JP7184997B2 - State determination device and state determination method - Google Patents

Description

The present invention relates to a state determination device and state determination method, and more particularly to a state determination device and state determination method for assisting maintenance of an injection molding machine.

Maintenance of industrial machines such as injection molding machines is performed regularly or when an abnormality occurs. When maintaining an industrial machine, a maintenance person determines whether there is an abnormality in the operating state of the industrial machine by using the physical quantity that indicates the operating state of the industrial machine, which is recorded during the operation of the industrial machine. Carry out maintenance work such as replacement of defective parts.

For example, as a maintenance work for the check valve of the injection cylinder of an injection molding machine, which is a type of industrial machinery, there is a known method of periodically removing the screw from the injection cylinder and directly measuring the dimensions of the check valve. there is However, this method has the problem that the production must be temporarily stopped and the measurement work must be performed, resulting in a decrease in productivity.

As a conventional technique for solving such a problem, abnormality is diagnosed by indirectly detecting the amount of wear of the check valve of the injection cylinder without temporarily stopping production by work such as extracting the screw from the injection cylinder. As a method, there is known a method of diagnosing an abnormality by detecting the rotational torque applied to the screw or by detecting a phenomenon in which the resin flows backward to the rear of the screw.

For example, Japanese Patent Laid-Open No. 2002-200002 discloses that the rotational torque acting on the method of rotating the screw is measured, and if it exceeds the allowable range, it is judged to be abnormal. Further, Patent Documents 2 and 3 disclose that an abnormality is diagnosed by supervised learning of the load of the drive unit, the resin pressure, and the like. Furthermore, Patent Documents 4 and 5 disclose techniques for machine learning using time-series data.

JP-A-01-168421 JP 2017-030221 A JP 2017-202632 A JP 2018-097616 A JP 2017-188030 A

However, in machines with different specifications of the elements that make up the driving part of the injection molding machine, the difference between the measured values obtained from the machine and the numerical values of the learning data input during machine learning is large, and correct diagnosis by machine learning cannot be performed. There is a problem. In addition, the type of resin, which is the raw material of the molded product manufactured by the injection molding machine, the type of mold, mold temperature controller, resin dryer, etc. Then, there arises a problem that a correct diagnosis cannot be made by machine learning.

In order to solve such problems and increase the diagnostic accuracy of machine learning, it is necessary to prepare a wide variety of learning conditions and carry out machine learning when creating a learning model for machine learning. However, machine learning with a wide variety of injection molding machines, resins, and incidental equipment requires a lot of costs. In addition, raw materials such as resins and workpieces must be prepared when operating the machine, and the cost of the raw materials required to acquire the learning data is high. Moreover, it takes a lot of time to acquire the learning data. Therefore, there was a problem that learning data could not be efficiently collected.

Here, we replace the resin that is the raw material of the molded product manufactured by the injection molding machine, replace the mold that is ancillary equipment of the injection molding machine, and perform peripheral equipment such as mold temperature controllers and resin dryers. When the operation of the injection molding machine is started or immediately after it is started, when the operating conditions such as the injection condition or holding pressure condition related to the operation of the injection molding machine are changed, or when the injection molding machine does not operate normally in an alarm state, the injection Time-series data obtained from molding machines may be inappropriate for machine learning. However, in the prior art, there is a problem in that the operating state of the machine cannot be correctly diagnosed, such as by performing machine learning on inappropriate learning data to derive a learning model, or by diagnosing inappropriate learning data.

Therefore, there is a demand for a state determination device and a state determination method that can easily eliminate inappropriate learning data, perform machine learning with high accuracy, and use the learning results to assist maintenance of various industrial machines. ing.

In the state determination device according to one aspect of the present invention, learning data to be input to machine learning is, for example, time-series data during an alarm, time-series data immediately after starting operation of the machine or immediately after replacing the mold, Or, time-series data immediately after changing the set values of molding conditions such as injection conditions and holding pressure conditions related to machine operation, changes in the operation state or operation state of the injection molding machine, or a state in which molding is unstable By excluding the time-series data from the learning data and performing machine learning, a highly accurate learning model is derived to solve the above problems.

Further, one aspect of the present invention is a management device that manages a plurality of injection molding machines, and is a data acquisition unit that acquires time-series data acquired in one cycle process related to the injection molding machine as one piece of data. and an extraction condition for extracting data to be used for processing related to machine learning from the data acquired by the data acquisition unit by associating a change in the operating state of the injection molding machine or a change in the operating state with a predetermined number of cycles. an extraction condition storage unit for storing; a learning data extraction unit for extracting data used for processing related to machine learning from the data acquired by the data acquisition unit according to the extraction conditions stored in the extraction condition storage unit; is a management device with
Another aspect of the present invention is a control device for controlling an injection molding machine, comprising: a data acquisition unit that acquires time-series data acquired in one cycle process of the injection molding machine as one piece of data; Extraction that stores an extraction condition for extracting data used for processing related to machine learning from the data acquired by the data acquisition unit in association with a change in the operating state or the change in the operating state of the injection molding machine and a predetermined number of cycles. a condition storage unit; and a learning data extraction unit for extracting data used for processing related to machine learning from the data acquired by the data acquisition unit according to the extraction conditions stored in the extraction condition storage unit. It is a control device.

Another aspect of the present invention is a data extraction method related to machine learning in a management device that manages a plurality of injection molding machines, wherein time-series data acquired in one cycle of processes related to the injection molding machine is a data acquisition step of acquiring data, and data acquired from the injection molding machine to be used for processing related to machine learning is associated with changes in the operating state of the injection molding machine or changes in the operating state and a predetermined number of cycles. and a learning data extraction step of extracting data used for processing related to machine learning from the data acquired by the data acquisition step according to an extraction condition to be extracted.
Another aspect of the present invention is a data extraction method related to machine learning in a control device that controls an injection molding machine, wherein time-series data acquired in one cycle process of the injection molding machine is used as one data. a data acquisition step for acquiring data to be used for processing related to machine learning from the data acquired from the injection molding machine, and extracting data by associating a change in the operating state of the injection molding machine or a change in the operating state with a predetermined number of cycles; and a learning data extraction step of extracting data used for processing related to machine learning from the data acquired by the data acquisition step according to an extraction condition.

According to one aspect of the present invention, machine learning is performed by excluding data acquired when a change occurs in the operating state or operation state of an injection molding machine or data acquired when molding is unstable from learning data. This makes it possible to expect improvements in the accuracy of machine learning decisions.

1 is a schematic hardware configuration diagram of a state determination device according to an embodiment; FIG. 1 is a schematic functional block diagram of a state determination device according to a first embodiment; FIG. It is a figure which shows the example of extraction conditions. It is a figure which shows the example of extraction of the data for learning by a learning data extraction part. FIG. 10 is a diagram showing another example of extraction of learning data by a learning data extraction unit; FIG. 10 is a diagram showing another example of extraction of learning data by a learning data extraction unit; It is a schematic functional block diagram of the state determination apparatus by 2nd Embodiment. It is a figure which shows the example of a display of an abnormal state.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a schematic hardware configuration diagram showing essential parts of a state determination device equipped with a machine learning device according to one embodiment. The state determination device 1 of this embodiment can be mounted, for example, on a control device that controls an industrial machine. In addition, the state determination device 1 of the present embodiment includes a personal computer installed together with a control device that controls industrial machinery, a management device 3 connected to the control device via a wired/wireless network, an edge computer, a fog computer, and so on. , a computer such as a cloud server. This embodiment shows an example in which the state determination device 1 is implemented as a computer connected via a network to a control device that controls an injection molding machine as an industrial machine. In each of the following embodiments, an injection molding machine will be described as an industrial machine. Examples of industrial machines whose states are to be judged by the state judging device 1 of the present invention include injection molding machines, machine tools, Robots, mining machines, woodworking machines, agricultural machines, construction machines, etc. can be targeted.

The CPU 11 included in the state determination device 1 according to this embodiment is a processor that controls the state determination device 1 as a whole. The CPU 11 reads the system program stored in the ROM 12 via the bus 20, and controls the entire state determination device 1 according to the system program. Temporary calculation data, various data input by the operator via the input device 71, and the like are temporarily stored in the RAM 13 .

The non-volatile memory 14 is composed of, for example, a memory backed up by a battery (not shown) or an SSD (Solid State Drive), and retains the stored state even when the power of the state determination device 1 is turned off. The nonvolatile memory 14 includes a setting area for storing setting information related to the operation of the state determination device 1, data input from the input device 71, and static data obtained from the injection molding machine 2 via the network 7. (model, mold mass and material, type of resin, etc.), physical quantities detected in the molding operation of the injection molding machine 2 (nozzle temperature, position, speed, acceleration, current, voltage, torque of the prime mover that drives the nozzle) , mold temperature, resin flow rate, flow velocity, pressure, etc.), information indicating the operation state and operation state of the injection molding machine 2 pressure holding process, weighing process, mold opening process, ejection process, information identifying cycle start and cycle end, information indicating alarm occurrence status, etc.), other data read from a computer or the like is stored. The programs and various data stored in the nonvolatile memory 14 may be developed in the RAM 13 at the time of execution/use. In addition, the ROM 12 is pre-written with a system program including a known analysis program for analyzing various data and a program for controlling communication with the machine learning device 100, which will be described later.

The state determination device 1 is connected to a wired/wireless network 7 via an interface 16 . Connected to the network 7 are at least one injection molding machine 2, a management device 3 that manages the manufacturing work by the injection molding machine 2, and the like, and exchange data with the state determination device 1. .

The injection molding machine 2 is a machine that manufactures products molded from resin such as plastic, and is a machine that melts resin as a material and fills (injects) it into a mold for molding. The injection molding machine 2 is composed of various equipment such as a nozzle, a prime mover (motor, etc.), a transmission mechanism, a speed reducer, and movable parts. controlled. As the prime mover used in the injection molding machine 2, for example, an electric motor, a hydraulic cylinder, a hydraulic motor, an air motor, etc. are used. Also, as a transmission mechanism used in the injection molding machine 2, a ball screw, a gear, a pulley, a belt, and the like are used.

On the display device 70, various data read into the memory, data obtained as a result of executing a program, etc., data output from the machine learning device 100, which will be described later, etc., are output and displayed via the interface 17. be done. An input device 71 composed of a keyboard, a pointing device, and the like passes commands, data, and the like based on operations by the operator to the CPU 11 via the interface 18 .

The interface 21 is an interface for connecting the state determination device 1 and the machine learning device 100 . The machine learning device 100 includes a processor 101 that controls the entire machine learning device 100, a ROM 102 that stores system programs and the like, a RAM 103 that temporarily stores data in each process related to machine learning, and a storage of learning models and the like. and a non-volatile memory 104 used for The machine learning device 100 acquires various types of information that can be acquired by the state determination device 1 via the interface 21 (for example, the model of the injection molding machine 2, the mass and material of the mold, various data such as the type of resin, the temperature of the nozzle, Time-series data of various physical quantities such as the position, speed, acceleration, current, voltage, torque, temperature of the mold, flow rate of the resin, flow velocity, and pressure of the prime mover that drives the nozzle, operating state and operation state of the injection molding machine 2, etc. time-series data, etc.) can be observed. Further, the state determination device 1 acquires the processing result output from the machine learning device 100 via the interface 21, stores the acquired result, displays it, and communicates with other devices via the network 7 or the like. to send.

FIG. 2 is a schematic functional block diagram of the state determination device 1 and the machine learning device 100 according to the first embodiment. The state determination device 1 of the present embodiment has a configuration required when the machine learning device 100 performs learning at the stage of performing machine learning (learning mode). Each functional block shown in FIG. 2 is implemented by the CPU 11 provided in the state determination device 1 and the processor 101 of the machine learning device 100 shown in FIG. It is realized by controlling the operation of each part of 100 .

The state determination device 1 of this embodiment includes a data acquisition unit 30 , a learning data extraction unit 32 and a preprocessing unit 34 , and the machine learning device 100 included in the state determination device 1 includes a learning unit 110 . In addition, on the nonvolatile memory 14, an acquired data storage unit 50 for storing data acquired from an external machine or the like, and an extraction condition storage unit 52 for storing conditions for extracting learning data from the acquired data are stored. A learning model storage unit 130 that stores a learning model constructed by machine learning by the learning unit 110 is provided on the nonvolatile memory 104 of the machine learning device 100 .

The data acquisition unit 30 is functional means for acquiring various data input from the injection molding machine 2, the input device 71, and the like. The data acquisition unit 30 acquires, for example, the model of the injection molding machine 2, the mass and material of the mold, static data such as the type of resin, the temperature of the nozzle, the position, speed, acceleration, current, and voltage of the prime mover that drives the nozzle. , torque, temperature of the mold related to the molding operation of the injection molding machine 2, flow rate of resin, flow velocity, time series data of various physical quantities such as pressure, during operation, during stop, during temperature increase, completion of temperature increase, mold change Information indicating the mechanical status of the injection molding machine 2, such as middle, mold exchange completion, alarm, production completion status, injection conditions indicating the operation status of the injection molding machine 2, holding pressure conditions, weighing conditions, mold opening/closing conditions, protrusion conditions, information for identifying whether or not temperature conditions have been changed, mold closing process, mold clamping process, injection process, holding pressure process, weighing process, mold opening process, ejection process, standby process, which are molding processes of the injection molding machine 2; Various types of data such as information identifying cycle start and cycle end, information indicating an alarm occurrence state, and information related to maintenance work of the injection molding machine input by the operator are acquired and stored in the acquired data storage unit 50 . When acquiring the time-series data, the data acquisition unit 30 acquires a predetermined time range (for example, one cycle of molding process range) are stored in the acquired data storage unit 50 as one piece of time-series data. The data acquisition unit 30 may acquire data from the management device 3 or another computer via an external storage device (not shown) or the wired/wireless network 7 .

The learning data extraction unit 32 acquires the data acquired by the data acquisition unit 30 (and stored in the acquired data storage unit 50) based on the extraction conditions stored in the extraction condition storage unit 52 in the stage of machine learning by the learning unit 110. 2) It is a functional means for extracting the acquired data to be used for machine learning from the acquired data. In other words, based on the extraction conditions stored in the extraction condition storage unit 52, the learning data extraction unit 32 excludes acquired data that is not suitable for use in machine learning from the acquired data acquired by the data acquisition unit 30. do.

FIG. 3 is a diagram illustrating extraction conditions stored in the extraction condition storage unit 52. As shown in FIG. The extraction condition storage unit 52 stores at least one extraction condition organized and managed by condition type, for example. The extraction conditions stored in the extraction condition storage unit 52 may be conditions for specifying acquired data to be used for machine learning, or may be conditions for specifying acquired data not to be used for machine learning (excluded). . The extraction conditions stored in the extraction condition storage unit 52 include at least a condition for classifying acquired data based on a predetermined data value included in the acquired data, and whether the acquired data classified according to the condition is used as data for learning. Including whether to use or not.

FIG. 4 is a diagram for explaining an example in which the learning data extraction unit 32 extracts acquired data based on the extraction conditions related to the machine state stored in the extraction condition storage unit 52. As shown in FIG. Consider a case where the acquired data illustrated in FIG. 4 is stored in the acquired data storage unit 50 and the learning data extraction unit 32 extracts the waveform data of the current value for each cycle as data for learning. In this case, if an extraction condition is set in the extraction condition storage unit 52 that "obtained data during an alarm is excluded from learning data", the learning data extraction unit 32 generates an alarm during the cycle of the molding process. If so, the current value data acquired in the cycle is not extracted as learning data. Specifically, in the case of the example of FIG. 4, the learning data extraction unit 32 extracts current value data acquired in (i+2)-cycle and (i+3)-cycle in which the occurrence of an alarm is detected as The current value data acquired before (i+1)-cycle and after (i+4)-cycle are extracted as learning data without being extracted as learning data.

FIG. 5 is a diagram for explaining an example in which the learning data extraction unit 32 extracts acquired data based on the extraction conditions related to the operation state stored in the extraction condition storage unit 52. As shown in FIG. Consider a case where the acquired data illustrated in FIG. 5 is stored in the acquired data storage unit 50 and the learning data extraction unit 32 extracts waveform data of voltage values for each cycle as data for learning. In this case, if an extraction condition is set in the extraction condition storage unit 52 that "acquisition data for 10 cycles after the injection condition change is excluded from learning data", the learning data extraction unit 32 If the injection condition is changed during the cycle (injection condition change signal is ON), voltage value data acquired during 10 cycles from the cycle is extracted as learning data. Do not behave. Specifically, in the case of the example of FIG. 5, the learning data extracting unit 32 acquires data for 10 cycles (up to (i+10)-cycle) from (i+1)-cycle where the injection condition is changed. Voltage value data obtained before (i)-cycle and after (i+11)-cycle are extracted as learning data.

FIG. 6 is a diagram for explaining an example in which the learning data extraction unit 32 extracts acquired data based on the extraction conditions related to the molding process stored in the extraction condition storage unit 52. As shown in FIG. Consider a case where the acquired data illustrated in FIG. 6 is stored in the acquired data storage unit 50 and the learning data extraction unit 32 extracts the waveform data of the current value for each cycle as data for learning. In this case, if the extraction condition storage unit 52 is set with an extraction condition that "extracts only the acquired data of the injection and holding pressure processes as data for learning", the learning data extraction unit 32 stores data for each molding process. The current value data obtained during the injection process and the holding pressure process are extracted as data for learning. Specifically, in the case of the example of FIG. 6, the learning data extractor 32 identifies the periods of the injection process and the holding pressure process in each molding process based on the start signal and end signal of each process. , the current value data acquired during the period is extracted as data for learning.

A plurality of extraction conditions can be set in the extraction condition storage unit 52 . In this case, it is possible that two or more extraction conditions conflict with each other in specifying whether or not to use data for learning. In that case, the learning data extraction unit 32 may give priority to designations that are not used as data for learning, or the priority among the extraction conditions is stored in the extraction condition storage unit 52 together with the extraction conditions. The learning data extracting unit 32 may specify whether or not to use data for learning based on the order of priority to resolve conflicts.

At the stage of machine learning by the machine learning device 100 , the preprocessing unit 34 creates learning data used for learning by the machine learning device 100 based on the learning data extracted by the learning data extraction unit 32 . The preprocessing unit 34 converts the data input from the learning data extraction unit 32 into a uniform format handled by the machine learning device 100 (digitization, sampling, etc.) to create learning data. For example, when the machine learning device 100 performs unsupervised learning, the preprocessing unit 34 creates state data S in a predetermined format as learning data, and when the machine learning device 100 performs supervised learning, , a set of state data S in a predetermined format and label data L in the learning is created as learning data, and when the machine learning device 100 performs reinforcement learning, state data S in a predetermined format in the learning and A set of judgment data D is created as learning data.

The learning unit 110 performs machine learning using the learning data created by the preprocessing unit 34 based on the learning data extracted by the learning data extraction unit 32 . The learning unit 110 generates a learning model by performing machine learning using data acquired from the injection molding machine 2 by a known machine learning method such as unsupervised learning, supervised learning, or reinforcement learning. The learned model is stored in the learning model storage unit 130 . Unsupervised learning methods performed by the learning unit 110 include, for example, the autoencoder method and k-means method, and supervised learning methods include, for example, the multilayer perceptron method, the recurrent neural network method, the Long Short-Term Memory method, the convolutional The neural network method and the like, and the reinforcement learning method includes, for example, Q-learning and the like.

The learning unit 110 performs unsupervised learning based on learning data obtained by processing acquisition data acquired from the injection molding machine 2 in a normal operating state by the learning data extraction unit 32 and the preprocessing unit 34, for example. It is possible to perform learning and generate a distribution of data acquired in a normal state as a learning model. Using the learning model generated in this manner, the estimating unit 120 described later determines that the state data S obtained by processing the acquired data acquired from the injection molding machine 2 by the preprocessing unit 34 is the normal state. It is possible to estimate how much the state data deviates from the state data acquired during operation, and calculate the degree of abnormality as the estimation result.
For example, the learning unit 110 assigns a normal label to acquired data acquired from an injection molding machine that is operating normally, and an abnormal label to acquired data acquired from the injection molding machine 2 before and after an abnormality occurs. is given, and supervised learning is performed using the learning data obtained by processing the acquired data by the learning data extraction unit 32 and the preprocessing unit 34, and the discrimination boundary between normal data and abnormal data is generated as a learning model. be able to. Using the learning model generated in this manner, the estimating unit 120, which will be described later, inputs the state data S obtained by processing the acquired data acquired from the injection molding machine 2 by the preprocessing unit 34 into the learning model. Then, whether the state data S belongs to normal data or abnormal data can be estimated, and the label value (normal/abnormal) as the estimation result and its reliability can be calculated.

In the state determination device 1 having the above configuration, the learning data extraction unit 32 extracts the acquired data stored in the extraction condition storage unit 52 from the acquired data included in the acquired data, for the acquired data acquired from the injection molding machine 2. Extract training data according to conditions. In the extraction condition storage unit 52, the operator can set extraction conditions so that appropriate data can be extracted as data for learning according to the purpose of machine learning at that time. The learning data extracted by the learning data extraction unit 32 in this way is, for example, time-series data during an alarm, time-series data immediately after starting operation of the machine or immediately after exchanging the mold, or Time-series data immediately after changing the setting values of molding conditions such as injection conditions and holding pressure conditions related to machine operation. By excluding the time-series data from the learning data, it is possible to machine-learn only the time-series data belonging to a predetermined process necessary for determining the operating state. Judgment of the state of the injection molding machine 2 using the learning model generated in this manner has a higher accuracy in judging the operation state of the injection molding machine 2 than the case of using the learning model generated by the conventional method. expected to improve.

FIG. 7 is a schematic functional block diagram of the state determination device 1 and the machine learning device 100 according to the second embodiment. The state determination device 1 of this embodiment has a configuration required when the machine learning device 100 performs estimation (estimation mode). Each functional block shown in FIG. 7 is executed by the CPU 11 provided in the state determination device 1 and the processor 101 of the machine learning device 100 shown in FIG. It is realized by controlling the operation of each part of 100 .

The state determination device 1 of the present embodiment includes a data acquisition unit 30, a learning data extraction unit 32, and a preprocessing unit 34, similarly to the first embodiment. , an estimator 120 . In addition, on the nonvolatile memory 14, an acquired data storage unit 50 for storing data acquired from an external machine or the like, and an extraction condition storage unit 52 for storing conditions for extracting learning data from the acquired data are stored. A learning model storage unit 130 is provided on the nonvolatile memory 104 of the machine learning device 100 and stores a learning model constructed by machine learning by the learning unit 110 .

The data acquisition unit 30 according to this embodiment has the same functions as the data acquisition unit 30 according to the first embodiment.
The basic operation of the learning data extraction unit 32 according to the present embodiment is the same as that of the learning data extraction unit 32 in the first embodiment. This differs from the first embodiment in that it is estimation data used to estimate the state of the aircraft 2 .

In the stage of estimating the state of the injection molding machine 2 using the learning model by the machine learning device 100, the preprocessing unit 34 according to the present embodiment stores the data for estimation extracted by the learning data extraction unit 32 in the machine learning device 100. is converted (digitized, sampled, etc.) into a uniform format handled in , and state data S in a predetermined format used for estimation by the machine learning device 100 is created.

The estimation unit 120 estimates the state of the injection molding machine using the learning model stored in the learning model storage unit 130 based on the state data S created by the preprocessing unit 34 . In the estimating unit 120 of the present embodiment, the state data S obtained from the preprocessing unit 34 is input to the learning model generated (the parameters are determined) by the learning unit 110, so that the injection molding machine It is calculated by estimating the degree of abnormality related to the state, or by estimating and calculating the class (normal/abnormal, etc.) to which the operation state of the injection molding machine belongs. The results estimated by the estimation unit 120 (such as the degree of abnormality related to the state of the injection molding machine and the class to which the operating state of the injection molding machine belongs) are displayed on the display device 70 or sent to the host via a wired/wireless network (not shown). The data may be transmitted and output to a computer, a cloud computer, or the like for use. Further, when the result estimated by the estimating unit 120 is in a predetermined state (for example, when the degree of abnormality estimated by the estimating unit 120 exceeds a predetermined threshold value, the estimating unit 120 (e.g., when the class to which the operating state of the injection molding machine estimated by ) belongs to is "abnormal", for example, as shown in FIG. Alternatively, a command to stop operation, decelerate, or limit the torque of the prime mover that drives the injection molding machine may be output to the injection molding machine.

In the state determination device 1 having the above configuration, the learning data extraction unit 32 stores the acquired data acquired from the injection molding machine 2 in the extraction condition storage unit 52 from among the acquired data included in the acquired data storage unit 50. Extract the data for estimation according to the specified extraction conditions. In this extraction condition storage unit 52, the operator can set extraction conditions so that appropriate data is extracted as data for estimation according to the purpose of judging the state of the injection molding machine 2 at that time. The data for estimation extracted by the learning data extraction unit 32 in this way is, for example, time-series data during an alarm, time-series data immediately after starting operation of the machine or immediately after exchanging the mold, or Time-series data immediately after changing the setting values of molding conditions such as injection conditions and holding pressure conditions related to machine operation. Time-series data is excluded, and only time-series data suitable for determining the operating state of the injection molding machine 2 can be used for state determination, improving the determination accuracy of the operating state of the injection molding machine 2 by machine learning. is expected.

The state determination apparatus 1 according to the first and second embodiments described above can be applied when determining the state of industrial machines such as robots and machine tools. It can be suitably used in industrial machines that exhibit unstable behavior within the expected range when starting up. Injection molding machines, in particular, often operate erratically within the expected range at the start of machine operation or immediately after changing injection conditions, even when manufacturing is performed under the same injection conditions. Such an operation state is not regarded as an abnormal state and is not subject to maintenance and inspection. Therefore, the state determination device 1 of the present invention is particularly useful for injection molding machines having such characteristics.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described examples of the embodiments, and can be implemented in various modes by adding appropriate modifications.
For example, in the above-described embodiment, the state determination device 1 and the machine learning device 100 are described as devices having different CPUs (processors). It may also be realized by a system program. Further, when a plurality of injection molding machines 2 are connected to each other via a network, the operation states of the plurality of injection molding machines may be determined by one state determination device 1, or the control provided in the injection molding machines may be determined. The state determination device 1 may be mounted on the device.

1 state determination device 2 injection molding machine 3 management device 11 CPU
12 ROMs
13 RAM
14 nonvolatile memory 16, 17, 18 interface 20 bus 21 interface 30 data acquisition unit 32 learning data extraction unit 34 preprocessing unit 50 acquisition data storage unit 52 extraction condition storage unit 70 display device 71 input device 100 machine learning device 101 processor 102 ROMs
103 RAM
104 non-volatile memory 110 learning unit 120 estimating unit 130 learning model storage unit

Claims (14)

A management device for managing a plurality of injection molding machines,
a data acquisition unit that acquires time-series data acquired in one cycle process of the injection molding machine as one piece of data;
An extraction condition for extracting data to be used for processing related to machine learning from the data acquired by the data acquisition unit is stored in association with a change in the operating state or operation state of the injection molding machine and a predetermined number of cycles. an extraction condition storage unit;
a learning data extraction unit for extracting data used for processing related to machine learning from among the data acquired by the data acquisition unit according to the extraction conditions stored in the extraction condition storage unit;
management device with Further comprising a machine learning device that executes processing related to machine learning using the data extracted by the learning data extraction unit,
The management device according to claim 1. The machine learning device includes a learning unit that performs machine learning using the data extracted by the learning data extraction unit and generates a learning model.
The management device according to claim 2. the learning unit performs at least one machine learning of supervised learning, unsupervised learning, and reinforcement learning;
The management device according to claim 3. The machine learning device
a learning data storage unit that stores a learning model generated by machine learning using the data extracted by the learning data extraction unit;
an estimation unit that estimates the state of the injection molding machine using the learning model based on the data extracted by the learning data extraction unit;
comprising
The management device according to claim 2. The estimating unit estimates the degree of abnormality related to the operation state of the injection molding machine,
The management device displays a warning message on a display device when the degree of anomaly estimated by the estimation unit exceeds a predetermined threshold.
The management device according to claim 5. The estimating unit estimates the degree of abnormality related to the operation state of the injection molding machine,
The management device displays a warning icon on a display device when the degree of anomaly estimated by the estimation unit exceeds a predetermined threshold.
The management device according to claim 5. The estimating unit estimates the degree of abnormality related to the operation state of the injection molding machine,
The management device outputs at least one command to the injection molding machine to stop operation, decelerate, or limit the torque of the prime mover when the degree of abnormality estimated by the estimation unit exceeds a predetermined threshold. ,
The management device according to claim 5.
The data acquired by the data acquisition unit represents the mechanical state of the injection molding machine, which is any of the following states: operating, stopping, temperature rising, temperature rising completed, mold changing, mold changing complete, alarm in progress, and production completed. information identifying whether the injection condition indicating the operating state of the injection molding machine, pressure holding condition, weighing condition, mold opening/closing condition, ejection condition, temperature condition is changed, information identifying whether the injection molding machine's molding process At least one of the information identifying any of the mold closing process, mold clamping process, injection process, holding pressure process, weighing process, mold opening process, projecting process, and standby process,
The management device according to claim 1. The data acquired by the data acquisition unit is at least one of data acquired from a plurality of injection molding machines connected by a wired or wireless network.
The management device according to claim 1. A control device for controlling an injection molding machine,
a data acquisition unit that acquires time-series data acquired in one cycle process of the injection molding machine as one piece of data;
An extraction condition for extracting data to be used for processing related to machine learning from the data acquired by the data acquisition unit is stored in association with a change in the operating state or operation state of the injection molding machine and a predetermined number of cycles. an extraction condition storage unit;
a learning data extraction unit for extracting data used for processing related to machine learning from among the data acquired by the data acquisition unit according to the extraction conditions stored in the extraction condition storage unit;
control device with Further comprising a machine learning device that executes processing related to machine learning using the data extracted by the learning data extraction unit,
Control device according to claim 11 . A data extraction method related to machine learning in a management device that manages a plurality of injection molding machines,
a data acquisition step of acquiring time-series data acquired in one cycle process of the injection molding machine as one piece of data;
Acquisition of data according to an extraction condition for extracting data used for processing related to machine learning from the data acquired from the injection molding machine by associating a change in the operating state or operation state of the injection molding machine with a predetermined number of cycles. A learning data extraction step of extracting data used for processing related to machine learning from the data acquired by the step;
data extraction method to perform. A data extraction method related to machine learning in a control device that controls an injection molding machine,
a data acquisition step of acquiring time-series data acquired in one cycle process of the injection molding machine as one piece of data;
Acquisition of data according to an extraction condition for extracting data used for processing related to machine learning from the data acquired from the injection molding machine by associating a change in the operating state or operation state of the injection molding machine with a predetermined number of cycles. A learning data extraction step of extracting data used for processing related to machine learning from the data acquired by the step;
data extraction method to perform.

Images