the feature statistical quantity index, k is the number of nearest neighbor samples, and T is the feature statistical index T of each characteristic output. Thus, the feature parameters with poor correlation and poor sensitivity to tool wear can be removed, the data dimension is reduced, and the operation speed and accuracy of the monitoring method are improved; dimension reduction can eliminate redundant information.

[0129] In the present embodiment, step S3 comprises the following steps:

[0130] S3-1, initial probability distribution vector π i is expressed as:

[0131] ;

[0132] S3-2, probability a of transferring from state i to state j based on the current state ij is expressed as:

[0133] ,

[0134] The state transition probability matrix A is expressed as:

[0135] ;

[0136] S3-3, probability b of occurrence of observation value k based on state j jk is expressed as:

[0137] ,

[0138] The observation value probability matrix B is expressed as:

[0139] ;

[0140] S3-4, the HMM model is expressed as:

[0141] ;

[0142] S3-5, the HMM model parameter re-estimation calculation formula is expressed as:

[0143]

[0144] The HMM model parameters in step S3-4 are re-estimated using the HMM model parameter re-estimation calculation formula until the HMM model converges, and a tool wear state model G is obtained;

[0145] Wherein, q1 represents the state of initial time 1, N represents the number of Markov chain states in the HMM model, O is the observation sequence, V is the observable state set, alpha is the forward variable, beta is the backward variable, L is the number of observation value sequence.

[0146] In the embodiment, step S4 comprises the following steps:

[0147] S4-1, judging the pressure signal P input by the tool, if the pressure P is less than the set threshold, the host computer 3 issues an alarm, and the loose tool is tightened, if the pressure P is greater than the set threshold, the wear state of the tool is continuously monitored;

[0148] S4-2, inputting the input tool vibration signal v and the pressure signal P into the tool wear state model G, and outputting the likelihood probability P1O|λ, P2O|λ, P3O|λ;

[0149] S4-3, if P1O|λ>P2O|λ and P1O|λ>P3O|λ, it indicates that the tool is in the initial wear stage, and the wear state of the tool is continuously monitored;

[0150] S4-4, if P2O|λ>P1O|λ and P2O|λ>P3O|λ, it indicates that the tool is in the normal wear stage, and the wear state of the tool is continuously monitored;

[0151] S4-5, if P3O|λ>P1O|λ and P3O|λ>P2O|λ, it indicates that the tool is in the sharp wear stage, and the host computer 3 issues an alarm, and the tool needs to be replaced. Thus, whether the tool is loose can be judged between the tool wear state monitoring, the influence of the tool loosening on the tool wear state monitoring is avoided, and the precision of the tool wear state monitoring is further improved.

[0152] In the embodiment, in step S2-2, the characteristic parameters include:

[0153] Mean , Root mean square , Standard deviation , Peak value , Absolute value , Square root amplitude , Kurtosis factor , Margin factor , Peak factor , Pulse factor , Waveform factor , Barycentric frequency ;

[0154] Mean :

[0155] ;

[0156] Root mean square :

[0157] ;

[0158] Standard deviation :

[0159] ;

[0160] Peak value :

[0161] ;

[0162] Absolute value :

[0163] ;

[0164] Square root amplitude :

[0165] ;

[0166] Kurtosis factor :

[0167] ;

[0168] Margin factor :

[0169] ;

[0170] Peak factor :

[0171] ;

[0172] Impulse factor :

[0173] ;

[0174] Waveform factor :

[0175] ;

[0176] Center of gravity frequency :

[0177] ;

[0178] Where: x i is the signal point, N is the number of signal points, f i is the signal frequency, pfi Power spectrum of the signal.

[0179] In this embodiment, the overall data sample E includes:

[0180] Data sample E1 of the initial wear stage, data sample E2 of the normal wear stage and data sample E3 of the sharp wear stage.

[0181] In this embodiment, step S2-6 includes the following steps:

[0182] S2-6-1, calculating the mean value of each feature and the standard deviation ;

[0183] S2-6-2, normalizing the features to obtain a feature matrix X:

[0184] ;

[0185] S2-6-3, calculating the covariance matrix R of the normalized feature matrix X:

[0186] .

[0187] In this embodiment, in step S2-1, the wear degree of the tool in the initial wear stage is [0mm, 0.10mm, the wear degree of the tool in the normal wear stage is [0.10mm, 0.18mm] and the wear degree of the tool in the sharp wear stage is 0.18mm, 0.30mm].

[0188] In this embodiment, in step S4-2, the tool wear state model G includes:

[0189] Tool wear state model G1 of the initial wear stage, tool wear state model G2 of the normal wear stage and tool wear state model G3 of the sharp wear stage.

[0190] In summary, the present application monitors the tool wear state through the vibration signal and the pressure signal of the tool, can quickly and accurately judge whether the tool is firmly installed and obtain the wear state of the tool, improves the monitoring precision of the tool wear state and reduces the economic loss caused by excessive tool wear or tool loosening.

[0191] The above describes the ideal embodiments of the present application as an inspiration, and through the above description, relevant personnel can make various changes and modifications without deviating from the technical idea of the present application. The technical scope of the present application is not limited to the contents of the specification, and must be determined by the scope of the claims.

Claims

1. A method for monitoring the wear condition of cutting machine tool tools, characterized in that, Includes the following steps: S1. Use a vibration sensor (1) to collect the vibration signal of the tool and a pressure sensor (2) to collect the pressure signal of the tool. Combine the collected vibration signal and pressure signal into a raw signal dataset. S2. Analyze, filter, and reduce the dimensionality of the original signal dataset in step S1 to obtain the analyzed, filtered, and dimensionality-reduced vibration signal dataset. S3. Construct an HMM model and train the HMM model using the vibration signal dataset that has been analyzed, filtered, and dimensionality reduced in step S2 to obtain the tool wear state model G. S4. Input the vibration sensor (1) to collect the vibration signal v of the tool in real time and the pressure sensor (2) to collect the pressure signal P of the tool in real time, and process it using the tool wear state model G in step S3. The host computer (3) outputs the tool wear state result under the vibration signal v and pressure signal P. Step S4 includes the following steps: S4-1. Judge the pressure signal P input by the tool. If the pressure P is less than the set threshold, the host computer (3) will issue an alarm and tighten the loose tool. If the pressure P is greater than the set threshold, continue to monitor the wear status of the tool. S4-2. Input the input tool vibration signal v and pressure signal P into the tool wear state model G, and output the likelihood probabilities P1(O|λ), P2(O|λ), and P3(O|λ). S4-3, if P1(O|λ)>P2(O|λ) and P1(O|λ)>P3(O|λ), it indicates that the tool is in the initial wear stage, and the wear status of the tool should continue to be monitored. S4-4, if P2(O|λ)>P1(O|λ) and P2(O|λ)>P3(O|λ), it indicates that the tool is in the normal wear stage, and the wear status of the tool should continue to be monitored. S4-5, if P3(O|λ)>P1(O|λ) and P3(O|λ)>P2(O|λ), it indicates that the tool is in a stage of rapid wear, and the host computer (3) will issue an alarm, requiring the tool to be replaced.

2. The method for monitoring the wear condition of cutting machine tool tools according to claim 1, characterized in that, Step S2 includes the following steps: S2-1. Based on typical wear rate changes, tool wear is divided into initial wear stage, normal wear stage, and rapid wear stage. S2-2. Obtain the vibration sensor (1) collect the vibration signals v1, v2, v3 of the tool on the X-axis, Y-axis, and Z-axis, and perform time-domain and frequency-domain analysis on the vibration signals v1, v2, v3 to obtain the characteristic parameters of the vibration signals v1, v2, v3; S2-3, Obtain the pressure signal P of the cutting tool collected by the pressure sensor (2); S2-4. Combine the feature parameters from step S2-2 and the pressure signal P from step S2-3 to form the original sample D. Then, use the ReliefF algorithm to filter the feature parameters of the original sample D, and assign weights W (A) to each feature parameter. i ): ； S2-5, Select W(A) i Features with a value greater than 0.1 constitute the overall data sample E: ； S2-6. Perform principal component analysis on the overall data sample E from step S2-5 to reduce its dimensionality, and obtain the covariance matrix R: ； S2-7. Calculate the p eigenvalues ​​of the covariance matrix R in step S2-6, and arrange them in ascending order to obtain m1, m2, ..., mp and the corresponding p eigenvectors L1, L2, ..., Lp. S2-8, the i-th principal component is Several principal components are selected to form training samples to replace data samples E, such that the cumulative contribution rate ψ ≥ 99%. The formula for calculating the cumulative contribution rate ψ is: ； In steps S2-4, R is a sample data arbitrarily drawn from the feature set, and H... j (j=1,2,…,k) are the k most nearest neighbors H found from the sample set of the same class in R. j (j=1,2,…,k), find the k most nearest neighbors M from each set of features that are inconsistent across categories. j (C) (j=1,2,…,k), P(C) is the proportion of this category, and P(Class(R)) is the proportion of a randomly selected sample category; Where: m is the number of sample samplings. is the threshold for the feature statistics index, k is the number of nearest neighbor samples, and T is the feature statistics index T of each output characteristic.

3. The method for monitoring the wear condition of cutting machine tool tools according to claim 2, characterized in that, Step S3 includes the following steps: S3-1, Initial probability distribution vector π i Represented as: ； S3-2, Based on the probability a of transitioning from the current state i to the current state j ij Represented as: ， The state transition probability matrix A is expressed as: ； S3-3, Based on the probability b of observation k occurring in state j jk Represented as: ， The observation probability matrix B is represented as follows: ； S3-4, HMM model is represented as follows: ； S3-5, The formula for re-estimating the parameters of the HMM model is expressed as follows: The HMM model parameters in step S3-4 are re-estimated using the HMM model parameter re-estimation formula until the HMM model converges, thus obtaining the tool wear state model G. Where q1 represents the state at initial time 1, N represents the number of Markov chain states in the HMM model, O is the observation sequence, V is the set of observable states, α is the forward variable, β is the backward variable, and L is the number of observation sequences.

4. The method for monitoring the wear condition of cutting machine tool tools according to claim 2, characterized in that, In step S2-2, the characteristic parameters include: mean Root mean square Standard deviation Peak absolute value Root amplitude kurtosis factor Margin factor Peak factor Pulse factor Waveform factor Center of gravity frequency .

5. The method for monitoring the wear condition of cutting machine tool tools according to claim 2, characterized in that, In step S2-2, the mean : ； Root Mean Square : ； Standard deviation : ； peak : ； absolute value : ； Root amplitude : ； kurtosis factor : ； margin factor : ； Peak factor : ； Pulse factor : ； Waveform factor : ； Center of gravity frequency : ； Where: x i Let N be the number of signal points, and f be the signal point. i p(f) is the signal frequency. i The power spectrum of the signal.

6. The method for monitoring the wear condition of cutting machine tool tools according to claim 2, characterized in that, In steps S2-5, the overall data sample E includes: Data sample E1 for the initial wear stage, data sample E2 for the normal wear stage, and data sample E3 for the rapid wear stage.

7. The method for monitoring the wear condition of cutting machine tool tools according to claim 2, characterized in that, Steps S2-6 include the following steps: S2-6-1. Calculate the mean of each feature. and standard deviation ; S2-6-2. Standardize the features to obtain the feature matrix X: ； S2-6-3, Calculate the covariance matrix R of the standardized characteristic matrix X: 。 8. The method for monitoring the wear condition of cutting machine tool tools according to claim 2, characterized in that, In step S2-1, the wear degree of the tool in the initial wear stage is [0mm, 0.10mm], the wear degree of the tool in the normal wear stage is [0.10mm, 0.18mm], and the wear degree of the tool in the rapid wear stage is (0.18mm, 0.30mm).

9. The method for monitoring the wear condition of cutting machine tool tools according to claim 1, characterized in that, In step S4-2, the tool wear state model G includes: Tool wear state model G1 in the initial wear stage, tool wear state model G2 in the normal wear stage, and tool wear state model G3 in the rapid wear stage.

Images