Abnormality detection device and machining system

Abstract

To provide a technology that can suppress a decrease in abnormality detection accuracy.SOLUTION: An abnormality detection device 13 continuously performs grinding processing on a plurality of workpieces W. The abnormality detection device 13 comprises: a sound collecting microphone 10 for collecting a sound of a grinding device 2; and a processing unit 4 to which an output of the sound collecting microphone 10 is given. The processing unit 4 comprises a processing unit 14 that executes: a period specifying process 14a for specifying a grinding processing period for one of the plurality of workpieces W based on the sound data obtained from the output; and an abnormality determination process 14b that determines a presence or absence of an abnormality based on partial sound data in the grinding processing period among the sound data.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an abnormality detection device and a machining system. [Background technology]

[0002] Among abnormality detection devices for machining equipment, there are those that detect the presence or absence of abnormalities based on status data such as the motor status and machining sounds during the period during which the tool comes into contact with the workpiece and actual machining is performed (actual machining period) during the machining work period (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2019-169003 Summary of the Invention [Problem to be solved by the invention]

[0004] In the above-described conventional anomaly detection device, in order to acquire status data for the actual machining period, it is necessary to perform processing for identifying the actual machining period included in the work period. In order to identify the actual machining period included in the working time, it is conceivable to record sounds made during work and identify the actual machining period based on the recorded sound data.

[0005] However, when specifying the actual machining period using sound data, it may be difficult to obtain the necessary sound due to the influence of the operating sound of the motor that rotates the workpiece or tool, the sound of the coolant being supplied, and the like. For this reason, the accuracy in identifying the actual processing period included in the working time using sound data is not very high, and as a result, there is a risk that the accuracy in detecting abnormalities will decrease. [Means for solving the problem]

[0006] (1) An embodiment of an anomaly detection device is an anomaly detection device for a machining device that continuously performs machining processing on multiple workpieces. The anomaly detection device includes a sound collection unit that collects sounds from the machining device and a processing device to which the output of the sound collection unit is applied. The processing device includes a processing unit that executes a period identification process that identifies a machining processing period for one of the multiple workpieces based on sound data obtained from the output, and an anomaly determination process that determines whether or not an anomaly exists based on partial sound data from the sound data for the machining processing period.

[0007] (6) From another perspective, the present embodiment is a machining system including a machining device that continuously performs machining processes on a plurality of workpieces, and the abnormality detection device described in (1) above. [Effects of the Invention]

[0008] According to the present disclosure, it is possible to suppress a decrease in anomaly detection accuracy. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is an external view of a machining system according to an embodiment. [Figure 2] FIG. 2 is a diagram showing devices placed on a bed. [Figure 3] FIG. 3 is a diagram showing an example of a process for grinding one workpiece. [Figure 4] FIG. 4 is a block diagram illustrating an example of the configuration of the processing device. [Figure 5] FIG. 5 is a diagram illustrating an example of a period specifying process performed by the processing unit. [Figure 6] FIG. 6 is a flowchart showing an example of the determination process. [Figure 7] FIG. 7 is a flowchart showing an example of the filtering process. [Figure 8] FIG. 8 is a diagram illustrating an example of an abnormality determination process performed by the processing unit. [Figure 9]Figure 9(a) is a diagram showing an example of sound data acquired when grinding processing was performed on multiple workpieces, Figure 9(b) is a diagram showing an enlarged view of one of the partial sound data (grinding processing periods) in Figure 9(a), and Figure 9(c) is a diagram showing an enlarged view of another grinding processing period. [Figure 10] FIG. 10 is a diagram showing an example of sound data during a carry-in / out period before and after a grinding processing period. DETAILED DESCRIPTION OF THE INVENTION

[0010] First, the contents of the embodiment will be listed and explained. [Outline of the embodiment] (1) An embodiment of an anomaly detection device is an anomaly detection device for a machining device that continuously performs machining processing on multiple workpieces. The anomaly detection device includes a sound collection unit that collects sounds from the machining device and a processing device to which the output of the sound collection unit is applied. The processing device includes a processing unit that executes a period identification process that identifies a machining processing period for one of the multiple workpieces based on sound data obtained from the output, and an anomaly determination process that determines whether or not an anomaly exists based on partial sound data from the sound data for the machining processing period.

[0011] (2) In the above-mentioned abnormality detection device, when the machining processing period of the one workpiece includes an actual machining period in which the tool of the machining device is in contact with the one workpiece and a non-machining period in which the tool is not in contact with the one workpiece, it is possible to determine whether or not there is an abnormality in the one workpiece without distinguishing between the actual machining period and the non-machining period. In other words, since the presence or absence of an abnormality in one workpiece is determined without specifying the actual machining period, which is difficult to specify with high accuracy, a decrease in accuracy when detecting an abnormality can be suppressed.

[0012] (3) In the above-described anomaly detection device, when the sound data is sound pressure, the period identification process preferably includes a sampling process for dividing the sound data into a plurality of data blocks at a predetermined sampling interval, a determination process for calculating an evaluation value indicating the sound pressure of each of the plurality of data blocks and outputting a determination result as a binary value by comparing the evaluation value with a predetermined threshold value, and a processing period identification process for identifying the machining processing period based on the determination result. During the machining process period of a workpiece, sound pressure above a certain level continues. Therefore, by outputting the evaluation value indicating the sound pressure of the data block and the judgment result relative to the threshold value as a binary value, it becomes easy to determine whether sound pressure above a certain level continues. As a result, the accuracy in identifying the machining process period is improved.

[0013] (4) In the anomaly detection device, the period identification process may further include a filtering process that performs a median filter process on the determination results corresponding to the data blocks. In this case, momentary work sounds other than the machining process, such as the opening and closing sound of the cover of the machining device and positioning work, can be eliminated as noise.

[0014] (5) In grinding, the sound generated when the tool contacts the workpiece is smaller than in cutting, etc., making it more difficult to distinguish between an actual machining period and a non-machining period. For this reason, it is preferable that the machining device is a grinding device. In this case, the deterioration of the accuracy of anomaly detection can be more suitably suppressed.

[0015] (6) From another perspective, the present embodiment is a machining system including a machining device that continuously performs machining processes on a plurality of workpieces, and the abnormality detection device described in (1) above.

[0016] [Details of the embodiment] Preferred embodiments will now be described with reference to the drawings. [Overall structure] FIG. 1 is an external view of a machining system according to an embodiment. The machining system 1 according to the embodiment includes a grinding device 2, a workpiece transport device 3, and a processing device 4. The grinding device 2, which is a machining device, includes a bed 2a and a cover 2b. A headstock, grinding stone, etc., which will be described later, are placed on the upper surface of the bed 2a. The cover 2b houses the headstock, grinding stone, etc., which are placed on the upper surface of the bed 2a. The cover 2b prevents the cooling liquid (coolant) from splashing outside during machining. The workpiece transport device 3 is disposed above the cover 2b. The workpiece transport device 3 has an arm that grips and transports the workpiece W inside the cover 2b of the grinding device 2. The workpiece transport device 3 has the function of removing the processed workpiece W from the grinding device 2 using the arm and supplying a new workpiece W to the grinding device 2 before processing. The cover 2b has an openable / closable shutter on its upper surface. By opening the shutter of the cover 2b, the workpiece transport device 3 can access the workpiece W on the bed 2a. The processing device 4 is, for example, a computer. The processing device 4 has a function of executing a process for determining whether or not an abnormality occurs during grinding.

[0017] FIG. 2 is a diagram showing the devices arranged on the bed 2a. The grinding device 2 further includes a table 5 , a headstock 6 , a grindstone 7 , and a tailstock 8 . The headstock 6 rotatably supports the spindle 6a and the chuck 6b. The chuck 6b grips one end of the workpiece W. The chuck 6b can be controlled to open and close by an actuator or the like. The tailstock 8 holds the other end of the workpiece W. The spindle 6a, the chuck 6b, and the workpiece W gripped by the chuck 6b are rotated by a motor possessed by the headstock 6. The grinding wheel 7 is supported by a grinding wheel head (not shown). The grinding wheel 7 can be rotated by a motor possessed by the grinding wheel head. The grinding wheel 7 can also be moved by an actuator or the like. The headstock 6 and tailstock 8 are provided on a table 5. The table 5 can be moved longitudinally (left and right on the page) by an actuator or the like. The table 5 moves the headstock 6, tailstock 8, and the workpiece W held thereon relative to the grinding wheel 7. This allows the grinding wheel 7 to grind the required portion of the workpiece W in the axial direction. The grinding device 2 also has a nozzle 9 for supplying a coolant to the location where the workpiece W and the grindstone 7 come into contact.

[0018] A sound collecting microphone 10 (sound collecting unit) is fixed to the tailstock 8. The sound collecting microphone 10 is an ultrasonic microphone. The sound collecting microphone 10 is fixed to the tailstock 8 by a fixing base 11. The sound collecting microphone 10 is placed inside the cover 2b. Therefore, a waterproof cover 12 is attached to the sound collecting microphone 10 to prevent the coolant from splashing on the sound collecting microphone 10. The sound collecting microphone 10 is disposed inside the cover 2b and collects sounds inside the cover 2b. The sound collecting microphone 10 is connected to the processing device 4. The output of the sound collecting microphone 10 is provided to the processing device 4.

[0019] The grinding device 2 further has a control unit (not shown) that controls each unit. The control unit controls the motor that rotates the spindle 6a and the motor that rotates the grinding wheel 7. The control unit also controls actuators for opening and closing the chuck 6b, moving the table 5 and grinding wheel 7, and opening and closing the shutter of the cover 2b. The control unit also controls the supply of coolant according to a preset procedure. This control unit has the function of operating each unit according to a preset procedure and grinding the workpiece W supplied from the workpiece transport device 3.

[0020] The machining system 1 has a function of grinding a plurality of workpieces W continuously. FIG. 3 is a diagram showing an example of a process for grinding one workpiece W. First, the workpiece conveying device 3 carries the unprocessed workpiece W into the cover 2b and supplies it to the grinding device 2. The grinding device 2 then holds the supplied workpiece W with the chuck 6b and tailstock 8 (step S100 in Figure 3) and closes the shutter of the cover 2b. Thereafter, the grinding device 2 starts rotating the workpiece W (spindle 6a) and the grinding wheel 7, and also starts supplying the coolant (step S101 in FIG. 3).

[0021] Next, the grinding device 2 moves the workpiece W and the grinding wheel 7, and performs grinding on the workpiece W (step S102 in FIG. 3). After completing the processing operation according to the preset procedure, the grinding device 2 stops the supply of the coolant and the rotation of the workpiece W and grindstone 7 (step S103 in FIG. 3). Thereafter, the grinding device 2 opens the shutter of the cover 2b and releases the workpiece W. The released workpiece W is gripped by the workpiece transport device 3 and carried out to the outside of the cover 2b (step S104 in FIG. 3).

[0022] After carrying out the processed workpiece W, the workpiece transport device 3 carries the unprocessed workpiece W back into the cover 2b and supplies it to the grinding device 2 (step S100 in FIG. 3). The machining system 1 repeats these operations to continuously grind a plurality of workpieces W. In addition, in this embodiment, the process from the start of rotation of the workpiece W and the supply of cooling liquid (step S101 in Figure 3) to the stop of rotation of the workpiece W and the supply of cooling liquid (step S103 in Figure 3) is called the grinding process (machining process) for one workpiece W.

[0023] FIG. 4 is a block diagram showing an example of the configuration of the processing device 4. In this embodiment, the processing device 4 and the sound collecting microphone 10 constitute an abnormality detection device 13 for the grinding device 2. That is, the abnormality detection device 13 includes the processing device 4 and the sound collecting microphone 10. As shown in FIG. 4, the processing device 4 includes a processing unit 14 including a processor or the like, and a storage unit 16 including a memory and a hard disk.

[0024] The storage unit 16 stores computer programs to be executed by the processing unit 14, necessary information, and the like. The processing unit 14 executes a computer program stored in a computer-readable non-transitory recording medium such as the storage unit 16 to realize various processing functions of the processing unit 14 . The storage unit 16 also stores a trained model 16a and sound data 16b, which will be described later.

[0025] By executing the above-mentioned computer program, the processing unit 14 can execute a period specification process 14a and an abnormality determination process 14b, which will be described later.

[0026] [Regarding period-specific processing] FIG. 5 is a diagram showing an example of the period specification process 14a performed by the processing unit 14. As shown in FIG. The period specification process 14a is a process for specifying the grinding processing period of one of the plurality of workpieces W based on the sound data obtained from the output of the sound collecting microphone 10. 5, the period specification process includes a sampling process S1, a determination process S2, a filtering process S3, and a processing period specification process S4. The processing unit 14 executes these processes in parallel.

[0027] The sampling process S1 is a process of acquiring sound data based on the output from the sound collection microphone 10 and dividing the sound data into a plurality of data blocks Bn. The determination process S2 is a process for outputting a determination result value Rn based on a plurality of data blocks Bn. The filtering process S3 is a process of performing median filtering on the determination result value Rn and outputting a filtering result value FRn. The machining period specification process S4 is a process for specifying a grinding processing period (machining processing period) based on the filter result value FRn. The grinding processing period refers to the period from the start to the end of grinding processing for one workpiece W. In other words, it refers to the period from the start of rotation of the workpiece W and the supply of coolant (step S101 in FIG. 3) to the stop of rotation of the workpiece W and the stop of supply of coolant (step S103 in FIG. 3).

[0028] In the sampling process S1, the processing unit 14 obtains sound pressure as sound data based on the output from the sound collection microphone 10. The processing unit 14 acquires the output of the sound collection microphone 10 over time at a predetermined sampling rate (e.g., 192 kHz). The sound data acquired by the processing unit 14 is a group of discrete values ​​of sound pressure arranged in time series. The acquired sound data is stored in the memory unit 16. The processing unit 14 divides the acquired sound data into a plurality of data blocks Bn at a predetermined sampling interval, where n is the block number. The block number n is an integer equal to or greater than 0 and indicates the order in which the data blocks Bn are divided. The predetermined sampling interval is, for example, 0.1 seconds. Each data block Bn contains discrete values ​​of sound pressure within the sampling interval. When the processing unit 14 receives an output from the sound collecting microphone 10, it performs the sampling process S1 continuously without waiting for the supply of the output to be stopped, and therefore the processing unit 14 generates data blocks Bn continuously. The data block Bn generated by the sampling process S1 is given to the determination process S2.

[0029] FIG. 6 is a flowchart showing an example of the determination process S2. In the determination process S2, the processing unit 14 first determines whether or not a data block Bn has been provided from the sampling process S1 to the determination process S2 (step S21 in FIG. 6). The processing unit 14 repeats step S21 until it determines that the data block Bn has been given.

[0030] When it is determined that a data block Bn has been given, the processing unit 14 calculates the RMS (Root Mean Square) of the sound pressure as an evaluation value indicating the sound pressure of the data block Bn. The processing unit 14 calculates the RMS of the discrete sound pressure values ​​included in the data block Bn. The processing unit 14 determines whether the RMS of the obtained data block Bn is greater than the threshold L (step S22 in FIG. 6). When the processing unit 14 determines that the RMS of the data block Bn is greater than the threshold L, it sets the determination result value Rn to 1 (step S23 in FIG. 6). On the other hand, when the processing unit 14 determines that the RMS of the data block Bn is not greater than the threshold L (equal to or less than the threshold L), it sets the determination result value Rn to 0 (step S24 in FIG. 6). The judgment result value Rn is a value indicating the judgment result obtained by comparing the RMS of the data block Bn with the threshold L, and is output as a binary value of 1 or 0. The n in the judgment result value Rn is the block number of the data block Bn being judged. The threshold value L is set appropriately depending on the magnitude of the sound pressure so that it is possible to determine whether the sound pressure is during the grinding process.

[0031] In FIG. 6, after step S23 or step S24, the processing unit 14 proceeds to step S25, and provides the determination result value Rn to the filter processing S3 (step S25 in FIG. 6). The processing unit 14 sequentially outputs a determination result value Rn each time a data block Bn is given, gives the determination result value Rn to the filtering process S3, and returns to step S21.

[0032] FIG. 7 is a flowchart showing an example of the filtering process S3. In the filtering process S3, the processing unit 14 first determines whether or not the determination result value Rn has been provided from the determining process S2 to the filtering process S3 (step S31 in FIG. 7). The processing unit 14 repeats step S31 until it determines that the determination result value Rn has been given. If it is determined that the determination result value Rn has been given, the processing unit 14 performs median filtering on the determination result value Rn (step S32 in FIG. 7). When a judgment result value Rn for block number n=N is given, the processing unit 14 performs median filtering on the judgment result value Rn for block number n=N within the range of 2t+1 judgment result values ​​Rn where block number n is from Nt to N+t, and outputs the filter result value FRn.

[0033] Here, t is a region setting value for setting a neighborhood region of the judgment result value Rn to be processed in the median filter process. The region setting value t is appropriately set to a value that can remove noise based on the approximate time of the grinding processing period. In this embodiment, the region setting value t is set to, for example, about 7. Furthermore, n of the filter result value FRn is n (block number) of the judgment result value Rn to be filtered.

[0034] Since the judgment result value Rn is a binary value of 0 or 1, if there are t+1 or more judgment result values ​​Rn that have a value of 0 out of the 2t+1 judgment result values ​​Rn, the median will be 0. In this case, the filter result value FRn for n=N will be 0. Also, if there are t+1 or more judgment result values ​​Rn that have a value of 1 out of the 15 judgment result values ​​Rn, the median will be 1. In this case, the filter result value FRn for n=N will be 1. Therefore, even if the determination result value Rn for n=N is set to 1 due to an instantaneous increase in sound pressure, if the majority of the determination result values ​​Rn before and after that are 0, the filter result value FRn for n=N will be set to 0. This makes it possible to remove instantaneous noise that appears in the determination result value Rn.

[0035] When the median filter process for the determination result value Rn is completed in step S32, the processing unit 14 proceeds to step S33 and provides the filter result value FRn to the processing period specification process S4 (step S33 in FIG. 7). The processing unit 14 sequentially outputs a filter result value FRn each time a judgment result value Rn is given, gives the filter result value FRn to the filtering process S3, and returns to step S31.

[0036] In FIG. 2, in the processing period specification process S4, the processing unit 14 specifies the grinding processing period based on the filter result values ​​FRn that are sequentially given. The processing unit 14 refers to a plurality of filter result values ​​FRn arranged over time, and identifies a period during which the filter result value FRn of 1 continues. When the period during which the filter result value FRn is 1 continues is equal to or longer than a preset threshold, the processing unit 14 identifies this period as a grinding processing period. The processing unit 14 outputs the grinding processing period by the block number n. That is, the grinding processing period is represented by a plurality of consecutive block numbers n.

[0037] When grinding processing is performed continuously on multiple workpieces W, the sound data at that time includes data acquired during the grinding processing periods of the multiple workpieces W. By executing the above-mentioned period identification process 14a, the processing unit 14 can identify the grinding processing period of one workpiece W based on the sound data including data on the grinding processing periods of the multiple workpieces W.

[0038] [Regarding abnormality determination processing] FIG. 8 is a diagram showing an example of the abnormality determination process 14b performed by the processing unit 14. The abnormality determination process 14b is a process for determining whether or not an abnormality occurs during grinding, based on the partial sound data during the grinding process. The processing unit 14 determines whether or not the grinding processing period has been specified by the period specifying process 14a (step S51 in FIG. 8). The processing unit 14 repeats step S51 until it determines that the grinding processing period has been specified.

[0039] When it is determined that the grinding processing period has been identified, the processing unit 14 acquires partial sound data (step S52 in FIG. 8). The partial sound data is sound data included in the grinding processing period among the sound data acquired by the sampling processing S1. The processing unit 14 acquires sound data (partial sound data) corresponding to multiple block numbers n indicating the grinding processing period identified by the period identification process 14a from the sound data 16b (Figure 4) stored in the memory unit 16.

[0040] Next, the processing unit 14 performs an abnormality determination based on the partial sound data (step S53 in FIG. 8). The processing unit 14 uses the trained model 16a (FIG. 4) stored in the storage unit 16 to determine whether or not there is an abnormality in the grinding process.

[0041] The trained model 16a is a model obtained by prior machine learning using partial sound data acquired during normal grinding processing. In this embodiment, an autoencoder is used as a machine learning algorithm, and the trained model 16a is obtained by unsupervised learning using only partial sound data acquired during normal grinding. However, this is not limited to this, and a model may be obtained by supervised learning, or other algorithms may be adopted.

[0042] The processing unit 14 provides the partial sound data acquired in step S52 to the trained model 16a and determines whether or not there is an abnormality (step S53 in FIG. 8). When determining whether or not there is an abnormality, the processing unit 14 outputs the determination result to the outside (step S54), and returns to step S51 again.

[0043] [About the sound data] Fig. 9(a) is a diagram showing an example of sound data acquired when grinding was performed on multiple workpieces W. In Fig. 9(a), the horizontal axis represents time and the vertical axis represents sound pressure. The dark colored parts in Fig. 9(a) are graphs showing sound pressure. The light colored parts surrounded by dark colored parts above and below are graphs showing RMS (volume). As shown in Fig. 9(a), the sound pressure appears higher during the grinding processing period than during other periods. The period between adjacent grinding processing periods is a loading / unloading period during which the workpiece W is loaded and unloaded. The sound data during the grinding process is acquired by the processing unit 14 as partial sound data.

[0044] FIG. 9(b) is an enlarged view of one of the partial sound data (grinding processing period) in FIG. 9(a). During the grinding process, the coolant is supplied and the workpiece W and grindstone 7 are rotating. Therefore, the sound of the coolant supply and the rotation sounds of the workpiece W, grindstone 7, motor, etc. are included throughout the entire range of the partial sound data. In addition, in FIG. 9(b), the grinding processing period includes a period in which dressing of the grinding wheel 7 is performed ("Dress" in FIG. 9(b)). The grinding processing period also includes a period ("air purge" in FIG. 9(b)) during which air purging (the workpiece from the previous processing is carried out by air) is performed. Furthermore, the grinding processing period includes an actual processing period AP during which the grinding wheel 7 (tool) is in contact with the workpiece W.

[0045] 9(b), six actual machining periods AP1 to AP6 are included. In this embodiment, grinding is performed on three machining locations on one workpiece W. The actual machining periods AP1 to AP3 are periods during which the grinding wheel 7 is brought into contact with the three machining locations to perform rough machining. The actual machining periods AP4 to AP6 are periods during which the grinding wheel 7 is brought into contact with the three machining locations to perform finish machining. In the grinding processing period, periods other than the actual processing periods AP1 to AP6 are non-processing periods in which the grindstone 7 is not in contact with the workpiece W. The non-machining period includes a period during which other work is being performed, a period during which the workpiece W is moved from one of the three machining locations to another machining location, and the like. 9(b), in the non-machining periods located between the actual machining periods AP2 to AP6, a slight drop in sound pressure can be seen compared to the actual machining periods AP2 to AP6. Therefore, the non-machining periods located between the actual machining periods AP2 to AP6 can be identified from the change in sound pressure. In this way, the grinding processing period includes the actual processing periods AP1 to AP6 and the non-processing period.

[0046] FIG. 9(c) is an enlarged view of another grinding processing period. The partial sound data included in the grinding processing period in FIG. 9(c) allows the period during which air purging is performed to be distinguished from other periods by the difference in sound pressure. However, in the partial sound data of FIG. 9(c), it is not possible to recognize the actual machining period AP or the non-machining period other than the period during which air purging is performed. In this way, even if partial sound data is acquired in the same way, it may differ greatly if the processing conditions are different. For this reason, it is difficult to accurately determine the actual processing period from the sound pressure or the like.

[0047] In this regard, in this embodiment, the grinding processing period of one workpiece W is identified without distinguishing between the actual processing period and the non-processing period, and abnormality determination is performed using partial sound data during this period, so that it is possible to determine whether or not there is an abnormality in one workpiece W. In other words, since the presence or absence of an abnormality in one workpiece is determined without specifying the actual machining period, which is difficult to specify with high accuracy, a decrease in accuracy when detecting an abnormality can be suppressed.

[0048] Furthermore, in this embodiment, abnormality determination is performed using partial sound data during the grinding processing period, which includes the actual processing period and the non-processing period, so even if there is some fluctuation in the partial sound data, the effect of that fluctuation on abnormality determination is suppressed. Furthermore, when abnormality determination is performed in units of actual machining periods, it is not possible to detect abnormalities that occur when the grinding process for one workpiece W is viewed as a whole. However, in this embodiment, abnormality determination is performed for each workpiece W, so that abnormalities occurring when the grinding process for one workpiece W is viewed as a whole can be detected.

[0049] Furthermore, as shown in FIGS. 9(a) to 9(c), during the grinding processing period (partial sound data) of one workpiece W, sound pressure continues at a certain level or above. In contrast, in this embodiment, the judgment result of the comparison between the RMS of the data block Bn and the threshold value L is output as a binary value by the judgment process S2 (FIGS. 5 and 6), making it easy to judge whether or not a certain level of sound pressure is continuing, thereby improving the accuracy in identifying the grinding processing period.

[0050] FIG. 10 is a diagram showing an example of sound data during a carry-in / out period before and after a grinding processing period. In Figure 10, the part P1 where the sound pressure appears to be greater than the partial sound data during the loading / unloading period at the front of the grinding processing period is the closing sound when the work transport device 3 loads the work W into the cover 2b and then closes the cover 2b (shutter). Furthermore, a portion P2 where the sound pressure increases between the portion P1 and the partial sound data is work noise generated by the positioning work of the workpiece W. Furthermore, a portion P3 in which the sound pressure appears to be greater than the partial sound data during the carry-in / out period at the end of the grinding processing period is the opening sound when the cover 2b (the shutter thereof) is opened.

[0051] A sound with a relatively low sound pressure like that of the portion P2 will not be identified as a grinding processing period if the sound pressure is lower than the threshold value L in the determination process S2. Furthermore, the sound pressure of the opening and closing of the cover 2b, such as the portions P1 and S3, increases instantaneously, and can be removed as noise by the filtering process S3 (FIGS. 5 and 7). In this manner, in this embodiment, the grinding processing period can be determined with high accuracy by the determination process S2 and the filtering process S3.

[0052] 〔others〕 The embodiments disclosed herein are illustrative in all respects and are not restrictive. In the above embodiment, the machining system 1 is configured to include a grinding device 2 as the machining device. However, the machining system 1 may also be configured to include a device that performs cutting, such as a lathe or a milling machine. However, in grinding, the volume of noise generated when the tool comes into contact with the workpiece is smaller than in cutting, etc., making it more difficult to distinguish between an actual machining period and a non-machining period. For this reason, it is preferable that the machining system 1 include a grinding device as the machining device.

[0053] Furthermore, in the present embodiment, the case where sound pressure is acquired as sound data from the output of the sound collecting microphone 10 has been exemplified, but the frequency spectrum of the sound may also be acquired as sound data and used. In addition, in this embodiment, an example has been given in which the RMS of sound pressure is used as the evaluation value indicating the sound pressure of data block Bn, but the maximum value among the discrete values ​​of sound pressure contained in data block Bn may also be used as the evaluation value indicating the sound pressure of data block Bn.

[0054] The scope of the present invention is not limited to the above-described embodiments, but includes all modifications within the scope of equivalents to the configurations described in the claims. [Explanation of symbols]

[0055] 1 Machining System 2 Grinding equipment 4 Processing equipment 10 Sound collection microphone 13 Anomaly detection device 14 Processing section 14a Period-specific processing 14b Abnormality determination processing S1 Sampling process S2 Judgment process S3 Filtering S4 Processing period specific processing double work

Claims

1. An abnormality detection device for a machining device that continuously performs machining processes on a plurality of workpieces, a sound collecting unit that collects sounds from the machining device; a processing device to which the output of the sound collection unit is given, The processing device includes: a period determination process for determining a machining processing period of one of the plurality of workpieces based on sound data obtained from the output; an abnormality determination process for determining whether or not there is an abnormality based on partial sound data during the machining process period among the sound data, The machining processing period of the one workpiece includes an actual machining period in which a tool of the machining device is in contact with the one workpiece, and a non-machining period in which the tool is not in contact with the one workpiece. Anomaly detection device.

2. the sound data is sound pressure, The period specification process includes: a sampling process for dividing the sound data into a plurality of data blocks at a predetermined sampling interval; a determination process for determining an evaluation value indicating the sound pressure of each of the plurality of data blocks, comparing the evaluation value with a predetermined threshold, and outputting a determination result as a binary value; and a processing period specification process for specifying the machining processing period based on the determination result. The anomaly detection device according to claim 1 .

3. The period specification process includes: a filtering process for performing a median filter process on the plurality of determination results corresponding to the plurality of data blocks. The anomaly detection device according to claim 2 .

4. The machining device is a grinding device. The anomaly detection device according to any one of claims 1 to 3.

5. a machining device that continuously performs machining processes on a plurality of workpieces; A machining system comprising: the abnormality detection device according to claim 1 .

Images