Machining process monitoring device and machining process monitoring method

Abstract

To provide a processing process monitoring device capable of determining more accurately than the prior art whether or not a processing machinery is operating normally, even when there is fluctuation in a processing time of each single cycle in cycle processing.SOLUTION: A processing process monitoring device comprises a data cut-out unit, a data length adjustment unit, and a determination processing unit. The data cut-out unit determines when processing begins and ends for each single cycle, and cuts out first measurement data from the start of processing to the end of processing for each single cycle from a first physical quantity that changes with time during cycle processing in the processing machinery. The data length adjustment unit generates second measurement data by adjusting a data length of the first measurement data to match a data length of determination reference data indicating a change in the first physical quantity in each single cycle as a reference when the processing machinery is operating normally. The determination processing unit compares the second measurement data with the determination reference data to determine whether or not the processing machinery is operating normally.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a machining process monitoring device and a machining process monitoring method. [Background technology]

[0002] In fields such as press working, injection molding, NC processing, and industrial robots, processes that repeat a relatively short single cycle are used. A technology has been proposed that uses sensors to detect the state of such processing machines and analyzes the detection results, that is, the time-series processing sensing data, to determine whether the processing machine is operating normally or not.

[0003] For example, the method of creating judgment data for manufacturing equipment diagnosis disclosed in Patent Document 1 provides accurate judgment results by dividing a state quantity representing normal operation of the equipment and a state quantity representing abnormal operation to obtain a judgment value and using it to judge processed sensing data. The data analysis device and data analysis method disclosed in Patent Document 2 provide a method of detecting anomalies by varying the positional relationship of the waveform of the processed sensing data so as to maximize the similarity to the master waveform and to make the positional relationship with the reference waveform appropriate. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Utility Model Application Publication No. 9-120365 [Patent Document 2] JP 2020-86843 A Summary of the Invention [Problem to be solved by the invention]

[0005] In the conventional technology, there is a problem that when fluctuations (time fluctuations) occur in the processing time of each single cycle in cyclic processing, it becomes impossible to accurately detect an abnormality.

[0006] An object of the present disclosure is to provide a machining process monitoring device and a machining process monitoring method that can determine more accurately than conventional techniques whether a machining machine is operating normally, even if fluctuations occur in the machining time of each single cycle in cyclic machining. [Means for solving the problem]

[0007] One aspect of the present disclosure is A processing process monitoring device that monitors a processing machine that performs cyclic processing by repeating a single cycle and determines whether the processing machine is operating normally, comprising: a data extraction unit that determines a machining start time and a machining end time of each single cycle, and extracts first measurement data from a first physical quantity that changes over time during the cycle machining in the machining machine from the machining start time to the machining end time for each single cycle; a data length adjusting unit that adjusts a data length of the first measurement data to match a data length of judgment reference data that indicates a change in the first physical quantity in the single cycle when the processing machine is operating normally, to generate second measurement data; a determination processing unit that compares the second measurement data with the determination reference data to determine whether the processing machine is operating normally; The present invention provides a processing process monitoring device comprising:

[0008] Another aspect of the present disclosure is 1. A machining process monitoring method for monitoring a machining machine that performs cyclic machining by repeating a single cycle, and determining whether the machining machine is operating normally, comprising the steps of: a first physical quantity acquisition step of sequentially acquiring a first physical quantity that changes over time during the cycle machining in the machining machine; a data extraction step of determining a processing start time and a processing end time of each single cycle and extracting first measurement data from the first physical quantity for each single cycle from the processing start time to the processing end time; a data length adjustment step of generating second measurement data by adjusting a data length of the first measurement data so as to coincide with a data length of judgment reference data that indicates a change in the first physical quantity in the single cycle when the processing machine is operating normally, and a determination processing step of comparing the second measurement data with the determination reference data to determine whether the processing machine is operating normally; The present invention provides a method for monitoring a processing step, comprising: Effect of the Invention

[0009] According to the machining process monitoring device and machining process monitoring method disclosed herein, even if fluctuations occur in the machining time of each single cycle in cyclic machining, it is possible to determine whether or not the machining machine is operating normally more accurately than with conventional techniques. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a block diagram showing a configuration example of a machining process monitoring device 100 according to a first embodiment of the present disclosure. [Diagram 2] FIG. 2 is a schematic diagram illustrating a data extraction process executed by the data extraction unit 8 of FIG. 1; [Diagram 3] FIG. 2 is a schematic diagram for explaining the oversampling process performed by the oversampling filter 9 in FIG. 1 and the resampling process performed by the resampling unit 10 in FIG. 1; [Figure 4] 2A to 2C are schematic diagrams illustrating shot data before and after a resampling process executed by the resampling unit 10 of FIG. 1. [Diagram 5] Graph showing the results of the resampling process performed by the resampling unit 10 of FIG. 1 [Figure 6] FIG. 11 is a block diagram showing a configuration example of a machining process monitoring device 200 according to a second embodiment of the present disclosure. [Figure 7] FIG. 7 is a schematic diagram for explaining a compression process executed by the resampling unit 210 in FIG. 6; [Figure 8]FIG. 11 is a schematic diagram for explaining a modified example of the second embodiment of the present disclosure. [Figure 9] FIG. 11 is a block diagram showing a configuration example of a machining process monitoring device 300 according to a third embodiment of the present disclosure. [Figure 10] Graph showing waveforms of sensing data during punching using a conventional press machine DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] (Background to this disclosure) In the conventional technology, when fluctuations occur in the machining time of cycle machining using a machining machine, there is a risk that an operational abnormality in the machining machine cannot be accurately detected. This problem will be described below.

[0012] An example of cyclic processing in which the same operation is repeated is a processing method in which a cycle of applying a load to a workpiece such as metal is repeated, for example, press processing in which the workpiece is punched. A die incorporating tools such as a punch and a die is set in a press machine used in press processing, and the press machine repeatedly punches the workpiece into a predetermined shape while feeding it with a slider that moves up and down at an almost constant cycle.

[0013] The press machine includes, for example, a crank-type press machine having a crank controlled to rotate at a constant cycle. In the crank-type press machine, when high-speed processing is required to improve productivity, the SPM (Shots Per Minute), which indicates the number of processing times per minute, is set to a high value. The press machine rotates the crank at high speed to achieve the set SPM, converts this rotational motion into up-and-down motion of a sliding member (slider), and processes the workpiece at high speed.

[0014] However, even if an attempt is made to control the machining at a constant speed, it is difficult to keep the machining speed constant with high precision, and there is a problem that the machining speed fluctuates. As a result of the fluctuation in the machining speed, the machining time also fluctuates. As the press becomes larger, the energy required for control increases, making it difficult to maintain the speed with high precision, so the fluctuation in the machining speed becomes more noticeable the larger the press is. In particular, immediately after the start of the press, the slider and the heavy die attached to the slider need to be moved from a stopped state, and the slider speed becomes lower than the set SPM.

[0015] Figure 10 is a graph showing the waveform of sensing data during punching by a conventional press machine. Figure 10(a) is a graph showing the change over time in the height of the stripper plate (e.g., the height from the die plate) measured by a height sensor. Figure 10(b) is a graph showing the change over time in the load [N] applied to the punch measured by a load sensor. The data group in Figures 10(a) and 10(b) is a superposition of multiple data acquired by sensors attached to the die of the press machine.

[0016] The acquisition of data in Figures 10(a) and 10(b) begins at the start of processing when the height of the stripper plate in Figure 10(a) falls below the start threshold, and ends at the end of processing when the height exceeds the end threshold. The start threshold and end threshold are set to the same threshold in Figure 10(a), but they may be set to different values. Figure 10 also shows a graph in which multiple measurement data are superimposed with the start of processing aligned.

[0017] In FIG. 10, only the measurement data in a normal machining state, in which the processing machine is normally executing cycle processing, is shown. From FIG. 10, it can be seen that even in a normal machining state, the time when the processing ends is not constant. This is because the time of the measurement data, i.e., the length of the processing time, changes due to the fluctuation in the processing time described above. In addition, with regard to the punch load data shown in FIG. 10(b), the punching period during which the workpiece is punched is short, and the load data changes sharply during the punching period. Therefore, even in a normal machining state, there is a time position shift in the waveform shape during the punching period, which characterizes the punching process, between multiple load data. This position shift appears with a large variation in the physical quantity direction immediately after the start of the press machine, while the SPM is not yet stable.

[0018] Therefore, in the conventional technology proposed in Patent Document 1, which detects anomalies based on whether they fall within a reference value range, the reference value needs to be set to a wide value that takes into account the variation in waveforms in normal machining conditions, and therefore it is not possible to detect minute anomalies.

[0019] In addition, even in the conventional technology proposed in Patent Document 2 in which the measured waveform is shifted in the time axis direction so that the peak position is the same as that of the reference waveform, the reference waveform and the measured waveform can be partially aligned, but cannot be aligned as a whole. In addition, in the case of a physical quantity with a large peak, the positional relationship can be aligned even partially, but in the case of a physical quantity with a large peak, for example, in the case of various physical quantities obtained by various sensors, the positional relationship cannot be aligned.

[0020] Furthermore, in the conventional technology, in order to accurately capture the steep changes in the load data during the punching period, it is necessary to set the sampling frequency in the A / D conversion that converts analog signals into digital signals high and perform sampling more frequently. However, when the sampling frequency is increased, the load data in the time domain other than the punching period, which changes slowly, and the height data of the stripper plate, which changes slowly overall, are also sampled more frequently, and the amount of data increases. Therefore, there is a problem that the amount of data processing required to detect an abnormality increases, and the processing time increases.

[0021] The machining process monitoring device according to the embodiment described below solves the above problems.

[0022] Hereinafter, the embodiments will be described in detail with reference to the drawings as appropriate. However, more detailed explanations than necessary may be omitted. For example, detailed explanations of already well-known matters or duplicate explanations of substantially identical configurations may be omitted. This is to avoid the following explanation becoming unnecessarily redundant and to facilitate understanding by those skilled in the art. Note that the applicant provides the accompanying drawings and the following explanation so that those skilled in the art can fully understand the present disclosure, and does not intend for them to limit the subject matter described in the claims.

[0023] (First embodiment) [composition] 1 is a block diagram showing a configuration example of a machining process monitoring device 100 according to a first embodiment of the present disclosure. The machining process monitoring device 100 includes a height sensor 1, a load sensor 2, controllers 3 and 4, an A / D converter 5, and filters 6 and 7. The height sensor 1 and the load sensor 2 are attached to a processing machine that performs cycle processing. The processing machine is, for example, a crank-type press machine in which a die incorporating tools such as a punch and a die is set.

[0024] The height sensor 1 is, for example, an optical, radio wave, ultrasonic distance sensor. In this embodiment, the height sensor 1 detects the height of the stripper plate of the press, for example, the distance between the die plate and the stripper plate. The controller 3 outputs a voltage signal according to the height detected by the height sensor 1. The load sensor 2 is, for example, a piezoelectric force sensor or an electric force sensor such as a strain gauge type, and detects the load applied to the punch of the press when punching the workpiece. The controller 4 outputs a voltage signal according to the load detected by the load sensor 2. The height detected by the height sensor 1 and the load detected by the load sensor 2 are examples of the "first physical quantity" of the present disclosure. The height detected by the height sensor 1 can also be an example of the "second physical quantity" of the present disclosure. The second physical quantity is information representing the position of a member of the processing machine, for example, the stripper plate, which moves to apply a load to the workpiece in a single cycle of the processing machine. The controllers 3 and 4 are examples of the "physical quantity acquisition unit" of the present disclosure.

[0025] The A / D converter 5 converts the input voltage signal into a digital signal. The A / D converter 5 operates at a sampling frequency Fs (samples / second). That is, the A / D converter 5 converts the voltage signal received from the controllers 3 and 4 into a digital signal at a sampling period of 1 / Fs (seconds) and outputs it.

[0026] The filters 6 and 7 perform filtering on the digital signals corresponding to the height sensor 1 and the load sensor 2, respectively. The filters 6 and 7 are composed of a low-pass filter (LPF), a median filter, etc., and have the function of removing noise from the input data. The filters 6 and 7 remove noise and the like that is mixed in the height sensor 1 and the load sensor 2 themselves, and in the paths from the outputs of the height sensor 1 and the load sensor 2 to the inputs to the filters 6 and 7, respectively. If the noise is small and can be ignored, at least one of the filters 6 and 7 may be omitted. When the filters 6 and 7 are omitted, the phase between the signals is aligned, for example, by using a delay device, etc., so that only one of the signals is not delayed.

[0027] The machining process monitoring device 100 further includes a data extracting unit 8, an oversampling filter 9, a resampling unit 10, a determination processing unit 11, an output unit 12, and a storage unit 13.

[0028] The data extraction unit 8 executes a process for aligning the multiple signals s1 having the height sensor 1 as a source and the multiple signals s2 having the load sensor 2 as a source so that the processing start and processing end conditions match each other to generate first shot data. Details of the data extraction process executed by the data extraction unit 8 will be described later. The data extraction unit 8 outputs the first shot data to an oversampling filter 9. The first shot data is an example of "first measurement data" in the present disclosure.

[0029] The oversampling filter 9 oversamples each of the digital signals included in the input first shot data at a sampling frequency of 8Fs. The digital signals included in the input first shot data include a signal whose source is the height sensor 1 and a signal whose source is the load sensor 2. Details of the oversampling (sometimes called upsampling) performed by the oversampling filter 9 will be described later. The oversampling filter 9 is an example of a "data interpolation unit" in the present disclosure. The oversampling filter 9 is, for example, a low-pass filter, such as a FIR (Finite Impulse Response) filter.

[0030] The storage unit 13 is a recording medium for recording various information including programs and data necessary for implementing each function of the machining process monitoring device 100. The storage unit 13 may be, for example, any of a hard disk drive (HDD), an optical drive, and a solid state drive (SSD). The storage unit 13 may be any of an internal type, an external type, and a NAS (network-attached storage) type, and may be realized by cloud computing.

[0031] The memory unit 13 stores, for example, judgment criteria data 131. The judgment criteria data 131 is information indicating, as a criterion, a change in measurement data in a single cycle when the processing machine is operating normally. The judgment criteria data 131 is, for example, waveform data such as height data and load data in a normal processing state, and serves as a criterion for judging anomaly detection and the like. The judgment criteria data 131 may include associated data such as upper and lower limits of such waveform data. The judgment criteria data 131 includes a reference data length indicating the data length of the judgment criteria data 131. The "data length" represents the length of the data, for example, as the number of samples. For example, the greater the number of samples, the longer the data length.

[0032] The resampling unit 10 receives the first shot data that has passed through the oversampling filter 9 and a reference data length based on the judgment reference data 131. The resampling unit 10 executes a resampling process on the first shot data that has passed through the oversampling filter 9 to match the data length with the reference data length, and generates second shot data. The resampling unit 10 outputs the second shot data to the judgment processing unit 11. The resampling unit 10 is an example of a "data length adjustment unit" in the present disclosure, and the second shot data is an example of "second measurement data" in the present disclosure.

[0033] The judgment processor 11 executes judgment processing such as anomaly detection based on the second shot data, and outputs the judgment result. For example, the judgment processor 11 compares the second shot data with judgment reference data 131, and if the comparison result is not between the upper limit value and the lower limit value, it judges that an anomaly exists.

[0034] The output unit 12 is an output interface for outputting the judgment result. The output unit 12 is, for example, a light-emitting device such as an LED that emits light to notify a user of certain information. The output unit 12 may be a display device such as a liquid crystal display or an organic EL display that can display information. The output unit 12 may be an audio output device such as a speaker that notifies information by sound. The output unit 12 may be a communication interface that enables communication between the machining process monitoring device 100 and an external device. Such a communication interface performs communication according to an existing wired communication standard or wireless communication standard.

[0035] Each component of the machining process monitoring device 100 may be realized by a circuit corresponding to each component, or may be realized by an arithmetic circuit. Such an arithmetic circuit includes a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), etc., and controls the operation of the machining process monitoring device 100, for example, the data extracting unit 8, the oversampling filter 9, the resampling unit 10, and the judgment processing unit 11, according to information processing. Such information processing is realized by the arithmetic circuit executing a program. The arithmetic circuit may be realized by one or more dedicated processors. Furthermore, functions of the components of the arithmetic circuit may be omitted, replaced, or added as appropriate depending on the embodiment.

[0036] 1 shows various components that the machining process monitoring device 100 may have, but it is not essential that the machining process monitoring device 100 has all of these components. The machining process monitoring device 100 only needs to have at least a data extracting unit 8, an oversampling filter 9, a resampling unit 10, a judgment processing unit 11, and judgment reference data 131 during operation. In other words, the machining process monitoring device 100 does not need to have some or all of the sensors 1 and 2, the controllers 3 and 4, the A / D converter 5, the filters 6 and 7, and the output unit 12. It is only necessary that the machining process monitoring device 100 can acquire signals from the sensors 1 and 2, the controllers 3 and 4, the A / D converter 5, and the filters 6 and 7 during operation.

[0037] [Operation] FIG. 2 is a schematic diagram illustrating the data extraction process executed by the data extraction unit 8 in FIG. 1. FIG. 2(a) shows a signal s1 input to the data extraction unit 8 and having the height sensor 1 as a source. The signal s1 is a signal before being extracted by the data extraction unit 8. FIG. 2(b) shows a signal output from the data extraction unit 8 corresponding to the signal s1. The signal shown in FIG. 2(b) is generated as a result of executing the data extraction process by the data extraction unit 8 on the signal s1. The signal shown in FIG. 2 is a discrete signal because it passes through the A / D converter 5, but is shown in FIG. 2 as a continuous signal.

[0038] Specifically, the data extractor 8 monitors the signal s1 whose source is the height sensor 1 shown in Fig. 2(a) and determines the start of processing when the intensity of the signal s1 falls below the start threshold and the end of processing when it exceeds the end threshold. Next, the data extractor 8 extracts the waveform of the signal s1 from the start of processing to the end of processing. The start threshold and the end threshold are set to the same threshold in Fig. 2(a), but may be set to different values.

[0039] 2 only describes the data extraction process for signal s1 whose source is the height sensor 1, but the data extraction unit 8 also executes data extraction process for signal s2 whose source is the load sensor 2. That is, for signal s2, the data extraction unit 8 extracts the same waveform from the start of processing to the end of processing as for signal s1. The signals s1 and s2 extracted by the data extraction unit 8 constitute the first shot data. As described above, since the signals s1 and s2 are aligned to have the same data from the start of processing to the end of processing, the data lengths of all the waveforms constituting the first shot data are the same.

[0040] Fig. 3 is a schematic diagram for explaining the oversampling process executed by the oversampling filter 9 in Fig. 1 and the resampling process executed by the resampling unit 10 in Fig. 1. Fig. 3(a) shows an example of shot data output from the oversampling filter 9. Fig. 3(b) shows an example of second shot data output from the resampling unit 10.

[0041] In Fig. 3(a), the sampling points sampled at the sampling frequency Fs are indicated by black circles, and the oversampled sampling points are indicated by white circles. Correspondingly, in Fig. 3(a), the measurement sampling timing at the sampling frequency Fs is indicated by a straight line, and the oversampling timing at the eight-times sampling frequency 8Fs is indicated by a dashed line.

[0042] As described above, the oversampling filter 9 is, for example, a low-pass filter using FIR. It is preferable that the oversampling filter 9 does not cause a phase delay in the output signal relative to the input signal. In addition, it is preferable that the filter constant of the oversampling filter 9 is set so that the cutoff frequency is approximately equal to half the sampling frequency Fs (Nyquist frequency) obtained as the reciprocal of the reference data length, for example.

[0043] As shown in FIG. 3(a), between sampling points (black circle symbols) sampled at a sampling frequency Fs, seven points (white circle symbols) that lie on a smooth curve with high frequency components above the cutoff frequency blocked are interpolated by the oversampling filter 9.

[0044] Next, the resampling unit 10 performs a resampling process on the first shot data as shown in Fig. 3(a) that has passed through the oversampling filter 9 to make the data length match the reference data length, generating the second shot data as shown in Fig. 3(b). The resampling unit 10 resamples the measurement sampling timing with a period of 1 / Fs shown by a solid line in Fig. 3(a) to the judgment sampling timing shown by a solid line in Fig. 3(b). In Fig. 3(b), the resampling frequency is shown as Fr. The judgment sampling timing in Fig. 3(b) is calculated as the reciprocal of the reference data length.

[0045] 3 illustrates a case where the processing time for the first shot data, i.e., the data length of the first shot data, is slightly shorter than the processing time for the judgment reference data 131, i.e., the reference data length of the judgment reference data 131. Even if the sampling period is the same, the reference data length and the data length of the first shot data do not necessarily match due to the above-mentioned temporal fluctuation of processing, and there are cases where the data length of the first shot data is shorter than the reference data length as in the example shown in FIG.

[0046] Therefore, the resampling unit 10 executes a resampling process to make the data lengths of both data coincide with each other. Specifically, the sampling period of the first shot data is adjusted so that the processing start and processing end of the first shot data coincide with the processing start and processing end of the judgment reference data 131, respectively. As a result, the data length of the first shot data and the reference data length of the judgment reference data 131 become the same. In the example shown in FIG. 3, the processing start and processing end of the first shot data are adjusted to coincide with the processing start and processing end of the judgment reference data 131, respectively. As a result, in this embodiment, the measurement sampling timing interval shown by the solid line in FIG. 3(a) is converted to be slightly wider.

[0047] Then, from among the sampling points oversampled by 8 times, i.e., from among the black circle symbols and white circle symbols in Fig. 3(a), the one closest in time to the judgment sampling timing shown by the solid line in Fig. 3(b) is selected, and becomes the resampled second shot data shown by the black circle symbol in Fig. 3(b). This makes it possible to obtain an interpolated value with a small error that corresponds to the judgment sampling timing shown by the solid line in Fig. 3(b).

[0048] In this way, by performing the oversampling process, it is not necessary to set the sampling frequency Fs in the A / D conversion by the A / D converter 5 extremely high compared to the speed of change of the physical quantity for the resampling process. Therefore, the sampling frequency Fs by the A / D converter 5 can be kept low, the hardware configuration can be simplified, the amount of data to be processed can be reduced, and the judgment process can be accelerated. Furthermore, even if the data length is not constant due to fluctuations in the processing time, the resampling unit 10 can convert the first shot data into second shot data that matches the reference data length, and always match the processing start time and processing end time.

[0049] FIG. 4 is a schematic diagram illustrating shot data before and after resampling processing executed by the resampling unit 10 of FIG. 1. FIG. 4(a) shows shot data before resampling by the resampling unit 10. The shot data of FIG. 4(a) is signal information whose source is the load sensor 2. FIG. 4(b) shows second shot data obtained by executing resampling processing by the resampling unit 10 on the shot data of FIG. 4(a). FIG. 4(c) shows an example of the judgment criterion data 131. The data shown in FIGS. 4(a), (b), and (c) is, for example, time-series data arranged at equal intervals along the time axis.

[0050] When the start of processing in the shot data of Fig. 4(a) is made to coincide with the start of processing in the criterion data 131 of Fig. 4(c), a processing time difference ΔT due to temporal fluctuations in processing occurs between the end of processing Ta in the shot data of Fig. 4(a) and the end of processing Tc in the criterion data 131 of Fig. 4(c). In addition, when the shot data of Fig. 4(a) is compared with the criterion data 131 of Fig. 4(c), the number of samples from the start of processing to the end of processing also differs.

[0051] On the other hand, as shown in FIG. 4(b), the second shot data obtained by executing the resampling process by the resampling unit 10 has the same data length as the judgment reference data 131 in FIG. 4(c), and the number of samples from the start of processing to the end of processing is also the same for both. In this way, the resampling unit 10 matches the data length of the second shot data to be judged with the data length of the judgment reference data 131 that is the basis of judgment. This allows the resampling unit 10 to match the peak position appearing in the second shot data with the peak position in the judgment reference data 131. Here, the peak position is, for example, the time when a peak appears in the data, and represents the time based on the start of processing. Therefore, the judgment processing unit 11 can highly accurately judge the second shot data based on the comparison with the judgment reference data 131 even if the processing time fluctuates.

[0052] FIG. 5 is a graph showing that the variation in the waveform characterizing the punching process in the second shot data is reduced as a result of the resampling process. FIG. 5(a) is a graph showing an overlapping display of the second shot data as shown in FIG. 4(b) acquired in various cycles of the cycle process. In FIG. 5(a), only the peak position and its surrounding area of ​​the second shot data are enlarged and displayed. FIG. 5(b) is a graph showing the frequency of the load at the sampling time t1 for the multiple second shot data shown in FIG. 5(a). The horizontal axis of FIG. 5(b) shows the frequency, and the vertical axis shows the load or the signal level corresponding to the load.

[0053] The sampling time t1 shown in Fig. 5(a) is the rising edge of the waveform, which in the conventional technology is the part where a noticeable change in the waveform appears when an abnormality occurs in the cycle machining or when a fluctuation occurs in the machining time of the cycle machining.

[0054] As described above, the resampling unit 10 matches the data length of the second shot data to be judged with the data length of the judgment reference data 131 on which the judgment is based. This allows the resampling unit 10 to match the waveforms, such as peak positions, characteristic of the cyclic processing, among a plurality of waveform data corresponding to a plurality of cycles, even if fluctuations occur in the processing time of the cyclic processing. Therefore, as shown in FIG. 5(b), the frequency of the load at sampling time t1 among the plurality of waveform data has a steep peak. In other words, the resampling unit 10 can reduce the standard deviation of the load at sampling time t1 among the plurality of waveform data.

[0055] As a result, the machining process monitoring device 100 can perform highly accurate judgment even if the difference between the upper limit value and the lower limit value is set narrow when performing judgment such as abnormality detection on the second shot data by setting, for example, upper and lower limit values ​​based on the waveform data of the judgment reference data 131. Even when performing judgment based on statistical data such as cosine similarity and Mahalanobis distance other than judgment using the upper limit value and the lower limit value, the machining process monitoring device 100 can reduce the influence of fluctuations in machining time and perform judgment with high accuracy.

[0056] The output unit 12 outputs the judgment result by the judgment processing unit 11. For example, the output unit 12 is a display that displays the judgment result. This allows a user operating the processing machine to know the judgment result. Alternatively, the output unit 12 may output the judgment result to the processing machine. For example, when a judgment result indicating an abnormality is input, the processing machine stops operation, and it is possible to prevent the production of abnormal processed products, breakdowns of the processing machine, accidents caused by abnormal operation, and the like.

[0057] [Effects, etc.] As described above, the machining process monitoring device 100 according to this embodiment monitors a machining machine that performs cyclic machining by repeating a single cycle, and judges whether the machining machine is operating normally. The machining process monitoring device 100 includes a data extracting unit 8, a resampling unit 10, which is an example of a data length adjusting unit, and a judgment processing unit 11. The data extracting unit 8 determines the start and end of machining for each single cycle, and extracts first shot data from the start to end of machining for each single cycle from a first physical amount that changes over time during cyclic machining in the machining machine. The resampling unit 10 adjusts the data length of the first shot data to match the data length of the judgment criterion data 131 to generate second shot data. The judgment criterion data 131 is information that indicates the change in the first physical amount in a single cycle when the machining machine is operating normally as a reference. The judgment processing unit 11 compares the second shot data with the judgment criterion data 131 to judge whether the machining machine is operating normally. The first physical amount is, for example, a load.

[0058] According to the above configuration, even if fluctuations occur in the processing time of each single cycle in cyclic processing, the resampling unit 10 matches the data length of the second shot data with the data length of the judgment reference data 131, making it possible to more accurately determine whether the processing machine is operating normally than with conventional techniques.

[0059] The data extraction unit 8 may determine the start and end of processing based on a second physical quantity that changes at the start and end of processing in the processing machine. The second physical quantity is information that represents the position of a member of the processing machine that moves to apply a load to the workpiece in a single cycle, such as the height of a stripper plate of a press machine.

[0060] According to the above configuration, the start and end times of each single cycle in cyclic processing can be aligned with high precision, and the resampling unit 10 can match the data length of the second shot data with the data length of the judgment reference data 131 with high precision.

[0061] The machining process monitoring device 100 may further include an oversampling filter 9. The oversampling filter 9 interpolates data between a plurality of data points indicating a temporal change in a first physical quantity included in the first shot data. The resampling unit 10 uses the plurality of data points and the data interpolated by the oversampling filter 9 to adjust the data length of the first shot data to match the data length of the judgment reference data 131, thereby generating second shot data.

[0062] According to the above configuration, since the data interpolation process is executed, the amount of data of the first physical quantity to be acquired can be reduced. Therefore, the configuration of the machining process monitoring device 100 can be simplified to reduce costs, and the amount of data can be reduced, reducing the amount of processing, and speeding up the determination process.

[0063] Second embodiment Fig. 6 is a block diagram showing a configuration example of a machining process monitoring device 200 according to a second embodiment of the present disclosure. Compared to the machining process monitoring device 100 of Fig. 1, the machining process monitoring device 200 of Fig. 6 includes a resampling unit 210 instead of the resampling unit 10. Moreover, the memory unit 13 of the machining process monitoring device 200 of Fig. 6 stores judgment criterion data 231 instead of the judgment criterion data 131 of Fig. 1.

[0064] The resampling unit 210 compresses at least a portion of the data output from the oversampling filter 9. For example, the resampling unit 210 does not compress the waveform portion of the first shot data that represents the processing characteristics, but compresses only the other waveform portions.

[0065] Fig. 7 is a schematic diagram for explaining the compression process executed by the resampling section 210 of Fig. 6. Fig. 7(a) is a graph showing data input to the resampling section 210.

[0066] In this embodiment, the first shot data is divided into a plurality of n sections (where n is an integer equal to or greater than 2). In the example shown in FIG. 7A, the first shot data is a0 From time t a1 The first section S a1 Data D1 at time t a1 From time t a2 The second section S a2 Data D2 at time t a2 From the end of processing t a3 The third section S a3 The division is performed by the data cutout unit 8 or the resampling unit 210. In the example shown in FIG. 7(a), the first section S a1 represents the section from the start of processing until the punch descends and reaches the workpiece. a2 corresponds to the section where the punch punches through the material. a3 corresponds to the section from after the punch has punched through the material to the end of processing.

[0067] Each section S a1 ,S a2 ,S a3 The length of time t a1 ,t a2 ,t a3 The value of is set in advance according to the type of processing by the processing machine. For example, in cycle processing by a press machine, the timing when the punch punches the material in each processing cycle is determined by the structure of the die. a0 Using this as a reference, we can fix it to a certain range. Therefore, the interval S a1 ,S a2 ,S a3 The ratio of the lengths of the sections S and S can be set to a fixed value. a1 ,S a2 ,S a3 The length ratio is set to 2:1:7.

[0068] The first shot data is a portion where there is a large change over time (in the example shown in FIG. 7(a), the second section S a2) and the part with small changes (in the example shown in FIG. 7(a), the first section S a1 and Section 3 S a3 ) is preferably divided into

[0069] 7B is a graph showing compressed data obtained by compressing the oversampled first shot data of FIG. 7A by the resampling unit 210. The compressed data is a graph showing the compressed data obtained by compressing the first shot data of FIG. b0 From time t b1 The first section S b1 and time t b1 From time t b2 The second section S b2 and time t b2 From the end of processing t b3 The third section S b3 The compressed section S b1 ,S b2 ,S b3 is the section S before compression in Fig. 7(a). a1 ,S a2 ,S a3 correspond to the following:

[0070] The resampling unit 210 compresses the portion of the oversampled first shot data that has a small temporal change, while not compressing the portion of the oversampled first shot data that has a large temporal change, or compresses it at a smaller compression rate than the portion of the oversampled first shot data that has a small temporal change. a1 ,S a2 ,S a3 The waveform data in the sections S1, S2, and S3 are compressed at compression rates C1, C2, and C3, respectively, to obtain the waveform data in the sections S1, S2, and S3 in FIG. b1 ,S b2 ,S b3 Here, the compression ratio is the ratio of the amount of data before compression to the amount of data after compression.

[0071] For example, the resampling unit 210 compresses the data of each section at different compression rates by allocating a number of samples corresponding to a different length of time to the data of each section.

[0072] In the example of FIG. 7, C1=4, C2=1, and C3=3. That is, in the first section S a1 The waveform data in is compressed to 1 / 4 and the first section S in FIG. b1 The compressed data is the third section S in FIG. a3 The waveform data in is compressed to 1 / 3 and shown in the third section S in Fig. 7(b). b3 In these sections, the number of samples is reduced. In contrast, in the second section S in FIG. a2 The waveform data in is uncompressed.

[0073] The judgment criteria data 231 in FIG. 6 is b1 ,S b2 ,S b3 6 are divided into reference intervals corresponding to the reference data 131 in FIG. 1. That is, the judgment criterion data 231 is also divided into a plurality of reference intervals, n in number. Furthermore, when compared with the judgment criterion data 131 in FIG. 1, each reference interval of the judgment criterion data 231 in FIG. 6 is divided into a reference interval S a1 ,S a2 ,S a3 are compressed at compression rates C1, C2, and C3, respectively. Each reference section of the judgment reference data 231 has a reference data length, which are called the first reference data length, the second reference data length, and the third reference data length, respectively.

[0074] The first to third reference data lengths of the judgment reference data 231 are input to the resampling unit 210. The resampling unit 210 performs the following operations on the compressed data: b1 ,S b2 ,S b3 A resampling process is then performed to make the data length of the second shot data coincide with the first to third reference data lengths, respectively, to generate second shot data.

[0075] In this embodiment, the resampling section 210 can execute the resampling process even when the sampling periods before and after resampling are set to be significantly different.

[0076] As described above, in the machining process monitoring device 200 according to this embodiment, the judgment reference data 131 may have a plurality of n reference sections divided at a predetermined time ratio. The resampling unit 210 divides the first shot data into a plurality of n sections at a predetermined time ratio. The resampling unit 210 also generates second measurement data by adjusting each of the data lengths of the first shot data in the plurality of n sections so that they match the data lengths of the judgment reference data 131 in the corresponding plurality of n reference sections.

[0077] The above-mentioned machining process monitoring device 200 exerts a higher effect when the time of the machining point that characterizes the machining (for example, the punching period) is short with respect to the entire machining time of each single cycle in the cycle machining. That is, the machining process monitoring device 200 measures physical quantities such as the punching load applied to a tool such as a punch in a punching process by a press machine, and can not only grasp the state of the tool in detail, but also grasp the state of the operation such as pressing the workpiece by the stripper plate and pushing the scrap into the die throughout the entire machining time. In addition, as in this embodiment, by compressing data for an appropriate portion of the entire machining time of each single cycle, it is possible to grasp both the partial detailed state and the entire state. Furthermore, by compressing data, the amount of data acquired and the amount of data processing can be reduced, and the judgment process can be accelerated.

[0078] Therefore, the machining process monitoring device 200 does not compress data of the large change portion that characterizes the machining, and compresses data of the gradual change portion to a small amount of data, and can execute the judgment process, and can obtain the judgment result more quickly without impairing the judgment accuracy. Also, the machining process monitoring device 200 makes it possible to perform the entire abnormality detection from the start to the end of machining at once. The machining process monitoring device 200 can obtain a higher effect by the judgment method using the Mahalanobis distance, which is a judgment method based on the correlation of the sampling values ​​at two different times.

[0079] The resampling unit 210 may compress the first measurement data in at least one of the multiple n sections, thereby adjusting the data length of the first measurement data in at least one section so that it matches the data length of the judgment criteria data 131 in the corresponding reference section.

[0080] According to the above configuration, the amount of data to be acquired and the amount of data to be processed can be reduced by compressing the data, thereby speeding up the determination process.

[0081] (Modification of the second embodiment) Hereinafter, a modified example of the second embodiment of the present disclosure will be described with reference to FIG. 8. In this modified example, the height sensor 1 of FIG. 6 measures the height of a slider of a crank-type press. For example, the height sensor 1 measures a die height. The die height is, for example, the distance from the lower surface of the slider to the upper surface of the bolster plate. The height of the slider is an example of the "second physical quantity" of the present disclosure.

[0082] Fig. 8(a) is a graph showing the change over time in the height of the slider measured by the height sensor 1. Fig. 8(b) is a graph showing the change over time in the load applied to the punch measured by the load sensor 2.

[0083] As shown in Fig. 8(a), the change in slider height over time of the crank press is close to a sine wave. The punching of the workpiece is performed at the point between the start of the punching and the point when the slider descends to the bottom dead center.

[0084] The data extractor 8 or the resampling unit 210 extracts the slider height from the first intermediate threshold at time t c1 and the time t when the second intermediate threshold, which is smaller than the first intermediate threshold, is exceeded. c2 The first and second intermediate thresholds are stored in advance in, for example, the storage unit 13. As described in the first embodiment, at the processing start time t when the intensity of the signal s1 from the height sensor 1 falls below the start threshold, c0and the end time t c3 is determined by the data extractor 8.

[0085] As a result, the first shot data is c0 From time t c1 The first section S c1 Data D1 at time t c1 From time t c2 The second section S c2 Data D2 at time t c2 From the end of processing t c3 The third section S c3 In the example shown in FIG. 8, the first section S c1 represents the section from the start of processing until the punch descends and reaches the workpiece. c2 corresponds to the section where the punch punches through the material. c3 corresponds to the section from after the punch has punched through the material to the end of processing.

[0086] As described above, in this modification, the resampling unit 210 divides the first shot data at each of multiple time points when the measured slider height falls below multiple different thresholds, for example, the first intermediate threshold and the second intermediate threshold.

[0087] In the second embodiment, each section S in FIG. a1 ,S a2 ,S a3 The length of time t a1 ,t a2 ,t a3 The value of each section S is preset, whereas in this modification, it is determined using the detection result of the slider height by the height sensor 1. a1 ,S a2 ,S a3Since the division is performed in synchronization with the waveform detected by the height sensor 1, even if there is a fluctuation in the machining time of each single cycle in the cyclic machining, each divided area can be accurately obtained. Therefore, by dividing in synchronization with the waveform detected by the height sensor 1, it is possible to obtain the effect of correcting the time fluctuation for each section, so that the frequency of the load at a specific sampling time as shown in Fig. 5(b) can be made steeper with less variation, that is, with a smaller standard deviation. Therefore, according to this modified example, anomaly detection can be performed with higher accuracy.

[0088] Third embodiment Fig. 9 is a block diagram showing a configuration example of a machining process monitoring device 300 according to a third embodiment of the present disclosure. While the machining process monitoring device 200 in Fig. 6 includes one oversampling filter 9, the machining process monitoring device 300 in Fig. 9 includes multiple oversampling filters. In the illustrated example, the machining process monitoring device 300 includes three oversampling filters 9a, 9b, and 9c.

[0089] The oversampling filters 9a, 9b, and 9c are, for example, low-pass filters having cutoff frequencies set independently of one another. That is, the cutoff frequencies of the oversampling filters 9a, 9b, and 9c may be different from one another or may be the same.

[0090] The oversampling filter 9a has a processing start time t a0 From time t a1 The first section S a1 The data D1 at time t a1 From time t a2 The second section S a2 The oversampling filter 9c receives data D2 at time t a2 From the end of processing t a3 The third section S a3 The data D3 in is input.

[0091] The cutoff frequencies of the oversampling filters 9a, 9b, and 9c are adjusted according to the compression rates C1, C2, and C3 for the input data D1, D2, and D3. For example, in the first section S a1 When the waveform data in the first section S is compressed to 1 / 4 (when C1=4), the cutoff frequency of the oversampling filter 9a is set to 1 / 4 of the Nyquist frequency. b1 When the waveform data in the first section S is not compressed (when C2=1), the cutoff frequency of the oversampling filter 9b is set to the Nyquist frequency. c1 When the waveform data in is compressed to 1 / 3 (when C3=3), the cutoff frequency of the oversampling filter 9c is set to 1 / 3 of the Nyquist frequency.

[0092] When there is only one oversampling filter, high frequency components due to the sampling frequency before compression are included, and when obtaining the value closest on the time axis to the sampling point before resampling to find the resampling point, a value containing high frequency components may be obtained. In contrast, according to this embodiment, which includes multiple oversampling filters 9a, 9b, and 9c, it is easy to prepare oversampling filters 9a, 9b, and 9c that satisfy the sampling definition, and it is possible to obtain accurate values ​​from which high frequency components have been removed during resampling.

[0093] Therefore, the machining process monitoring device 300 can obtain measurement data containing correct frequency components by compressing the amount of data while satisfying the sampling definition, and the determination processing unit 11 can perform anomaly detection more accurately.

[0094] In this embodiment, the type of oversampling filter is the same as the number of sections, and the two correspond one-to-one, but this embodiment is not limited to this. For example, if the compression ratios applied to multiple sections are approximately the same, one oversampling filter may be provided for the multiple sections.

[0095] Furthermore, when a high compression ratio is applied to a certain section, for example a compression ratio of 8 times or more, the compression itself functions as an oversampling rate, so that the data for that section does not necessarily need to be oversampled to a frequency equal to or higher than the sampling frequency Fs.

[0096] As described above, the machining process monitoring device 300 according to this embodiment may further include a plurality of n oversampling filters corresponding to the plurality of n sections, respectively. Each of the plurality of n oversampling filters performs an oversampling process on a plurality of data points indicating a temporal change in the first physical quantity, which are included in the first measurement data in each of the plurality of n sections. The plurality of n oversampling filters have cutoff frequencies independent of each other.

[0097] According to the above configuration, even if the input data is compressed, the resampling unit 210 can resample the input data while satisfying the sampling definition. Therefore, the machining process monitoring device 300 can easily obtain measurement data containing correct frequency components by compressing the amount of data while satisfying the sampling definition, and the judgment processing unit 11 can more accurately detect anomalies.

[0098] The resampling unit 210 may compress the first measurement data in at least one of the multiple n sections at a predetermined compression rate to adjust the data length of the first measurement data in at least one section to match the data length of the judgment reference data 131 in the corresponding reference section. The cutoff frequency of each of the multiple n oversampling filters is set based on the compression rate.

[0099] According to the above configuration, in addition to the effects of the above-mentioned machining process monitoring device 300, by compressing data for an appropriate portion of the entire machining time of each single cycle, it is possible to grasp both partial detailed conditions and overall conditions. Furthermore, by compressing data, the amount of data acquired and the amount of data processing can be reduced, thereby speeding up the judgment process.

[0100] (Other embodiments) As described above, the above embodiment has been described as an example of the technology disclosed in this application. However, the technology in this disclosure is not limited to this, and can be applied to embodiments in which modifications, substitutions, additions, omissions, etc. are appropriately performed. In addition, it is also possible to combine the components described in the above embodiment to create a new embodiment. Other embodiments are exemplified below.

[0101] In the above embodiment, a configuration for processing two signals detected by the height sensor 1 and the load sensor 2 has been described, but the present disclosure is not limited to this, and any configuration capable of detecting the timing of the start and end of processing based on at least one signal may be used. Therefore, the number of sensors for detecting physical quantities is not limited to two, and may be one or three or more. The at least one signal is not limited to those obtained by the height sensor 1 or the load sensor 2, and may be, for example, a signal from a controller such as a PLC (Programmable Logic Controller), or a signal indicating the rotation angle, position, etc. of the press machine.

[0102] In the above embodiment, an example has been described in which the first shot data is oversampled by 8 times, but the oversampling factor is not limited to 8 and may be any factor greater than 1. Although a larger oversampling factor increases the amount of data and calculation, it is possible to obtain an interpolated value with a smaller error. Considering this trade-off, it is desirable to set the oversampling factor to about 8.

[0103] In the above embodiment, an oversampling filter has been described as an example of the data interpolation unit, but the data interpolation unit is not limited to this and may be any unit capable of obtaining sampled values ​​obtained by interpolating between a plurality of data points. For example, the data interpolation unit may perform linear interpolation between two sampled points, or curved interpolation based on sampled values ​​of three or more points.

[0104] In the above embodiment, the judgment reference data 131 has been described as waveform data serving as a reference for making a judgment, but the judgment reference data 131 is not limited to this and may be any data having a reference data length. The judgment reference data 131 may be a trained model generated by learning a plurality of time-series data having the reference data length by machine learning. The judgment processing unit 11 may use such a trained model to output a judgment result in response to an input of time-series data having the same data length as the reference data length, i.e., second shot data.

[0105] The effect of performing the judgment process using compressed and reduced data as in the second and third embodiments is remarkable when a trained model is used for the judgment process. This is because, in machine learning of time series data, the size of the trained model generated tends to increase as the data length of the time series data increases, and the amount of inference processing by the trained model increases accordingly. In addition, there is a problem that it becomes difficult to make the model learn appropriately as the data length increases. By using compressed waveform data without impairing the processing characteristics, the above problem can be solved and judgment can be performed at high speed.

[0106] As described above, the embodiments have been described as examples of the technology in the present disclosure. For this purpose, the accompanying drawings and detailed description have been provided.

[0107] Therefore, among the components described in the attached drawings and the detailed description, not only are there components essential for solving the problem, but there may also be components that are not essential for solving the problem in order to illustrate the above technology. Therefore, the fact that such non-essential components are described in the attached drawings or the detailed description should not be interpreted as immediately indicating that such non-essential components are essential.

[0108] Furthermore, since the above-described embodiments are intended to illustrate the technology in the present disclosure, various modifications, substitutions, additions, omissions, and the like can be made within the scope of the claims or their equivalents. [Industrial Applicability]

[0109] The present disclosure is applicable to a machining process monitoring device and a machining process monitoring method that monitor a machining machine that performs cyclic machining that repeats a single cycle and determines whether the machining machine is operating normally. [Explanation of symbols]

[0110] 1 Height Sensor 2 Load Sensor 3,4 Controller 5. A / D Converter 6,7 Filters 8 Data Extraction Section 9 Oversampling filter (data interpolation section) 10,210 Resampling section (data length adjustment section) 11 Judgment processing unit 12 Output section 13 Storage section 100,200,300 Process monitoring device 131,231 Criteria data

Claims

1. A processing process monitoring device that monitors a processing machine that performs cyclic processing by repeating a single cycle and determines whether the processing machine is operating normally, comprising: a data extraction unit that extracts first measurement data from a first physical quantity that changes over time during the cyclic machining in the machining machine, from a machining start time to a machining end time of each single cycle; a data length adjusting unit that adjusts a data length of the first measured data to match a data length of the judgment reference data to generate second measured data; a determination processing unit that compares the second measurement data with the determination reference data to determine whether the processing machine is operating normally; Equipped with The data extracting unit determines the machining start time and the machining end time of each single cycle based on a signal related to the operation of the machining machine. Process monitoring device.

2. The data extracting unit determines the machining start time and the machining end time based on a second physical quantity that changes at the machining start time and the machining end time in the machining machine. The machining process monitoring device according to claim 1.

3. The processing machine applies a load to a workpiece in the single cycle to perform processing, 3. The machining process monitoring device according to claim 2, wherein the second physical amount is information representing a position of a member of the machining device that moves to apply a load to the workpiece in the single cycle.

4. a data interpolation unit that interpolates data between a plurality of data points that indicate a change over time of the first physical quantity included in the first measurement data, the data length adjustment unit adjusts a data length of the first measured data to match a data length of the reference data by utilizing the plurality of data points and data interpolated by the data interpolation unit, thereby generating the second measured data. The machining process monitoring device according to claim 1 or 2.

5. 5. The machining process monitoring device according to claim 4, wherein the data interpolation section interpolates data between the plurality of data points by oversampling the plurality of data points.

6. The judgment criteria data has a plurality of n reference intervals divided at a predetermined ratio in time, The data length adjustment unit Dividing the first measurement data into a plurality of n sections at the predetermined ratio in time; generating the second measured data by adjusting a data length of the first measured data in the plurality of n sections so as to match a data length of the judgment reference data in the corresponding plurality of n reference sections; The machining process monitoring device according to any one of claims 1 to 5.

7. The data length adjustment unit divides the first measurement data at a plurality of time points when a value of a second physical quantity that changes at the start of the machining and at the end of the machining is below a plurality of different threshold values. The machining process monitoring device according to claim 6.

8. 8. The machining process monitoring device according to claim 6, wherein the data length adjustment unit adjusts the data length of the first measurement data in at least one section among the plurality of n sections so as to match the data length of the judgment reference data in a corresponding reference section by compressing the first measurement data in the at least one section.

9. a plurality of n oversampling filters corresponding to the plurality of n sections, the plurality of n oversampling filters performing an oversampling process on a plurality of data points indicating a temporal change in the first physical quantity included in the first measurement data in each of the plurality of n sections; The plurality of n oversampling filters have cutoff frequencies independent of each other. The machining process monitoring device according to any one of claims 6 to 8.

10. the data length adjustment unit adjusts a data length of the first measured data in at least one section among the plurality of n sections at a predetermined compression rate so as to match a data length of the judgment reference data in a corresponding reference section; a cutoff frequency of each of the n oversampling filters is set based on the compression ratio; The machining process monitoring device according to claim 9.

11. The machining process monitoring device according to any one of claims 1 to 10, wherein the data length adjustment unit adjusts the data length of the first measurement data to match the data length of the judgment reference data by resampling the first measurement data.

12. The machining process monitoring device according to any one of claims 1 to 11, further comprising an output section which outputs a result of the determination made by the determination processing section.

13. The processing process monitoring device according to any one of claims 1 to 12, wherein the processing machine is a press machine.

14. The machining process monitoring device according to any one of claims 1 to 13, wherein the first physical amount is a load detected by a load sensor.

15. 1. A machining process monitoring method for monitoring a machining machine that performs cyclic machining by repeating a single cycle, and determining whether the machining machine is operating normally, comprising the steps of: a first physical quantity acquisition step of sequentially acquiring a first physical quantity that changes over time during the cycle machining in the machining machine; a data extraction step of extracting first measurement data from the first physical quantity from a processing start time to a processing end time of each single cycle; a data length adjusting step of adjusting a data length of the first measured data so as to match a data length of the judgment reference data to generate second measured data; a determination processing step of comparing the second measurement data with the determination reference data to determine whether the processing machine is operating normally; Including, the data extraction step includes determining the start and end of each single cycle of machining based on signals related to the operation of the machining machine; Process monitoring method.

16. A machining process monitoring device for monitoring a machining machine that performs cyclic machining by repeating a single cycle and determining whether the machining machine is operating normally, comprising: a data extraction unit that extracts first measurement data from a first physical quantity that changes over time during the cyclic machining in the machining machine, from a machining start time to a machining end time for each single cycle; a data interpolation unit that interpolates data between a plurality of data points that indicate a temporal change in the first physical quantity included in the first measurement data by oversampling; a data length adjusting unit that uses the plurality of data points and the data interpolated by the data interpolating unit to adjust a data length of the first measured data to match a data length of the judgment reference data, thereby generating second measured data; a determination processing unit that compares the second measurement data with the determination reference data to determine whether the processing machine is operating normally; Equipped with The judgment criteria data has a plurality of n reference intervals divided at a predetermined ratio in time, The data length adjustment unit divides the first measurement data into a plurality of n sections at the predetermined ratio in time, the data length adjustment unit compresses the first measured data in at least one section among the plurality of n sections, thereby adjusting each data length of the first measured data in the plurality of n sections so as to match a data length of the judgment reference data in the corresponding plurality of n reference sections. Process monitoring device.

17. A machining process monitoring method for monitoring a machining machine that performs cyclic machining by repeating a single cycle and determining whether the machining machine is operating normally, comprising the steps of: a first physical quantity acquisition step of sequentially acquiring a first physical quantity that changes over time during the cycle machining in the machining machine; a data extraction step of extracting first measurement data from the first physical quantity for each single cycle from a processing start to a processing end; a data interpolation step of interpolating data between a plurality of data points that indicate a temporal change in the first physical quantity included in the first measurement data by oversampling; a data length adjusting step of adjusting a data length of the first measurement data to match a data length of a judgment reference data by utilizing the plurality of data points and the data interpolated in the data interpolation step, thereby generating second measurement data; a determination processing step of comparing the second measurement data with the determination reference data to determine whether the processing machine is operating normally; a determination processing step of comparing the second measurement data with the determination reference data to determine whether the processing machine is operating normally; Including, The judgment criteria data has a plurality of n reference intervals divided at a predetermined ratio in time, the data length adjustment step includes dividing the first measurement data into a plurality of n sections at the predetermined ratio in time; the data length adjusting step includes compressing the first measurement data in at least one section among the plurality of n sections, thereby adjusting each of the data lengths of the first measurement data in the plurality of n sections so as to match a data length of the judgment reference data in the corresponding plurality of n reference sections. Process monitoring method.

Images