JP2020044836A - Injection molding machine, state notification system of injection molding machine, and state notification method of injection molding machine - Google Patents

Abstract

To provide an injection molding machine, a state notification system of the injection molding machine and a state notification method of the injection molding machine capable of appropriately notifying a maintenance state.SOLUTION: An injection molding machine 11 comprises: detection means 35 which detects that an operating state of the injection molding machine 11 is a non-steady value V; and counting means 29a which counts an integrated value N composed of at least one of the number of conversion hours and the number of conversion frequencies. The injection molding machine 11 further comprises notification means 29, 30 which notify maintenance states A, B of the injection molding machine 11 when a value M calculated from both the non-steady value V and the integrated value N reaches prescribed threshold values ma, mb or exceeds the threshold values ma, mb.SELECTED DRAWING: Figure 1

Description

The present invention relates to an injection molding machine that reports a maintenance status, a status reporting system of the injection molding machine that reports a maintenance status, and a status reporting method of the injection molding machine that reports a maintenance status.

2. Description of the Related Art Conventionally, as an information management apparatus for an injection molding machine capable of accurately detecting an abnormality of a part used for injection molding, one described in Patent Document 1 is known. In Patent Document 1, in a cycle operation for repeatedly manufacturing a molded article under the same setting conditions, an n-th (n is a natural number of 2 or more) shot and an m-th shot (where m is a predetermined positive natural number obtained by subtracting a predetermined positive natural number from n). By calculating the difference between the actual number of shots and the actual value of the physical quantity representing the load on the component, an abnormality of the component is detected. More specifically, as described in (0058) to (0064) of the specification, the abnormality is detected based on the physical quantity representing the load. Then, as described in (0074) to (0077), when a slight abnormality has occurred, the cycle operation may be continued. When a severe abnormality has occurred, the cycle operation is interrupted.

In addition, by diagnosing the information of the injection molding machine, it is possible to estimate and prevent the occurrence of abnormalities, to prevent erroneous diagnosis of abnormalities, and to prevent unexpected production stoppage as an injection molding system. Is described in Patent Document 2. In Patent Document 2, the injection molding machine and the diagnostic server are connected by a communication line, and the measured value of the heat value of the motor is compared with the estimated value, and the difference between the measured values is outside the allowable range or exceeds the allowable range. It is described that the diagnostic request data is transmitted to the diagnostic server when the number of times is determined to be the allowable number of abnormalities (0038). This makes it possible to prevent abnormal development due to delays in maintenance management (0045).

What is common to the two patent documents 1 and 2 is not to stop an injection molding machine immediately after detecting an abnormal state, but to notify while continuing operation when a slight abnormality is detected. Thus, countermeasures are taken before a severe abnormality occurs.

JP-A-2017-87588 (Claim 1, (0058) to (0064), (0074) to (0077), FIG. 1) Japanese Patent Application Laid-Open No. 2014-184641 (Claims 1, (0038), (0045), (FIG. 1), (FIG. 3)

However, it is very difficult to accurately detect the occurrence of a slight abnormality that can continue the operation as described in Patent Literature 1 and Patent Literature 2. That is, it is difficult to accurately detect minor abnormalities under any conditions according to the change of the material to be formed, the change of the forming condition accompanying the change, and the change of disturbance factors such as weather conditions such as temperature and humidity. was there. Further, Patent Documents 1 and 2 do not disclose at what time an abnormality detected in the injection molding machine is transmitted or not transmitted.

Therefore, an object of the present invention is to provide an injection molding machine, a status reporting system of an injection molding machine, and a method of reporting a state of an injection molding machine, which can appropriately report a maintenance state.

An injection molding machine according to claim 1 of the present invention is an injection molding machine for notifying a maintenance state, wherein a detecting means for detecting that an operation state of the injection molding machine is an unsteady value, a conversion time or a conversion count. Counting means for counting an integrated value consisting of at least one of the numbers,
A notifying means is provided for notifying a maintenance state of the injection molding machine when a value calculated from both the unsteady value and the integrated value becomes a predetermined threshold value or exceeds a threshold value.

In the injection molding machine according to a second aspect of the present invention, in the first aspect, the unsteady value and the integrated value are each converted into a numerical value, and a value obtained by adding both numerical values becomes a predetermined threshold or exceeds the threshold. In this case, a notifying means for notifying the maintenance state of the injection molding machine is provided.

According to a third aspect of the present invention, in the injection molding machine according to the first or second aspect, the value calculated from both the unsteady value and the integrated value is weighted by a machine learning device using a neural network. It is characterized by having been done.

The status notification system for an injection molding machine according to claim 4 of the present invention is a status notification system for an injection molding machine for notifying a maintenance status, wherein the central control device in a company in which the injection molding machine is disposed or the injection molding. The control device is connected to a control device outside the company where the machine is arranged, and is provided so as to be able to transmit a maintenance state of the injection molding machine to the control device outside the company.

In the state notification system for an injection molding machine according to claim 5 of the present invention, in claim 4, the maintenance state is notified at least on a screen of the injection molding machine, and when the operator permits transmission from the screen or another device. The maintenance state of the injection molding machine is transmitted to a control device outside the company.

The status notification system for an injection molding machine according to claim 6 of the present invention, according to claim 4 or 5, wherein the control device inside the central control device inside the company or outside the company where the injection molding machine is arranged. A machine learning device provided inside, or a machine learning device provided directly connected to a central control device inside the company or directly connected to a control device outside the company where the injection molding machine is arranged. I do.

According to a seventh aspect of the present invention, there is provided a method for informing a state of an injection molding machine for notifying a maintenance state, wherein the detecting means detects that an operation state of the injection molding machine is an unsteady value. In addition, when the integrated value of at least one of the number of conversion hours or the number of conversions is counted by the counting means, when a value calculated from both the unsteady value and the integrated value becomes a predetermined threshold or exceeds the threshold The maintenance state is reported by the reporting means.

According to a seventh aspect of the present invention, in the method for informing a state of an injection molding machine, different maintenance states are reported in accordance with a plurality of threshold values.

According to a ninth aspect of the present invention, in the method of informing a state of an injection molding machine according to the seventh or eighth aspect, the notification of the different maintenance state uses a machine learning device that performs supervised learning. .

The injection molding machine of the present invention is an injection molding machine for notifying a maintenance state, comprising: a detection unit for detecting that an operation state of the injection molding machine is an unsteady value; and at least one of a conversion time or a conversion count. Counting means for counting an integrated value, and a notification for notifying a maintenance state of the injection molding machine when a value calculated from both the unsteady value and the integrated value becomes a predetermined threshold or exceeds the threshold. Since the means is provided, the maintenance state can be properly notified.

1 is a schematic diagram of an injection molding machine and a state notification system of the injection molding machine of the present embodiment. It is explanatory drawing which shows the calculation method of the maintenance state of the injection molding machine of this embodiment, the state notification system of an injection molding machine, and the state notification method of an injection molding machine. It is a flow chart figure of an injection molding machine of this embodiment, a state reporting system of an injection molding machine, and a state reporting method of an injection molding machine. It is the schematic of the injection molding machine of 2nd Embodiment, and the state notification system of an injection molding machine. It is a block diagram of an injection molding machine of a 3rd embodiment, and a control device of a state reporting system of an injection molding machine. It is explanatory drawing which shows the neural network of the machine learning apparatus of the control apparatus of the injection molding machine of 3rd Embodiment, and the state notification system of an injection molding machine.

The injection molding machine 11 and the state notification system 12 of the injection molding machine of the present invention will be described with reference to FIG. The injection molding machine 11 is an injection molding machine capable of notifying a maintenance state, and is connected to a central control device 14 in a factory 13 or an external control device 16 such as an injection molding machine manufacturer 15.

The injection molding machine 11 is provided with an injection device 17 for injecting a molten resin and a mold clamping device 19 for clamping a molding die 18. The injection device 17 has a built-in screw 22 having a check valve 21 in a heating cylinder 20. The screw 22 is rotated by a servomotor 23 for measurement, and is moved forward and backward by a servomotor 24 for injection and back pressure control. Is performed. The mold clamping device 19 includes a mold clamping mechanism 25 and a mold opening / closing mechanism 26. A fixed mold 18 a is attached to the fixed board 27, and a movable mold 18 b is attached to the movable board 28. However, the injection molding machine 11 may be operated by another mechanism, and the type thereof is not limited.

Further, the injection molding machine 11 includes a control device 29, and a display device 30 also serving as an operation screen is connected to the control device 29. The display device 30 includes a touch panel of a liquid crystal panel. In addition to displaying various information on the injection molding machine, the operator touches the touch panel to input molding conditions and the like. In addition to the display device 30 including the touch panel, button keys such as numeric keys and various function keys may be provided. The notifying means for notifying the maintenance state of the injection molding machine 11 includes a control device 29, a display device 30, and the like.

Further, the control device 29 is connected to various sensors (not shown) which are attached to the injection molding machine 11 and are detection means for detecting the state of the injection molding machine 11. Since the sensors attached to the injection molding machine 11 are known, they will not be described in detail, but typical ones of the servo motors such as a servo motor 23 for measurement, a servo motor 24 for injection, and a servo motor of a mold opening / closing mechanism 25 are exemplified. A rotary encoder that detects the rotation angle and the position of the movable part moved by each servomotor, a position sensor such as a linear scale that detects the position of the movable part, a current sensor and a voltage sensor that detects the load and torque of the servomotor, Temperature sensors such as thermocouples that detect heating of parts heated by heaters and switchboards, force sensors such as load cells that detect the force applied to movable parts, and hydraulic pressures of various parts of hydraulic mechanisms such as mold clamping cylinders Hydraulic sensors, etc. Alternatively, a sensor such as a vibration sensor or a sound level meter may be separately attached to detect an unsteady state of the injection molding machine 11.

These sensors detect signals at the time of sequence control or the like at the time of molding of the injection molding machine 11, but in the present invention, they also serve as detection means for detecting that the operation state of the injection molding machine 11 is in an unsteady state. . Further, the detecting means for detecting the state of the injection molding machine 11 may be a peripheral device included in the injection molding machine in a broad sense or a device attached near the injection molding machine. In the present embodiment, an unloader 34 is mounted on the upper part of the fixed platen 27, and the unloader 34 is provided with a sensor 35 composed of a load cell for detecting the weight of the molded product. The apparatus for measuring the weight of a molded article as the detecting means may be arranged in the vicinity of the injection molding machine 11.

The control device 29 includes a storage unit, a calculation unit, an input / output unit, and the like. The storage unit stores a program for sequence control at the time of performing injection molding, and also stores data and a program for state detection of the present embodiment. In addition, the arithmetic unit 29a determines the state of the injection molding machine 11 of the present embodiment in addition to various data processing during injection molding. Further, the input / output unit is connected to the actuator and the sensor. The transmission of the output signal from the computing device to the actuator and the transmission of the input signal from the sensor or the like to the computing unit 29a are performed via the input / output unit. Further, in the present embodiment, the timer is provided for the arithmetic unit 29a or the arithmetic unit 29a, and detects the operation time of each unit. In the present invention, the calculation unit 29a including a timer corresponds to a counting unit that counts the operation time and processes the result as the integrated value N.

The control device 29 of the injection molding machine 11 is connected via a line 31 to the central control device 14 in the factory 13 where the injection molding machine 11 is located or in the company via an input / output unit. In these, the line 31 between the personal computer 14a including the server of the central control unit 14 and each of the injection molding machines 11, 11, 11 in the company may be a dedicated communication line, or an intranet constructed inside the company. Or an Ethernet (registered trademark) line. The line connection between the injection molding machine 11 and the central controller 14 may be a connection via a wireless LAN or WiFi, and the communication method is not limited.

The function of calculating the maintenance index value M of the present invention, which will be described in detail later, may be held by the control device 29 of the injection molding machine 11 or may be held by the central control device 14. You may. In general, when the calculation of the maintenance index value M is limited to a slight calculation process, the calculation can be performed by the control device 29 of the injection molding machine 11, but when performing a high-level calculation process, the calculation is performed by the central control device 14. Preferably, processing is performed. These are also closely related to the system construction status of the company that owns and operates the injection molding machine 11.

The central control device 14 in the company where the injection molding machine 11 is operating and the control device 16 of the injection molding machine manufacturer 15 outside the company are connected by a line 32 such as the Internet. The default gateway 16b of the personal computer 16a of the control device 16 outside the company is connected via the default gateway 14b of the personal computer 14a which is the central control device 14 for performing data transmission in the company in which the injection molding machine 11 is arranged, In each case, intrusion of virus software and the like from outside is prevented.

In the present embodiment, the control device 16 outside the company refers to the control device 16 of the injection molding machine maker 15, but may be a control device of a company that specializes in maintenance. The control device 16 on the side of the injection molding machine maker 15 may be located at a site such as a head office having a function of supervising the entire service operation, but may be located at each service site in the area of the company that owns the injection molding machine 11. It may be. The control device 16 can also be linked to the mobile phone of each service person. Further, the communication between the central control device 14 inside the company where the injection molding machine 11 is arranged and the control device 16 outside the company is not limited to the Internet, but may be a telephone line (including facsimile) other than the Internet or a dedicated line between two companies. And the communication method is not limited.

Next, a method of reporting the state of the injection molding machine 11 using the state reporting system 12 of the injection molding machine will be described with reference to FIG. A first feature of the present invention is that the operating state of the injection molding machine 11 is determined by combining both an unsteady value V detected by a detecting means such as a sensor and an integrated value N such as a molding time counted by a timer. The point is to judge. These determinations are made for each part and each part of the injection molding machine 11. More specifically, the detection and notification of the operation state of the injection molding machine 11 is performed by distributing various motors, such as a switchboard such as a motor control device, a speed reducer, a ball screw mechanism, a mold clamping cylinder, a hydraulic mechanism, a hydraulic pump, a movable plate 28, and the like. , The screw 22, the check valve 21, the heater, the load cell for injection and mold clamping, etc. are separately performed. Further, in the above description, each part indicates that the judgment and notification for two or more parts are performed collectively.

First, detection and use of the unsteady value V during operation of the injection molding machine 11 will be described. The type of the sensor used for detecting the unsteady value V of each part or each part is not limited as described above, and it is preferable that the type used for controlling the injection molding machine 11 is also used. May be provided exclusively for the detection of. Further, a portion where the sensor 35 is provided is not limited.

The determination as to whether or not the unsteady value V has been detected in the injection molding machine 11 in the operating state is made by comparison with the steady value. A general method of setting a steady value indicates the state of the molding machine, such as vibration, noise, molding time, and current value, when the injection molding machine 11 is newly built or when the parts are replaced or when the operation is performed within a certain period of time after the new construction or replacement. This is performed by, for example, storing the average value of the values detected in a state where the value is within the setting and a good molded article is molded in the storage unit. The storage of the steady-state value is performed automatically or by a worker storing it in a storage unit. Further, the steady value may be set using a calculated value or a value specified by a part maker. Further, the steady value may be calculated using an average value, a differential value, or the like of measured values during a molding cycle for a certain period in which molding is performed stably when the steady value differs depending on molding. As an example, the temperature of a switchboard or a motor measured by a temperature sensor, the resistance value of a heater measured by a current sensor, the vibration of a movable part or a pump measured by a vibration sensor, the clamping cylinder measured by a hydraulic sensor. As the state of the seal member or the like, a steady value that is fixedly set in advance may be used. Measuring the motor current value (torque) with a movable part such as a movable platen including a ball screw mechanism, a screw 22 including a check valve 21, a reduction gear, etc., is caused by fluctuations due to molding material, mold weight, cycle time, etc. The steady-state value may be calculated from the average value of a plurality of measured values during the molding cycle.

Then, a threshold (unsteady value judgment value) having a certain width with respect to the steady value is set for the judgment of the unsteady value V, and is stored in the storage unit together with the steady value. The extent to which a value deviating from the steady value is set as the unsteady value determination value differs depending on each component or part. In order to catch a minor abnormality of a part or a part or a precursor thereof, it is necessary to catch an extremely small unsteady state, and the unsteady value V needs to be set to a value slightly deviated from the steady value. If the unsteady value V is set to a small value, the unsteady value V may exceed the unsteady value V due to a disturbance element even though the parts and parts of the injection molding machine 11 are operating normally. However, in the present invention, in most cases, the maintenance state is not determined based on exceeding the unsteady value V only once. Therefore, it is desirable to count the fact that the unsteady value V is detected for any reason, because there is a possibility of minor abnormality of parts or parts. The determination of the unsteady state using the unsteady value V is usually performed by cumulatively adding the number of times the unsteady value V is detected to a converted value. However, the value or amount of the unsteady value V deviating from the steady value may be added to the calculation for the converted value, or a combination of the two may be used.

Further, it is often the case that it is suspected that one of the plurality of components is abnormal due to the detection of one unsteady value V. For example, when an unsteady value is detected in the current value of the mold opening / closing servomotor that moves the movable platen 28, a problem is caused in a mechanical unit such as a linear guide, a tie bush, or a ball screw mechanism (including loosening of a bolt in a mounting part). May be occurring, or a problem may be occurring in the servomotor itself or its control system. In these cases, even if it is not clear which part has an abnormality, the operator is first notified of the state by notifying the display device 30 of the state, and the operator is informed of which part may be abnormal. The operator sees the notification and determines whether to contact the injection molding machine maker 15 or the like outside the company.

Further, when it is suspected that any of the plurality of components has an abnormality, it may be possible to estimate which component is likely to have a problem by combining the values with other sensor values. For example, in the case of the mold opening / closing movement of the movable platen 28, whether an unsteady value occurs in the entire mold opening / closing section or an unsteady value occurs only in a specific section depending on the combination with the position sensor and the timer. By knowing the parts, it may be possible to estimate which part is likely to have a problem. Further, the history of the abnormality notification of the injection molding machine 11 is stored in the storage unit of the injection molding machine 11 or the storage unit of the central control unit 14, and by referring to the data when the same abnormality has occurred last time, In some cases, it is possible to predict whether the possibility of a component abnormality is high.

Next, detection and use of the integrated value N will be described. The integrated value N is a value that is gradually increased even if there is no abnormality during molding, and generally the molding time is most often used. The integrated value N is counted by a timer, which is a coefficient means of the calculation unit 29a. As the integrated value N, generally, an operation time from when the power of the entire apparatus is turned on to when the power is turned off is used. The start of counting by the timer may be any time, such as when the injection molding machine starts operating by any method, when automatic operation starts, or when the operation of each part is started, other than when the power is turned on.

When the operating time is used for the integrated value N, 1 hour = 1 is usually counted as the conversion time. If the injection molding machine is used under special conditions such as overload due to high acceleration or use of recycled materials, the load on each part will increase, so multiply the molding time by a constant coefficient such as 1.5 Then, the conversion time may be calculated. In addition, the conversion time, which is the integrated value N, may be integrated based on the number of days of molding or the value added to the unit of minutes or seconds below the time.

Further, the integrated value N may be a number such as the number of molding cycles. When a molding cycle is used as the integrated value N, normally, one molding cycle = 1 is counted as the number of conversions. However, the number of conversions during molding is such that the number of times the mold is opened and closed is counted as two times with the mold open and the mold closed in one molding cycle, and the number of rotations of the screw 22 such as the screw 22 and the measuring servomotor 23 is counted. May be counted. Also in the case where these conversion times are used for the integrated value N, a value obtained by multiplying a certain coefficient or the like in consideration of the load according to the state of use may be used as the conversion number. The integrated value N may use both the conversion time and the number of conversions.

The integrated value N such as the conversion time and the conversion count is counted for each of a plurality of components and stored in the storage unit. In other words, even when the injection molding machine 11 is new, a component for calculating the conversion time with the molding time = 1 as described above and a component for obtaining the conversion time by multiplying a constant coefficient according to the load as described above have an integrated value of the conversion time. N differs from one another. Further, the integrated value N such as the conversion time and the number of conversions is reset to 0 when the parts are replaced. As a result, even when there are a plurality of parts to be counted based on the integrated value N of the molding time = 1 when the injection molding machine 11 is shipped, the integrated values N constituting the maintenance index value M by the replacement of each part are respectively different values. Become. Various data such as the replacement date or replacement time of the parts, the numerical values of the sensors at the time of replacement, etc. are stored in the storage unit of the central control unit 14 automatically or by the input of the operator. Has become.

In FIG. 2, the conversion index V1 of the non-stationary value V is added to the integrated value N gradually increasing to the right, and the accumulated value is a maintenance index value M which is a judgment value for notifying a message about maintenance. It has become. In FIG. 2, the unsteady value V is detected four times, and the converted value V1 is added to the maintenance index value M based on the integrated value N each time. An example is shown in which the converted value V1 is added to the value M, and the first message A is reported beyond the threshold ma for reporting the first message A.

Next, the relationship with the manufacturer's warranty period and legal inspections is described. Generally, the warranty period of each part of the injection molding machine 11 is determined by a part maker. Each part also deteriorates due to the accumulation of the molding time or the number of moldings, or simply aging. However, each component can be used for a considerable period of time, even though the warranty period of the component maker has passed, although the probability of failure gradually increasing due to gradual deterioration gradually increases. Accordingly, the manufacturer's warranty period is an important factor in determining the maintenance state message notification based on the integrated value N, but is generally not used as notification indicating actual component replacement. However, if the manufacturer specifies the replacement deadline, the display device 30 is notified of a date that is a predetermined number of days before the replacement deadline and the replacement deadline. It is needless to say that maintenance (inspection and replacement of parts) in accordance with laws and regulations is required, as well as those that require legal inspections and periodic replacements, such as those that use steam and compressors Notify on the date a predetermined number of days before the deadline and the replacement deadline.

In addition, the present invention can also be realized by the injection molding machine 11 that notifies the maintenance state without being connected to a communication such as an in-house LAN or the Internet. That is, the maintenance information based on the maintenance index value M obtained by combining the unsteady value V detected by the sensor and the integrated value N derived from at least one of the conversion time or the conversion count is displayed on the screen of the injection molding machine or by voice. May be notified only. Upon receiving the notification, the operator can take necessary measures such as contacting the injection molding machine manufacturer 15 and securing spare parts.

Next, a state notification system of the injection molding machine 11 for notifying the maintenance state and a state notification method of the injection molding machine 11 for notifying the maintenance state will be described with reference to a flowchart of FIG. As described above, the maintenance state of the injection molding machine 11 is managed for each part or each part. Here, detection of wear of the check ring 21 of the screw 22 will be described as an example.

First, when the power of the injection molding machine 11 is turned on, the maintenance index value M accumulated up to the previous molding is read out for each component (here, the backflow prevention valve 21) for which the abnormality sign notification is set in advance (s1). . When the power is turned on, the counting of the operation time (integrated value N) is started (s2). Then, the molding conditions are read out and the molding cycle starts (s3). As described above, the accumulation of the operation time (integrated value N) may be started from the start of the molding cycle.

The weight of the continuously molded injection molded product is detected by a load cell (sensor 35) for measuring the weight of an unloader 34 which is a peripheral device included in the injection molding machine 11 in a broad sense. Then, a variation in the weight of the molded product is detected with respect to the weight of the molded product (steady value) determined in advance as a passed product, and whether the weight of the molded product exceeds the predetermined range (threshold) and the unsteady value V is detected. (S4) is determined. When the unsteady value V is detected, the controller 29 replaces and calculates the unsteady value V with the converted value V1 (s5). Then, the conversion value V1 including the unsteady value V is added to the maintenance index value M to obtain a new maintenance index value M (M + V1 = M) (s6). The magnitude of the converted value V1 is determined in relation to the converted value of the integrated value N. As an example, if the integrated value N is 1 for a molding time of 1 hour, the converted value V1 when the unsteady value V is detected once is 100 times the converted value V1 and 100 is added. Note that the difference between the steady value at the time of measuring the weight of the molded product (the value of the unsteady value judgment value) may be the same as the value determined as a defective product. It is also desirable to capture numerical fluctuations within the specified range. Instead of counting the number of abnormalities as one and always adding a fixed number, a conversion value may be determined according to the degree of deviation from the comparison value, and the conversion value composed of variables may be added. Further, the above-mentioned calculation method may use a part or all of calculation methods other than addition.

Next, it is determined whether or not the new maintenance index value M is a value for performing abnormality notification. Specifically, it is determined whether the new maintenance index value M is smaller than the threshold value ma for displaying the first message A (s7). Also, in the case where the unsteady value V is not detected in (s4) (s4 = N), the threshold value ma for displaying the first message A by the maintenance index value M to which the converted value V1 is not added is used. Determine if it is small. This is because the threshold value ma may be reached only by the accumulation of the integrated value N during the continuous molding.

When the maintenance index value M is smaller than ma in step (s7) (s7 = Y), it is next determined whether or not the molding is completed (s8). If the molding is not completed (s8 = N), the calculation of the integrated value N is performed next. As a calculation of the integrated value N, it is determined whether or not the molding time has been counted by the timer (s9). For example, when the counting has been performed after one hour, 1 is added and N + 1 = N (s10). Then, the new integrated value N is also added to the maintenance index value M to obtain a new maintenance index value M (M ← M + N) (s11). When the timer counts (s9 = Y) or does not exist (s9 = N), the process proceeds to the next molding cycle, and it is determined whether the unsteady value V is detected again in the next molding cycle (s4). to decide.

If the maintenance index value M is equal to or larger than the threshold value ma in step (s7) (s7 = N), then in step (s12), determination of ma ≦ M <mb is performed. In step (s12), mb is also a threshold value for determining the state of the injection molding machine 11. In the case of (s12 = Y), the first message A and the maker contact button 33 are displayed (notified) on the display device 30 of the injection molding machine 11 (s13). The display of the first message A is displayed only when ma falls within the range ma ≦ M <mb for the first time, and the display is continued until the operator deletes it.

The first message A is to notify a sign of a slight abnormality, and the following notification is made as an example.
"We recommend preparing spare parts for the check valve." Or "Checking the check valve is recommended."
In addition to the screen display, the notification may be a sound or a character printed from a printer, or a combination of at least two of them. The notification may be displayed on the display screen of the personal computer 14a of the central control device 14 in the factory (or in the company) by communication instead of the display device 30 on the side of the injection molding machine 11.

Although it is suspected that the check valve is worn with respect to the weight variation of the molded product, there is no other cause. Although not described in detail here for simplicity, the screw drive mechanism is deteriorated, the screw drive mechanism is poorly controlled, the screw 22 other than the backflow prevention valve 21 is worn, the heating cylinder inner hole is worn, and the heating cylinder and the mold are damaged. There is an inappropriate temperature. The defective state or unsuitable state depends on the combination of the detection value of another sensor and the lapse of time (integrated value N) from the exchange of the check valve or other parts. In many cases can be estimated. If it is not possible to specify which of the plurality of components has a sign of abnormality, the first message A is notified by listing the parts that may have a sign of abnormality.

Further, on the display device 30 of the injection molding machine 11 or the like, a maker contact button 33 for transmitting to the injection molding machine manufacturer 15 by pressing a button together with the first message A is displayed. Only when the operator presses the maker contact button 33 of the display device 30 of the injection molding machine 11 and an ON signal is transmitted, the maintenance information of the check valve of the injection molding machine 11 is stored in the personal computer 14a of the central control device 14. The data is transferred to the personal computer 14a outside the company such as the injection molding machine manufacturer 15 via the line 32 such as the Internet via the (default gateway 14b). In the present embodiment, the data sent includes contact information including the company name and factory name, the name of the person in charge, the model and serial number of the injection molding machine 11, the part number of the check valve, the replacement date and time, the molding time, and the weight of the molded product. Data on distribution (variation), molding history at the time of each molding, and the like.

At this time, the line 31 between the injection molding machine 11 and the central controller 14 may be connected by a cable such as a wired LAN, or may be connected by wireless such as a wireless LAN including WiFi. Also, the central control device 14 inside the company and the control device 16 outside the company may be connected by a line means other than the Internet such as a dedicated line. In the personal computer 14a of the company in which the injection molding machine 11 is arranged, the history information can be stored in the server directly or via the personal computer 14a, and the stock including ordering of spare parts can be secured. In addition, the injection molding machine manufacturer 15 can receive the information, manufacture spare parts, secure inventory, send out, arrange service personnel, and the like.

Although the present invention does not exclude a case where the maintenance state is automatically transmitted from the injection molding machine 11 to the injection molding machine maker 15, the maker communication button 33 is provided and displayed in the present embodiment for the following reason. The transmission to the injection molding machine maker 15 is performed only when the operator who has seen the apparatus 30 or the like permits the maintenance state to be transmitted from the display apparatus 30 or the central control apparatus 14 as another apparatus. That is, a molding maker that operates the injection molding machine 11 to perform molding often has a feeling of resistance to automatically sending all data to the injection molding machine manufacturer 15. In addition, in the state where the early signs of the slight abnormality are notified, there is a case where it is not necessary to contact the injection molding machine manufacturer 15 if there is a spare part. Further, by limiting the number of times of communication, the possibility of virus infection or data leakage can be further reduced.

When the ON signal is not transmitted from the injection molding machine 11 without the operator pressing the maker communication button 33 in response to the first message A displayed on the display device 30 (s14 = N), the injection molding machine is not used. No communication with the manufacturer 15 is made. In this case, the data may be sent to the central control device 14 in the company, or the data may be accumulated as a history only in the injection molding machine 11. Then, it is determined whether or not the molding is completed (s8), and the process proceeds to the next molding cycle through the steps of calculating the integrated value N of (s9), (s10), and (s11) described above.

If the maintenance index value M is equal to or larger than mb in step (s12) (s12 = N), then in step (s16), mb ≦ M <mc is determined. The process proceeds to step (s16) in the case where the first message A has already been notified and the integrated value N has been added to the maintenance index value M and the case where the unsteady value V1 has been added to the maintenance index value M. As shown in FIG. 2, as a result of adding the maintenance index value M and the non-stationary value V2a as a result of adding the maintenance index value M and the unsteady value V2a as shown in FIG. This is the case. When the maintenance index value M is equal to or larger than mb and smaller than mc (s16 = Y), a second message B is displayed (notified) on the display device 30 of the injection molding machine 11 (s17). The display of the second message B is displayed only when mb ≦ M <mc for the first time, and the display is continued until the operator deletes it.

The second message B is for notifying the presence of a minor abnormality, and the following notification is given as an example.
"We recommend replacing the check valve.""The check valve needs to be checked."
Also, the notification may be made by voice or a character printed from a printer, or a combination of at least two of them. The notification may be displayed on the screen of the personal computer 14a of the central control device 14 in the factory 13 (or in the company) by communication instead of the screen of the display device 30 on the side of the injection molding machine 11.

Similarly to the case of the first message A, the second message B is displayed when the maker contact button 33 is displayed on the display device 30 and the maker contact button is pressed by the operation of the operator and the ON signal is transmitted (s18). = Y), the maintenance information of the injection molding machine 11 is transferred to a personal computer outside the company such as the injection molding machine manufacturer 15 via the line 32 such as the Internet (s19). If the transmission is not performed and the operator does not stop the apparatus due to the completion of the molding quantity molding (s8 = N), the molding is continued. That is, when the maintenance index value indicates a value within the range of mb ≦ M <mc, it does not immediately lead to any of the complete failure of the injection molding machine 11, molding failure, danger to the operator, and the like. Specifically, the injection molding machine 11 may not be stopped, and the molding for a preset number of products may be continued.

If M is equal to or larger than mc in step (s16) (s16 = N), a message C indicating that an abnormality has occurred in the display device 30 and a maker contact button 33 are displayed (notified) in (s20). . The process proceeds to step (s20) in the case where the second message B has already been notified and the unsteady value V1 is added to the maintenance index value M including the integrated value N, as shown in FIG. This is a case where the calculated new maintenance index value M suddenly becomes equal to or more than mc as a result of adding the maintenance index value M and the non-stationary value converted value V2b using the constant value converted value of the unsteady value. When an abnormal value is detected, the injection molding machine 11 performs molding immediately or until the molding cycle is completed, and then stops. When the abnormality of the check valve is detected by detecting the weight of the molded product, the number of times the unsteady value V is detected may be added as a fixed conversion value. For example, when the abnormality of the servomotor or the movable portion is detected, it is desirable that the converted value V of the unsteady value is a dependent value, and that the injection molding machine 11 is emergency stopped when the abnormal value is detected.

Also in the case where the abnormal stop is notified, the maker contact button 33 is displayed on the display device 30 as in the case of the first message A and the second message B, and the maker contact button is pressed by the operator. When the ON signal is transmitted (s21 = Y), the maintenance information including the abnormality of the injection molding machine 11 is transferred to a personal computer outside the company such as the injection molding machine maker 15 via the line 32 such as the Internet ( s22). Regardless of the presence or absence of the transfer, the molding cycle ends without proceeding to the step (s8). In the flowchart of FIG. 3 of the present embodiment, the notification based on the maintenance index value M includes a first message A recommending the preparation of parts and the like, a second message B recommending replacement of the parts, and an abnormality. There are three stages of message C for stopping the device. However, the number of message stages for each part or component only needs to be such that different maintenance statuses are reported in accordance with a plurality of threshold values. Good.

In the above, when the maintenance index value M has reached a predetermined threshold value ma, mb, bc (M = ma, M = mb, M = mc) or when the maintenance index value M has exceeded the threshold value (M> ma, M>). Whether the first message A, the second message B, and the third message C are broadcast to mb, M> mc) is a slight difference, and any method may be used.

The present invention accumulates both the converted value V1 obtained by converting the unsteady value V detected by the various sensors as described above, and the integrated value N such as the molding time (including the case of the converted molding time) and the molding number. The maintenance state of the injection molding machine 11 is notified when the maintenance index value M obtained by performing the above calculation becomes a predetermined threshold value (ma, mb, etc.) or exceeds the threshold value (ma, mb, etc.). Therefore, the state of the injection molding machine 11 can be detected, judged, and notified in consideration of the aging of the parts or parts constituting the injection molding machine 11 and the state change detected by the sensor.

Next, an example in which the molding time or the number of molding cycles is multiplied by a predetermined value according to various conditions and the integrated value is converted will be described with reference to the example of the state notification system of the vertical rotary injection molding machine 41 of the second embodiment in FIG. explain. The vertical rotary injection molding machine 41 is a molding machine in which a rotating disk 43 on which a mold 42 is mounted is rotatably provided by an electric motor such as a servomotor 44 as described in Japanese Patent No. 5518565. The vertical rotary injection molding machine 41 has a basic part including a mold clamping device 45 and a first injection device 46 and a second injection device 47 provided on both sides thereof. In the mold clamping device 45, four tie bars 49 are fixed in an upright direction to a fixed plate 48 provided below, and a movable plate 50 is inserted through the tie bar 49, so that the movable plate 50 can move in the vertical direction. ing. A rotary plate 43 (rotary table) is rotatably mounted on the lower surface of the movable plate 50. A fixed mold 42a is attached to the fixed plate 48, and a movable mold 42b is attached to the rotating plate 43.

The rotation of the turntable 43 is performed by a servomotor 44, and a timing belt 53 is stretched over a toothed pulley 52 attached to a drive shaft 51 of the servomotor 44 and a toothed portion on the outer peripheral surface of the turntable 43. I have. Further, a speed reducer 54 is attached to the servo motor 44. The vertical rotary injection molding machine 41 includes a control device 56. The control device 56 controls the driving of the servo motor 44 and the like, and is connected to each sensor including an encoder 44 a of the servo motor 44 and a current sensor for measuring a current value transmitted to the servo motor 44. The control device 56 is connected to a display device 57 that displays maintenance information and the like. Further, the control device 56 is also connected to a control device outside the company (not shown) via a central control device 58 inside the factory.

The weight of the mold 42 attached to the turntable 43 of the vertical rotary injection molding machine 41 may be different depending on the molded product. Further, the rotation speed of the rotating disk 43 is controlled so that the rotating disk 43 is reversed or rotated at almost the maximum speed within the capability of the servo motor 44 so that the molding time is not unnecessarily extended. At this time, the rotation speed of the rotating disk 43 is set such that the acceleration and the deceleration are within the allowable range of the servomotor 44 and the speed reducer 54, and the load on the servomotor 44 and the speed reducer 54 is reduced. ing. However, the load on the servomotor 44 and the reducer 54 differs for each molding condition, such as when the heaviest mold 42 is used and when the lightest mold 42 is used or when the rotation speed (acceleration) is different. I do.

Therefore, in the second embodiment of FIG. 4, the peak torque is detected from the current value sent to the servomotor 44, and the peak torque is converted into a coefficient value. That is, the coefficient value a increases as the peak torque increases. Then, the conversion time aN used for the integrated value N is obtained by multiplying the coefficient value a by the molding time N. Alternatively, the coefficient value a is multiplied by the molding cycle number n to obtain an integrated value an of the converted molding number, which is the number of conversions. When the load of the servomotor 44 and the reduction gear 54 is large, the integrated value N is closer to the actual load accumulation than the simple use of the molding time N or the molding cycle number n as the integrated value N. Can be.

In the second embodiment shown in FIG. 4, the coefficient value is used in addition to the peak torque as long as the numerical value is changed according to the magnitude of the load. , The weight of the mold, the temperature of the servomotor or the speed reducer, and the like. Also in the case of the second embodiment of FIG. 4, the maintenance index value M is calculated by adding the integrated value N such as the conversion time aN calculated as described above and the unsteady value V detected by the sensor. Used for notification. For the unsteady value V, the peak torque of the servomotor 44, the deviation of the entire current value, the deviation of the time until the rotation is completed, and the like are measured by sensors, respectively, and when the servomotor 44 and the speed reducer 54 are new or close to it. Compare with the detected value (steady value).

Further, the reduction gear 54 is calculated from the normal value of the detection value of the encoder 44a of the servo motor 44 when the rotating disk 43 is rotated and comes into contact with a stopper (not shown) or when a cam or the like is detected by a proximity switch (not shown). Is detected. Alternatively, when the stopper 55 for fixing the rotating plate 43 is advanced at the stop position and inserted into the positioning hole 43a of the rotating plate 43, the phenomenon that the tip end side of the tapered stopper hits the side surface of the positioning hole 43a and the rotating plate 43 rotates. It is detected by the encoder 44a of the servomotor 44. Thus, it is detected that backlash is increasing due to wear of gears in the speed reducer 54 or the like. As for the method of calculating the unsteady value when the unsteady value V is detected, it is desirable to use the metered value according to the deviation from the steady value as described above.

The driving systems such as the servo motor 44 for rotating the rotating disk, the speed reducer 54, and the timing belt 53 of the vertical rotary injection molding machine 41 of the second embodiment shown in FIG. However, it may not be possible to judge that the abnormality is a sign of a particular one component. In such a case, the display may be performed in the order of parts having a high possibility that the sign of the abnormality has occurred in consideration of the accumulated time from the replacement of each part and the record at the time of the occurrence of the previous abnormality. The notification of the abnormality in FIG. 4 is basically the same as that in the flowchart of FIG.

Next, a control device of an injection molding machine and a state notification system of the injection molding machine according to the third embodiment shown in FIGS. 5 and 6 will be described. The mechanism such as the injection device and the mold clamping device of the injection molding machine of the third embodiment may be the injection molding machine of the first embodiment or the injection molding machine of the second embodiment. The mechanism of the molding machine is not limited. Accordingly, in the following description, the same elements or parts having the same functions will be denoted by the same reference numerals as in FIGS. 1 to 3 and redundant description will be omitted. The third embodiment has the same technical idea of determining and reporting the maintenance state, but is characterized in that the maintenance state is determined and reported by using an AI (Artificial Intelligence) technique.

The block diagram shown in FIG. 5 conceptually shows an example in which the machine learning device 61 is mounted on the arithmetic device 29a of the control device 29 of the injection molding machine 11. The control device 29 is also connected to a display device 30 having a setting / input function and each sensor of the injection molding machine 11. The control device 29 includes an input unit 62 for inputting signals from the display device 30 and sensors. The input unit 62 is connected to the machine learning device 61 of the arithmetic device 29a.

The machine learning device 61 includes a data input processing unit 63, a learning unit 64, a learning result output unit 65, a storage unit 66 including a weight storage unit 66a, and the like. Here, the storage unit 66 also serves as a storage unit for the entire control device 29. The storage unit 66 stores data of the injection molding machine at the time of part replacement and molding. Then, the integrated value and the accumulated value of the unsteady value for each component for notifying the maintenance information are also stored. Further, the arithmetic unit 29a of the control unit 29 has a part for judging and reporting the maintenance state of the normal injection molding machine 11. Specifically, the arithmetic unit 29a includes a timer 67, and the timer 67 is connected to the integrated value arithmetic processing unit 68. Further, the arithmetic unit 29a includes an unsteady value arithmetic processing unit 69 connected to the input unit 62. The integrated value calculation processing section 68 and the non-stationary value calculation processing section 69 are connected to the weight correction section 70, respectively.

The weight correction unit 70 is connected to the weight storage unit 66a of the storage unit 66 of the machine learning device 61. Further, the weight correction unit 70 is connected to the maintenance index value calculation unit 71, and the maintenance index value calculation unit 71 is connected to the maintenance information notification determination unit 72. The maintenance information notification determination unit 72 is connected to the data input processing unit 63 and the storage unit 66 of the machine learning device 61, and is further connected to the output unit 73 of the control device 29. The output unit 73 of the control device 29 is connected to the display device 30 of the injection molding machine 11 and other drive system servo amplifiers and solenoids. The output unit 73 is connected to a central control device 14 in the factory 13 or an external control device 16 such as an injection molding machine maker 15 outside the factory in some cases.

Next, the functions of the calculation unit 29a of the control device 28 and the machine learning device 61 that notify the maintenance state will be described. The integrated value calculation processing section 68 generates an integrated value N for the maintenance state calculation from the time measured by the timer 67 and the like. The integrated value N may take into account the operating state of the injection molding machine 11, and may be multiplied by a certain coefficient when the injection molding machine 11 is used in a high load state or in a poor environment. The unsteady value calculation processing unit 69 generates an unsteady value V for maintenance state calculation from a data signal input from an external sensor or the like via the input unit 62.

The integrated value N and the unsteady value V are multiplied by a coefficient value in the weight correction unit 70, and corrected to a weighted integrated value Na and unsteady value Va. The method of correcting the weight is not limited to the multiplication, but may be another calculation method. The calculation and correction of the coefficient value when weighting the integrated value N and the unsteady value V will be described later in the description of the machine learning device 61. Then, the integrated value Na and the unsteady value Va whose weights have been corrected are sent to the next maintenance index value calculation unit 71 and added to generate the maintenance index value M. Then, in the next maintenance state notification judging section 72, the first message A which is compared with the threshold values ma, mb, and mc stored in the storage section 66 to notify a sign of a minor abnormality and recommend an inspection, Output any one of a second message B notifying the existence of the component and recommending replacement of the part, and a third message C notifying the abnormality, or no message indicating that it is not yet in the stage of notifying the maintenance state. Do not announce

When any one of the first message A, the second message B, and the third message C is output in the maintenance state notification determination unit 72, the maintenance state notification determination unit 72 performs maintenance The data including the index value M and its calculation process is sent to the data input processing unit 63 of the machine learning device 61. It is also sent to the storage unit 66 and stored.

When any one of the first message A, the second message B, and the third message C is output, the operator checks the degree of deterioration such as wear of parts. If the first message A, the second message B, and the third message C have been output at the optimal time, the operator inputs information indicating that the output of the message was optimal from the display device 30. I do. The input information is sent to the data input processing unit 63 of the machine learning device 61 via the input unit 62. Then, the information is linked with the data sent from the maintenance state notification determining unit 72 and a maintenance index value for issuing any one of the first message A, the second message B, and the third message C ( (Including the weighting of the integrated value and the unsteady value) is regarded as appropriate teacher data.

Also, the operator checks the degree of deterioration of the component and issues any of the first message A, the second message B, and the third message C too early or too late. Information indicating that the timing of issuing the message A or the message B from the display device 30 was not optimal. Then, the input information is sent to the data input processing unit 63 of the machine learning device 61. Then, the information is linked with the data sent from the maintenance state notification determination unit 72, and a maintenance index value (integration) for issuing any one of the first message A, the second message B, and the third message C. (Including weighting of values and unsteady values) is regarded as inappropriate teacher data.

When the degree of component deterioration is visually determined by an operator, inputting information using a push button or the like for proceeding to the next stage is advantageous because the labor of the operator can be saved. Specifically, in the case of the first message A, buttons such as “parts have no problem at all”, “parts have very slight deterioration but molding can be continued”, and “parts need to be replaced immediately” are alternatively provided. . If the operator presses the button of "Parts have no problem", the message is displayed too quickly, and if the button of "Partially deteriorated but molding can be continued" is pressed, the message is displayed properly. Yes, the message may be displayed too late when the "immediate component replacement required" button is pressed. The information on the degree of deterioration of parts may be obtained from evaluation of a plurality of items, or may be obtained by using label data obtained by processing data by an operator. Further, the information on the degree of deterioration of the parts may be numerically determined by mechanically judging the degree of deterioration of the parts by using a camera or a sensor so that the worker does not judge the degree of wear or the like.

In addition, the information when the message B recommending the replacement is displayed before the first message A recommending the inspection is issued, or before the first message A or the second message B is issued, When the third message C indicating the abnormality is displayed, the maintenance index value M including the information of the weighted integrated value N and the unsteady value V is automatically determined without the operator's judgment. It is considered appropriate teacher data.

The external information processed in the data input processing unit 63 of the machine learning device 61 is sent to the learning unit 64, and the learning unit 64 uses a neural network that imitates the mechanism of the brain neural circuit as shown in FIG. Mainly supervised learning is performed. Specifically, the training data in which the maintenance index value M including the information of the weighted integrated value N and the unsteady value V is appropriate, and the training data in which the maintenance index value M is inappropriate are each a neural network. Are input from the input layer.

Then, in the neural network, a regression analysis is performed using each of the teacher data (what type of integrated value N and unsteady value V were used to give an optimal message notification), and a functional expression (generally, a plurality of Is derived. The supervised learning method may use a decision tree in addition to the regression analysis. The function calculation formula may be used as it is or may be used as an approximate formula. FIG. 6 shows a neural network in which only one intermediate layer 76 (hidden layer) is provided between the input layer 74 and the output layer 75, but between the input layer and the output layer. It does not exclude those having a plurality of intermediate layers (hidden layers) and performing deep learning. In particular, when there is little data used for learning in abnormality detection or failure prediction, it is desirable to perform deep learning including inference performed by the machine learning device 61.

Then, the learning unit 64 uses the neural network function formula to weight the integrated value N such as the time serving as the basis of the maintenance index value and the unsteady value N other than the normal state of each sensor in the output layer 75. A numerical value is output. That is, in the present invention, the integrated value N and the non-stationary value V are weighted by the machine learning device 61 using a neural network. To illustrate this point in a simpler example, the fact that the time element (integrated value N) is closely related to the element that issues the second message B recommending the replacement of a certain part is supervised learning from many learning opportunities. If so, the weight of the integrated value N is increased, and the weight of the unsteady value V detected from a sensor or the like is reduced. On the other hand, if it is learned that the relationship between the integrated value N such as the time is small and the relationship with the detected value (unsteady value V) of the sensor or the like is large, the weighting of the unsteady value V of the sensor or the like is performed. To increase. The adjustment of the weighting is performed in a relationship between a plurality of sensors, and also in a relationship between the number of times the unsteady value V is detected and the magnitude (frequency) of the unsteady value V. In addition, the values of the thresholds ma, mb, and mc for notifying the first message A and the second message B are often changed in accordance with the adjustment of the weights.

The learning result (the numerical value of the weighting of the integrated value N and the unsteady value V) output to the output layer 75 using the neural network of the learning unit 64 is weighted in the storage unit 66 via the learning result output unit 65. The information is temporarily stored in the storage unit 66a. Note that the block diagram of FIG. 5 conceptually illustrates the functions of the control device 29 and the machine learning device 61, and various types of actual control are assumed.

Next, a state notification method of the injection molding machine using the machine learning device 61 that performs supervised learning by AI will be further described. The calculation method using the machine learning device 61 also uses the method of FIG. 2, and utilizes the result of the machine learning for weighting the integrated value N and the unsteady value V. Also, the flowchart of FIG. 3 is used for the flowchart of the state notification method of the injection molding machine 11. However, when the predictive ability of the AI is improved, the message may be a two-stage message of a second message B recommending replacement of a part and a third message C indicating an abnormality.

The integrated value N detected by the timer 67 or the like and subjected to numerical processing by the integrated value calculation processing unit 68 is based on data machine-learned by the learning unit 64 of the machine learning device 61 and stored in the weight storage unit 66a of the storage unit 66. Further, the weighted correction unit 70 performs weighting and correction. In the first stage, the weight of the integrated value is preferably set to 50% or more. The unsteady value V numerically processed by the unsteady value calculation processing unit 69 detected by the sensor of the injection molding machine 11 is further weighted and corrected by the weight correction unit 70 based on the data stored in the weight storage unit 66a. . The corrected integrated value Na and the unsteady value Va are added by the maintenance index value calculation unit 71. Then, the maintenance information notification determination unit 72 compares the threshold value with the predetermined threshold value stored in the storage unit 66, and notifies the maintenance state of the injection molding machine 11 when the threshold value is reached or exceeds the threshold value ma, mb, mc. As described above, the threshold values ma, mb, and mc are often changed as a result of machine learning.

The weighting of the integrated value N and the unsteady value V and the corresponding change of the threshold values ma, mb, and mc are performed for each machine learning corresponding to each message display. However, machine learning may be performed after accumulating a certain number of data amounts, and weighting and threshold values may be changed. Further, when the number of data is small, the AI may change the weighting of each of the integrated value N and the non-stationary value V and change the threshold value in accordance with the inference by deep learning or the like.

Further, in the above description, the machine learning device 61 mainly describes a device that performs supervised learning using a neural network, but does not exclude a device that performs unsupervised learning such as a k-means method. Unsupervised learning is also possible if the state of the parts that need to be replaced can be objectively grasped by a sensor or the like. That is, the combination of the integrated value N and the unsteady value V indicates what kind of combination has caused a problem in the part and the replacement should be recommended, or in what kind of combination the part has a problem. Create a classification if there is no need for replacement. The accuracy of the machine learning can be improved by further verifying whether the classification is correct using the machine learning device 61. Further, reinforcement learning is performed by giving a negative reward to the data when the third message C indicating the abnormality is issued from the injection molding machine 11, and the data of the reinforcement learning is accumulated to thereby improve the accuracy of the failure prediction. May be improved.

In the above description, the example in which the machine learning device 61 is provided in the injection molding machine 11 has been described. However, in the case of the injection molding machine 11 alone, the frequency of needing component failure prediction is not so high. Therefore, the machine learning device 61 is provided inside the central control device inside the company or inside the control device outside the company where the injection molding machine is arranged, or is directly connected to the central control device inside the company or the injection molding machine. It is also preferable to provide a direct connection to the arranged control device outside the company. When the machine learning device 61 is provided in the central control device 14, the external control device 16, or the external control device 16 as described above, the opportunity for learning of data of each injection molding machine 11 other than the steady state increases, and the supervised learning is performed. Opportunities can be increased and accuracy can be improved. Further, even when the machine learning device 61 is provided in the control device 29 of the injection molding machine 11, the injection molding machine 11 having the same model or the same model or a mechanism that approximates the same model and having a common degree of component deterioration is provided. It is also possible to input data to another injection molding machine 11 and use it.

In particular, in current injection molding machines, the number of occurrences of the problem that the apparatus (parts) actually stops and the apparatus stops is small, so even if the AI is used, there are few opportunities for machine learning at the time of the abnormality and the failure prediction. It is said that it is difficult to improve the accuracy of the data. However, in the present invention, not only the occurrence of an abnormality in the apparatus is detected, but the information at the stage of the first message A for notifying a sign of a minor abnormality or the second message B for recommending replacement of parts is used, so that the data amount is reduced. There are many opportunities for machine learning. Further, as described above, since the data of many injection molding machines 11 can be collected by the central control device 14 and the external control device 16, the opportunity of machine learning is further increased, and the accuracy of the notification of the maintenance state is improved. I do.

In the failure prediction of parts of the injection molding machine, there is a case where abnormality of a plurality of parts (including necessity of inspection and replacement) is suspected from one abnormal value. However, in the third embodiment using the AI, it is recommended to replace or inspect any one of a plurality of parts and locations based on one or more command values and detection values (unsteady value V) of one or more sensors. In this case, when the unsteady value V of the combination of the command value and the sensor value or the value between the sensors is generated, it is necessary to learn which component is likely to have a problem. It is possible to specify. Alternatively, in specifying the parts, the respective integrated values N from the replacement time of each part are also considered. If it is determined that an abnormality is suspected in a plurality of parts among some of the parts, a list may be displayed in descending order of the possibility of the part abnormality. If a part having a problem is actually found as a result of the inspection, the fact is further machine-learned (supervised learning), whereby the accuracy of the list in the order of high possibility of the part abnormality can be improved.

In addition, the recommendation of component replacement not only captures and reports only the optimal recommended timing for each component, but also reports the inspection recommendation and replacement recommendation for a single component or multiple components in relation to other elements. You may do so. Specifically, when it is determined that replacement of the component A is recommended, and the component B and C are notified, the non-stationary value V does not appear or the occurrence of the non-stationary value V is small and the replacement is recommended slightly before the recommended replacement. Even if the replacement is recommended at the same time. At this time, the following points are considered. In contrast to the cost increase due to the replacement of parts before the recommended replacement time, the cost reduction by reducing the cost such as the number of parts replacement work and time (which may include the travel expenses of workers). If it is larger, it is reasonable to replace multiple parts at the same time. Alternatively, considering the number of parts in stock and the number of days required for parts procurement at the manufacturer of the injection molding machine or the factory that owns the injection molding machine, etc. It can also be said that those that perform Further, even when it is known that a busy state in which the operation of the injection molding machine cannot be stopped at all is found, it is also possible to collectively take measures such as hastening the timing of recommending replacement of parts.

When the AI is used in these cases, it is desirable to use a mathematical programming in machine learning for optimization. More specifically, a mathematical model such as an objective function is created, and optimization is performed to determine under what conditions the cost related to component replacement is minimized. The algorithm for optimization is not limited to a specific algorithm, and a linear programming method such as a branch and bound method, a trust region method, or the like may be used.

The present invention will not be enumerated one by one, but the present invention is not limited to the above-described embodiment, and a person skilled in the art modified based on the gist of the present invention or obtained by multiplying each description of the present embodiment. It goes without saying that the above also applies.

Although the present invention has been described with respect to an injection molding machine, the present invention can also be applied to notification of a maintenance state of a press apparatus or a vacuum laminating apparatus using a diaphragm.

11. Injection molding machine 12 Injection molding machine status notification system 13 Factory (company where injection molding machine is located)
14 Central control unit 15 Injection molding machine manufacturer (outside the company where the injection molding machine is located)
16, 29 Controller 29a Operation unit (timekeeping means)
30 display device 35 sensor (detection means)
N Integrated value V Unsteady value M Maintenance index value ma, mb, mc Threshold

Claims (9)

In an injection molding machine that reports the maintenance status,
Detecting means for detecting that the operation state of the injection molding machine is an unsteady value,
Counting means for counting an integrated value consisting of at least one of the number of conversion hours or the number of conversions,
Injection molding, characterized in that a notifying means for notifying a maintenance state of the injection molding machine when a value calculated from both the unsteady value and the integrated value becomes a predetermined threshold or exceeds a threshold is provided. Machine. The unsteady value and the integrated value are each converted into a numerical value, and a notifying means for notifying the maintenance state of the injection molding machine when a value obtained by adding both values becomes a predetermined threshold value or exceeds the threshold value is provided. The injection molding machine according to claim 1, characterized in that: 3. The injection molding machine according to claim 1, wherein a value calculated from both the unsteady value and the integrated value is weighted by a machine learning device using a neural network. . In the state notification system of the injection molding machine that notifies the maintenance state,
The injection molding machine according to claim 1, wherein the injection molding machine is connected to a central control device inside the company where the injection molding machine is arranged or a control device outside the company where the injection molding machine is arranged. A status notification system for an injection molding machine, wherein a maintenance status of the injection molding machine is provided so as to be transmitted. The maintenance status is notified at least on the screen of the injection molding machine,
The status notification of the injection molding machine according to claim 4, wherein the maintenance status of the injection molding machine is transmitted to a control device outside the company when an operator permits transmission from the screen or another device. system. A machine learning device is provided inside the central control device inside the company or inside the control device outside the company where the injection molding machine is arranged, or directly connected to the central control device inside the company or the injection molding machine. The status notification system for an injection molding machine according to claim 4 or 5, wherein a machine learning device is provided directly connected to a control device outside the company disposed. In a state notification method of an injection molding machine for notifying a maintenance state,
While the detection means detects that the operation state of the injection molding machine is an unsteady value,
The integrated value taken from at least one of the number of conversion hours or the number of conversion times is counted by a counting means,
A state notifying method for an injection molding machine, wherein a notifying unit notifies a maintenance state when a value calculated from both the unsteady value and the integrated value becomes a predetermined threshold value or exceeds a threshold value. 8. The method for informing a state of an injection molding machine according to claim 7, wherein notification of different maintenance states is made in accordance with a plurality of threshold values. 9. The method for informing a state of an injection molding machine according to claim 7, wherein the notification of the different maintenance states is performed by using a machine learning device that performs supervised learning.