Tool anomaly detection system

Abstract

To provide a tool abnormality detecting system that can be prevented from erroneously determining abnormality of a tool.SOLUTION: The tool abnormality detecting system comprises a detecting part 10 that detects a sound wave generated when processing a material 200 to be ground using a tool 140 and outputs a detection signal based on the sound wave, and a control part 30 that determines the abrasion of the tool 140 on the basis of a detection signal. The control part 30 generates a determination signal indicating intensity of a basic frequency band of an abrasion sound wave that is generated when the tool 140 has abrasion and a determination signal indicating intensity of a frequency band which is higher than the basic frequency band, on the basis of the detection signal, and compares the respective intensities of the determination signals with respective abrasion thresholds set for the frequency bands to determine the abrasion of the tool.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a tool abnormality detection system.

Background Art

[0002] Conventionally, a tool abnormality detection system for detecting abnormalities of tools in machine tools has been proposed (see, for example, Patent Document 1). Specifically, this tool abnormality detection system includes a detection unit that detects sound waves generated when machining a workpiece using a tool, and a control unit. Then, the control unit extracts the intensity of the sound waves in a predetermined frequency band in order to reduce the influence of noise sound waves caused by background noise, and then determines the abnormality (that is, wear) of the tool by comparing the intensity of the extracted frequency band with a threshold value. Note that the noise sound waves caused by background noise include the voices of workers monitoring around the machine tool, the operating sounds of other machine tools installed around the target machine tool, the chime of the factory where the machine tool is arranged, and the like. In addition, the sound waves caused by background noise also include the sound waves of devices provided in the target machine tool, and examples thereof include the operating sounds of a motor, an automatic tool changer (that is, ATC (abbreviation for Automatic Tool Changer)), a coolant device, an oil mist collector, and the like.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, even in the tool abnormality detection system as described above, if the set frequency band includes noise sound waves caused by background noise, there is a possibility of misjudging the abnormality of the tool.

[0005] In view of the above points, the present invention aims to provide a tool abnormality detection system that can make it difficult to misidentify tool abnormalities. [Means for solving the problem]

[0006] Claim 1, for achieving the above objective, is a tool abnormality detection system comprising: a detection unit (10) that detects sound waves generated when a workpiece (200) is machined using a tool (140, 410) and outputs a detection signal based on the sound waves; and a control unit (30) that determines tool wear based on the detection signal. The control unit generates, based on the detection signal, a determination signal indicating the intensity of the fundamental frequency band of wear sound waves generated when the tool is worn, and a determination signal indicating the intensity of a frequency band higher than the fundamental frequency band. The control unit then determines tool wear by comparing the intensity of each determination signal with a wear threshold set for each frequency band. Furthermore, if the control unit determines that wear has occurred in the tool, it compares the intensity of each determination signal with a breakage threshold that is smaller than the wear threshold to determine if the tool is broken. .

[0007] According to this system, the control unit generates judgment signals for the fundamental frequency band of wear sound waves, as well as for frequency bands higher than this band. The control unit then compares each judgment signal with a wear threshold to determine wear. This helps to suppress misjudgments of tool wear.

[0008] The reference numerals in parentheses attached to each component indicate an example of the correspondence between that component and the specific components described in the embodiments described later. [Brief explanation of the drawing]

[0009] [Figure 1] This is a schematic diagram of the tool abnormality detection system in the first embodiment. [Figure 2] This is a block diagram of the control unit. [Figure 3] This diagram shows the relationship between frequency and intensity. [Figure 4] This is a flowchart of the actions performed by the control unit. [Figure 5] This is a schematic diagram of the tool abnormality detection system in the second embodiment. [Figure 6] This is a schematic diagram of the control unit in the third embodiment. [Figure 7] This figure shows the relationship between frequency and intensity in the fourth embodiment. [Figure 8] This is a schematic diagram showing the relationship between time, frequency, abrasion sound waves, and noise sound waves in another embodiment. [Modes for carrying out the invention]

[0010] The embodiments of the present invention will be described below with reference to the drawings. In the following embodiments, parts that are the same or equivalent to each other will be denoted by the same reference numerals.

[0011] (First Embodiment) The tool abnormality detection system of the first embodiment will be described with reference to Figure 1. The tool abnormality detection system of this embodiment is used in cutting processes and detects abnormalities in the tools of the target machine tool. In this embodiment, an abnormality detection system used when drilling holes in a workpiece as a cutting process will be used as an example.

[0012] The machine tool 100 in question is, for example, a machining center having a machining chamber. In this embodiment, the machine tool 100 includes a stage 110, a vise 120 that is placed on the stage 110 and fixes the workpiece 200, a spindle holder 130, a drill 140 that is mounted on the spindle holder 130 and processes the workpiece 200, and so on. The machine tool 100 then displaces the drill 140 to form a hole 201 in the workpiece 200. In this embodiment, the drill 140 corresponds to the tool. The workpiece 200 can be a material used in metal cutting, such as SUS or AL.

[0013] The tool abnormality detection system 1 includes a detection unit 10 and a control unit 30, etc.

[0014] The detection unit 10 outputs a detection signal corresponding to the detected sound wave, and is composed of a highly waterproof and dustproof sound-collecting microphone, etc., and is positioned near the workpiece 200. In this embodiment, the detection unit 10 detects processing sound waves generated when the workpiece 200 is cut, as will be described later. The processing sound waves in this embodiment include not only direct sound waves generated directly when the workpiece 200 is cut by the drill 140, but also wear sound waves generated when the drill 140 wears down and the workpiece 200 vibrates, as will be described later. The detection signal is an analog signal.

[0015] The control unit 30 is composed of a microcomputer or the like, which includes a CPU and a storage unit composed of non-transitional physical storage media such as ROM, RAM, flash memory, and HDD. As shown in Figure 2, the control unit 30 of this embodiment includes an amplification unit 31, an AD conversion unit 34, a signal processing unit 32, a determination unit 33, and the like. Note that CPU is an abbreviation for Central Processing Unit, ROM is an abbreviation for Read Only Memory, RAM is an abbreviation for Random Access Memory, and HDD is an abbreviation for Hard Disk Drive. Storage media such as ROM are non-transitional physical storage media.

[0016] The amplification unit 31 is connected to the detection unit 10 and amplifies the detection signal detected by the detection unit 10. The amplification unit 31 may be connected to the detection unit 10 by wireless communication or by wired communication. The detection signal transmitted from the detection unit 10 is a signal based on processed sound waves, but it also includes the influence of noise sound waves caused by background noise.

[0017] The AD conversion unit 34 samples the detection signal amplified by the amplification unit 31 at a predetermined sampling frequency and converts the detection signal, which is an analog signal, into a digital signal. Note that the sampling frequency in the present embodiment is set to a frequency higher than the highest frequency band from which the determination signal is derived by the signal processing unit 32 described later. Also, the Nyquist frequency, which is a value half of the sampling frequency, is set to a frequency higher than the highest frequency band from which the determination signal is derived by the signal processing unit 32 described later, similar to the sampling frequency. Note that the highest frequency band from which the determination signal is derived by the signal processing unit 32 described later is, in other words, the highest frequency band determined by the determination unit 33 described later, and in the present embodiment, it is a frequency band four times as high.

[0018] The signal processing unit 32 derives a determination signal indicating the intensity in a predetermined frequency band. The signal processing unit 32 in the present embodiment has a configuration including an FFT (Fast Fourier Transform) circuit unit that executes a discrete Fourier transform (that is, Discrete Fourier Transform) at high speed and a band-pass filter or the like so as to be able to derive a determination signal.

[0019] Here, the wear sound wave includes, in addition to the sound wave of the fundamental frequency, sound waves of harmonics that are integer multiples of 2 or more. Then, the signal processing unit 32 derives a plurality of determination signals indicating the respective intensities in a plurality of different frequency bands. Specifically, the signal processing unit 32 derives a determination signal indicating the intensity of the frequency band that is the fundamental of the wear sound wave and the frequency bands that are integer multiples of 2 or more of the fundamental frequency band. Note that the fundamental frequency band is set to a predetermined range including the natural frequency of the workpiece 200, and the natural frequency of the workpiece 200 is derived in advance by experiments or the like.

[0020] When the determination unit 33 receives the determination signal derived by the signal processing unit 32, it performs a wear determination as an abnormal determination to determine wear of the drill 140 by comparing the intensity of the determination signal with the wear threshold. Specifically, as described above, the signal processing unit 32 derives determination signals in multiple different frequency bands. In this case, since the wear sound waves and noise sound waves contain different harmonics, the proportion of noise sound waves that affect the determination signal also differs, as shown in Figure 3. Therefore, the determination unit 33 performs a wear determination as an abnormal determination to determine wear of the drill 140 by comparing the intensity of the determination signal in each frequency band with the wear threshold.

[0021] For example, the determination unit 33 in this embodiment compares the intensity of the determination signal with the wear threshold in the basic frequency band, the 2x frequency band, the 3x frequency band, and the 4x frequency band. In other words, the signal processing unit 32 in this embodiment derives determination signals for the basic frequency band, the 2x frequency band, the 3x frequency band, and the 4x frequency band. Note that the intensity of harmonics decreases as the multiple increases. For this reason, a wear threshold is set for each frequency band, and it decreases as the frequency increases. The determination unit 33 in this embodiment determines that the drill 140 is worn if it determines that the intensity of the determination signal in all frequency bands is equal to or greater than the wear threshold. However, the determination unit 33 may also determine that the drill 140 is worn if, for example, it determines that the intensity of the determination signal in a predetermined proportion of the frequency band to be determined is equal to or greater than the wear threshold. In this way, by comparing the determination signal with the wear threshold in each of the multiple frequency bands, it is possible to suppress misjudgments of drill 140 wear.

[0022] Furthermore, if the determination unit 33 of this embodiment determines that the drill 140 is worn in the wear determination, it performs a damage determination after the wear determination to determine whether the intensity of the determination signal has fallen below a damage threshold, which is smaller than the wear threshold. Specifically, when the drill 140 is damaged, the friction between the drill 140 and the workpiece 200 decreases, the wear sound waves decrease, and the intensity of the determination signal decreases. For this reason, the determination unit 33 determines that the drill 140 is damaged when the intensity of the determination signal falls below the damage threshold. However, the damage threshold, like the wear threshold, is set for each frequency band, and becomes smaller as the frequency increases. The determination unit 33 then compares the intensity of the determination signal in each frequency band with the respective damage threshold to determine whether the drill 140 is damaged. In this embodiment, in the damage determination, as with the wear determination, the drill 140 is determined to be damaged when it is determined that the intensity of the determination signal in all frequency bands is below the damage threshold. However, the determination unit 33 may, for example, determine that the drill 140 is damaged when it is determined that the intensity of the determination signal in a predetermined proportion of the frequency bands to be determined is below the damage threshold.

[0023] In this embodiment, the determination unit 33 is connected to a notification unit 40 which has a display unit, an audio unit, etc. If it determines that an abnormality such as wear or damage has occurred in the drill 140, it transmits an abnormality signal to the notification unit 40 indicating that an abnormality has occurred in the drill 140. In this embodiment, if the determination unit 33 determines that the drill 140 is worn, it transmits a wear abnormality signal to the notification unit 40. If the determination unit 33 determines that the drill 140 is damaged, it transmits a damage abnormality signal to the notification unit 40. The notification unit 40 then transmits a notification according to the content of the abnormality signal to the worker to inform them that an abnormality has occurred in the drill 140.

[0024] The determination unit 33 may operate independently of the numerical control device (i.e., the NC (Numerical Control) device) that is generally provided in the machine tool 100, but it may also send an abnormality signal to the numerical control device. If an abnormality signal is sent to the numerical control device, the numerical control device may, for example, readjust the set movement direction and amount of movement based on the wear of the drill 140.

[0025] The above describes the configuration of the tool abnormality detection system 1 in this embodiment. Next, a method for detecting abnormalities using the tool abnormality detection system 1 will be described. In this embodiment, as described above, a method for detecting abnormalities when drilling holes in a workpiece 200 using a drill 140 will be described.

[0026] First, when drilling a hole in the workpiece 200, the workpiece 200 is fixed to the vise 120, and the spindle holder 130 is rotated and displaced downward to press the drill 140 against the workpiece 200, thereby forming a hole 201 in the workpiece 200. As the drilling process continues and the cutting edge of the drill 140 wears down, the contact friction between the cutting edge of the drill 140 and the workpiece 200 at the bottom of the hole 201 increases, increasing the cutting resistance.

[0027] As contact friction increases, the spindle holder 130, which displaces the drill 140 downward while rotating, becomes less able to operate properly, causing the spindle holder 130 to vibrate slightly and the drill 140 to rub against the side wall of the hole 201. When the drill 140 rubs against the side wall of the hole 201, the workpiece 200 begins to vibrate freely due to its natural vibration. In other words, as the drill 140 wears down, wear sound waves are generated from the workpiece 200 due to the wear of the drill 140.

[0028] The detection unit 10 detects sound waves and outputs a detection signal. Specifically, if the drill 140 is not worn, it detects direct sound waves and noise sound waves, and if the drill 140 is worn, it detects wear sound waves in addition to direct sound waves and noise sound waves and transmits a detection signal.

[0029] The determination unit 33 then compares the intensity of the determination signal based on the detection signal with the wear threshold to determine wear of the drill 140. In this embodiment, wear determination is performed by comparing the intensity of the determination signal with the wear threshold in each frequency band. This suppresses the misjudgment of drill 140 wear due to noise sound waves. Furthermore, if the determination unit 33 determines that the drill 140 is worn, it performs damage determination by comparing the intensity of the determination signal with the damage threshold.

[0030] The above describes the abnormality detection method in this embodiment. Next, the operations performed by the control unit 30 will be explained with reference to Figure 4. The control unit 30 performs the following operations, for example, when machining of the workpiece 200 is started.

[0031] First, in step S101, the control unit 30 receives a detection signal from the detection unit 10 and generates a determination signal. In this embodiment, the determination signal is generated for the frequency band of harmonics that are integer multiples of 2 or more, in addition to the fundamental frequency band of the abrasion sound wave.

[0032] Next, in step S102, the control unit 30 determines whether or not it has transmitted a wear abnormality signal. In other words, the control unit 30 determines whether or not it has already determined that the drill 140 is in a worn state. If the control unit 30 determines that it has transmitted a wear abnormality signal (i.e., step S102: YES), it proceeds with the processing described later in step S105 and beyond.

[0033] If the control unit 30 determines in step S102 that it has not transmitted a wear abnormality signal (i.e., step S102: NO), it determines in step S103 whether the intensity of the determination signal is equal to or greater than the wear threshold. Specifically, it performs a wear determination by comparing the intensity of the determination signal in each frequency band with the wear threshold.

[0034] If the control unit 30 determines that the intensity of the judgment signal is less than the wear threshold (i.e., step S103: NO), it terminates the process because the drill 140 is not worn. Conversely, if the control unit 30 determines that the intensity of the judgment signal is equal to or greater than the wear threshold (i.e., step S103: YES), it transmits a wear abnormality signal to the notification unit 40 in step S104. As a result, the notification unit 40 performs a process to notify the operator that wear has occurred in the drill 140. In this embodiment, the determination in step S103 is made if the intensity of the judgment signal in all frequency bands is equal to or greater than the wear threshold, in which case it is determined that the drill 140 is worn.

[0035] If the control unit 30 transmits a wear abnormality signal in step S104, or if it determines in step S102 that a wear abnormality signal has been transmitted (i.e., step S102: YES), then in step S105, it determines whether the intensity of the determination signal is below the damage threshold. Specifically, it performs a damage determination by comparing the intensity of the determination signal in each frequency band with the damage threshold.

[0036] If the control unit 30 determines that the intensity of the judgment signal is below the damage threshold (i.e., step S105: YES), it determines that the drill 140 is damaged and sends a damage abnormality signal to the notification unit 40. As a result, the notification unit 40 processes a notification to the operator that the drill 140 is damaged. Conversely, if the control unit 30 determines that the intensity of the judgment signal is greater than the damage threshold (i.e., step S105: NO), it terminates the process. In this embodiment, the determination in step S105 is that the drill 140 is damaged if the intensity of the judgment signal in all frequency bands is below the damage threshold.

[0037] According to the embodiment described above, judgment signals are generated for the fundamental frequency band of the wear sound wave and for frequency bands that are integer multiples of two or more of the fundamental frequency band. Then, wear is determined by comparing each judgment signal with a wear threshold. This makes it possible to suppress misjudgments of wear of the drill 140. Furthermore, since the tool abnormality detection system 1 of this embodiment can suppress misjudgments by performing judgments for each frequency band, it can be applied even when the diameter of the hole 201 is small or when the wear sound wave is small due to the material of the workpiece 200.

[0038] (Modification of the first embodiment) A modification of the first embodiment described above will now be explained. In the first embodiment, an example was described in which the control unit 30 sends a wear abnormality signal in step S104 and then performs the processing from step S105 onwards. However, the control unit 30 may, for example, perform the processing from step S105 onwards if it sends the wear abnormality signal multiple times.

[0039] (Second Embodiment) A second embodiment will now be described. This embodiment is modified from the first embodiment to detect abnormalities in the tool during turning, which is a cutting process. Other aspects are the same as in the first embodiment, so a detailed explanation will be omitted here.

[0040] The tool abnormality detection system 1 of this embodiment is for detecting abnormalities in tools used for turning as a cutting process. Therefore, the machine tool 100 of this embodiment, as shown in Figure 5, is equipped with a cutting tool 410 for turning the workpiece 200, and a vise 420 for holding the cutting tool 410. The cutting tool 410 is rod-shaped with a blade at one end, and the other end opposite to the one end is fixed by the vise 420.

[0041] As will be described in more detail later, in this embodiment, when the cutting edge of the tool 410 wears down, the tool 410 vibrates at its natural frequency, generating wear sound waves. For this reason, the detection unit 10 is positioned near the tool 410. Figure 5 shows the state in which the tool 410 is worn and vibrating.

[0042] The control unit 30 performs wear determination by comparing the determination signal with the wear threshold, similar to the first embodiment described above. The control unit 30 also performs damage determination by comparing the determination signal with the damage threshold. However, in this embodiment, the control unit 30 (i.e., the signal processing unit 32) has a predetermined frequency band that is fundamental to the wear sound wave and includes the natural frequency of the byte 410. The control unit 30 then performs abnormality determination by comparing the determination signal in each frequency band with the wear threshold and the damage threshold, similar to the first embodiment described above.

[0043] The above describes the configuration of the tool abnormality detection system 1 in this embodiment. Next, a method for detecting abnormalities using the tool abnormality detection system 1 described above will be explained. In this embodiment, as described above, a method for detecting abnormalities when turning a workpiece 200 using a cutting tool 410 will be explained.

[0044] First, during turning, the workpiece 200 is brought into contact with the cutting tool 410 to form the desired shape. In this case, as the cutting edge of the cutting tool 410 wears down as the turning process continues, the contact friction between the cutting edge and the workpiece 200 increases, and the turning resistance increases. When the force relationship between the cutting tool 410 and the workpiece 200 becomes unbalanced and the relative speed becomes zero, the cutting tool 410 undergoes a stick-slip phenomenon, and the spring force when the cutting tool 410 returns to its original position causes the cutting tool 410 to begin free vibration due to its natural vibration. In other words, as the cutting tool 410 wears down, it generates wear sound waves characteristic of wear.

[0045] The stick-slip phenomenon is an intermittent motion in which two contacting objects slide against each other, but instead of continuous, smooth sliding, sliding and sticking occur alternately.

[0046] The detection unit 10 detects direct sound waves and noise sound waves if the byte 410 is not worn, and if the byte 410 is worn, it detects wear sound waves in addition to direct sound waves and noise sound waves and transmits a detection signal.

[0047] Then, similar to the first embodiment described above, the control unit 30 compares the intensity of the determination signal based on the detection signal with the wear threshold and the damage threshold to determine if there is an abnormality in the byte 410.

[0048] According to the embodiment described above, since a determination signal is generated for the frequency band of harmonics that are integer multiples of 2 or more, in addition to the fundamental frequency band of the abrasion sound wave, the same effects as in the first embodiment can be obtained.

[0049] (Third embodiment) A third embodiment will now be described. This embodiment is a modification of the configuration of the control unit 30 compared to the first embodiment. Other aspects are the same as in the first embodiment, so their explanation will be omitted here. In this embodiment, the wear determination in the first embodiment will be described, but damage determination may also be performed.

[0050] The control unit 30 of this embodiment determines wear of the drill 140 using the self-learning of an anomaly detection system already filed by the applicant in Japanese Patent Application Publication No. 2020-129233.

[0051] As shown in Figure 6, the control unit 30 of this embodiment includes a signal acquisition unit 301, a learning target data storage unit 302, a monitoring target data storage unit 303, a state observer generation unit 304, a state observer information storage unit 305, a normal model generation unit 306, a normal model parameter storage unit 307, an abnormality degree calculation unit 308, a determination unit 309, a factor analysis unit 310, and the like.

[0052] The signal acquisition unit 301 acquires data to be learned and data to be monitored. In this embodiment, the detection signal based on sound waves detected by the detection unit 10 is acquired as the data to be monitored and the data to be learned.

[0053] Furthermore, the data to be learned may be obtained by downloading it from a database containing reference values, rather than from the detection signal. Also, the data to be monitored in this embodiment may be any data that is to be monitored, and it is optional whether or not it is used for other purposes; for example, it may be data that is also used for learning. Also, the data to be learned in this embodiment may be any data that is to be used for learning, and it is optional whether or not it is used for other purposes.

[0054] The learning target data storage unit 302 stores the learning target data acquired by the signal acquisition unit 301. The stored learning target data is then input to the normal model generation unit 306, which will be described later.

[0055] The monitored data storage unit 303 stores the monitored data acquired by the signal acquisition unit 301. The stored monitored data is then input to the abnormality calculation unit 308, which will be described later.

[0056] The learning target data storage unit 302 and the monitored data storage unit 303 are expected to be composed of HDDs, flash memory, etc., but they may also be RAM, etc. Furthermore, the learning target data storage unit 302 and the monitored data storage unit 303 may be either volatile memory or non-volatile memory.

[0057] The state observer generation unit 304 generates a state observer using the variables included in the input variable configuration. The state observer is a value that includes predetermined coefficients generated by linearly combining the variables included in the input variable configuration with a pre-set mathematical formula. Preferably, the state observer reflects the state of the drill 140, which is detected as abnormal by the tool abnormality detection system 1, and preferably reflects the wear of the drill 140. In this embodiment, the variables are variables corresponding to the learning target data and monitoring target data acquired by the signal acquisition unit 301, and the variable configuration is one variable or a combination of multiple variables.

[0058] The state observer information storage unit 305 stores the coefficients of the state observer obtained by the state observer generation unit 304. It also simultaneously stores the variable configuration used to generate the state observer.

[0059] The state observer generation unit 304 only needs to be provided when generating the state observer. That is, the state observer generation unit 304 may be disconnected from the control unit 30 after it has generated the state observer and stored the variable configuration and coefficients of the state observer in the state observer information storage unit 305.

[0060] The normal model generation unit 306 derives a first state observation value obtained by inputting the data to be learned into the state observer. In this embodiment, the variable configuration and coefficients of the state observer are read from the state observer information storage unit 305, and the first state observation value is obtained by applying the data to be learned read from the data to be learned storage unit 302 to these variables. Then, the normal model generation unit 306 generates a normal model by combining the derived first state observation value and the data to be learned and inputting them into a competitive neural network (i.e., NN). For example, in this embodiment, the material of the drill 140 and the diameter of the hole are input into the competitive neural network (NN) as the data to be learned.

[0061] Furthermore, when providing initial values ​​to a competitive neural network, if there are multiple combinations of attributes in the input data, such as the season and time of measurement, it is desirable to sample them evenly or randomly. This allows for faster convergence of the neuron weight vectors during training on the map in the competitive neural network.

[0062] In this context, a competitive neural network is a network consisting only of an input layer and an output layer, comprising multiple input layer neurons and multiple output layer neurons that are fully connected to the input layer neurons.

[0063] The normal model generation unit 306 then calculates the anomaly score, which is the difference between the training data and first state observation values ​​input to the competitive neural network and the neuron weight data of the winning unit, and uses the set of differences to determine the wear threshold. For example, the wear threshold is set to a constant multiple of the 99.9th percentile of the set of differences (i.e., absolute values). The wear threshold is set not only for the fundamental frequency band of the wear sound wave, but also for the harmonic frequency bands that are integer multiples of 2 or more.

[0064] The normal model parameter storage unit 307 stores the wear threshold value obtained by the normal model generation unit 306.

[0065] The anomaly score calculation unit 308 derives a second state observation value obtained by inputting the monitored data into the state observer. In this embodiment, the variable configuration and coefficients of the state observer are read from the state observer information storage unit 305, and the monitored data read from the monitored data storage unit 303 is applied to these variables to obtain the second state observation value. The anomaly score calculation unit 308 then combines the derived second state observation value and the monitored data and inputs them into a competitive neural network (i.e., NN) to calculate the anomaly score using the weight data of the output layer neurons. The anomaly score is calculated for the fundamental frequency band of the wear sound wave, as well as for the frequency bands of harmonics that are integer multiples of 2 or more. For example, in this embodiment, the material of the drill 140 and the diameter of the hole are input into the competitive neural network (NN) as monitored data.

[0066] The determination unit 309 performs wear determination by comparing the wear threshold read from the normal model parameter storage unit 307 with the abnormality level output from the abnormality level calculation unit 308. Specifically, if the abnormality level is equal to or greater than the wear threshold, it is determined that the drill 140 is worn; if the abnormality level is less than the wear threshold, it is determined that there is no wear.

[0067] If the determination result of the judgment unit 309 is that wear is present, the factor analysis unit 310 identifies the cause of the wear using the second state observation value and monitored data that caused the determination that wear was present. In this case, the factor analysis unit 310 may further identify the cause of the abnormality using the second state observation value and monitored data that preceded, preceded, or preceded and preceded in time, based on the second state observation value and monitored data that caused the determination that wear was present.

[0068] According to the embodiment described above, since a determination signal is generated for the frequency band of harmonics that are integer multiples of 2 or more, in addition to the fundamental frequency band of the abrasion sound wave, the same effects as in the first embodiment can be obtained.

[0069] (1) As in this embodiment, the control unit 30 may perform wear determination of the drill 140 while self-learning. This is expected to further suppress misjudgments because wear determination can be performed according to the situation.

[0070] (Fourth Embodiment) A fourth embodiment will now be described. This embodiment differs from the first embodiment in that the sampling frequency and Nyquist frequency are changed. Other aspects are the same as in the first embodiment, so a detailed explanation will be omitted here.

[0071] The basic configuration of this embodiment is the same as that of the first embodiment described above. In this embodiment, however, the sampling frequency and Nyquist frequency of the AD conversion unit 34 have been changed. Specifically, the sampling frequency is set to a frequency higher than the highest frequency band determined by the determination unit 33, while the Nyquist frequency is set to a frequency that falls within the range of the frequency band determined by the determination unit 33. In this embodiment, as shown in Figure 7, the sampling frequency and Nyquist frequency are set such that the sampling frequency is higher than the frequency band four times higher, and the Nyquist frequency is between the frequency band twice and the frequency band three times higher.

[0072] In this embodiment, the Nyquist frequency being included in the frequency band range determined by the determination unit 33 means that the Nyquist frequency is a frequency between the fundamental frequency band and a frequency band four times that frequency. However, as described above, the sampling frequency is set to a frequency higher than the highest frequency band determined by the determination unit 33. Therefore, the Nyquist frequency is actually set appropriately according to the highest frequency band determined by the determination unit 33.

[0073] The AD conversion unit 34 then generates a digital signal that includes frequency components higher than the Nyquist frequency as aliasing distortion when converting the detection signal from an analog signal to a digital signal. In this embodiment, since the Nyquist frequency is set to a frequency between the 2x frequency band and the 3x frequency band, the signal converted to a digital signal includes the detection signal in the 3x frequency band and the 4x frequency band as aliasing distortion. Hereinafter, the frequency band obtained by aliasing the 3x frequency band will also be called the 3x aliasing frequency band, and the frequency band obtained by aliasing the 4x frequency band will also be called the 4x aliasing frequency band.

[0074] The signal processing unit 32 derives multiple determination signals indicating the respective strengths in multiple different frequency bands. However, in this embodiment, as described above, the detection signals for the 3x frequency band and the 4x frequency band are folded back at the Nyquist frequency. Therefore, the signal processing unit 32 derives determination signals for the basic frequency band, the 2x frequency band, the 3x folded frequency band, and the 4x folded frequency band. In other words, instead of deriving determination signals for the 3x frequency band and the 4x frequency band, the signal processing unit 32 in this embodiment derives determination signals for the 3x folded frequency band and the 4x folded frequency band.

[0075] In determining wear, the determination unit 33 compares the intensity of the determination signal with the wear threshold. However, in this embodiment, as described above, the signal processing unit 32 derives determination signals for the basic frequency band, the 2x frequency band, the 3x aliased frequency band, and the 4x aliased frequency band. Therefore, the determination unit 33 compares the intensity of the determination signal with the wear threshold in the basic frequency band, the 2x frequency band, the 3x aliased frequency band, and the 4x aliased frequency band.

[0076] Furthermore, in the case of damage determination, the determination unit 33 compares the intensity of the determination signal with the wear threshold in the basic frequency band, twice the frequency band, three times the aliased frequency band, and four times the aliased frequency band, similar to the wear determination.

[0077] According to the embodiment described above, since a determination signal is generated not only for the fundamental frequency band of the abrasion sound wave but also for the harmonic frequency band, the same effects as in the first embodiment can be obtained.

[0078] (1) In this embodiment, the sampling frequency is set to a frequency higher than the highest frequency band determined by the determination unit 33, and the Nyquist frequency is set to a frequency that falls within the range of the frequency band determined by the determination unit 33. Signal components with frequencies higher than the Nyquist frequency become aliasing distortion, which is folded symmetrically with respect to the Nyquist frequency. Therefore, the signal processing unit 32 can derive a determination signal similar to the case where the determination signal is derived based on signal components with frequencies higher than the Nyquist frequency by deriving the determination signal based on the aliasing distortion. Thus, the range of the frequency band for generating the determination signal and the frequency band for determining the determination signal can be narrowed. This reduces the computational processing. Furthermore, although the communication of the determination signal, the wear determination result, and the damage determination result is performed using digital signals, the number of bits representing the digital signal can be reduced because the range of the frequency band for deriving the determination signal and the frequency band for determining the determination signal can be narrowed. This reduces the amount of data required for processing and improves responsiveness.

[0079] (Other embodiments) This disclosure is described in accordance with embodiments, but it is understood that this disclosure is not limited to such embodiments or structures. This disclosure also includes various modifications and variations within the scope of equivalents. In addition, various combinations and forms, as well as other combinations and forms that include only one, more, or fewer of those elements, fall within the scope and concept of this disclosure.

[0080] For example, in the first and second embodiments described above, wear detection and damage detection are performed, but it is also possible to perform only wear detection.

[0081] Furthermore, in the first and third embodiments described above, a drill 140 was used as an example of a tool, and in the second embodiment described above, a cutting tool 410 was used as an example. However, the tool can be changed as appropriate and may be an end mill, grinding wheel, or the like.

[0082] Furthermore, in each of the above embodiments, for example, the time evolution of the sound wave in the fundamental frequency band of the wear sound wave and in twice the frequency band is shown as in Figure 8. That is, as shown in Figure 8, when the drill 140 or the cutting tool 410 wears down, the wear sound wave increases compared to the normal case. Therefore, for example, if it is determined in step S103 that the determination signal is above the wear threshold, it may be determined whether the determination signal from a predetermined period prior was below the wear threshold. This further enables highly accurate determination of wear on the drill 140 or the cutting tool 410.

[0083] Furthermore, the above embodiments can be combined. For example, the fourth embodiment may be combined with the second and third embodiments so that the Nyquist frequency is a frequency that falls within the range of the frequency band determined by the determination unit 33.

[0084] The control unit and its method described herein may be implemented by a dedicated computer provided by configuring a processor and memory programmed to perform one or more functions embodied by a computer program. Alternatively, the control unit and its method described herein may be implemented by a dedicated computer provided by configuring a processor by one or more dedicated hardware logic circuits. Alternatively, the control unit and its method described herein may be implemented by one or more dedicated computers configured by a combination of a processor and memory programmed to perform one or more functions and a processor configured by one or more hardware logic circuits. Furthermore, the computer program may be stored as instructions executed by the computer on a computer-readable non-transitional tangible recording medium. [Explanation of symbols]

[0085] 10 Detection unit 30 Control Unit 140 Drill (Tool) 200 Work material

Claims

1. A tool anomaly detection system, A detection unit (10) detects sound waves generated when processing a workpiece (200) using tools (140, 410) and outputs a detection signal based on the sound waves, The system includes a control unit (30) that determines the wear of the tool based on the detection signal, The control unit generates, based on the detection signal, a determination signal indicating the intensity of the fundamental frequency band of wear sound waves generated when the tool wears down, and a determination signal indicating the intensity of the frequency band higher than the fundamental frequency band, and determines the wear of the tool by comparing the intensity of each determination signal with the wear threshold set for each frequency band. The control unit further determines, if wear occurs in the tool, that the tool is damaged by comparing the intensity of each determination signal with a damage threshold smaller than the wear threshold, thereby determining the tool's damage.

2. The tool abnormality detection system according to claim 1, wherein the control unit generates the determination signals for the basic frequency band and for frequency bands that are two or more integer multiples of the basic frequency band.

3. The detection unit outputs an analog signal as the detection signal. The control unit has an AD conversion unit (34) that converts the analog signal into a digital signal, and the sampling frequency is set to a frequency higher than the highest frequency band of the frequency band in which the determination signal is generated, and the Nyquist frequency, which is half the sampling frequency, is set to a frequency that falls within the range of the frequency band in which the determination signal is generated, and the control unit generates the determination signal in a frequency band of the same or less than the Nyquist frequency based on the digital signal converted by the AD conversion unit, as described in claim 1 or 2.

Images