Abnormality detection system

Abstract

To provide an abnormality detection system capable of easily detecting an abnormality in an injection molding machine.SOLUTION: There is provided an abnormality detection system, comprising: a sensor 1 that is installed at a predetermined position in an injection molding machine 9 and measures sensing data for detecting an abnormality of an injection molding, which occurs at the predetermined position; and a processing device 3 that acquires the sensing data and detects an abnormality in the injection molding, which occurs at the predetermined position using the sensing data.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an abnormality detection system for injection molding. [Background technology]

[0002] Injection molding using thermoplastic resins is used in a wide range of fields as a manufacturing method that can mass-produce parts with complex shapes. Technology for detecting abnormalities in injection molding machines is being researched (Patent Documents 1 and 2). Technology for determining defects in molded products molded by injection molding machines is also being researched (Patent Documents 3 and 4). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-122955 [Patent Document 2] International Publication No. 2019 / 188665 [Patent Document 3] JP 2023-68058 A [Patent Document 4] Patent No. 7250072 Summary of the Invention [Problem to be solved by the invention]

[0004] Conventionally, skilled technicians detect injection molding abnormalities based on their many years of experience, intuition, or knowledge. Specifically, skilled technicians detect injection molding machine abnormalities by checking the sounds and vibrations emitted by the injection molding machine, or by visually inspecting the condition of the parts of the injection molding machine.

[0005] However, for engineers with little experience, it is difficult to detect abnormalities in injection molding, and it takes a long time to detect the abnormalities. Products molded by injection molding machines with abnormalities are likely to have molding defects such as flash and short shots.

[0006] The present invention has been made in consideration of the above problems, and an object of the present invention is to easily detect abnormalities that are factors that cause molding defects when performing injection molding. [Means for solving the problem]

[0007] In order to achieve the above-mentioned objective, one aspect of the present invention is an abnormality detection system comprising a sensor that is installed at a predetermined position within an injection molding machine and measures sensing data for detecting an abnormality in the injection molding machine that occurs at the predetermined position, and a processing device that acquires the sensing data and uses the sensing data to detect an abnormality in the injection molding that occurs at the predetermined position. Effect of the Invention

[0008] According to the present invention, an abnormality in injection molding can be easily detected. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a schematic configuration diagram of an anomaly detection system according to an embodiment. [Diagram 2] FIG. 2 is a schematic diagram showing an outline of an injection molding machine. [Figure 3A] FIG. 3A is a diagram illustrating a three-axis acceleration sensor. [Figure 3B] FIG. 3B is a diagram for explaining the three axes of the three-axis acceleration sensor. [Figure 4] Figure 4 shows the gravitational acceleration data on the X-axis when there are no abnormalities in the die. [Diagram 5] Figure 5 shows the gravitational acceleration data when there are no abnormalities in the die. [Figure 6] Figure 6 shows the gravitational acceleration data when the die is out of oil. [Figure 7] Figure 7 shows the gravitational acceleration data when there is an abnormality in the die installation. [Figure 8] FIG. 8 is a schematic diagram of the air pressure sensor and the degassing unit. [Figure 9]Figure 9 shows air pressure data measured by the air pressure sensor. [Figure 10] FIG. 10 is a schematic diagram showing an axial tension bolt installed in a locking block. [Figure 11] Figure 11 shows the strain data measured by the axial tension bolts. [Figure 12] FIG. 12 is a spectrogram measured by a stereo ultrasonic detector. [Figure 13] FIG. 13 is a table that associates abnormalities with control instructions for the injection molding machine. [Figure 14] FIG. 14 shows an example of a hardware configuration. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0011] 1 is a schematic diagram of an abnormality detection system according to an embodiment of the present invention. The abnormality detection system shown in the figure includes at least one sensor 1 and a processing device 3 arranged in an injection molding machine 9, and detects abnormalities in injection molding.

[0012] The sensor 1 is installed at a predetermined position in the injection molding machine 9, and measures sensing data for detecting an abnormality in injection molding that occurs at the predetermined position. The number of sensors 1 installed in the injection molding machine 9 may be one or more. In the case of multiple sensors 1, each sensor 1 may be a different type of sensor. Furthermore, each sensor 1 is installed at a predetermined position in the injection molding machine 9 corresponding to its type.

[0013] The processing device 3 acquires (collects) sensing data measured by the sensor 1, and uses the sensing data to detect an abnormality in injection molding that occurs at a predetermined position where the sensor 1 is installed. A computer such as a server or a personal computer can be used as the processing device 3. The processing device 3 shown in the figure includes an acquisition unit 31, a detection unit 32, an output unit 33, and a storage unit 34.

[0014] The acquiring unit 31 acquires sensing data from the sensor 1. The anomaly detection system of this embodiment includes a microcomputer 2 as shown in Fig. 1, and the processing device 3 may acquire the sensing data from the sensor 1 via the microcomputer 2. The microcomputer 2 is connected to the sensor 1 and the processing device 3 using, for example, a USB cable.

[0015] The microcomputer 2 controls the sensor 1 and converts the format of the sensing data measured by the sensor 1. For the microcomputer 2, a microcomputer board such as Arduino can be used. A program that specifies the operation of the microcomputer 2 is created by the processing device 3, and the compiled program can be installed in the microcomputer via a USB cable or the like. Note that if the acquisition unit 31 of the processing device 3 has the functions of the microcomputer 2, the processing device 3 may acquire the sensing data directly from the sensor 1 without going through the microcomputer 2.

[0016] The detection unit 32 detects an abnormality in injection molding occurring at a predetermined position where the sensor 1 is installed, using the sensing data measured by the sensor 1. For example, the detection unit 32 may detect an abnormality in injection molding occurring at a position where the sensor 1 is installed by comparing the acquired sensing data with normal sensing data of the corresponding sensor 1 stored in the storage unit 34. The detection unit 32 may also detect an abnormality in injection molding using the acquired sensing data and a threshold value or peak value in a predetermined process of the corresponding sensor 1 stored in the storage unit 34. The detection unit 32 may also input the acquired sensing data to a trained model stored in the storage unit 34 to detect an abnormality in the injection molding machine 9.

[0017] When the detection unit 32 detects an abnormality, the output unit 33 outputs to a display (not shown) that an abnormality has been detected at the position where the sensor 1 is installed in the injection molding machine 9. The output unit 33 may output the type of abnormality detected by the detection unit 32 to the display.

[0018] The storage unit 34 stores normal sensing data for each of the various sensors 1. It is assumed that the normal sensing data is set in association with each process of the injection molding machine 9. The storage unit 34 may also store a threshold value or a peak value in a corresponding process for each of the various sensors 1. The storage unit 34 may also store a trained model.

[0019] The above-mentioned units 31 to 34 of the processing device 3 may be developed using a flow-based programming tool such as Node-RED.

[0020] 2 is a schematic diagram showing an overview of an injection molding machine 9. The injection molding machine 9 melts a molding material (such as a resin) with heat, performs injection molding, and injects the molten resin into a mold to produce a molded product. The injection molding machine 9 has a mold clamping unit 4, a mold 5, and an injection unit 6. The injection unit 6 melts the molding material with heat and injects it into the mold 5. The injection unit 6 shown in the figure includes a nozzle 60, a cylinder 61, a check valve 62, a heater 63, a screw 64, a hopper 65, a motor 66, and the like.

[0021] The mold 5 includes tie bars 51, a fixed mold 52, and a movable mold 53. The mold clamping unit 4 performs a mold clamping step of bringing the movable mold 53 of the mold 5 close to the fixed mold 52 to close the mold 5. The mold clamping unit 4 includes a crosshead 41, an ejector mechanism 42, and the like.

[0022] In this embodiment, a sensor 1 is newly attached to a commercially available injection molding machine 9, and sensing data measured by each sensor 1 is used to detect an abnormality occurring at a predetermined position of the injection molding machine 9. In the examples described below, the cases where the sensors 1 are an acceleration sensor 11, an air pressure sensor 12, an axial tension bolt 13, an ultrasonic detector 14, and a gas concentration detection sensor 15 will be described. However, the sensor 1 is not limited to these. Furthermore, the sensor 1 attached to the injection molding machine 9 may be one, or a combination of multiple sensors 1 of different types. Furthermore, the position at which the sensor 1 is attached is a position according to the type of the sensor 1.

[0023] <Example 1: Acceleration sensor> In this embodiment, an acceleration sensor 11 installed on a mold 5 of an injection molding machine is used as the sensor 1. The acceleration sensor 11 is installed on the top surface of the mold 5 here, but may be installed on a location other than the top surface. Here, gravitational acceleration data (sensing data) measured by the acceleration sensor 11 is used to detect abnormalities in the mold 5.

[0024] Fig. 3A is a schematic diagram showing a three-axis acceleration sensor 11. As shown in Fig. 3B, the acceleration sensor 11 is installed on the top surface of the mold 5 so that the mold opening direction is the X-axis plus direction 301, the operation side direction is the Y-axis plus direction 302, and the mold top surface direction is the Z-axis plus direction 303.

[0025] 4 shows an example of gravitational acceleration data on the X-axis measured by the acceleration sensor 11 when there is no abnormality in the mold 5. The gravitational acceleration on the X-axis increases slightly in the positive direction of the X-axis from the start of mold closing by the mold clamping unit 4 to mold touch, and then remains approximately zero until mold opening begins. Then, from the start of mold opening by the mold clamping unit 4 to completion of mold opening, the gravitational acceleration decreases slightly in the negative direction of the X-axis and then remains approximately zero. The illustrated gravitational acceleration data has small amplitudes (fluctuations) from the start of mold closing to mold touch and from the start of mold opening to completion of mold opening.

[0026] Fig. 5 is an example of three-axis gravitational acceleration data measured by acceleration sensor 11 when there is no abnormality in mold 5. The gravitational acceleration data on the X-axis is the same as that in Fig. 4. When there is no abnormality in mold 5, the amplitudes of the gravitational acceleration data on the Y-axis and Z-axis due to mold clamping (mold closing) and mold opening are small, similar to the X-axis.

[0027] FIG. 6 is an example of gravitational acceleration data measured by the acceleration sensor 11 when there is an oil shortage in the mold 5. Due to oil shortage in the mold 5, there are some points where the gravitational acceleration on the X-axis fluctuates significantly between the start of mold closing and mold touch, and between the start of mold opening and the completion of mold opening, indicating that vibrations are occurring in the X-axis direction due to mold closing and mold opening. Similarly, there are some points where the gravitational acceleration on the Z-axis fluctuates significantly between the start of mold closing and mold touch, and between the start of mold opening and the completion of mold opening, indicating that a collision is occurring in the Z-axis direction due to oil shortage. A collision refers to a state in which, for example, a guide pin and a guide bush (not shown) are in contact with each other.

[0028] 7 shows gravitational acceleration data measured by the acceleration sensor 11 when there is an abnormality in the installation of the mold 5. The example shown is gravitational acceleration data when one of the multiple clamps (mold fixtures) that attach the mold 5 is not attached (or is loose). Here, gravitational acceleration data is shown when only seven clamps are attached, although eight clamps should be attached.

[0029] Because some clamps are not attached, the gravitational acceleration on the X-axis fluctuates greatly in the positive direction of the X-axis from the start of mold closing to mold touch, exceeding the specified threshold (0.4g in this case). Also, the gravitational acceleration on the X-axis fluctuates greatly in the negative direction of the X-axis from the start of mold opening to the completion of mold opening, exceeding the specified threshold (-0.4g in this case) at some points. This shows that mold closing and opening are causing large vibrations in the X-axis direction.

[0030] Similarly, the gravitational accelerations on the Y and Z axes also fluctuate significantly between the start of mold closing and mold touch, and between the start of mold opening and the completion of mold opening, and there are points where they exceed predetermined thresholds.

[0031] In this embodiment, the detection unit 32 of the processing device 3 detects an abnormality in the mold 5 by using normal gravitational acceleration data (see FIG. 5 ) stored in the memory unit 34. Specifically, the detection unit 32 detects an abnormality in the mold 5 by comparing the gravitational acceleration data measured by the acceleration sensor 11 with the normal gravitational acceleration data in the memory unit 34.

[0032] The detection unit 32 determines that the mold 5 is running out of oil when the difference between the amplitude (amplitude, amount of fluctuation) of the gravitational acceleration from the start of mold closing to mold touch and from the start of mold opening to the completion of mold opening in the gravitational acceleration data of the X-axis and Z-axis output by the acceleration sensor 11 and the amplitude of normal gravitational acceleration data stored in the memory unit 34 is greater than a predetermined threshold. Note that the amplitude may be the maximum amplitude or the average. The threshold value of the difference in amplitude on the X-axis and the threshold value of the amplitude on the Z-axis may be different values ​​or the same values.

[0033] Furthermore, when the gravitational acceleration data for the X-axis, Y-axis and Z-axis output by the acceleration sensor 11 indicates that the amplitude of the gravitational acceleration from the start of mold closing to the time of mold touch, and from the start of mold opening to the completion of mold opening, is large and exceeds a predetermined threshold value at any point, the detection unit 32 determines that there is an abnormality in the installation of the mold 5.

[0034] In addition, since the normal gravitational acceleration data in the memory unit 34 is set in correspondence with each process of the injection molding machine 9 (start of mold closing, mold touch, start of mold opening, completion of mold opening), it is possible to identify each process of the acquired gravitational acceleration data.

[0035] When the detection unit 32 detects that the mold 5 is out of oil or that the mold 5 is not installed properly, the output unit 33 displays the abnormality on a display (not shown). This allows the user to repair the mold 5 according to the type of abnormality displayed on the display.

[0036] For example, if a shortage of oil in the die 5 is detected, the user applies lubricant to parts of the die 5, such as guide pins and bushes. By applying lubricant, the shortage of oil in the die 5 is resolved. Furthermore, if an installation abnormality is detected, the user installs a clamp that is not installed, or tightens a loose clamp with a torque wrench. This resolves the installation abnormality in the die 5.

[0037] The sensor 1 of this embodiment described above is an acceleration sensor 11 installed in the mold 5 of the injection molding machine 9, which measures the gravitational acceleration, and the processing device 3 detects a shortage of oil in the mold 5 or an abnormality in the installation of the mold 5 using the amplitude of the gravitational acceleration.

[0038] Until now, a skilled technician has detected running out of oil in mold 5 by sensing sounds and vibrations during the mold closing and opening processes, and has also detected an abnormality in the installation of mold 5 by checking each clamp using a torque wrench. However, according to this embodiment, running out of oil and an abnormality in the installation of mold 5 can be easily detected using the gravitational acceleration data from acceleration sensor 11, and defects in molded products produced by injection molding machine 9 caused by such abnormalities can be reduced.

[0039] <Example 2: Barometric pressure sensor> In this embodiment, an air pressure sensor 12 is used as the sensor 1. The air pressure sensor 12 is installed in the mold 5 and measures the air pressure inside the mold 5 (see FIG. 2). Specifically, the air pressure sensor 12 is installed at the end of the cavity of the mold 5 and measures the air pressure in the cavity space. The cavity space is filled with molten molding material from the injection unit 6. The gas vent is provided to exhaust air or gas present in the cavity space of the mold 5 to the outside space, and is a passage that connects the cavity space to the outside space. Here, air pressure data (sensing data) measured by the air pressure sensor 12 is used to detect clogging (abnormality) of the gas vent.

[0040] The atmospheric pressure sensor 12 may be, for example, an atmospheric pressure sensor for an automobile exhaust system as shown in Fig. 8. Here, a degassing unit 121 is attached to the tip of the atmospheric pressure sensor 12 to prevent the molding material from entering the atmospheric pressure sensor 12.

[0041] Figure 9 shows an example of air pressure data measured by air pressure sensor 12 during the injection molding process from the start of injection to the start of mold opening. In the example shown, normal air pressure data 911, where the gas vent is not clogged, has a peak air pressure value of about 5 PSI immediately after the start of injection. On the other hand, abnormal air pressure data 912, where the gas vent is clogged, has a peak air pressure value of over 7 PSI immediately after the start of injection. The valve shown in Figure 9 is disposed between air pressure sensor 12 and the cavity inside mold 5.

[0042] In this embodiment, the detection unit 32 of the processing device 3 detects a clogged gas vent by using normal air pressure data stored in the memory unit 34. Specifically, the detection unit 32 detects a clogged gas vent by comparing the air pressure data measured by the air pressure sensor 12 with the normal air pressure data in the memory unit 34.

[0043] When the detection unit 32 acquires the air pressure data, it compares the peak value of the air pressure immediately after the start of injection with the peak value of the normal air pressure data immediately after the start of injection to detect the clogging of the gas vent. For example, if the peak value of the acquired air pressure data is greater than the peak value of the normal air pressure data by a predetermined threshold, the detection unit 32 determines that the gas vent is clogged and detects an abnormality. On the other hand, if the difference between the peak value of the acquired air pressure data and the peak value of the normal air pressure data is equal to or less than the predetermined threshold, the detection unit 32 determines that the gas vent is not clogged, i.e., is normal. Note that the normal air pressure data in the memory unit 34 is set in correspondence with each process of the injection molding machine 9, so the detection unit 32 can identify each process of the acquired air pressure data.

[0044] When the detector 32 detects a clogged gas vent, the output unit 33 displays the abnormality on the display, which prompts the user to clean the mold 5 and remove the clogged gas vent.

[0045] The sensor 1 of this embodiment described above is an air pressure sensor installed in the mold 5 of the injection molding machine, which measures the air pressure inside the mold 5, and the processing device 3 detects clogging of the gas vent of the mold 5 using the peak air pressure value measured by the air pressure sensor 12 at the start of injection.

[0046] Until now, a skilled technician has detected a clogged gas vent by visually inspecting the mold 5 or the molded product, but according to this embodiment, a clogged gas vent can be easily detected using the air pressure data from the air pressure sensor 12, thereby reducing defects in molded products produced by an injection molding machine.

[0047] <Example 3: Axial tension bolt> In this embodiment, an axial force bolt 13 with an embedded strain gauge is used as the sensor 1. The axial force bolt 13 detects the elongation and loosening of the bolt as a strain amount (strain data). The axial force bolt 13 is installed in the locking block of the mold 5 as shown in FIG. 10. The axial force bolt 13 may be embedded in the locking block by, for example, screwing it in. The locking block is a component for suppressing deformation of the mold 5. Here, the strain amount (strain data) measured by the axial force bolt 13 is used to detect abnormalities in the locking block. Abnormalities in the locking block include deformation of the locking block, galling (mold wear, chipping), and the like.

[0048] 11 shows an example of strain data (sensing data) measured by the axial tension bolt 13 during the injection molding process from the completion of mold clamping to the start of mold opening. In the example shown, strain data 111 of a normal locking block and strain data 112 of an abnormal locking block are shown.

[0049] In the illustrated example, there are two locations 113, 114 where abnormal strain data 112 exists. In the first location 113, the normal strain data 111 shows an increase in strain amount after a predetermined time has elapsed since the start of injection, whereas the abnormal strain data 112 shows an increase in strain amount immediately after the start of injection. The abnormal strain data 112 in the first location 113 indicates deformation of the locking block due to burrs or the like in the molded product.

[0050] Furthermore, in the case of abnormal strain data 112 at the second location 114 at the start of mold opening, if an attempt is made to forcibly open the mold in a state in which the locking block is deformed due to the occurrence of burrs or the like, the amount of strain increases because scraping or wear (galling) of the mold occurs. On the other hand, in the case of normal strain data 111 where scraping or wear (galling) of the mold does not occur, the amount of strain does not increase at the start of mold opening.

[0051] In this embodiment, the detection unit 32 of the processing device 3 detects an abnormality in the locking block by using normal strain data stored in the memory unit 34. Specifically, the detection unit 32 detects an abnormality in the locking block by comparing the strain data measured by the axial tension bolt 13 with the normal strain data in the memory unit 34.

[0052] When the detector 32 acquires the strain data, it detects an abnormality in the locking block by determining, for example, whether the amount of strain increases immediately after the start of injection or whether the amount of strain increases after a predetermined time has passed since the start of injection. Note that, since the normal strain data in the memory 34 is set in association with each process of the injection molding machine 9, each process of the acquired strain data can be identified.

[0053] When the detection unit 32 detects an abnormality (deformation, galling, etc.) in the locking block, the output unit 33 displays the abnormality on a display, which allows the user to repair the locking block according to the type of abnormality displayed on the display.

[0054] The sensor 1 of the present embodiment described above is an axial force bolt 13 with an embedded strain gauge that is installed in the locking block of the mold 5 of the injection molding machine and measures the amount of strain in the locking block. The processing device 3 detects abnormalities in the locking block using the amount of strain in the axial force bolt during injection molding.

[0055] According to this embodiment, abnormalities in the locking block can be easily detected using the strain data of the axial tension bolt 13, and defects in molded products produced by the injection molding machine can be reduced.

[0056] <Example 4: Stereo ultrasonic detector> In this embodiment, a stereo ultrasonic detector 14 (bat detector, ultrasonic octaver) and a piezoelectric pickup 15 are used as the sensor 1 to detect an abnormality in the injection unit 6 during injection molding. That is, the sensor 1 includes the piezoelectric pickup 15 installed on the screw 64 of the injection unit 6, and the stereo ultrasonic detector 14 connected to the piezoelectric pickup 15. The stereo ultrasonic detector 14 and the piezoelectric pickup 15 are connected by electrical wiring.

[0057] The piezo pickup 15 is disposed in the injection unit 6. The piezo pickup 15 may be disposed near the screw 64, for example, between the hopper 65 and the screw 64 (see FIG. 2). In the example shown in FIG. 2, the stereo ultrasonic detector 14 is disposed in the injection unit 6, but the location of the stereo ultrasonic detector 14 is not limited to within the injection unit 6.

[0058] The piezo pickup 15 is an element that outputs vibration as an electric signal, and converts the vibration of the screw 64 into an ultrasonic electric signal. Specifically, the piezo pickup 15 detects vibration including ultrasonic waves up to 40,000 Hz of the injection unit 6, converts it into an electric signal (voltage) including ultrasonic waves, and outputs the converted electric signal to the stereo ultrasonic detector 14 via electric wiring.

[0059] The stereo ultrasonic detector 14 is an ultrasonic modulator that converts an electrical signal containing ultrasonic waves into an electrical signal in the audible range and generates a spectrogram of the converted electrical signal in the audible range. Here, the stereo ultrasonic detector 14 converts the electrical signal detected by the piezo pickup 15 into an electrical signal up to 20,000 Hz that can be heard by the human ear. By performing machine learning using the spectrogram of the electrical signal in which vibrations (sounds) up to 40,000 Hz have been converted into up to 20,000 Hz, it becomes possible to determine whether sounds in a frequency band that cannot actually be heard by the human ear are normal or abnormal.

[0060] By attaching the piezo pickup 15 near the screw 64, abnormalities in the injection unit 6 such as wear of the screw 64, measurement errors, and clogging of the nozzle 60 can be efficiently detected.

[0061] FIG. 12 shows an example of a spectrogram (sensing data) of an audible electric signal converted by the stereo ultrasonic detector 14 during injection molding. In the example shown, two types of spectrograms are shown, one for a frequency band of 0 to 20,000 Hz and the other for a frequency band of 0 to 40,000 Hz. The spectrogram of the audible range converted by the stereo ultrasonic detector 14 (the spectrogram used in machine learning) is a spectrogram of a frequency band of 0 to 20,000 Hz. The spectrogram of the frequency band of 0 to 40,000 Hz shows a spectrogram of an electric signal including ultrasonic waves before conversion, for reference. The horizontal axis indicates time, the vertical axis indicates frequency, and the color (brightness) indicates volume (frequency intensity). Here, a spectrogram showing an abnormality of the injection unit 6 in the injection molding process is shown as an example.

[0062] In the spectrogram when the screw 64 is worn, a low tone appears periodically. This indicates that the rotation of the screw 64 is undergoing precession. For such a spectrogram, the processing device 3 detects the wear of the screw 64 as an abnormality.

[0063] Furthermore, in the spectrogram in which a metering failure occurs, a characteristic double tone appears. The processing device 3 detects an abnormality in such a spectrogram as indicating the occurrence of a metering failure.

[0064] Furthermore, an impact sound during injection appears in the spectrogram when clogging occurs in the nozzle 60. The processing device 3 detects an abnormality in such a spectrogram as indicating that clogging has occurred in the nozzle 60.

[0065] In this embodiment, the processing device 3 detects an abnormality in the injection unit 6 by using the trained model. The trained model is stored in the memory unit 34 of the processing device 3. In addition, the detection unit 32 inputs the spectrogram output by the stereo ultrasonic detector 14 into the trained model generated by machine learning, and determines whether or not an abnormality has occurred in the injection unit 6. Specifically, it determines whether or not various abnormalities have occurred, such as wear of the screw 64, poor metering, or clogging of the nozzle 60.

[0066] When a spectrogram is input, the trained model outputs the probability of normality and abnormality. The trained model may output a probability for each type of abnormality. The trained model is generated by performing machine learning using training data. The training data is, for example, a correspondence between input data (spectrogram) and a label indicating whether or not each type of abnormality is abnormal.

[0067] The trained model includes a neural network and parameters (weighting values, bias). Machine learning is a process of optimizing the parameters of a neural network. Before learning, the parameters are set to initial values. In this embodiment, machine learning is performed by a neural network configured by combining neurons. Specifically, a set of input data and a label included in the teacher data is given to the neural network, and learning is repeated while changing the weighting value and bias of each neuron so that the output of the neural network is the same as the label. In this way, the input data of the teacher data is trained, and a trained model for detecting anomalies from the input data is generated. Note that the neural network includes an input layer, an intermediate layer, and an output layer, and each layer is composed of neurons. The intermediate layer may be composed of one layer or multiple layers.

[0068] The detection unit 32 detects an abnormality in the injection unit 6 using the probability output by the trained model. Specifically, the detection unit 32 determines that no abnormality has occurred when the normality probability is at a maximum. Furthermore, the detection unit 32 detects an abnormality in the wear of the screw 64 when the wear probability of the screw 64 is at a maximum.

[0069] When the detection unit 32 detects an abnormality, the output unit 33 displays the abnormality on a display, which allows the user to repair the injection unit 6 according to the type of abnormality displayed on the display.

[0070] The sensor 1 of this embodiment described above includes a piezoelectric pickup 15 installed on the screw 64 of the injection unit 6, and a stereo ultrasonic detector 14 connected to the piezoelectric pickup 15. The piezoelectric pickup 15 converts the vibration of the screw 64 into an ultrasonic electrical signal, and the stereo ultrasonic detector 14 converts the ultrasonic electrical signal into an audible electrical signal and generates a spectrogram of the converted electrical signal. The processing device 3 inputs the spectrogram into a trained model to detect an abnormality in the injection unit 6.

[0071] Until now, a skilled technician has detected abnormalities by the sound of the injection unit 6, but according to this embodiment, abnormalities in the injection unit 6 can be easily detected by using the spectrogram of the stereo ultrasonic detector 14, thereby reducing molding defects.

[0072] In addition, in Examples 1 to 3 and Example 5 described later, the processing device 3 may generate a trained model in the same manner as in this example, and detect anomalies using the trained model.

[0073] <Example 5: Gas concentration detection sensor> In this embodiment, a gas concentration detection sensor 16 installed in a hopper 65 of an injection unit 6 is used as the sensor 1 to detect abnormalities in the preliminary drying of the molding material charged into the hopper 65 .

[0074] Usually, prior to injection molding, the molding material (such as resin pellets) is pre-dried, and the pre-dried molding material is charged into the hopper 65. However, there are cases where abnormalities in the pre-drying of the molding material occur due to a breakdown in the dryer, insufficient drying time, etc., and the molding material with the abnormalities in the pre-drying is charged into the hopper 65 and injection molded. When abnormalities in the pre-drying occur, it can cause deterioration of the physical properties of the molded product, flow marks, burrs, gas burns, peeling, swelling (blisters) in the molded product, discoloration, measurement failure, etc. In this embodiment, abnormalities in the pre-drying of the molding material are detected using a gas concentration detection sensor 16.

[0075] The gas concentration detection sensor 16 of this embodiment measures the concentration of at least one of volatile organic compound gas (hereinafter, VOC gas) and carbon dioxide (hereinafter, CO2) in the hopper 65.

[0076] The table below shows the VOC gas and CO2 concentrations measured when molding material without pre-drying was charged into each hopper 65 of two injection molding machines (molding machine A and molding machine B), and the VOC gas and CO2 concentrations measured when molding material with pre-drying was charged.

[0077] [Table 1]

[0078] As shown in the table above, in both molding machine A and molding machine B, the VOC gas concentration is higher when the molding material is not pre-dried before injection molding (abnormal pre-drying) than when the molding material is pre-dried. Similarly, the CO2 concentration is higher when the molding material is not pre-dried before injection molding (abnormal pre-drying) than when the molding material is pre-dried.

[0079] That is, it can be seen that the concentrations of VOC gases and CO2 in the hopper 65 are high when the molding material is abnormally pre-dried, and the concentrations of VOC gases and CO2 in the hopper 65 are low when the molding material is normally pre-dried.

[0080] The humidity inside the hopper 65 without pre-drying is also higher than the humidity inside the hopper 65 with pre-drying.

[0081] The processing device 3 of this embodiment detects abnormality in the preliminary drying of the molding material put into the hopper 65, using the concentration measured by the gas concentration detection sensor 16. Specifically, the detection unit 32 of the processing device 3 detects abnormality in the preliminary drying of the molding material, using threshold values ​​set for each type of gas (VOC gas, CO2) stored in the memory unit 34.

[0082] For example, the detection unit 32 may detect a preliminary drying abnormality when the concentration of the VOC gas is equal to or greater than a first threshold value and the concentration of the CO2 is equal to or greater than a second threshold value. The detection unit 32 may also detect a preliminary drying abnormality using only the concentration of the VOC gas when the concentration of the VOC gas is equal to or greater than a third threshold value. The detection unit 32 may also detect a preliminary drying abnormality using only the concentration of the CO2 when the concentration of the CO2 is equal to or greater than the third threshold value.

[0083] When the detection unit 32 detects a preliminary drying abnormality, the output unit 33 displays the abnormality on the display. This allows the user to replace the molding material put into the hopper 65.

[0084] The sensor 1 of this embodiment described above is a gas concentration detection sensor 16 installed in the hopper 65 of the injection unit 6, which measures the concentration of at least one of the VOC gases and CO2 in the hopper 65, and the processing device 3 detects abnormalities in the pre-drying of the molding material charged into the hopper 65 using the concentrations measured by the gas concentration detection sensor 16.

[0085] According to this embodiment, the gas concentration detection sensor 16 installed inside the hopper 65 can be used to easily detect abnormalities in the preliminary drying of the molding material, and defects in the molded products molded by the injection molding machine 9 can be easily detected.

[0086] <Modification> When the processing device 3 detects an abnormality in the injection molding machine 9, it may control the injection molding machine 9. In this case, the processing device 3 and a control unit (not shown) of the injection molding machine 9 are connected via a wired or wireless network.

[0087] 13 shows an example of a table in which an abnormality detected by the detection unit 32 of the processing device 3 is associated with a control instruction for the injection molding machine 9. The table is stored in advance in the storage unit 34 of the processing device 3.

[0088] For example, when the detection unit 32 detects an abnormality (oil shortage, installation abnormality) of the mold 5 using the gravitational acceleration data of the acceleration sensor 11, it transmits a control signal to the control unit of the injection molding machine 9 to reduce the mold opening / closing speed to 1 / 2. When the detection unit 32 detects an abnormality of the mold using the air pressure data of the air pressure sensor 12, it transmits a control signal to the control unit of the injection molding machine 9 to reduce the final stage of the injection speed to 1 / 2 and increase the VP switching position by 10%. When the detection unit 32 detects an abnormality in which the mold release resistance exceeds the upper limit using the sensing data of the ejection pressure sensor (not shown), it transmits a control signal to the control unit of the injection molding machine 9 to reduce the ejection speed to 1 / 2 and extend the cooling time by 20%. When the detection unit 32 detects an abnormality in the mold opening amount using the sensing data of the mold opening sensor (not shown), it transmits a control signal to the control unit of the injection molding machine 9 to reduce the injection pressure to 1 / 2.

[0089] In this modified example, when the processing device 3 uses sensing data to detect an abnormality in the injection molding machine 9 occurring at a specified position, it sends a control instruction corresponding to the abnormality in the sensing data to the injection molding machine 9 and controls the injection molding machine 9.

[0090] As a result, in this modified example, measures against abnormalities in the injection molding machine 9 can be automatically executed.

[0091] The above-described processing device 3 may be, for example, a general-purpose computer system as shown in Fig. 14. The computer system shown in the figure includes a CPU (Central Processing Unit, processor) 901, a memory 902, a storage 903 (HDD: Hard Disk Drive, SSD: Solid State Drive), a communication device 904, an input device 905, and an output device 906. The memory 902 and the storage 903 are storage devices. In this computer system, the CPU 901 executes a predetermined program loaded on the memory 902, thereby realizing each function of the processing device 3.

[0092] The processing device 3 may be implemented in one computer or in multiple computers. The processing device 3 may be a virtual machine implemented in a computer. The program of the processing device 3 may be stored in a computer-readable recording medium such as a HDD, SSD, a Universal Serial Bus (USB) memory, a Compact Disc (CD), or a Digital Versatile Disc (DVD), or may be distributed via a network. The computer-readable recording medium is, for example, a non-transitory recording medium.

[0093] The present invention is not limited to the above-described embodiment, and various modifications are possible within the scope of the present invention. [Explanation of symbols]

[0094] 1: Sensor 11: Acceleration sensor 12: Barometric pressure sensor 13: Axial tension bolt 14: Stereo ultrasonic detector 15: Piezo pickup 16: Gas concentration detection sensor 2: Microcomputer 3: Processing equipment 31: Acquisition part 32: Detection unit 33: Output section 34: Storage part

Claims

1. a sensor that is installed at a predetermined position in the injection molding machine and measures sensing data for detecting an abnormality of the injection molding machine occurring at the predetermined position; a processing device that acquires the sensing data and detects an abnormality of the injection molding machine occurring at the predetermined position by using the sensing data. Anomaly detection systems.

2. The sensor is a three-axis acceleration sensor installed in a mold of the injection molding machine, and measures the gravitational acceleration; The processing device detects a shortage of oil in the mold or an abnormality in the installation of the mold by using the amplitude of the gravitational acceleration. The anomaly detection system of claim 1 .

3. the sensor is an air pressure sensor installed in a mold of the injection molding machine and measures the air pressure inside the mold; The processing device detects clogging of the gas vent of the mold using the peak value of the air pressure measured by the air pressure sensor at the start of injection. The anomaly detection system of claim 1 .

4. the sensor is an axial tension bolt having an embedded strain gauge that is installed in a locking block of a mold of the injection molding machine and measures the amount of strain of the locking block; The processing device detects an abnormality in the locking block by using a strain amount of the axial tension bolt during injection molding. The anomaly detection system of claim 1 .

5. The sensor includes a piezo pickup installed on the screw of the injection unit and a stereo ultrasonic detector connected to the piezo pickup; The piezoelectric pickup converts the vibration of the screw into an ultrasonic electrical signal, The stereo ultrasonic detector converts the ultrasonic electrical signal into an electrical signal in an audible range and generates a spectrogram of the converted electrical signal; The processing device inputs the spectrogram into a trained model to detect an anomaly in the injection unit. The anomaly detection system of claim 1 .

6. The sensor is a gas concentration detection sensor installed in a hopper of the injection unit, and measures the concentration of at least one of a volatile organic compound gas and carbon dioxide in the hopper; The processing device detects abnormality in preliminary drying of the molding material put into the hopper by using the concentration measured by the gas concentration detection sensor. The anomaly detection system of claim 1 .

7. The processing device acquires the sensing data from at least one of the sensors via a microcomputer. The anomaly detection system of claim 1 .

8. When the processing device detects an abnormality of the injection molding machine occurring at the predetermined position using the sensing data, the processing device transmits a control instruction corresponding to the abnormality of the sensing data to the injection molding machine. The anomaly detection system of claim 1 .

Images