Abnormality detection device, abnormality detection method and computer program product

By installing a vibration sensor on the ball screw of the injection molding machine and performing frequency analysis, the problem of difficulty in accurately detecting ball screw abnormalities is solved by simply monitoring the vibration strength, and the accurate identification of ball screw abnormalities and the determination of fault status is achieved.

CN114845853BActive Publication Date: 2024-12-06THE JAPAN STEEL WORKS LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202080076151.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2019-11-06
Filing Date
2020-10-30
Publication Date
2024-12-06
Estimated Expiration
2040-10-30

AI Technical Summary

Technical Problem

In injection molding machines, the vibration waveform of the ball screw is complex, and it is difficult to detect abnormalities of the ball screw with good accuracy by simply monitoring the vibration strength.

Method used

By installing a vibration sensor on the ball screw, detecting and performing frequency analysis, the frequency component of the vibration intensity when abnormal is compared with the frequency component of the normal is detected to detect abnormality in the ball screw.

Benefits of technology

Accurate detection of ball screw abnormalities is achieved, detection accuracy is improved, and vibration abnormalities and fault status can be better identified.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114845853B_ABST
    Figure CN114845853B_ABST
Patent Text Reader

Abstract

The present invention provides an abnormality detection device for detecting abnormalities of a ball screw installed on an injection molding machine, comprising: a vibration sensor, which detects the vibration of the ball screw; and a calculation unit, which detects the abnormality of the ball screw based on information on the distribution of frequency components obtained by frequency analysis of the vibration intensity detected by the vibration sensor and information on the distribution of frequency components obtained by frequency analysis of the vibration intensity detected during normal operation of the ball screw.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to an abnormality detection device, an abnormality detection method and a computer program product for detecting an abnormality of a ball screw provided in an injection molding machine. Background Art

[0002] Patent Document 1 discloses a monitoring method that monitors the vibration and operating sound of each movable part of an injection molding machine using an acceleration sensor and a microphone to detect the state of a molding process and the occurrence of abnormality in the movable part.

[0003] Patent Document 2 discloses a load state display device that calculates an average load based on the instantaneous current and instantaneous voltage of each motor of an injection molding machine to detect load abnormality in a drive system and issue an alarm.

[0004] Prior art literature

[0005] Patent Literature

[0006] Patent Document 1: Japanese Patent Application Laid-Open No. 5-50480

[0007] Patent Document 2: Japanese Patent Application Laid-Open No. 9-262889 Summary of the invention

[0008] Problems to be solved by the invention

[0009] However, there is a problem that the vibration waveform generated in the ball screw installed in the injection molding machine is complex, and simply monitoring the vibration intensity cannot accurately detect abnormalities in the ball screw.

[0010] Patent Document 1 does not disclose a specific analysis method or abnormality detection method for acceleration data.

[0011] An object of the present invention is to provide an abnormality detection device, an abnormality detection method, and a computer program product capable of detecting abnormality of a ball screw by performing frequency analysis on the intensity of vibration generated in the ball screw.

[0012] Means for solving problems

[0013] The abnormality detection device of the present invention detects the abnormality of a ball screw installed on an injection molding machine, and includes: a vibration sensor, which detects the vibration of the ball screw; and a calculation unit, which detects the abnormality of the ball screw based on the distribution information of the frequency components obtained by performing a frequency analysis on the vibration intensity detected by the vibration sensor and the distribution information of the frequency components obtained by performing a frequency analysis on the vibration intensity detected during normal operation of the ball screw.

[0014] The abnormality detection method of the present invention detects the abnormality of a ball screw installed on an injection molding machine, including: detecting the vibration of the ball screw; performing frequency analysis on the detected vibration intensity; detecting the abnormality of the ball screw based on the distribution information of the frequency components obtained by the frequency analysis and the distribution information of the frequency components obtained by performing frequency analysis on the vibration intensity detected during normal operation of the ball screw.

[0015] The computer program product of the present invention includes a computer program for causing a computer to perform a process for detecting an abnormality of a ball screw provided on an injection molding machine. When the computer program is run on a computer, the computer is caused to perform the following processes: detecting the vibration of the ball screw; performing a frequency analysis on the detected vibration intensity; and detecting the abnormality of the ball screw based on information on the distribution of frequency components obtained by frequency analysis and information on the distribution of frequency components obtained by performing a frequency analysis on the vibration intensity detected during normal operation of the ball screw.

[0016] Effects of the Invention

[0017] According to the present invention, abnormality of the ball screw can be detected by performing frequency analysis on the intensity of vibration generated in the ball screw. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a partial cross-sectional front view showing a structural example of the injection molding machine according to the first embodiment.

[0019] Figure 2 This is a block diagram showing a configuration example of the abnormality detection device according to the first embodiment.

[0020] Figure 3 This is a flowchart showing the processing procedure of the vibration-based abnormality detection method.

[0021] Figure 4 This is a flowchart showing the processing procedure for reading measurement data.

[0022] Figure 5 This is a flowchart showing the processing procedure of analysis region clipping.

[0023] Figure 6 It is an explanatory diagram showing the clipping process of the analysis area.

[0024] Figure 7 This is a flowchart showing the processing procedure of frequency analysis.

[0025] Fig. 8A It is an explanatory diagram showing an example of a frequency analysis waveform.

[0026] Figure 8B It is an explanatory diagram showing an example of a frequency analysis waveform.

[0027] Fig. 9 This is a flowchart showing the processing procedure of operating frequency calculation.

[0028] Fig.10 This is a flowchart showing the processing procedure of feature amount calculation.

[0029] Fig.11A It is an explanatory diagram showing the designation process of the analysis area.

[0030] Fig. 11B It is an explanatory diagram showing the designation process of the analysis area.

[0031] Fig. 12A It is an explanatory diagram showing the center of gravity calculation process.

[0032] Fig. 12B It is an explanatory diagram showing the center of gravity calculation process.

[0033] Fig.13 This is a flowchart showing the processing procedure of the failure state determination process.

[0034] Fig.14 It is an explanatory diagram showing the fault state determination process.

[0035] Fig.15 It is an explanatory diagram showing the history of the failure state determination result.

[0036] Fig.16 This is a flowchart showing the processing procedure of the abnormality detection method based on the motor current.

[0037] Fig.17 This is a flowchart of the fundamental wave extraction process.

[0038] Fig.18 It is a flow chart of the normalization process.

[0039] Fig.19 It is an explanatory diagram showing a normalized frequency analysis waveform.

[0040] Fig. 20 This is a flowchart showing the processing procedure of the failure state determination process.

[0041] Fig.21A It is an explanatory diagram showing the fault state determination process.

[0042] Fig. 21B It is an explanatory diagram showing the fault state determination process.

[0043] Fig. 22 This is a functional block diagram of the calculation unit of the second embodiment. DETAILED DESCRIPTION

[0044] Specific examples of the abnormality detection device, abnormality detection method, and computer program product according to the embodiments of the present invention are described below with reference to the accompanying drawings. The present invention is not limited to these examples but is indicated by the claims, and includes all modifications within the meaning and scope equivalent to the claims.

[0045] Hereinafter, the present invention will be described in detail based on the drawings showing embodiments of the present invention.

[0046] (Implementation Method 1)

[0047] 〈Injection molding machine〉

[0048] Figure 1 1 is a partial cross-sectional front view showing a structural example of an injection molding machine according to Embodiment 1. The injection molding machine according to Embodiment 1 includes an injection device 1, a front portion ( Figure 1 The mold clamping device 50 (left side in the middle), a control device (not shown), and an abnormality detection device 7 of this embodiment 1. The operation of the injection device 1 and the mold clamping device 50 is controlled by the control device.

[0049] The injection device 1 includes an injection unit IU, a screw driving unit SDU that drives the screw 3 of the injection unit IU, and a nozzle contact device NT.

[0050] The injection unit IU includes a heating barrel 2 and a screw 3, which is arranged inside the heating barrel 2 and can be driven in the plasticizing direction and the injection direction. The outer periphery of the heating barrel 2 is provided with a plurality of heaters 4, 4, ... whose heating temperature is controlled. An injection nozzle 5 is provided at the front tip of the heating barrel 2. A hopper 6 for temporarily storing resin materials is provided at the rear of the heating barrel 2. The bolt groove shaft 3a extending backward along the axial direction from the rear end of the screw 3 extends outward from the rear end of the heating barrel 2 and is connected to the screw drive unit SDU.

[0051] The screw drive unit SDU includes an intermediate plate 10 for moving the screw 3 in the axial direction, an electric motor 11 for plasticization, and an electric motor 29 for injection.

[0052] The intermediate plate 10 is supported in a guideable manner in the axial direction of the screw 3 in a state where the rotation is restrained by a plurality of guide rods 16, 16 parallel to the axial direction of the screw 3. A bearing seat for axially supporting the rear end of the bolt groove shaft 3a is formed in the approximate center of the intermediate plate 10, and a thrust bearing 15 is mounted on the bearing seat. The rear end of the bolt groove shaft 3a is supported on the intermediate plate 10 in a freely rotatable manner by the thrust bearing 15 that bears the load in the radial and thrust directions. When the intermediate plate 10 is moved in the front-to-back direction ( Figure 1 When the screw rod 3 moves in the axial direction (in the middle lateral direction), that is, in the axial direction, the screw rod 3 is driven in the axial direction.

[0053] The intermediate plate 10 is connected to the ball screw 30 for injection. The ball screw 30 for injection is a mechanism for driving the screw 3 in the axial direction. The ball screw 30 for injection includes a screw shaft 31 for injection and a nut 32. The nut 32 that moves linearly along the screw shaft 31 is threadedly connected to the screw shaft 31 for injection. The nut 32 is mounted on the intermediate plate 10 through a nut fixing component 33.

[0054] The injection molding machine includes an injection transmission mechanism 20 for transmitting the driving force of the injection electric motor 29 to the ball screw 30. The injection transmission mechanism 20 is composed of an injection drive pulley 21, an injection driven pulley 22, and an injection timing belt 23, wherein the injection drive pulley 21 is provided on the output shaft of the injection electric motor 29, the injection driven pulley 22 is provided on the end of the injection screw shaft 31, and the injection timing belt 23 is sleeved on the injection drive pulley 21 and the injection driven pulley 22. The injection drive pulley 21, the injection driven pulley 22, and the injection timing belt 23 each have a plurality of teeth for transmitting the driving force of the injection electric motor 29 by meshing.

[0055] When the injection electric motor 29 outputs a driving force, the driving force is transmitted to the ball screw 30 through the injection drive pulley 21 and the injection driven pulley 22. The screw shaft 31 of the ball screw 30 rotates under the driving force. Since the rotation of the intermediate plate 10 is constrained by the guide rods 16, 16, the rotation of the nut 32 is also constrained. Therefore, when the screw shaft 31 is driven to rotate, the nut 32 moves in the axial direction, in other words, the intermediate plate 10 and the screw 3 move in the axial direction.

[0056] A plasticizing drive pulley 12 is provided on the output shaft of the plasticizing electric motor 11. A plasticizing driven pulley 13 is provided on the bolt groove shaft 3a, which is restricted in the rotation direction and freely movable in the axial direction. A plasticizing timing belt 14 is sleeved on the plasticizing drive pulley 12 and the plasticizing driven pulley 13. When the plasticizing electric motor 11 outputs a driving force, the screw 3 is driven to rotate around the axis.

[0057] In order to drive the injection nozzle 5 of the injection device 1 configured as described above in a direction of contact with a positioning ring of a fixed mold 54 described later and in a direction away from the positioning ring, a nozzle contact device NT is provided. The nozzle contact device NT is composed of a pair of support members 42, 42 fixedly provided at a predetermined interval on an injection base, a nozzle contact screw shaft 40 supported by the support members 42, 42 via bearings that bear thrust and radial loads, a nozzle contact nut 41 threadedly connected to the nozzle contact screw shaft 40, and a nozzle contact electric motor 39 that drives the nozzle contact screw shaft 40 in forward and reverse directions. The nozzle contact nut 41 is fixed to a frame of the injection device 1 (not shown). Since a pair of supporting parts 42, 42 are fixed on the injection base, when the nozzle contact electric motor 39 rotates in the forward and reverse directions, the nozzle contact screw shaft 40 is driven to rotate in the forward and reverse directions through the nozzle contact transmission mechanism 43, and the nozzle contact nut 41 and the injection device 1 are driven in the direction in which the injection nozzle 5 contacts the fixed mold 54 and in the direction away from the fixed mold 54.

[0058] The mold clamping device 50 includes a mold device MM to which the mold can be freely detachably mounted, and a toggle mechanism TM for clamping the mold. The mold device MM includes a fixed platen 51 to which a fixed mold 54 can be freely detachably mounted, and a movable platen 52 to which a movable mold 55 can also be freely detachably mounted. The toggle mechanism TM is disposed between the movable platen 52 and the mold clamping housing 53. The toggle mechanism TM can be moved in the forward and backward directions ( Figure 1 The fixed mold 54 and the movable mold 55 are opened and closed by moving in the horizontal direction (in the middle and horizontal directions), and when the molten resin injected from the injection device 1 is filled into the mold, the mold is clamped so that the fixed mold 54 and the movable mold 55 will not open.

[0059] <Abnormality detection device 7>

[0060] Figure 2 This is a block diagram showing a configuration example of the abnormality detection device 7 according to the first embodiment.

[0061] The abnormality detection device 7 is a computer, and includes a computing unit 71, a storage unit 72, an input interface (input I / F) 73, and an output interface (output I / F) 74. The computing unit 71 has computing circuits such as a CPU (Central Processing Unit), a multi-core CPU, a GPU (Graphics Processing Unit), a GPGPU (General-purpose computing on graphics processing units), a TPU (Tensor Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), and an NPU (Neural Processing Unit), internal storage devices such as a ROM (Read Only Memory) and a RAM (Random Access Memory), and I / O terminals. The computing unit 71 performs the functions of the abnormality detection device 7 of the first embodiment by executing a computer program 72a stored in the storage unit 72 described later. In addition, each functional unit of the abnormality detection device 7 may be implemented in the form of software, or part or all of it may be implemented in the form of hardware.

[0062] The storage unit 72 , the input interface 73 , and the output interface 74 are connected to the calculation unit 71 .

[0063] The storage unit 72 is a nonvolatile memory such as a hard disk, an EEPROM (Electrically Erasable Programmable ROM), or a flash memory. The storage unit 72 stores a computer program 72a for causing a computer to implement the abnormality detection method of the first embodiment, normal vibration information 72b, and normal current information 72c. The details of the normal vibration information 72b and the normal current information 72c will be described later.

[0064] The computer program 72a of the present embodiment 1 may also be in a form recorded in the recording medium 8 in a computer-readable manner. The storage unit 72 stores the computer program 72a read from the recording medium 8 by a reading device not shown in the figure. The recording medium 8 is a semiconductor memory such as a flash memory. In addition, the recording medium 8 may also be an optical disk such as a CD (Compact Disc)-ROM, a DVD (Digital Versatile Disc)-ROM, or a BD (Blu-ray (registered trademark) Disc). The recording medium 8 may also be a magnetic disk such as a floppy disk or a hard disk, a magneto-optical disk, or the like. In addition, the computer program 72a of the present embodiment 1 may also be downloaded from an external server not shown in the figure connected to a communication network not shown in the figure, and stored in the storage unit 72.

[0065] The input interface 73 is connected to the vibration sensor 7a, the current sensor 7b, and the operation unit 7c.

[0066] The vibration sensor 7a is a sensor that measures the vibration generated on the ball screw 30 for injection and outputs the measured vibration intensity data to the operation unit 71 of the abnormality detection device 7 in a wired or wireless manner. As the vibration sensor 7a, for example, an acceleration sensor, a velocity sensor, and a displacement sensor can be used. The following describes the case where the vibration sensor 7a is an acceleration sensor. The vibration sensor 7a as an acceleration sensor outputs acceleration data indicating the vibration intensity to the operation unit 71. The operation unit 71 receives the acceleration data through the input interface 73 and stores the received acceleration data as a vibration measurement data file in the storage unit 72.

[0067] The vibration sensor 7a is preferably mounted on the nut 32. For vibrations caused by abnormalities in the ball screw 30, installing the vibration sensor 7a on the nut 32 rather than on the screw shaft 31 can more effectively measure the vibrations caused by abnormalities in the ball screw 30.

[0068] The vibration sensor 7a can be installed on the side of the cylindrical nut 32 in a posture that can measure the axial acceleration (linear movement direction) of the screw shaft 31. Between the screw shaft 31 and the nut 32, the ball not shown in the figure circulates and rotates, and the direction of the force acting on the groove, ball and nut 32 of the screw shaft 31 is roughly the axial direction of the screw shaft 31. Therefore, the vibration caused by the abnormality of the ball screw 30 is greater. Therefore, by installing the vibration sensor 7a in the above posture and position, the axial acceleration can be measured. In addition, compared with being set on the circumferential surface of the nut 32, it is easier to install.

[0069] The upper limit of the response frequency of the vibration sensor 7a, that is, the frequency at which vibration can be measured, is preferably about 2.5 kHz or more. By performing frequency analysis on data obtained by measuring acceleration at 2.5 kHz as described later, at least the frequency components of 0 to 1000 Hz of acceleration can be analyzed.

[0070] Furthermore, abnormality of the ball screw 30 can be detected by analyzing a frequency component of about 50 to 600 Hz based on the characteristics of the injection device 1. In this case, a vibration sensor 7a having a measurable frequency upper limit of about 1.5 kHz or more may be used.

[0071] The normal vibration information 72b stored in the storage unit 72 is a frequency component distribution obtained by frequency analysis of the acceleration of the vibration detected by the vibration sensor 7a when the ball screw 30 is running normally (see Fig. 8A and Figure 8B ) information. Hereinafter, the waveform of the frequency component distribution obtained by frequency analysis is appropriately referred to as the frequency analysis waveform. This information is, for example, information indicating the center of gravity of the frequency analysis waveform on the frequency-acceleration plane calculated based on multiple data (frequency, acceleration) obtained from a normal ball screw 30 (refer to Fig. 12A and Fig. 12B ).

[0072] The current sensor 7b is a sensor that measures the driving alternating current flowing to the injection electric motor 29 and outputs the measured current data to the calculation unit 71 of the abnormality detection device 7 in a wired or wireless manner. The calculation unit 71 receives the current data through the input interface 73 and stores the received current data in the storage unit 72 as a current measurement data file.

[0073] The above-mentioned normal current information 72c stored in the storage unit 72 is information on the frequency component distribution obtained by frequency analysis of the AC current detected by the current sensor 7b when the ball screw 30 is operating normally. Hereinafter, the waveform of the frequency component distribution obtained by frequency analysis is referred to as a frequency analysis waveform. In addition, the waveform after the frequency analysis waveform is normalized as described later is referred to as a normalized waveform (refer to Fig.19 ).

[0074] This information is, for example, a statistic representing the characteristics of a normalized waveform calculated based on data (frequency, current) of the alternating current obtained by a normal ball screw 30. The statistic is, for example, information related to variance. In addition, the statistic is an example, and any information representing the characteristics of the above-mentioned normalized waveform will suffice. For example, it may be image data representing an average waveform of a normalized waveform. In addition, since the abnormality detection device 7 of the present embodiment 1 is structured to use the statistics of the normalized waveform, the statistics of the normalized waveform obtained from a model of an injection molding machine different from that of the inspection object may also be used as the normal current information 72c.

[0075] In addition, if the abnormality detection device 7 is installed on the injection molding machine to detect abnormalities of the ball screw 30 during maintenance inspection of the injection molding machine, the vibration sensor 7a is preferably detachably fixed to the ball screw 30, and the current sensor 7b is also preferably detachably provided on the conductive wire (not shown). The detachable current sensor 7b is, for example, a clamp-type sensor.

[0076] If the structure is to always monitor the abnormality of the ball screw 30 of the injection molding machine, the vibration sensor 7a and the current sensor 7b can be fixed in an undetachable manner to any part of the driving circuit system of the ball screw 30 and the injection electric motor 29 respectively.

[0077] In the following description, the vibration sensor 7a and the current sensor 7b are mainly detachable and used to check abnormalities of the ball screw 30 during maintenance and inspection.

[0078] The operation unit 7c is an interface for operating the operation of the abnormality detection device 7 and inputting characteristic information indicating the characteristics of the injection transmission mechanism 20 as information for detecting abnormality of the ball screw 30. The operation unit 7c is, for example, an operation panel having buttons, a touch panel, etc. The operation unit 7c outputs the input characteristic information to the input interface 73.

[0079] Furthermore, if the output shaft of the injection electric motor 29 is directly connected to the bolt groove shaft 3a without passing through the injection transmission mechanism 20, characteristic information indicating the characteristics of the injection electric motor 29 can be input into the operating unit 7c.

[0080] The display unit 7d is connected to the output interface 74. The display unit 7d displays acceleration data and current data measured by the vibration sensor 7a and the current sensor 7b, a frequency analysis result described later, a fault state determination result, and the like.

[0081] 〈Abnormality detection based on vibration〉

[0082] Figure 31 is a flowchart showing the processing sequence of the vibration-based abnormality detection method. In the following description, an example is mainly described in which an operator who performs maintenance inspection of the injection molding machine starts the computer program 72a of the first embodiment and checks whether the ball screw 30 for injection has abnormalities. Some of the processing requires input operations by the operator, but all processing can also be determined and executed by the calculation unit 71.

[0083] First, the abnormality detection device 7 reads measurement data related to the vibration of the injection nut 32 (step S101).

[0084] Figure 4 2 is a flowchart showing the processing procedure of reading measurement data. The operator starts the computer program 72a of the first embodiment (step S111).

[0085] Furthermore, the computer program 72a may be configured to be actively activated by the calculation unit 71 when triggered by a predetermined condition.

[0086] Next, the operator selects a measurement data file storing acceleration data (step S112), and the calculation unit 71 reads the sampling period (step S113). The calculation unit 71 reads the measurement data, that is, the acceleration data, from the selected measurement data file (step S114).

[0087] It can also be a structure as follows: the file name and sampling period of the measurement data file recording the acceleration data are pre-stored in the storage unit 72 of the abnormality detection device 7, and the calculation unit 71 automatically reads the acceleration data according to the file name and sampling period stored in the storage unit 72.

[0088] Next, the calculation unit 71 causes the display unit 7d to display the measured acceleration waveform (see FIG. 14 ) based on the read measurement data. Figure 6 ) (Step S115). In addition, it is not essential to display the waveform of the acceleration. In particular, when the clipping of the analysis area described later is automatically performed, the waveform of the acceleration may not be displayed.

[0089] Back to Figure 3 After completing the processing of step S101, the abnormality detection device 7 cuts out the analysis area (step S102).

[0090] Figure 5 is a flowchart showing the processing sequence of analysis area clipping. Figure 6 7c is an explanatory diagram showing the cropping process of the analysis area. The operator selects the area to be analyzed from the displayed acceleration waveform using the cursor (step S121). The operation unit 71 receives the selection operation performed by the operator through the operation unit 7c, such as Figure 6 As shown, the display unit 7d displays the accepted selected area. Figure 6In the figure, the horizontal axis represents time, the vertical axis represents acceleration, and the change of acceleration over time is shown as a waveform. In addition, the area for selecting the analysis object is shown by a dotted line rectangular frame.

[0091] Next, the operator specifies the measurement data to be analyzed (step S122 ). The calculation unit 71 receives the operator's specification operation and temporarily stores the acceleration data, which is the measurement data corresponding to the area selected in step S121 , as the analysis target.

[0092] Furthermore, the calculation unit 71 may be configured to automatically perform a process of acquiring measurement data of an analysis region within a predetermined time range as an analysis target.

[0093] Back to Figure 3 After completing the processing of step S102, the abnormality detection device 7 performs frequency analysis on the measurement data to be analyzed (step S103).

[0094] Figure 7 is a flowchart showing the processing sequence of frequency analysis. Fig. 8A and Figure 8B : is an explanatory diagram showing an example of a frequency analysis waveform. The operation unit 71 performs frequency analysis processing on the measurement data, i.e., acceleration data, which is the analysis object (step S131), and obtains frequency analysis waveform data as the analysis result (step S132). Specifically, the operation unit 71 performs a fast Fourier transform (FFT). FFT is a well-known algorithm for performing a discrete Fourier transform quickly. Through frequency analysis, acceleration as a function of time is transformed into acceleration as a function of frequency.

[0095] Fig. 8A This is an example of a frequency analysis waveform (frequency distribution) of acceleration obtained from normal acceleration data of the ball screw 30. Figure 8B This is an example of a frequency analysis waveform (frequency distribution) of acceleration obtained from the acceleration data of the abnormal ball screw 30 . Fig. 8A and Figure 8B In the figure, the horizontal axis represents frequency and the vertical axis represents acceleration. Fig. 8A and Figure 8B It can be seen that the frequency distribution of acceleration has changed.

[0096] Back to Figure 3 After completing the processing of step S103, the abnormality detection device 7 calculates the driving operating frequency of the ball screw 30 for injection (step S104).

[0097] Fig. 97c reads the device specification (step S141). For example, the calculation unit 71 obtains the characteristic information of the injection transmission mechanism 20. The characteristic information of the injection transmission mechanism 20 includes, for example, the number of teeth of the injection driven pulley 22, the rotation speed [times / second] and other information.

[0098] When the injection electric motor 29 is directly connected to the bolt groove shaft 3a, the calculation unit 71 obtains characteristic information of the injection electric motor 29. The characteristic information of the injection electric motor 29 is a numerical value related to the frequency of vibration generated when the injection electric motor 29 is in operation.

[0099] Next, the calculation unit 71 performs a calculation process of the operating frequency based on the acquired characteristic information (step S142). The operating frequency is the main frequency of the vibration generated on the nut 32 by the screw shaft 31 due to the operation of the drive system. For example, the calculation unit 71 can calculate the operating frequency by multiplying the number of teeth of the injection driven pulley 22 by the rotation speed.

[0100] The calculation unit 71 that executes the process of step S142 functions as a functional unit that determines a predetermined frequency range based on the characteristic information acquired by the acquisition unit. The predetermined frequency range is, for example, a frequency range that is determined based on the above-mentioned operating frequency.

[0101] Furthermore, the calculation unit 71 may be configured to obtain characteristic information of the injection transmission mechanism 20 or the injection electric motor 29 in advance and store the information in the storage unit 72 to automatically calculate the operating frequency.

[0102] Back to Figure 3 After completing the process of step S104, the abnormality detection device 7 calculates the feature value (step S105).

[0103] Fig.10 is a flowchart showing the processing procedure of feature quantity calculation, Fig.11A and Fig. 11B is an explanatory diagram showing the designation process of the analysis area. Fig. 12A and Fig. 12B 2 is an explanatory diagram showing the center of gravity calculation process. The operation unit 71 reads the frequency analysis waveform data calculated in step S104 and step S142 (step S151). Then, the operation unit 71 performs noise elimination processing (step S152). For example, vibrations around 50 Hz or 60 Hz are eliminated as noise.

[0104] Next, if Fig.11A and Fig. 11BAs shown, the calculation unit 71 specifies the analysis area from the frequency distribution of acceleration (step S153). In Fig. 11, the horizontal axis represents frequency, the vertical axis represents acceleration, and the frequency distribution of acceleration is shown as a waveform. In addition, the analysis area is shown by a dotted line rectangular frame. Fig.11A This is the frequency analysis waveform of the normal vibration of the ball screw 30. Fig. 11B It is a frequency analysis waveform of the vibration of the abnormal ball screw 30. The analysis area is, for example, a frequency band of ±150 Hz based on the operating frequency. Alternatively, the analysis area may be manually selected by an operator.

[0105] Next, if Fig. 12A and Fig. 12B As shown, the calculation unit 71 calculates the center of gravity using the frequency analysis waveform data of the acceleration included in the designated analysis area (step S154). Fig. 12A and Fig. 12B In FIG. 1 , the horizontal axis represents frequency, the vertical axis represents acceleration, and the black dot represents the center of gravity. Specifically, the operation unit 71 calculates the following based on the multiple data (frequency, acceleration) included in the analysis area: Fig. 12A and Fig. 12B Center of gravity on the frequency-acceleration plane shown. Fig. 12A The center of gravity obtained by frequency analysis of the vibration waveform of the normal ball screw 30 is shown in FIG. Fig. 12B 9 shows the center of gravity obtained by frequency analysis of the vibration waveform of the abnormal ball screw 30 .

[0106] The frequency center of gravity is expressed by the following equation (1), and the acceleration center of gravity is expressed by the following equation (2).

[0107]

[0108]

[0109] in:

[0110] Fi represents the frequency of the i-th frequency component contained in the analysis area;

[0111] ai represents the acceleration of the i-th frequency component contained in the analysis area;

[0112] S represents the sum of the accelerations of all frequency components contained in the analysis area.

[0113] Back to Figure 3 After completing the processing of step S105, the abnormality detection device 7 performs a fault state determination process (step S106).

[0114] Fig.13 is a flowchart showing the processing sequence of the fault state determination process, Fig.14is an explanatory diagram showing the fault state discrimination process. The calculation unit 71 reads the center of gravity calculated in step S105 or step S154 and the multiple centers of gravity represented by the normal vibration information 72b stored in the storage unit 72 as feature quantities (step S161). Fig.14 As shown, the calculation unit 71 causes the display unit 7d to display the read feature value (step S162). Fig.14 In FIG. 1 , the horizontal axis represents frequency, the vertical axis represents acceleration, and the black dot represents the feature quantity read in step S161 , that is, the center of gravity.

[0115] Then, the calculation unit 71 calculates the fault information indicating the presence or absence of a fault or the degree of the fault based on the read feature quantity (step S163). Specifically, the calculation unit 71 calculates the statistical distance between the center of gravity calculated in step S154 and a plurality of centers of gravity in normal state, such as the Mahalanobis distance. In addition, the Mahalanobis distance is an example, and other statistical distances such as the Euclidean distance may also be used.

[0116] Then, the calculation unit 71 causes the display unit 7d to display an image indicating the fault information (step S164). For example, the calculation unit 71 generates an arrow image, which starts from the average position of the center of gravity in normal time on the frequency-acceleration plane and ends at the center of gravity calculated in step S154, and causes the display unit 7d to display the arrow image.

[0117] Furthermore, when the failure information calculated in the past is stored, the calculation unit 71 may display the history of the failure information on the display unit 7d.

[0118] Fig.15 This is an explanatory diagram showing the history of fault state judgment results. The horizontal axis represents the time such as the year, month, and day of the inspection, and the vertical axis represents the fault state, such as the above-mentioned Mahalanobis distance. The intermittent line represents the boundary between the normal state and the abnormal state of the ball screw 30. Multiple black dots represent the fault states calculated in the past. The operator confirms the fault state by checking the following information: Fig.15 The fault state history shown can provide detailed information about the state of the ball screw 30. Specifically, the operator can recognize how far the current ball screw 30 deviates from the normal state, whether it is close to an abnormal state, and estimate the timing of maintenance.

[0119] 〈Abnormality detection based on motor current〉

[0120] Fig.162 is a flowchart showing the processing sequence of the abnormality detection method based on motor current. The abnormality detection device 7 reads the measurement data of the alternating current of the injection electric motor 29, that is, the current data (step S201). Then, the abnormality detection device 7 cuts the analysis area (step S202) and performs frequency analysis on the measurement data as the analysis object (step S203). The processing content of steps S201 to S203 is the same as the reading, cutting, and frequency analysis processing of the measurement data related to vibration, except that the measurement data is current data.

[0121] After completing the process of step S203 , the calculation unit 71 of the abnormality detection device 7 performs fundamental wave extraction processing (step S204 ).

[0122] Fig.17 This is a flowchart of the fundamental wave extraction process. The operation unit 71 performs Fourier series expansion on the current measurement data (step S231) and performs a fundamental wave search process (step S232). In the fundamental wave search process, the operation unit 71 determines the frequency and amplitude value (wave height value) of the fundamental wave. The fundamental wave is usually near the basic frequency of the alternating current of the injection electric motor 29, for example, near 50 Hz or 60 Hz, and is a wave with a maximum wave height value. In addition, the operation unit 71 can also cause the display unit 7d to display the waveform of the fundamental wave.

[0123] Back to Fig.16 After completing the processing of step S204, the calculation unit 71 normalizes the frequency analysis waveform based on the frequency and amplitude value of the fundamental wave (step S205).

[0124] Fig.18 is a flowchart of the normalization process. Fig.19 is an explanatory diagram showing the normalized frequency analysis waveform. The operation unit 71 reads the frequency analysis waveform data (step S251), and reads the frequency and amplitude value of the fundamental wave (step S252). Then, Fig.19 As shown, the calculation unit 71 normalizes the frequency analysis waveform so that, for example, the frequency of the fundamental wave is "1" and the amplitude value of the fundamental wave is "1" ("0" in logarithmic expression) (step S253). Fig.19 The graph shown is a semi-logarithmic graph in which the horizontal axis represents frequency and the vertical axis represents motor current, and the motor current is expressed logarithmically. It is needless to say that the above-mentioned normalization method is an example. Next, the calculation unit 71 causes the display unit 7d to display the normalized waveform (step S254).

[0125] Back to Fig.16 After completing the processing of step S205, the abnormality detection device 7 performs a fault state determination process (step S206).

[0126] Fig. 20is a flowchart showing the processing sequence of the fault state determination process, Fig.21A and Fig. 21B It is an explanatory diagram showing the fault state determination process.

[0127] The calculation unit 71 reads the normalized waveform data, that is, the frequency analysis waveform data normalized in step S205 (step S261 ).

[0128] Fig.21A FIG. 3 shows the normalized waveform of the current when the ball screw 30 is normal. Fig. 21B The normalized waveform of the current when the ball screw 30 is abnormal is shown.

[0129] The operation unit 71 calculates information indicating the change of the frequency analysis waveform based on the normalized waveform data (step S262). As information indicating the change of the frequency analysis waveform, the operation unit 71 calculates a statistic of the normalized waveform, such as a variance. In addition, the operation unit 71 may also generate image data of the normalized waveform as information indicating the change of the frequency analysis waveform.

[0130] Next, the calculation unit 71 calculates the fault information indicating whether the ball screw 30 has a fault or the degree of the fault based on the information calculated in step S262 (step S263). Specifically, the calculation unit 71 calculates the difference between the statistic calculated in step S254 and the statistic represented by the normal vibration information 72b stored in the storage unit 72. The statistic is, for example, the variance.

[0131] When the operation unit 71 generates the image data of the normalized waveform in step S262, the normalized waveform can be compared with the normalized waveform indicated by the normal vibration information 72b to determine whether the ball screw 30 is faulty. For example, the operation unit 71 determines whether the ball screw 30 is abnormal by comparing the approximation of the normalized waveform with a predetermined threshold value.

[0132] Then, the calculation unit 71 causes the display unit 7d to display an image indicating the fault information (step S264). For example, the calculation unit 71 can display the abnormality of the ball screw 30, that is, the degree of deviation from the normal state, by the size of the difference in variance. In addition, the difference in variance when abnormal can also be displayed.

[0133] The calculation unit 71 may also be configured to compare the variance and the threshold value, and when the variance is smaller than the threshold value, display the main idea that the ball screw 30 is normal, and when the variance is larger than the threshold value, display the main idea that the ball screw 30 is abnormal. In addition, it may also be configured to compare the variance using multiple threshold values ​​to display the various states of the ball screw 30 from the normal state to the fault state, that is, the state that may reach the abnormal state.

[0134] The calculation unit 71 may be configured to cause the display unit 7d to display the frequency analysis waveform in a normal state and the frequency analysis waveform in an abnormal state superimposed on or side by side.

[0135] According to the abnormality detection device 7 , the abnormality detection method, and the computer program 72 a of the first embodiment configured in this manner, it is possible to detect abnormality of the ball screw 30 by analyzing the intensity of vibration generated in the ball screw 30 and the frequency of the motor current.

[0136] Furthermore, by providing the vibration sensor 7a on the side of the nut 32 to detect the axial acceleration of the screw shaft 31, abnormality of the ball screw 30 can be detected with higher accuracy than when the vibration sensor 7a is provided at other locations in other postures.

[0137] By comparing the center of gravity of the frequency distribution of acceleration with the center of gravity in a normal state, an abnormality of the ball screw 30 can be detected with higher accuracy than when simply monitoring the vibration intensity.

[0138] By comparing the center of gravity calculated based on the characteristics of the injection electric motor 29 or the injection transmission mechanism 20 with the center of gravity in normal state as the analysis target, abnormality of the ball screw 30 can be detected with higher accuracy than when analyzing other frequency bands.

[0139] The calculation unit 71 can obtain characteristic information of the injection electric motor 29 or the injection transmission mechanism 20, calculate the frequency band as the analysis object, and determine the fault state. The operator does not need to determine the frequency range as the analysis object, and the calculation unit 71 can automatically or semi-automatically calculate whether the ball screw 30 has an abnormality.

[0140] An abnormality in the ball screw 30 can be detected by a simple process of comparing statistics (for example, variance) of frequency analysis waveforms obtained by frequency analysis of the motor current.

[0141] By comparing the image of the frequency analysis waveform of the current with the frequency analysis waveform in a normal state, it is possible to detect abnormality in the ball screw 30 by taking into account information that the variance does not have.

[0142] By normalizing and comparing the frequency analysis waveforms of the current, the frequency analysis waveforms of injection molding machines of different models or manufacturing periods can be compared with each other, and the presence or absence of abnormality in the ball screw 30 can be better determined using more accurate frequency analysis waveforms.

[0143] In addition, in the present embodiment 1, an example of performing abnormality detection of the ball screw 30 based on vibration and abnormality detection of the ball screw 30 based on motor current is described, but it is also possible to use both the abnormality detection result based on vibration of the ball screw 30 and the abnormality detection result based on motor current to determine whether the ball screw 30 has an abnormality by logical OR or logical AND. In addition, the abnormality detection device 7 can also be configured to calculate the average of the degree of dissociation from the normal state indicated by each abnormality detection result, and output the fault state of the ball screw 30.

[0144] When detecting abnormality of the ball screw 30 using vibration intensity such as acceleration data and AC current of the injection electric motor 29, the calculation unit 71 preferably synchronously acquires vibration intensity data output by the vibration sensor 7a and current data output by the current sensor 7b.

[0145] In addition, the technical elements shown in the abnormality detection based on vibration can also be applied to the abnormality detection based on the motor current. Similarly, the technical elements shown in the abnormality detection based on the motor current can also be applied to the abnormality detection based on vibration.

[0146] In addition, in embodiment 1, an example of detecting an abnormality of the ball screw 30 used for injection is mainly described. The abnormality detection device 7 of this embodiment 1 can also be used to detect an unillustrated ball screw constituting the drive mechanism of the mold clamping device 50 and other ball screws constituting the drive mechanism of the injection molding machine.

[0147] In the first embodiment, the vibration sensor 7a is provided on the side surface of the nut 32, but it may be provided on the circumferential surface of the nut 32. In addition, the vibration sensor 7a may be provided inside the nut 32.

[0148] In the first embodiment, the example of driving the screw shaft 31 to linearly move the nut 32 is described, but the structure may also be such that the nut 32 rotates to linearly move the screw shaft 31. In this case, the vibration sensor 7a may be provided at an appropriate position of the screw shaft 31, such as an end portion.

[0149] In the first embodiment, a pulley and a timing belt are shown as an example of the injection transmission mechanism 20 , but the transmission mechanism of the driving force is not limited thereto, and a gear mechanism, a chain mechanism, a friction transmission mechanism, or other known transmission mechanisms may be used.

[0150] In the first embodiment, an example in which abnormality of the ball screw 30 is checked during maintenance and inspection work is mainly described. However, as described above, a configuration in which abnormality of the ball screw 30 is monitored in real time may be adopted.

[0151] Although an example of a rotary electric motor is shown as the driving source of the ball screw 30 , the power source is not limited to the electric motor, and other well-known driving sources can also be used.

[0152] (Implementation Method 2)

[0153] The abnormality detection device, abnormality detection method and computer program of the second embodiment have a fault determination method different from that of the first embodiment, so the following mainly describes the above differences. The other structures and effects are the same as those of the first embodiment, so the same figure marks are marked on the corresponding parts, and the detailed description is omitted.

[0154] Fig. 22 1 is a functional block diagram of the calculation unit 71 according to Embodiment 2. The calculation unit 71 according to Embodiment 2 includes a first learner 91 , a second learner 92 , and a fault determiner 93 as functional units.

[0155] The first learner 91 is, for example, a neural network including an input layer 91a, a hidden layer 91b, and an output layer 91c. The input layer 91a is input with frequency analysis waveform data related to the vibration of the ball screw 30. The hidden layer 91b performs a calculation based on a learned weight coefficient on the frequency analysis waveform data input to the input layer 91a. The output layer 91c outputs data indicating the degree of abnormality of the ball screw 30. In other words, the first learner 91 is a learned model that enables a computer to function, and outputs data indicating the degree of abnormality of the ball screw 30 when the vibration frequency analysis waveform data is input.

[0156] The learning of the first learner 91 can be completed as follows: prepare learning data, which includes multiple frequency analysis waveform data involving vibration and training data indicating the degree of abnormality of the ball screw 30 that outputs the frequency analysis waveform data, and use a back propagation algorithm, an error gradient descent method, etc. to optimize the weight coefficient of the hidden layer 91b of the neural network so that when the frequency analysis waveform data as the learning data is input, data indicating the degree of abnormality of the ball screw 30 represented by the training information is output.

[0157] The input layer 91a includes a plurality of neurons (nodes). The data of the vibration frequency analysis waveform is, for example, a plurality of data (frequency, acceleration), and the frequency and acceleration of each data are input to the plurality of neurons (nodes) constituting the input layer 91a.

[0158] In addition, the frequency analysis waveform data of the vibration may be image data representing the waveform or time series data. In the case where the frequency analysis waveform data is image data, the first learner 91 may be formed by a convolutional neural network. The convolutional neural network (CNN) has multiple sets of convolutional layers, pooling layers, and full integration layers not shown as hidden layers 91b. In the case where the data of the vibration frequency analysis waveform is time series data, a recurrent neural network (RNN) may be used.

[0159] The output layer 91c has one or more neurons, and data output from each neuron is input to the fault discriminator 93. Each neuron has, for example, a neuron (node) representing a first probability indicating that the ball screw 30 is normal and a neuron (node) representing a second probability indicating that the ball screw 30 is abnormal.

[0160] The second learner 92 is a neural network having an input layer 92a, a hidden layer 92b, and an output layer 92c. The input layer 92a is input with, for example, frequency analysis waveform data related to the motor current, or normalized waveform data after the data is normalized. The hidden layer 92b performs operations based on learned weight coefficients on the frequency analysis waveform data or normalized waveform data input to the input layer 92a. The output layer 92c outputs information indicating the degree of abnormality of the ball screw 30. In other words, the second learner 92 is a learned model that enables a computer to function, and when current frequency data or normalized waveform data is input, information indicating whether the ball screw 30 is abnormal is output.

[0161] The learning of the second learner 92 is completed as follows: prepare learning data, which includes multiple frequency analysis waveform data or normalized waveform data related to the motor current and training data indicating the degree of abnormality of the ball screw 30 that outputs the frequency analysis waveform data or normalized waveform data, and use a back propagation algorithm, an error gradient descent method, etc. to optimize the weight coefficient of the hidden layer 92b of the neural network so that when the frequency analysis waveform data or normalized waveform data as the learning data is input, data indicating the degree of abnormality of the ball screw 30 represented by the training information is output.

[0162] The input layer 92a includes a plurality of neurons (nodes). The data of the current frequency or the normalized waveform data is, for example, a plurality of data (frequency, acceleration), and the frequency and acceleration of each data are input to the plurality of neurons (nodes) constituting the input layer 92a.

[0163] In addition, the frequency analysis waveform data or normalized waveform data of the motor current input to the input layer 92a may be image data representing the waveform or time series data. In the case where the frequency analysis waveform data or the normalized waveform data is image data, the second learner 92 may be formed by a convolutional neural network. The convolutional neural network has multiple sets of convolutional layers, pooling layers, and full integration layers not shown as hidden layers 92b. In the case where the data of the current frequency is time series data, a recurrent neural network may be used.

[0164] The output layer 92c has one or more neurons, and data output from each neuron is input to the fault discriminator 93. Each neuron has, for example, a neuron (node) representing a first probability indicating that the ball screw 30 is normal and a neuron (node) representing a second probability indicating that the ball screw 30 is abnormal.

[0165] The fault discriminator 93 obtains the data output from the first learner 91 and the second learner 92, and comprehensively judges whether the ball screw 30 has a fault or the degree of the fault based on the obtained data. For example, the fault discriminator 93 can calculate the average value of the probability that the ball screw 30 is normal and the average value of the probability that the ball screw 30 is abnormal, which are output from the first learner 91 and the second learner 92.

[0166] In addition, a neural network may be used to constitute the fault discriminator 93, so that the neural network determines whether the ball screw 30 has a fault. In this case, the data output from the first learner 91 and the second learner 92 do not necessarily need to be data indicating whether the ball screw 30 has an abnormality, and data indicating the characteristics of the frequency analysis waveform of the vibration and the frequency analysis waveform or the normalized waveform of the current are sufficient.

[0167] According to the abnormality detection device 7 , the abnormality detection method, and the computer program 72 a of the second embodiment, it is possible to perform machine learning on the frequency analysis waveforms of vibrations in normal and abnormal conditions, and to discriminate the fault state of the ball screw 30 .

[0168] Furthermore, it is possible to perform machine learning on the frequency analysis waveforms of the current in normal and abnormal conditions, and to discriminate the fault state of the ball screw 30 .

[0169] The failure state of the ball screw 30 can be comprehensively determined using both the frequency analysis waveform data of the vibration in normal and abnormal times and the frequency analysis waveform data of the current.

[0170] In addition, in Implementation 2, an example of machine learning using neural networks such as CNN and RNN is mainly described, but the structure of the learned model is not limited to CNN, RNN, etc., and can also be composed of neural networks other than CNN and RNN, SVM (Support Vector Machine), Bayesian network, or regression tree.

[0171] Description of Reference Numerals

[0172] 1 Injection device

[0173] 2 Heating barrel

[0174] 3 Screw

[0175] 3a Grooved shaft

[0176] 4 Heater

[0177] 5 Injection nozzle

[0178] 6 Hopper

[0179] 7 Abnormality detection device

[0180] 7a Vibration sensor

[0181] 7b Current sensor

[0182] 7c Operation section

[0183] 7d Display

[0184] 8 Recording media

[0185] 10 Middle plate

[0186] 11 Electric motor for plasticizing

[0187] 12 Plasticizing drive pulley

[0188] 13 Plasticizing driven pulley

[0189] 14 Synchronous belt for plasticizing

[0190] 15 Thrust bearing

[0191] 16 Guide rod

[0192] 20 Transmission mechanism for injection

[0193] 21 Injection drive pulley

[0194] 22 Driven pulley for injection

[0195] 23. Synchronous belt for injection

[0196] 29 Electric motor for injection

[0197] 30 Ball Screw

[0198] 31 Screw shaft

[0199] 32 Nut

[0200] 33 Nut fixing parts

[0201] 39 Electric motor for nozzle contact

[0202] 40 Nozzle contact screw shaft

[0203] 41 Nozzle contact nut

[0204] 42 Supporting parts

[0205] 43. Transmission mechanism for nozzle contact

[0206] 50 Clamping device

[0207] 51 Fixed plate

[0208] 52 Movable plate

[0209] 53 mold shell

[0210] 54 Fixed mold

[0211] 55 Movable mold

[0212] 71 Operation unit

[0213] 72 Storage

[0214] 72a Computer programs

[0215] 72b Normal vibration information

[0216] 72c Normal current information

[0217] 73 Input Interface

[0218] 74 Output Interface

[0219] 91 First Learning Machine

[0220] 92 Second Learning Machine

[0221] 93 Fault Identifier

[0222] SDU Screw Drive Unit

[0223] IU Injection Unit

[0224] NT Nozzle Contact Device

[0225] MM mold device

[0226] TM Toggle Mechanism

Claims

1. An abnormality detection device for detecting abnormalities of a ball screw provided on an injection molding machine, comprising: a vibration sensor that detects vibration of the ball screw; an acquisition unit that acquires information indicating characteristics of a motor that drives the ball screw or characteristics of a transmission mechanism that transmits a driving force from the motor to the ball screw; and A calculation unit determines a specified frequency range based on the characteristics of the motor or the characteristics of the transmission mechanism based on the information obtained by the acquisition unit, calculates a statistical distance between a center of gravity of a distribution of the frequency components in the specified frequency range determined by performing a frequency analysis on the vibration intensity detected by the vibration sensor and a center of gravity of a distribution of the frequency components in the frequency range determined by performing a frequency analysis on the vibration intensity detected during normal operation of the ball screw, and detects an abnormality of the ball screw based on the calculated statistical distance.

2. The abnormality detection device according to claim 1, The ball screw comprises: a screw shaft rotated by a motor driving the ball screw; and As the screw shaft rotates, the nut moves linearly along the screw shaft. The vibration sensor is arranged on the nut.

3. The abnormality detection device according to claim 2, The nut is cylindrical. The vibration sensor is arranged on the end surface of the nut and detects at least the vibration intensity in the linear moving direction of the nut.

4. The abnormality detection device according to any one of claims 1 to 3, The computing unit includes a learner that performs machine learning and outputs information on the presence or absence of abnormality in the ball screw when information obtained by frequency analysis of the vibration intensity detected by the vibration sensor is input.

5. A method for detecting an abnormality of a ball screw provided on an injection molding machine, comprising: detecting vibration of the ball screw; Conduct frequency analysis on the detected vibration intensity; acquiring information indicating characteristics of a motor driving the ball screw or characteristics of a transmission mechanism transmitting a driving force from the motor to the ball screw; Based on the information obtained, a specified frequency range is determined based on the characteristics of the motor or the characteristics of the transmission mechanism, and a statistical distance is calculated between the center of gravity of the distribution of the frequency components in the specified frequency range determined by frequency analysis and the center of gravity of the distribution of the frequency components in the specified frequency range determined by frequency analysis of the vibration intensity detected during normal operation of the ball screw, and abnormalities of the ball screw are detected based on the calculated statistical distance.

6. A computer program product, comprising a computer program for causing a computer to perform a process for detecting an abnormality of a ball screw provided in an injection molding machine, When the computer program is executed on a computer, the computer is caused to perform the following processing: detecting vibration of the ball screw; Conduct frequency analysis on the detected vibration intensity; acquiring information indicating characteristics of a motor driving the ball screw or characteristics of a transmission mechanism transmitting a driving force from the motor to the ball screw; Based on the information obtained, a specified frequency range is determined based on the characteristics of the motor or the characteristics of the transmission mechanism, and a statistical distance is calculated between the center of gravity of the distribution of the frequency components in the specified frequency range determined by frequency analysis and the center of gravity of the distribution of the frequency components in the specified frequency range determined by frequency analysis of the vibration intensity detected during normal operation of the ball screw, and abnormalities of the ball screw are detected based on the calculated statistical distance.