JP2017097839A - Abnormality analysis system and analysis device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an abnormality analysis system capable of analyzing on the basis of information on many production facilities and promptly feeding back an analysis result to the production facilities.SOLUTION: A plurality of production facilities (11 to 13) and an analysis device (50) are connected through a fog network (31). The analysis device (50) performs a data analysis on the basis of detected information of detectors (112b, 114c, 114d, 115b,116, 117 and 118c) obtained through the fog network (31) and stores determination information, about respective abnormality in the plurality of production facilities (11 to 13) or production objects thereof, as a data analysis result. The plurality of production facilities (11 to 13) determine the abnormality therein or the abnormality in the production objects thereof on the basis of the determination information stored in the analysis device (50).SELECTED DRAWING: Figure 1

Description

  The present invention relates to an abnormality analysis system and an analysis apparatus used for the abnormality analysis system.

  Patent Document 1 describes a method for monitoring grinding burn of a workpiece. During grinding of a workpiece, the grinding load of the grinding wheel and the rotational speed of the workpiece are detected, and the presence or absence of occurrence of grinding burn is determined by comparing with a threshold value for the grinding load corresponding to the rotational speed. Here, the threshold value is set based on the grinding load of the grinding wheel with respect to the rotational speed of the workpiece when grinding burn of the workpiece occurs.

  Patent Document 2 describes performing trial grinding and setting a threshold value based on a grinding load in trial grinding. Thereafter, the presence or absence of occurrence of grinding abnormality is determined by comparing the grinding load detected in the actual grinding with a threshold value.

  Japanese Patent Application Laid-Open No. H10-228561 describes predicting a quality abnormality of a product based on the following quality tendency pattern. For example, when grinding the outer peripheral surface of a workpiece with a grinding wheel, the dimensional accuracy tends to deteriorate as the number of workpieces increases (see FIG. 4 of Patent Document 3). Further, the relationship between the number of workpieces and the average value of the grinding resistance is obtained from the relationship between the grinding time and the grinding resistance in one workpiece (see FIGS. 5 and 10 of Patent Document 3). Then, by considering the relationship between the number of workpieces and the dimensional accuracy, it is possible to set a threshold for the average value of the grinding resistance in the quality trend pattern indicating the relationship between the number of workpieces and the average value of the grinding resistance. That is, by grasping the grinding resistance and the number of workpieces, it is possible to predict the abnormality of the product based on the quality tendency pattern and the threshold value.

JP2013-129027A International Publication No. 2012/098805 JP 2014-154094 A

  In recent years, it is said to be an IoT (Internet of Things) era, and utilization of big data obtained by connecting many things to the Internet is expected. In production facilities, it is also expected to perform abnormality analysis of production objects based on a large amount of information obtained from the production facilities.

  In recent years, cloud computing has been known. Cloud computing is a form of using a computer connected via the Internet or the like. For example, instead of using data and applications stored in the local computer, the local computer uses data stored in a computer connected via the Internet, etc., or uses the computer application To do.

  The use of big data in production facilities can be considered using cloud computing. However, in cloud computing, too much data is communicated, so communication congestion may occur. Furthermore, when the distance to the cloud server is long, the communication time becomes long. For these reasons, when using cloud computing, it is not quick.

  If an abnormality analysis of a production facility is performed, if the analysis result can be fed back to the production facility at an early stage, an effect of suppressing the occurrence of an abnormality in the production object can be expected. Therefore, it is not sufficient to use cloud computing as it is as an abnormality analysis system for production facilities.

  An object of the present invention is to provide an abnormality analysis system capable of analyzing based on information on a large number of production facilities and feeding back the analysis results to the production facilities at an early stage, and an analysis apparatus used therefor.

(1. Anomaly analysis system)
An abnormality analysis system according to the present invention is a facility for producing a production object, and is connected to a plurality of production facilities each including one or a plurality of detectors, and constructs fog computing. A first network installed in a predetermined area; connected to the first network; performing data analysis based on detection information of the detector acquired via the first network; and a result of the data analysis And an analysis device that generates determination information regarding each abnormality of the plurality of production facilities or the abnormality of the production object. Each of the plurality of production facilities includes an abnormality determination device that determines each abnormality of the plurality of production facilities or abnormality of the production object based on the determination information generated in the analysis device.

  Detectors and analyzers in a plurality of production facilities are connected via a first network installed in a predetermined area for constructing fog computing. Fog computing is a network-connected system in a narrow area compared to cloud computing. That is, the first network for constructing fog computing is a network installed in a predetermined area narrower than the area for constructing cloud computing. Therefore, the occurrence of communication congestion is suppressed in data communication between the detector and the analysis device. Further, since the first network is constructed in a narrow predetermined area, the communication time between the production facility and the analysis device can be shortened. Therefore, the analysis apparatus can receive information detected by the detector at high speed.

  Since the analysis apparatus can acquire detection information in a plurality of production facilities at an early stage and perform analysis, the analysis apparatus can feed back the results of the analysis apparatus to the production facilities at an early stage. Since the analysis result can be fed back to the production facility at an early stage, the occurrence of an abnormality in the production target can be suppressed earlier and reliably.

(2. Analysis device)
The analysis apparatus according to the present invention is the analysis apparatus used in the above-described abnormality analysis system. According to the analysis device, the effect of the above-described abnormality analysis system can be obtained.

It is a figure which shows an abnormality analysis system. It is a figure which shows the structure of a grinding machine as an example of the production equipment of FIG. It is a block diagram of production equipment. It is a figure which shows the behavior of the power of the motor of a grinding wheel with respect to the elapsed time from the grinding | polishing start of one production target object. It is a figure which shows the structure of the analyzer of FIG. In 1st embodiment, the flow of the detailed process of an abnormality determination apparatus, an analysis apparatus, and a high-order analysis apparatus is shown. In 2nd embodiment, the flow of the detailed process of an abnormality determination apparatus, an analysis apparatus, and a high-order analysis apparatus is shown. In 2nd embodiment, it is a graph which shows the result of the frequency analysis by an abnormality determination apparatus. In 2nd embodiment, it is a figure which shows the 1st example of the abnormality determination by an abnormality determination apparatus, Comprising: It is a figure about the peak value (evaluation parameter) of the amplitude of a vibration in the time slot | zone (regular parameter) of a day. In 2nd embodiment, it is a figure which shows the 2nd example of the abnormality determination by an abnormality determination apparatus, Comprising: It is a figure about the peak value (evaluation parameter) of the amplitude of the vibration in the time of 1 year (regulated parameter). In 2nd embodiment, it is a figure explaining the production | generation of the pattern of the determination information in the 1st example by an analyzer. In 2nd embodiment, it is a figure explaining the production | generation of the pattern of the determination information in the 2nd example by an analyzer. In 3rd embodiment, the flow of the detailed process of an abnormality determination apparatus, an analysis apparatus, and a high-order analysis apparatus is shown. In 3rd embodiment, it is a figure which shows the abnormality determination by an abnormality determination apparatus, Comprising: It is a figure about the electric current value (evaluation parameter) of the motive power of the motor in environmental temperature (regulated parameter). In 3rd embodiment, it is a figure explaining the production | generation of the pattern of the determination information by an analyzer.

<1. First embodiment>
(1-1. Configuration of abnormality analysis system)
The configuration of the abnormality analysis system 1 of the present embodiment will be described with reference to FIG. As shown in FIG. 1, the abnormality analysis system 1 includes production facilities 11 to 13, other production facilities 21 to 23, a fog network 31 connected to the production facilities 11 to 13, and other production facilities 21 to 23. , A cloud network 40 connected to the fog networks 31, 32, an analysis device 50, another analysis device 60, and a higher-level analysis device 70. Here, the analysis devices 50 and 60 and the host analysis device 70 may be built-in systems such as a PLC (Programmable Logic Controller) and a CNC (Computerized Numerical Control) device, or may be a personal computer or a server. it can.

  The production facilities 11 to 13 (corresponding to the production facility of the present invention) are facilities for producing a predetermined production object. Other production facilities 21 to 23 (corresponding to other production facilities of the present invention) are facilities for producing a predetermined production object. The production object produced by the production facilities 11 to 13 and the production object produced by the other production facilities 21 to 23 may be the same or different.

  The production facilities 11 and 21 are, for example, machine tools that are in charge of the first machining process in the production line, and are grinders that grind the crankshaft. The production facilities 13 and 23 are machine tools in charge of the second machining process, and are grinders for grinding the crankshaft as described above. The production facilities 12 and 22 are conveyance machines that convey production objects between the production facilities 11 and 13 or between the production facilities 21 and 23.

  The production facilities 11 to 13 are installed in the same building or in a nearby building. The other production facilities 21 to 23 are installed in the same building or nearby buildings, and are installed in a building different from the production facilities 11 to 13. For example, when the production facilities 11 to 13 are installed in Japan and the other production facilities 21 to 23 are installed in countries other than Japan, the production facilities 11 to 13 and the other production facilities 21 to 23 are Both are installed in Japan, but may be installed in remote areas.

  That is, the production facilities 11 to 13 are installed in a predetermined area where fog computing described later can be constructed. Similarly, the other production facilities 21 to 23 are installed in a predetermined area where fog computing can be constructed. However, the production facilities 11 to 13 and the other production facilities 21 to 23 are installed in an area where fog computing cannot be constructed.

  Here, fog computing is a network-connected system in a narrow area as compared to cloud computing. That is, the network for constructing fog computing is a network installed in a predetermined area narrower than the area for constructing cloud computing. The fog computing is sometimes referred to as edge computing.

  The fog network 31 (corresponding to the first network of the present invention) is a network connected to the production facilities 11 to 13 and installed in a predetermined area for constructing fog computing. The fog network 31 is installed in the same building as the building where the production facilities 11 to 13 are installed, or is installed in a building where one of the production facilities 11 to 13 is installed and a neighboring building. Yes.

  The other fog network 32 is connected to the other production facilities 21 to 23 and is a network installed in a predetermined area where fog computing is constructed. The other fog network 32 is installed in the same building where the other production facilities 21 to 23 are installed, or is adjacent to the building where any of the other production facilities 21 to 23 is installed. It is installed in the building. The other fog network 32 is not directly connected to the fog network 31. Here, as the fog networks 31 and 32, the Internet, a local area network (LAN), a wide area network (WAN), or the like can be applied.

  The cloud network 40 (corresponding to the second network of the present invention) is a network connected to the fog networks 31 and 32. The cloud network 40 is a wider area network than the fog networks 31 and 32, for example, the Internet. Therefore, the cloud network 40 is a network that connects the production facilities 11 to 13 and the other production facilities 21 to 23.

  The analysis device 50 is directly connected to the fog network 31 and is installed in a building that is the same as or adjacent to the building where the production facilities 11 to 13 are installed. The analysis device 50 performs data analysis based on the detection information acquired from the production facilities 11 to 13. For example, the analysis device 50 acquires detection information for one day of the production facilities 11 to 13 and performs data analysis every day. Data analysis can be learned by repeating it many times. And the analysis apparatus 50 memorize | stores the determination information regarding the abnormality of the production facilities 11-13 or the abnormality of the production target object by the production facilities 11-13 as a result of a data analysis. Further, the analysis device 50 acquires the result of the higher-order data analysis by the higher-order analysis device 70 described later, so that the determination information is based on the result of the data analysis by the analysis device 50 and the result of the higher-order data analysis by the higher-order analysis device 70 Is determined and memorized.

  The other analysis device 60 is directly connected to the fog network 32, and is installed in a building that is the same as or adjacent to the building where the production facilities 21 to 23 are installed. The other analysis device 60 performs data analysis based on the detection information acquired from the other production facilities 21 to 23. The other analysis device 60 performs the same processing as the above-described analysis device 50 with the other production facilities 21 to 23 as targets.

  The host analysis device 70 is connected to the cloud network 40 and performs host data analysis based on the acquired information. That is, the higher-level analysis device 70 acquires information from the production facilities 11 to 13 and also acquires information from the other production facilities 21 to 23 via the cloud network 40 and the respective fog networks 31 and 32. The host analysis device 70 is intended for host data analysis that requires a longer time than the data analysis performed by the analysis devices 50 and 60, and is intended for host data analysis using a large amount of information. The host analysis device 70 acquires, for example, the detection information of the production facilities 11 to 13 and the other production facilities 21 to 23 for one week, several weeks, one month, or several months, and acquires it in the acquisition period. Perform data analysis accordingly. The host data analysis can also be learned by repeating it many times.

(1-2. Configuration of Production Equipment 11)
Next, an example of the configuration of the production facility 11 will be described with reference to FIGS. In the present embodiment, the production facility 11 is a grinding machine, for example. As an example of the grinder 11, a grindstone traverse type grinder that traverses (moves in the Z-axis direction) the grindstone table 114 with respect to the bed 111 will be described as an example. However, the grinding machine 11 can also be applied to a table traverse type grinding machine in which the spindle device 112 traverses (moves in the Z-axis direction) with respect to the bed 111.

  A production object (workpiece) by the grinding machine 11 is, for example, a crankshaft W. The parts to be ground by the grinder 11 are a crank journal, a crankpin, and the like of the crankshaft.

  The grinding machine 11 is configured as follows. A bed 111 is fixed to the installation surface, and a spindle device 112 and a tailstock device 113 that support the crankshaft W at both ends rotatably are attached to the bed 111. The crankshaft W is supported by the main shaft device 112 and the tailstock device 113 so as to rotate around the crank journal. The spindle device 112 includes a motor 112a that rotationally drives the crankshaft W. The spindle device 112 is provided with a detector (vibration sensor) 112b that detects the vibration of the spindle.

  Further, on the bed 111, a grindstone table 114 is provided that can move in the Z-axis direction (axial direction of the crankshaft W) and the X-axis direction (direction orthogonal to the axial line of the crankshaft W). The wheel head 114 is moved in the Z-axis direction by the motor 114a, and is moved in the X-axis direction by the motor 114b. Furthermore, the wheel head 114 is provided with a detector 114c that detects the position of the wheel head 114 in the Z direction and a detector 114d that detects the position of the wheel head 114 in the X direction. The detectors 114c and 114d are rotary encoders that measure the rotation of the motor 114b and the like, and may be linear position detectors such as a linear scale.

  A grinding wheel 115 for grinding a crankpin or a crank journal is rotatably provided on the grinding wheel base 114. The grinding wheel 115 is rotationally driven by a motor 115a. Further, the grindstone table 114 is provided with a detector 115b for detecting the power of the motor 115a. The detector 115b is, for example, a motor wattmeter, but may be a voltmeter or an ammeter that measures the voltage or current of the motor 115a. The power, voltage, current, etc. of the motor 115a of the grinding wheel 115 can indirectly obtain grinding resistance. In addition to the above, the detector 115b may be a load detector provided on the spindle device 112 or the grindstone table 114 so as to obtain a grinding resistance directly.

  Further, the bed 111 is provided with a sizing device 116 for measuring an outer diameter of a crank pin or a crank journal which is a grinding portion of the crankshaft W. Furthermore, the bed 111 is provided with a detector 117 that detects an environmental temperature (outside air temperature). The bed 111 further includes a pump 118a for supplying coolant to the grinding site, a valve 118b for switching ON / OFF of the coolant supply, and a detector 118c for detecting the state of the valve 118b. The detector 118c is a coolant flowmeter, but may be a pressure sensor that detects the pressure of the coolant.

  Furthermore, the grinding machine 11 includes a CNC device 121, a PLC 122, an abnormality determination device 123, and an operation panel 124. Here, the abnormality determination device 123 can be an embedded system of the CNC device 121 or the PLC 122, and can also be a personal computer, a server, or the like.

  As shown in FIG. 3, the CNC device 121 controls motors 112 a and 115 a that rotate the spindle device 112 and the grinding wheel 115, and controls motors 114 a and 114 b that move the grinding wheel 115 relative to the crankshaft W. . In the control, the CNC device 121 acquires the detectors 114c and 114d for the position of the grindstone table 114 and the detector 115b for the power of the motor 115a.

  The PLC 122 acquires detection information from the sizing device 116. Further, the PLC 122 controls the supply of the coolant by controlling the pump 118a and the valve 118b. In this control, the PLC 122 acquires detection information of the detector 118c that detects the state of the valve 118b. Furthermore, the PLC 122 acquires detection information of the detector 117 that detects the environmental temperature.

  Here, the sampling periods of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c are not all the same, but are at least partially different. For example, the sampling period of the power detector 115b of the motor 115a is several milliseconds, the sampling period of the sizing device 116 is several milliseconds, and the sampling period of the valve state detector 118c is several tens of milliseconds. The sampling period of the temperature detector 117 is several tens of msec. Each sampling period is appropriately adjusted according to the control method.

  The abnormality determination device 123 determines abnormality of the grinding machine 11 or abnormality of a production target (workpiece). The abnormality determination device 123 stores a threshold value according to the determination target, and compares the detection information from each detector 112b, 114c, 114d, 115b, 116, 117, 118c with the corresponding threshold value to determine abnormality. I do.

  For example, as shown in FIG. 4, the abnormality determination device 123 stores in advance threshold values Th11 and Th12 for comparing one piece of production object (workpiece) with information detected by the detector 115b for the power of the motor 115a. doing. The threshold values Th11 and Th12 are set so as to change in accordance with the behavior of the power of the motor 115a with respect to the elapsed time from the start of grinding for one production object (workpiece). The threshold value Th11 is an upper limit value, and the threshold value Th12 is a lower limit value.

  The abnormality determination device 123 determines the abnormality of the production object by comparing the power of the motor 115a with the threshold values Th11 and Th12. Specifically, when the power of the motor 115a exceeds the upper limit threshold Th11 or falls below the lower limit threshold Th12, the abnormality determination device 123 has a state in which the production object does not satisfy grinding burn or shape accuracy. The production object is determined to be abnormal. The power of the motor 115a of the grinding wheel 115 corresponds to grinding resistance. Therefore, instead of the power of the motor 115a of the grinding wheel 115, a grinding resistance detected by another detection method may be employed. For example, Japanese Patent Application Laid-Open No. 2013-129027 discloses a determination as to whether grinding burn or the like has occurred in a production object by comparing the grinding resistance with a threshold value.

  The abnormality determination device 123 determines the abnormality of the drive devices 112a, 114a, 114b, 115a, 118a, and 118b to be controlled by the CNC device 121 and the PLC 122. For example, the abnormality determination device 123 compares a use record value obtained from information such as the use state and use history of the motors 114a and 114b with a pre-stored threshold value, so that the ball screw used in the drive mechanism Judgment of abnormalities such as bearings. In addition, the abnormality determination device 123 determines the abnormality of the valve 118b by comparing a use record value obtained from information such as a use state and a use history of the valve 118b with a threshold value stored in advance. The abnormality of the driving mechanism and the abnormality of the valve 118b are used in the meaning including not only the failure of the driving mechanism and the valve 118b but also the state that requires life and maintenance.

  Here, the threshold value stored in the abnormality determination device 123 is a different value depending on the target grinding machine 11. Even if the production equipment 11 and the other production equipment 21 produce the same object in FIG. 1, the use environment is different or the material composition of the production object is different. In addition, there may be individual differences between the production facilities 11 and 21. Therefore, even when the same object is produced, the threshold value of the production facility 11 and the threshold value of the other production facility 21 may be set to different values.

  Moreover, in the above, although the production equipment 11 was demonstrated, it is the same also about the production equipment 13, 21, 23 as a grinding machine. Further, the production facilities 12 and 22 as the transfer device are similarly provided with an abnormality determination device 123. In this case, the abnormality determination device 123 compares the actual usage value obtained from information such as the usage state and the usage history of the production facilities 12 and 22 as the conveyance device with a threshold value stored in advance, for example, conveyance. It is possible to determine an abnormality of a part constituting the road (a state where failure, life, or maintenance is required). Although the abnormality determination device 123 is provided in the production facility 11 as shown in FIG. 2, it may be provided in the analysis device 50.

(1-3. Configuration of Analysis Device 50)
Next, the configuration of the analysis device 50 will be described with reference to FIG. The analysis device 50 is connected to the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c of the production facilities 11 to 13 via the fog network 31. The analysis apparatus 50 acquires the detection information of each detector 112b, 114c, 114d, 115b, 116, 117, 118c of the production facilities 11 to 13 via the fog network 31. Furthermore, the analysis device 50 is also connected to the CNC device 121 and the PLC 122 of the production facilities 11 to 13. The analysis device 50 acquires various control parameters via the fog network 31.

  The fog network 31 is constructed in a narrower area than the cloud network 40. Therefore, the analysis apparatus 50 can acquire the detection information of each detector 112b, 114c, 114d, 115b, 116, 117, 118c of the production facilities 11 to 13 at an early stage from the detection time point.

  As illustrated in FIG. 5, the analysis device 50 includes an analysis unit 51, a display unit 52, and an input unit 53. The analysis part 51 acquires the detection information of each detector 112b, 114c, 114d, 115b, 116, 117, 118c of the production facilities 11-13. Here, the analysis unit 51 acquires all detection information detected by each of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c. That is, the analysis unit 51 acquires all the detection information regardless of the sampling period of each detector 112b, 114c, 114d, 115b, 116, 117, 118c. Here, since the analysis unit 51 acquires all of the detection information, the analysis unit 51 has a huge amount of data. However, since the analysis unit 51 passes through the fog network 31, a delay in communication time does not matter.

  Further, the analysis unit 51 acquires various control parameters in the production facilities 11 to 13 in addition to the detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c. For example, the control parameters in the production facilities 11 and 13 include grinding process information such as the shape and material of the crankshaft W as a production object, the shape and material of the grinding wheel 115, the grinding cut amount, and the coolant flow rate. .

  The analysis unit 51 performs data analysis based on the acquired detection information and various control parameters. Data analysis is so-called data mining. In particular, the analysis unit 51 not only detects the detection information of each detector 112b, 114c, 114d, 115b, 116, 117, 118c in one production facility 11, but also each detector 112b in a plurality of production facilities 11-13. The detection information of 114c, 114d, 115b, 116, 117, 118c, etc. are acquired.

  And the analysis part 51 can produce | generate the determination information regarding abnormality of a production target object by data analysis, and memorize | stores this determination information. For example, the analysis unit 51 generates threshold values Th11 and Th12 (shown in FIG. 4) for determining the presence or absence of grinding burn of the production object as one piece of determination information by data analysis. Moreover, the analysis part 51 produces | generates the threshold value for determining the abnormality of each component of the production facilities 11-13 by data analysis as another one of determination information. Furthermore, after generating the determination information once, the analysis unit 51 updates the determination information by acquiring new detection information.

  The display unit 52 displays determination information as a result of data analysis by the analysis unit 51, and allows the operator to confirm the result of data analysis. The display unit 52 can also display the detection information and various control parameters acquired by the analysis unit 51. For example, the display unit 52 includes a threshold value for determining the presence or absence of grinding burn obtained by the analysis unit 51, information detected by the detector 115b of the power of the motor 115a in the production facility 11, and the power of the motor 115a in the production facility 13. The detection information by the detector 115b is superimposed and displayed.

  The input unit 53 receives input of determination information and the like by an operator. The input unit 53 can set determination information corresponding to each of the production facilities 11 to 13. The analysis unit 51 can obtain determination information corresponding to each of the production facilities 11 to 13, but the operator can further arbitrarily edit while referring to the obtained determination information. The edited determination information is stored in the analysis unit 51.

  And the production facilities 11-13 acquire and memorize | store the determination information memorize | stored in the analysis part 51 via the fog network 31. FIG. The abnormality determination device 123 of the production facilities 11 to 13 determines the abnormality of the production facilities 11 to 13 or the abnormality of the production target based on the acquired determination information.

(1-4. Detailed processing of the abnormality determination device 123, the analysis devices 50 and 60, and the host analysis device 70)
Next, detailed processing of the abnormality determination device 123, the analysis devices 50 and 60, and the host analysis device 70 will be described with reference to FIG. The analysis devices 50 and 60 and the host analysis device 70 acquire detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c, and generate various determination information. However, for ease of explanation, the following description will be given by taking an example of processing when the detection information of the detector 115b is used.

  The detector 115b detects the power of the motor 115a every time the production object (workpiece) is ground (S1). Subsequently, the abnormality determination device 123 collects data for one production object (S2). This data is, for example, the behavior indicated by the solid line in FIG. If the threshold values Th11 and Th12 as determination information are already stored in the abnormality determination device 123, the abnormality determination device 123 performs abnormality determination (S3). That is, the abnormality determination device 123 determines whether or not the production object is abnormal by comparing the data for one production object with the threshold values Th11 and Th12 that are already stored. .

  Furthermore, the abnormality determination device 123 collects data for a plurality of production objects (S4). The abnormality determination device 123 collects data of production objects for one day, for example. Data of a plurality of production objects collected by the abnormality determination device 123 is transmitted to the analysis devices 50 and 60 via the fog networks 31 and 32, for example, once a day. And the analysis apparatuses 50 and 60 acquire the detection information of the detector 115b about the production target object for several times, for example once a day (S5). Here, the analysis devices 50 and 60 acquire all the detection information of the detector 115b.

  The analysis devices 50 and 60 perform data analysis based on the detection information of the detector 115b for a plurality of production objects (S6). Then, the analysis devices 50 and 60 generate threshold values Th11 and Th12 as determination information by data analysis (S7). Further, when the analysis devices 50 and 60 newly acquire the detection information of the detector 115b, the analysis devices 50 and 60 update the threshold values Th11 and Th12 as determination information by performing data analysis again (S7). Then, the analysis devices 50 and 60 transmit threshold values Th11 and Th12 as determination information to the abnormality determination device 123 via the fog networks 31 and 32. Then, the abnormality determination device 123 stores the threshold values Th11 and Th12 as determination information while sequentially updating them (S8).

  Meanwhile, the analysis devices 50 and 60 extract only a part of the acquired detection information of the detector 115b in parallel with the data analysis for generating the threshold values Th11 and Th12 as the determination information (S9). . For example, the analysis devices 50 and 60 extract the power P of the motor 115a when performing steady machining in FIG. The analysis devices 50 and 60 transmit the extracted information to the upper analysis device 70 via the cloud network 40. The transmission may be performed once a day or once a month, for example.

  Then, the upper analysis device 70 acquires a part of the detection information of the detector 115b via the cloud network 40 (S10). Further, the host analysis device 70 acquires various control parameters from the analysis devices 50 and 60 as necessary. Various control parameters have a smaller data amount than the detection information.

  The data communication amount in the cloud network 40 is much smaller than the data communication amount in the fog networks 31 and 32. Even if the analysis device 50, the other analysis device 60, and the higher-order analysis device 70 are located far away, the problem of delay in communication speed via the cloud network 40 does not occur.

  The host analysis device 70 performs host data analysis based on a part of the detection information acquired from the analysis devices 50 and 60 and various control parameters (S11). Upper data analysis is so-called data mining. The host analysis device 70 performs host data analysis using information on the production facilities 11 to 13 and the production facilities 21 to 23 installed in different areas. Therefore, the host analysis device 70 can perform host data analysis using a large amount of information.

  If the installation locations of the production facilities 11 to 13 and the other production facilities 21 to 23 are different, the environmental temperatures of the two may be different. For example, the higher-order analysis device 70 can perform higher-order data analysis on the influence of the environmental temperature in more detail.

  The analysis devices 50 and 60 can acquire the result of the upper data analysis by the upper analysis device 70 via the cloud network 40. Accordingly, the analysis devices 50 and 60 update the threshold values Th11 and Th12 as determination information generated by their own data analysis with reference to the result of the upper data analysis (S7). Then, the analysis devices 50 and 60 transmit the threshold values Th11 and Th12 as updated determination information to the abnormality determination device 123 via the fog networks 31 and 32. In this way, the abnormality determination device 123 stores the threshold values Th11 and Th12 as determination information considering the result of the upper data analysis while sequentially updating them (S8).

  Here, the display unit 52 of the analysis device 50 displays the determination information as a result of the data analysis of the analysis device 50 itself, and displays the determination information as a result of the upper data analysis of the upper analysis device 70 in an overlapping manner. Can do. The operator can set the determination information used in the production facilities 11 to 13 through the input unit 53 while confirming the determination information of both. The same applies to the analysis device 60.

  The analysis devices 50 and 60 can transmit part or all of the collected data to the higher-level analysis device 70. The range of data transmitted by the analysis devices 50 and 60 (the range set by the value size, time, etc.) of the production facilities 11 to 13 is determined in cooperation with the analysis devices 50 and 60 themselves or the operation of the operator. It can be decided in the vicinity.

<2. Second embodiment>
Detailed processing of the abnormality determination device 123, the analysis devices 50 and 60, and the higher-order analysis device 70 in the second embodiment will be described with reference to FIGS. In the second embodiment, a case where the detection information of the detector 112b is used will be described as an example.

  As shown in FIG. 7, the detector 112b detects the vibration of the main shaft every time the production object (workpiece) is ground (S21). Subsequently, the abnormality determination device 123 collects data for one production object (S22). Subsequently, the abnormality determination device 123 performs frequency analysis (corresponding to the predetermined process of the present invention) on the vibration data for one production object (S23). The result of the frequency analysis is shown in FIG. And the abnormality determination apparatus 123 acquires the peak value (equivalent to the post-processing data of this invention) of the predetermined frequency band of the vibration data obtained by frequency analysis.

  Here, as shown in FIG. 8, the result of the frequency analysis has a peak value (encircled in FIG. 8) in a plurality of frequency bands. These frequency bands correspond to the cause of vibration of the main shaft. For example, when the outer ring of the bearing of the spindle device 112 is damaged, the inner ring is damaged, the rolling element is damaged, the frequency band is different. Therefore, the abnormality determination device 123 acquires the peak value of the frequency band corresponding to each cause of vibration.

  If the threshold values Th21 and Th22 are already stored as the determination information pattern, the abnormality determination device 123 performs abnormality determination (S24). The threshold values Th21 and Th22 as determination information patterns are, for example, patterns of peak values (evaluation parameters) of frequency analysis of vibration data with respect to a time zone (specified parameters) for one day, as shown in FIG. Here, even in one day, the magnitude of vibration varies depending on the elapsed time after starting the production facilities 11 to 13, the environmental temperature, and the like. Therefore, as shown in FIG. 9, the threshold values Th21 and Th22 as the determination information pattern are set such that the horizontal axis is the time zone of one day (specified parameter), and the vertical axis is the peak value (evaluation parameter) of frequency analysis of vibration data. Represent as

  That is, the abnormality determination device 123 determines whether the abnormality is based on the currently acquired actual time zone (specified parameter), the currently acquired actual peak value (evaluation parameter), and the stored determination information pattern. Judgment is made. Here, in FIG. 9, mark ■ is the peak value for the actual time zone that is currently acquired. The mark ■ is equal to or lower than the upper limit threshold Th21 and is equal to or higher than the lower limit threshold Th22, so that it is determined to be normal.

  Further, in the abnormality determination device 123, the threshold values Th31 and Th32 as other determination information patterns are, for example, as shown in FIG. 10, the peak value (evaluation of frequency analysis of vibration data for one year (prescribed parameter)). Parameter) pattern is stored. Here, even in one year, the magnitude of vibration differs due to the influence of the difference in environmental temperature. Therefore, as shown in FIG. 10, the threshold values Th31 and Th32 as other determination information patterns have a horizontal axis of one year (regulated parameter) and a vertical axis of a peak value (evaluation parameter) of frequency analysis of vibration data. ).

  That is, the abnormality determination device 123 determines the abnormality based on the currently acquired actual time (specified parameter), the currently acquired actual peak value (evaluation parameter), and the stored determination information pattern. Make a decision. Here, in FIG. 10, the mark ▲ is the peak value for the actual time currently acquired. The mark ▲ is determined to be normal because it is equal to or lower than the upper limit threshold Th31 and equal to or higher than the lower limit threshold Th32.

  Furthermore, the abnormality determination device 123 collects the peak values (processed data) of a plurality of production objects (S25). The abnormality determination device 123 collects, for example, the peak value of the production target for one day. The peak values of a plurality of production objects collected by the abnormality determination device 123 are transmitted to the analysis devices 50 and 60 via the fog networks 31 and 32, for example, once a day. And the analysis apparatuses 50 and 60 acquire the peak value of the frequency analysis of vibration data about the production target object for several once, for example once a day (S26). Here, the analysis devices 50 and 60 obtain a peak value with a much smaller data amount than the detection information of the detector 112b.

  The analysis devices 50 and 60 perform data analysis based on the peak values for a plurality of production objects (S27). For example, the distribution of peak values for two days is as shown in FIG. And the analyzers 50 and 60 analyze a normal tendency pattern from the peak value for several days. The normal tendency pattern may be an approximate curve of distributed data (for example, a least-square approximate curve), or a curve having a width including all of the distributed data. Then, the analysis devices 50 and 60 generate threshold values Th21 and Th22 as patterns of determination information, as indicated by broken lines in FIG. 11, based on the normal tendency pattern (S28).

  Further, when the analysis devices 50 and 60 newly acquire the detection information of the detector 112b, the analysis devices 50 and 60 update the threshold values Th21 and Th22 as the determination information pattern by performing data analysis again (S28). Then, the analysis devices 50 and 60 transmit threshold values Th21 and Th22 as determination information patterns to the abnormality determination device 123 via the fog networks 31 and 32. Then, the abnormality determination device 123 stores the threshold values Th21 and Th22 as the determination information pattern while sequentially updating them (S29).

  The analysis devices 50 and 60 perform data analysis based on the peak value for the production target for one year (S27). For example, the distribution of peak values for one year is as shown in FIG. And the analyzers 50 and 60 analyze a normal tendency pattern from the peak value for one year. Based on the normal tendency pattern, the analysis devices 50 and 60 generate threshold values Th31 and Th32 as determination information patterns, as indicated by the broken lines in FIG. 12 (S28).

  Similarly, in this case, when the detection devices 50 and 60 newly acquire the detection information of the detector 112b, the data Th is analyzed again to update the threshold values Th31 and Th32 as the determination information pattern ( S28). Then, the analysis devices 50 and 60 transmit the threshold values Th31 and Th32 as determination information patterns to the abnormality determination device 123 via the fog networks 31 and 32. Then, the abnormality determination device 123 stores the threshold values Th31 and Th32 as the determination information pattern while sequentially updating them (S29).

  By the way, in parallel with the data analysis for generating the threshold values Th21, Th22, Th31, Th32 as the determination information, the analysis devices 50, 60 extract only a part of the acquired peak value (S30). . For example, the analysis devices 50 and 60 extract the peak values of some production objects instead of the peak values of all production objects. For example, the analysis devices 50 and 60 extract the peak value of one production object from the same lot. The analysis devices 50 and 60 transmit the extracted information to the upper analysis device 70 via the cloud network 40. The transmission may be performed once a week or once a month, for example.

  Then, the higher-level analysis device 70 acquires part of the peak value information via the cloud network 40 (S31). Further, the host analysis device 70 acquires various control parameters from the analysis devices 50 and 60 as necessary. Various control parameters have a smaller data amount than the detection information.

  The host analysis device 70 performs host data analysis based on part of the peak values acquired from the analysis devices 50 and 60 and various control parameters (S32). Upper data analysis is so-called data mining. The host analysis device 70 performs host data analysis using information on the production facilities 11 to 13 and the production facilities 21 to 23 installed in different areas. Therefore, the host analysis device 70 can perform host data analysis using a large amount of information.

  The analysis devices 50 and 60 can acquire the result of the upper data analysis by the upper analysis device 70 via the cloud network 40. Accordingly, the analysis devices 50 and 60 update the threshold values Th21, Th22, Th31, and Th32 as the patterns of determination information generated by their own data analysis with reference to the result of the upper data analysis (S28). Then, the analysis devices 50 and 60 transmit the threshold values Th21, Th22, Th31, and Th32 as updated determination information to the abnormality determination device 123 via the fog networks 31 and 32. In this manner, the abnormality determination device 123 stores the threshold values Th21, Th22, Th31, and Th32 as determination information considering the result of the upper data analysis while sequentially updating them (S29).

<3. Third Embodiment>
Detailed processing of the abnormality determination device 123, the analysis devices 50 and 60, and the higher-order analysis device 70 in the third embodiment will be described with reference to FIGS. In the third embodiment, a case where the detection information of the detectors 115b and 117 is used will be described as an example.

  As shown in FIG. 13, the detector 115b detects the current value of the power of the motor 115a every time the production object (workpiece) is ground (S41). The detector 117 detects the environmental temperature every time the production object is ground (S42). Subsequently, the abnormality determination device 123 collects each data for one production target (S43).

  Subsequently, the abnormality determination device 123 extracts data when performing steady machining from the power data of the motor 115a for one production object (corresponding to the predetermined processing of the present invention) ( S44). For example, in FIG. 4, the current value of the power of the motor 115 a when performing steady machining is P. Then, the abnormality determination device 123 acquires data on the power current value P obtained by the extraction process and data on the environmental temperature (corresponding to post-processing data of the present invention).

  If the threshold values Th41 and Th42 are already stored as the determination information pattern, the abnormality determination device 123 performs abnormality determination (S45). The threshold values Th41 and Th42 as the determination information patterns are, for example, patterns of current values (evaluation parameters) of the power of the motor 115a with respect to the environmental temperature (specified parameters), as shown in FIG. Here, the current value of the power of the motor 115a changes depending on the environmental temperature. Therefore, as shown in FIG. 14, the threshold values Th41 and Th42 as the determination information pattern are represented by the horizontal axis as the environmental temperature (specified parameter) and the vertical axis as the current value P (evaluation parameter) of the power of the motor 115a.

  That is, the abnormality determination device 123 is based on the actual environmental temperature (specified parameter) acquired at present, the current value P (evaluation parameter) of the actual power acquired at present, and the stored determination information pattern. To determine abnormality. Here, in FIG. 14, the mark ■ is the current value P of the power with respect to the actual environmental temperature acquired at present. The mark ■ is equal to or lower than the upper threshold Th41 and is equal to or higher than the lower threshold Th42.

  Furthermore, the abnormality determination device 123 collects a plurality of motive power current value P data and environmental temperature data (post-processing data) (S46). The abnormality determination device 123 collects, for example, data on the current value P of the power of the production target and environmental temperature data for one day. A plurality of data collected by the abnormality determination device 123 is transmitted to the analysis devices 50 and 60 via the fog networks 31 and 32, for example, once a day. Then, for example, once a day, the analysis devices 50 and 60 acquire data on the power current value P and data on the environmental temperature for a plurality of production objects (S47). Here, the analysis devices 50 and 60 obtain data having a data amount far smaller than that of all the detection information of the detectors 115b and 117.

  The analysis devices 50 and 60 perform data analysis based on data on a plurality of production objects (S48). For example, the distribution of data for a plurality of days with different environmental temperatures is as shown in FIG. And the analyzers 50 and 60 analyze a normal tendency pattern from the data for several days. Based on the normal tendency pattern, the analysis devices 50 and 60 generate threshold values Th41 and Th42 as determination information patterns as indicated by broken lines in FIGS. 14 and 15 (S49).

  Furthermore, when the analysis devices 50 and 60 newly acquire the detection information of the detectors 115b and 117, the thresholds Th41 and Th42 as the determination information pattern are updated by performing data analysis again (S49). . Then, the analysis devices 50 and 60 transmit threshold values Th41 and Th42 as determination information patterns to the abnormality determination device 123 via the fog networks 31 and 32. Then, the abnormality determination device 123 stores the threshold values Th41 and Th42 as the determination information pattern while sequentially updating them (S50).

  By the way, in parallel with the data analysis for generating the threshold values Th41 and Th42 as the determination information, the analysis devices 50 and 60 extract only a part of the acquired data (S51). For example, the analysis devices 50 and 60 extract data of some production objects instead of data of all production objects. For example, the analysis devices 50 and 60 extract data of one production object from the same lot. The analysis devices 50 and 60 transmit the extracted information to the upper analysis device 70 via the cloud network 40. The transmission may be performed once a week or once a month, for example.

  Then, the higher-level analysis device 70 acquires partial information of data via the cloud network 40 (S52). Further, the host analysis device 70 acquires various control parameters from the analysis devices 50 and 60 as necessary. Various control parameters have a smaller data amount than the detection information.

  The host analysis device 70 performs host data analysis based on part of the data acquired from the analysis devices 50 and 60 and various control parameters (S53). Upper data analysis is so-called data mining. The host analysis device 70 performs host data analysis using information on the production facilities 11 to 13 and the production facilities 21 to 23 installed in different areas. Therefore, the host analysis device 70 can perform host data analysis using a large amount of information.

  The analysis devices 50 and 60 can acquire the result of the upper data analysis by the upper analysis device 70 via the cloud network 40. Accordingly, the analysis devices 50 and 60 update the threshold values Th41 and Th42 as the patterns of determination information generated by their own data analysis with reference to the result of the upper data analysis (S49). Then, the analysis devices 50 and 60 transmit the threshold values Th41 and Th42 as updated determination information to the abnormality determination device 123 via the fog networks 31 and 32. In this way, the abnormality determination device 123 stores the threshold values Th41 and Th42 as determination information considering the result of the higher-order data analysis while sequentially updating them (S50).

<4. Effects of the embodiment>
In the first to third embodiments, the abnormality analysis system 1 is a facility for producing a production target, and includes one or a plurality of detectors 112b, 114c, 114d, 115b, 116, 117, and 118c. A plurality of production facilities 11 to 13, a fog network (corresponding to the first network) 31 connected to the plurality of production facilities 11 to 13 and installed in a predetermined area for constructing fog computing, and 31 connections to the fog network The data analysis is performed based on the detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c acquired through the fog network 31, and the plurality of production facilities 11 are based on the data analysis result. Analysis for generating determination information on each abnormality of -13 or abnormality of production object And a location 50. Each of the plurality of production facilities 11 to 13 determines the abnormality of each of the plurality of production facilities 11 to 13 or the abnormality of the production target based on the determination information generated by the analysis device 50.

  The detectors 112b, 114c, 114d, 115b, 116, 117, 118c in the plurality of production facilities 11 to 13 and the analysis device 50 are connected via a fog network 31 installed in a predetermined area for constructing fog computing. Has been. Fog computing is a network-connected system in a narrow area compared to cloud computing. That is, the fog network 31 for constructing fog computing is a network installed in a predetermined area narrower than the area for constructing cloud computing. Therefore, the occurrence of communication congestion is suppressed in data communication between the detectors 112b, 114c, 114d, 115b, 116, 117, 118c and the analysis device 50. Moreover, since the fog network 31 is constructed in a narrow predetermined area, the communication time between the production facilities 11 to 13 and the analysis device 50 can be shortened. Therefore, the analysis device 50 can receive detection information from the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c at high speed.

  Since the analysis apparatus 50 can acquire detection information in the plurality of production facilities 11 to 13 at an early stage and can perform data analysis, the result of the analysis apparatus 50 can be fed back to the production facilities 11 to 13 at an early stage. Since the analysis result can be fed back to the production facilities 11 to 13 at an early stage, the occurrence of an abnormality in the production target can be suppressed more quickly and reliably.

  In the first embodiment, the analysis device 50 acquires all the detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c via the fog network 31, and based on all the detection information. Data analysis. In particular, the plurality of detectors 112b, 114c, 114d, 115b, 116, 117, and 118c acquire detection information at different sampling periods, and the analysis device 50 includes the plurality of detectors 112b, 114c, 114d, 115b, and 116. , 117, and 118c, all of the detection information is acquired, and data analysis is performed based on all of the detection information. Even if a large amount of data communication is performed on the fog network 31, the problem of communication delay does not occur. Therefore, the analysis device 50 acquires all the detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c. Therefore, the analysis device 50 can perform data analysis with high accuracy and in real time.

  In the second embodiment and the third embodiment, the abnormality determination device 123 performs predetermined processing on the detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c to generate post-processing data. At the same time, the abnormality is determined based on the determination information. Then, the analysis devices 50 and 60 obtain the processed data via the fog network 31, perform data analysis based on the processed data, and update the determination information based on the data analysis result.

  The abnormality determination device 123 performs abnormality determination, and the analysis devices 50 and 60 update the determination information. Here, the analysis devices 50 and 60 use post-processing data obtained by performing predetermined processing on the detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c. That is, the analysis devices 50 and 60 do not update the determination information based on all the detection information itself of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c. Therefore, the analysis devices 50 and 60 can process the determination information at a higher speed than when all the detection information itself is used. As described above, the abnormality analysis system 1 can reliably update the determination information while performing abnormality determination.

  In particular, the data capacity of the processed data that has been subjected to the predetermined processing by the abnormality determination device 123 is made smaller than the detection information before the processing. Therefore, since the traffic of the fog network 31 can be reduced, the analysis devices 50 and 60 can shorten the time for acquiring data for one day, for example. As a result, the analysis devices 50 and 60 can secure a sufficient time for the analysis.

  In the second embodiment, the abnormality determination device 123 performs abnormality determination based on the generated post-processing data and determination information, and the analysis devices 50 and 60 perform post-processing used by the abnormality determination device 123 for determination. The judgment information is updated based on the data. That is, the post-processing data is shared by the abnormality determination device 123 and the analysis devices 50 and 60.

  In particular, in the second embodiment, the detector 112b is a vibration detection sensor, and the predetermined processing by the abnormality determination device 123 is frequency analysis for the detection information of the detector 112b. Therefore, the abnormality determination device 123 does not generate dedicated data for use in the analysis devices 50 and 60, but only generates data used by itself. Therefore, the abnormality determination device 123 does not require a dedicated process, and thus the communication amount of the fog network 31 can be reduced while speeding up the processing of the abnormality determination device 123 itself.

  In the third embodiment, the predetermined processing by the abnormality determination device 123 is processing for extracting specific information from the detection information of the detectors 115b and 117. Also in this case, the abnormality determination device 123 does not generate dedicated data for use in the analysis devices 50 and 60, but only generates data used by itself. Therefore, the abnormality determination device 123 does not require a dedicated process, and thus the communication amount of the fog network 31 can be reduced while speeding up the processing of the abnormality determination device 123 itself.

  In the second embodiment and the third embodiment, the analysis devices 50 and 60 analyze the normal tendency pattern for the evaluation parameter with respect to the prescribed parameter by data analysis, and evaluate the evaluation parameter for the prescribed parameter based on the normal tendency pattern. The pattern of the judgment information about is updated. Then, the abnormality determination device 123 acquires an actual specified parameter and an actual evaluation parameter, and determines an abnormality based on the determination information pattern, the actual specified parameter, and the actual evaluation parameter.

  For example, in the second embodiment, as a first example, the specified parameter is a time zone in one day, and the evaluation parameter is a parameter that changes according to the time zone in one day. In the second embodiment, as a second example, the specified parameter is a time in one year, and the evaluation parameter is a parameter that changes according to the time in one year.

  The state of the components of the production facilities 11 to 13 or the state of the production object varies depending on, for example, the elapsed time after starting the production facilities 11 to 13 or the environmental temperature. The ambient temperature varies depending on the time of day or the time of year. Moreover, the elapsed time after starting the production facilities 11-13 changes with the time slot | zone in one day, when starting once per day. Therefore, by setting the specified parameters and the evaluation parameters as described above, the state of the production facilities 11 to 13 or the state of the production object can be reliably evaluated.

  In particular, the detector 112b detects vibrations of the production facilities 11 to 13 or the production object, and the evaluation parameter is a peak value of a predetermined frequency band in the vibration. The amplitude of the vibration is a parameter that varies depending on, for example, the elapsed time after starting the production facilities 11 to 13 or the environmental temperature. That is, the peak value is a parameter that changes depending on the elapsed time, the environmental temperature, and the like since the production facilities 11 to 13 are started. Therefore, by setting the evaluation parameter to the peak value, the state of the production facilities 11 to 13 or the state of the production object can be reliably evaluated.

  In the third embodiment, the specified parameter is the environmental temperature, and the evaluation parameter is a parameter that changes according to the environmental temperature. In this case, the state of the production facilities 11 to 13 or the state of the production object can be evaluated by evaluating the parameter that changes according to the environmental temperature by using the environmental temperature itself as the specified parameter.

  In the second embodiment and the third embodiment, the analysis devices 50 and 60 collectively acquire the processing results for the plurality of times by the abnormality determination device 123 after the predetermined processing by the abnormality determination device 123 is performed a plurality of times. ing. That is, the analysis devices 50 and 60 do not acquire data from the abnormality determination device 123 every time the abnormality determination device 123 acquires detection information from the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c. .

  Here, in the second embodiment and the third embodiment, the abnormality determination device 123 performs predetermined processing on the detection information, and the analysis devices 50 and 60 perform processing in which the data amount is reduced by the predetermined processing. After data is acquired. Therefore, even if the analysis devices 50 and 60 collectively acquire a plurality of results, the communication amount of the fog network 31 can be sufficiently small.

  In the first embodiment to the third embodiment, the predetermined area in which the fog network 31 is constructed is in the same building as the building in which any of the plurality of production facilities 11 to 13 is installed, or the production facility. It is in the building where 11-13 is installed and the neighboring building. Therefore, the production facilities 11 to 13 and the analysis device 50 can be reliably configured by the fog network 31.

  Further, in the first to third embodiments, since the analysis is performed near the production facilities 11 to 13, the operator can check for abnormalities and normality while checking the state of the production object and the production facilities 11 to 13. It is possible to determine a value to be determined (determination information). In addition, when a sudden abnormality occurs in the production facilities 11 to 13 or the production target, the analysis is performed in the vicinity of the production facilities 11 to 13, so that the data is analyzed by the cooperation of the operator and the analysis device 50. Can be performed immediately, and the result can be immediately reflected in the determination information for the target production facilities 11-13. Further, according to the analysis result of the analysis device 50, in the previous stage of abnormality determination or abnormality determination (a state that is not abnormal but is close to abnormality), the production facilities 11 to 13 or the analysis device 50 inform the operator of the abnormal state or automatically The operation of the production facilities 11 to 13 can be stopped.

  Moreover, in 1st embodiment to 3rd embodiment, the analysis apparatus 50 is provided with the display part 52 which displays the result of a data analysis, and the input part 53 which receives the input of the determination information by an operator. The determination information for the production facilities 11 to 13 can be manually set by the operator. The setting is not limited to manual setting by the operator, and automatic setting by the system is also possible.

  Further, in the first to third embodiments, the abnormality analysis system 1 is not directly connected to the fog network 31 and includes other detectors 112b, 114c, 114d, 115b, 116, 117, and 118c. Production network 21 to 23, a plurality of production facilities 11 to 13 and other production facilities 21 to 23, and a cloud network 40 (in the second network) for constructing cloud computing in a wider area than a predetermined area of the fog network 31 The detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, 118c and the other detectors 112b, 114c, 114d, 115b, which are connected to the cloud network 40 and acquired via the cloud network 40, Based on detection information of 116, 117, 118c And a higher-level analysis device 70 for upper data analysis.

  The analysis device 50 can also determine and store determination information based on the result of data analysis by the analysis device 50 and the result of higher-order data analysis by the higher-order analysis device 70. Therefore, better determination information can be obtained by performing higher-order data analysis using information that cannot be obtained by the production facilities 11 to 13 and feeding back to the production facilities 11 to 13.

  In the first to third embodiments, the analysis device 50 acquires all the detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c, and performs data analysis. On the other hand, the host analysis device 70 includes a part of the detection information of the detectors 112b, 114c, 114d, 115b, 116, 117, 118c of the production facilities 11-13 and the other detectors 112b of the other production facilities 21-23. 114c, 114d, 115b, 116, 117, 118c A part of the detection information is acquired, and higher-order data analysis is performed. Even if the high-order analysis device 70 is not required to perform high-speed processing, all of the detectors 112b, 114c, 114d, 115b, 116, 117, and 118c are temporarily transferred to the high-order analysis device 70 via the cloud network 40. If it is transmitted, there is a possibility that the influence of the communication delay of the cloud network 40 is given to others. Therefore, as described above, the amount of data communication of the cloud network 40 becomes part of the detection information, and the influence of communication delay in the cloud network 40 can be suppressed.

  In the first embodiment to the third embodiment, the plurality of production facilities 11 to 13 include a grinding machine that grinds the production target, and the determination information is, for example, determination information related to grinding abnormality of the production target. . Thereby, in a system including a grinding machine, occurrence of grinding abnormality such as grinding burn can be reliably suppressed.

  Moreover, in 1st embodiment to 3rd embodiment, determination information is good also as determination information regarding the component failure in any of the some production facilities 11-13, component lifetime, or the necessity for maintenance of components. Thereby, it is possible to predict a component failure in the production facilities 11 to 13 and to prepare a replacement part in advance. In the past, parts were often replaced based on the period of use of the parts. However, it is possible to replace parts after grasping the life of each part with a higher degree of accuracy. be able to. In addition, before the performance of the parts deteriorates, the parts can be maintained at an appropriate time. Thereby, the use period of components can be lengthened.

1: Abnormality analysis system, 11-13: Production equipment (11: Grinding machine), 21-23: Other production equipment, 31: Fog network (first network), 32: Fog network, 40: Cloud network (second) Network), 50: analysis device, 51: analysis unit, 52: display unit, 53: input unit, 60: other analysis device, 70: host analysis device, 112b, 114c, 114d, 115b, 116, 117, 118c: Detector, 121: CNC device, 122: PLC, 123: Abnormality judgment device, Th11, Th12, Th21, Th22, Th31, Th32: Threshold value, W: Crankshaft (production object)

Claims (20)

A facility for producing a production object, a plurality of production facilities each including one or a plurality of detectors,
A first network connected to the plurality of production facilities and installed in a predetermined area for constructing fog computing;
Connected to the first network, performs data analysis based on the detection information of the detector acquired via the first network, and each abnormality of the plurality of production facilities based on the result of the data analysis Or an analysis device for generating determination information regarding abnormality of the production object;
With
Each of the plurality of production facilities includes an abnormality determination device that determines each abnormality of the plurality of production facilities or abnormality of the production object based on the determination information generated by the analysis device. Analysis system. The analysis device includes:
All the detection information of the detector is obtained via the first network;
The abnormality analysis system according to claim 1, wherein the data analysis is performed based on all of the detection information. The plurality of detectors obtain detection information at different sampling periods,
The analysis device includes:
Obtaining all of the detection information in each of the plurality of detectors;
The abnormality analysis system according to claim 2, wherein the data analysis is performed based on all of the detection information. The abnormality determination device performs a predetermined process on the detection information of the detector to generate post-processing data, performs an abnormality determination based on the determination information,
The analysis device acquires the processed data via the first network, performs data analysis based on the processed data, and updates the determination information based on a result of the data analysis. Anomaly analysis system described in 1. The abnormality determination device performs an abnormality determination based on the generated post-processing data and the determination information,
The analysis device updates the determination information based on the post-processing data used for determination by the abnormality determination device,
The abnormality analysis system according to claim 4, wherein the post-processing data is shared by the abnormality determination device and the analysis device. The detector is a vibration detection sensor;
The abnormality analysis system according to claim 5, wherein the predetermined process is frequency analysis for detection information of the detector.   The abnormality analysis system according to claim 5, wherein the predetermined process is a process of extracting specific information from detection information of the detector. The analysis device includes:
By the data analysis, a normal tendency pattern for the evaluation parameter with respect to the specified parameter is analyzed,
Based on the normal tendency pattern, update the pattern of the determination information for the evaluation parameter with respect to the specified parameter,
The abnormality determination device is
Obtain the actual specified parameter and the actual evaluation parameter,
The abnormality analysis system according to any one of claims 4 to 7, wherein abnormality is determined based on the pattern of the determination information, the actual specified parameter, and the actual evaluation parameter. The prescribed parameter is a time zone in a day,
The abnormality analysis system according to claim 8, wherein the evaluation parameter is a parameter that changes according to a time zone in one day. The prescribed parameter is a time in a year,
The abnormality analysis system according to claim 8, wherein the evaluation parameter is a parameter that changes according to a time in a year. The detector detects vibrations of the production facility or the production object,
The abnormality analysis system according to claim 9 or 10, wherein the evaluation parameter is a peak value of a predetermined frequency band in the vibration. The specified parameter is an environmental temperature,
The abnormality analysis system according to claim 8, wherein the evaluation parameter is a parameter that changes according to the environmental temperature.   The analysis device according to any one of claims 4 to 12, wherein after the predetermined processing by the abnormality determination device is performed a plurality of times, a plurality of processing results by the abnormality determination device are collectively acquired. Anomaly analysis system.   The predetermined region is in a building that is the same as a building in which any of the plurality of production facilities is installed, or in a building that is adjacent to a building in which the production facilities are installed. The abnormality analysis system according to any one of the above. The analysis device includes:
A display unit for displaying the result of the data analysis;
An input unit for receiving input of the determination information by an operator;
The abnormality analysis system according to any one of claims 1 to 14, further comprising: The abnormality analysis system is
Other production equipment not directly connected to the first network and comprising other detectors;
A second network that is connected to the plurality of production facilities and the other production facilities and constructs cloud computing in a wider area than the predetermined area;
A host analyzer connected to the second network and performing host data analysis based on the detection information of the detector and the detection information of the other detectors acquired via the second network;
With
The said analysis apparatus produces | generates the said determination information based on the result of the said data analysis by the said analysis apparatus, and the result of the said high-order data analysis by the said high-order analysis apparatus, As described in any one of Claims 1-15 Anomaly analysis system. The analysis device acquires all the detection information of the detector, and performs the data analysis,
The abnormality analysis system according to claim 16, wherein the upper analysis apparatus acquires a part of detection information of the detector and a part of detection information of the other detector, and performs the upper data analysis. The plurality of production facilities include a grinding machine for grinding the production object,
The abnormality determination system according to any one of claims 1 to 17, wherein the determination information is determination information related to grinding abnormality of the production object.   The abnormality analysis system according to any one of claims 1 to 18, wherein the determination information is determination information regarding a component failure, component life, or necessity of component maintenance of any of the plurality of production facilities.   The said analysis apparatus used for the abnormality analysis system as described in any one of Claims 1-19.

Images