Machine learning device, abnormal vibration prediction device, and computer-readable recording medium

Abstract

An abnormal vibration prediction device according to the present disclosure comprises: a processing program reception unit that receives at least one processing program used for the operation of an industrial machine; a processing information reception unit that receives processing information including at least one from among processing conditions, the material shape before processing, material quality information, and tool information; an estimation unit that, on the basis of the processing program and the processing information, performs estimation using a trained model and that estimates information pertaining to whether abnormal vibrations have occurred and information pertaining to the cause of the occurrence of the abnormal vibrations; and an output unit that outputs the result of estimation.

Description

Machine learning device, abnormal vibration prediction device, and computer-readable recording medium

[0001] The present disclosure relates to a machine learning device, an abnormal vibration prediction device, and a computer-readable recording medium.

[0002] Abnormal vibrations that occur during cutting and turning processes have a negative impact on machining accuracy and tool life. If abnormal vibrations can be predicted before a machining program is run, the occurrence of abnormal vibrations can be avoided by reducing the cutting depth in advance. Up until now, simulators have been proposed that predict the presence or absence of self-excited chatter vibrations in end milling processes based on theoretical analysis and generate machining paths to avoid them (for example, Non-Patent Document 1).

[0003] Nishikawa, T., Himemiya, K., and Kikuta, K., "Development of an NC simulator for chatter analysis in end milling," Hiroshima Prefectural Technology Research Institute, Western Industrial Technology Center Research Report, 2020, No. 63, pp. 1-4.

[0004] Abnormal vibrations can be caused by a wide range of factors, including self-excited chatter vibration, interrupted cutting, chip leakage, and mechanical components. It is difficult to predict these abnormal vibrations. It is also difficult to predict the cause of abnormal vibrations. Therefore, it takes time and effort to change the machining program to one that does not generate abnormal vibrations. There is a demand for technology that supports countermeasures against abnormal vibrations in manufacturing sites.

[0005] The abnormal vibration prediction device according to the present disclosure uses a trained model that has learned the correspondence between the machining program and machining information and the vibration state of the industrial machine through supervised learning. The above-mentioned problem is solved by inputting the machining program and machining information to this trained model and outputting an estimation result of the occurrence state of abnormal vibration.

[0006] One aspect of the present disclosure is an abnormal vibration prediction device that includes: a machining program receiving unit that receives at least one machining program to be used for operating industrial machinery; a machining information receiving unit that receives machining information including at least any one of machining conditions, material shape before machining, material information, and tool information; an estimation unit that performs estimation using a trained model that, when a pair of the machining program and the machining information is input, outputs a pair of information related to whether or not abnormal vibration has occurred and information related to the cause of the abnormal vibration, and estimates the information related to whether or not abnormal vibration has occurred and information related to the cause of the abnormal vibration; and an output unit that outputs the results of the estimation.

[0007] 1 is a schematic hardware configuration diagram of a machine learning device according to an embodiment of the present disclosure. FIG. 2 is a block diagram showing the schematic functions of the machine learning device according to the first embodiment. FIG. 3 is a schematic diagram showing an example of a machining program. FIG. 4 is a block diagram showing the schematic functions of a machine learning device according to a second embodiment. FIG. 4 is a schematic diagram showing an example of a plurality of machining programs. FIG. 5 is a schematic diagram showing an example of a plurality of machining programs combined into one machining program. FIG. 6 is a schematic diagram showing an example of a machining program including a cycle command. FIG. 7 is a schematic diagram showing an example of a machining program in which a cycle command is replaced with a normal command. FIG. 5 is a schematic hardware configuration diagram of an abnormal vibration prediction device according to an embodiment of the present disclosure. FIG. 6 is a block diagram showing the schematic functions of an abnormal vibration prediction device according to a third embodiment. FIG. 7 is a screen diagram showing an example of a display output of information related to the presence or absence of abnormal vibration and information related to the cause of the occurrence of abnormal vibration. FIG. 8 is a block diagram showing the schematic functions of an abnormal vibration prediction device according to a fourth embodiment. FIG. 9 is a block diagram showing the schematic functions of an abnormal vibration prediction device according to a fifth embodiment. FIG. 10 is a schematic diagram showing an example of a change condition. FIG. 11 is a block diagram showing the schematic functions of an abnormal vibration prediction device according to a sixth embodiment. FIG. 12 is a schematic diagram showing another example of a change condition.

[0008] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the following description, components having the same or similar functions will be denoted by the same reference numerals. Duplicate descriptions of those components may be omitted.

[0009] In this application, "based on XX" means "based on at least XX," and includes cases where it is based on other elements in addition to XX. Furthermore, "based on XX" is not limited to cases where XX is used directly, but also includes cases where it is based on XX that has been calculated or processed. "XX" is any element (for example, any information).

[0010] 1 is a schematic hardware configuration diagram showing the main parts of a machine learning device according to a first embodiment of the present disclosure. The machine learning device 1 according to this embodiment is used to create a trained model used by an abnormal vibration prediction device 2. The machine learning device 1 according to this embodiment can be implemented on a computer such as a personal computer, a cell computer, a fog computer 6, or a cloud server 7. This embodiment shows an example in which the machine learning device 1 is implemented on a computer connected via a network to a control device that controls industrial machinery, the fog computer 6, the cloud server 7, etc.

[0011] The CPU 11 included in the machine learning device 1 according to this embodiment is a processor that provides overall control of the machine learning device 1. The CPU 11 reads a system program stored in the ROM 12 via the bus 22 and controls the entire machine learning device 1 in accordance with the system program. The RAM 13 temporarily stores temporary calculation data, display data, various data acquired from the outside, and the like.

[0012] The non-volatile memory 14 is composed of, for example, a battery-backed memory (not shown) or an SSD (Solid State Drive), and retains its stored state even when the machine learning device 1 is powered off. The non-volatile memory 14 stores programs and data read from an external device 72 via the interface 15, programs and data input via the input device 71, and programs and data acquired from the control device 3 that controls the industrial machine 4 or other devices via the network 5. The programs and data stored in the non-volatile memory 14 may be expanded into the RAM 13 when executed / used. In addition, various system programs such as known analysis programs are written in the ROM 12 in advance.

[0013] The interface 15 is an interface for connecting the CPU 11 of the machine learning device 1 to an external device 72 such as a USB device. For example, system programs, data, etc. are read from the external device 72. In addition, programs, data, etc. created or edited within the machine learning device 1 can be stored in external storage means via the external device 72.

[0014] The interface 20 is an interface for connecting the CPU 11 of the machine learning device 1 to a wired or wireless network 5. The network 5 may communicate using technologies such as serial communication such as RS-485, Ethernet (registered trademark), optical communication, wireless LAN, Wi-Fi (registered trademark), Bluetooth (registered trademark), etc. The abnormal vibration prediction device 2, a control device 3 that controls at least one industrial machine 4, a fog computer 6, a cloud server 7, etc. may be connected to the network 5.

[0015] The display device 70 displays data and the like obtained as a result of executing various data and programs loaded into the memory, output via the interface 17. An input device 71, which is composed of at least one input device such as a keyboard, a pointing device, a voice input device, an imaging device, etc., passes commands, data, etc. based on user operations to the CPU 11 via the interface 18.

[0016] 2 is a schematic block diagram illustrating functions of the machine learning device 1 according to the first embodiment of the present disclosure. Each function of the machine learning device 1 according to this embodiment is realized by the CPU 11 included in the machine learning device 1 shown in FIG. 1 executing a system program and controlling the operation of each unit of the machine learning device 1.

[0017] The machine learning device 1 of this embodiment includes a machining program acquisition unit 100, a machining information acquisition unit 110, an operation information acquisition unit 120, a learning data creation unit 130, a learning unit 140, and an output unit 180. In addition, the RAM 13 to the nonvolatile memory 14 of the machine learning device 1 are provided in advance with a model storage unit 160, which is an area for storing the trained model created by the learning unit 140.

[0018] The machining program acquisition unit 100 acquires a machining program to be executed by the control device 3 when the industrial machine 4 is in operation. The machining program acquired by the machining program acquisition unit 100 is, for example, a numerical control program written in G-code or M-code as illustrated in FIG. 3 . The machining program acquisition unit 100 may acquire a machining program input by a user via the input device 71. Alternatively, the machining program may be acquired via the network 5 from the control device 3 that controls the industrial machine 4 in accordance with a user operation. Furthermore, the machining program may be acquired from a machining program recorded in an external device 72 in accordance with a user operation. The machining program acquisition unit 100 outputs the acquired machining program to the learning data creation unit 130.

[0019] The machining information acquisition unit 110 acquires machining information related to machining in the control device 3 and the industrial machine 4 that operate the machining program acquired by the machining program acquisition unit 100. The machining information may include, for example, machining conditions set in the control device 3 and the industrial machine 4. It may also include information on the material shape and material quality of the workpiece to be machined in the industrial machine 4 before machining. In this case, the material shape may be indicated, for example, in text, or an image or a three-dimensional shape file (e.g., CAD data) showing the shape of the material before machining may be acquired. It may also include tool information related to a tool attached to the industrial machine 4. In this case, the tool information may be, for example, the tool model number, or an image or a three-dimensional shape file (e.g., CAD data) showing the shape of the tool. The machining information acquisition unit 110 may acquire machining information input by a user via the input device 71. It may also acquire machining information via the network 5 from the control device 3 that controls the industrial machine 4 in accordance with a user's operation. It may also acquire machining information recorded in an external device 72 in accordance with a user's operation. The processing information acquisition unit 110 outputs the acquired processing information to the learning data creation unit 130 .

[0020] The operation information acquisition unit 120 acquires information on whether or not abnormal vibration has occurred in the industrial machine 4 when the control device 3 that controls the industrial machine 4 executes the machining program acquired by the machining program acquisition unit 100 based on the machining information acquired by the machining information acquisition unit 110, and information on the cause of the abnormal vibration if abnormal vibration has occurred. The information on whether or not abnormal vibration has occurred may include, for example, information indicating whether abnormal vibration has occurred for each block of the machining program. Furthermore, the information on the cause of abnormal vibration may include wording indicating a specific cause of abnormal vibration, such as a large depth of cut, chip leakage, or intermittent cutting. The operation information acquisition unit 120 outputs the acquired information on whether or not abnormal vibration has occurred and information on the cause of abnormal vibration to the learning data creation unit 130.

[0021] The learning data creation unit 130 creates learning data to be used in supervised learning, using the machining program acquired by the machining program acquisition unit 100 and the machining information acquired by the machining information acquisition unit 110 as input data, and information on the occurrence or non-occurrence of abnormal vibration and information on the cause of abnormal vibration acquired by the operation information acquisition unit 120 as label data. The learning data creation unit 130 links each pair of machining program and machining information with a pair of information on the occurrence or non-occurrence of abnormal vibration and information on the cause of abnormal vibration. This linking may be performed manually by the user. The learning data creation unit 130 creates learning data by linking the pair of machining program and machining information with a pair of information on the occurrence or non-occurrence of abnormal vibration and information on the cause of abnormal vibration based on the user's operation via the input device 71. The learning data may be a pair of information on the occurrence or non-occurrence of abnormal vibration and information on the cause of abnormal vibration linked to one pair of machining program and machining information. The learning data creation unit 130 outputs the created learning data to the learning unit 140.

[0022] The learning unit 140 performs supervised learning using a predetermined model based on the learning data created by the learning data creation unit 130. Then, when a set of a machining program and machining information is input, a trained model is generated that outputs a set of information regarding the presence or absence of abnormal vibration and information regarding the cause of the abnormal vibration. The trained model outputs a set of information regarding the presence or absence of abnormal vibration and information regarding the cause of the abnormal vibration from the input machining program and machining information. The model used by the learning unit 140 may be, for example, an encoder / decoder model. In this case, for example, a convolutional neural network (CNN) or the like can be used as the encoder. Furthermore, the decoder may be a recurrent neural network (RNN) such as a gated recurrent unit (GRU) or a long short-term memory (LSTM). By using such a model, an intermediate representation vector can be generated from an input character string, voice, image, etc. using an encoder. Then, a set of information on the occurrence or non-occurrence of abnormal vibration and information on the cause of the abnormal vibration can be sequentially generated from the generated intermediate representation vector using a decoder. The learning unit 140 learns the encoding of general character strings, voice, and images and the decoding of the information on the occurrence or non-occurrence of abnormal vibration and information on the cause of the abnormal vibration using a general learning algorithm tailored to the respective adopted models. As a result, the trained model generated by the learning unit 140 outputs a set of information on the occurrence or non-occurrence of abnormal vibration and information on the cause of the abnormal vibration when a set of a machining program and machining information is input. The learning unit 140 stores the generated trained model in the model storage unit 160.

[0023] The output unit 180 outputs the trained model generated by the training unit 140. The output unit 180 may output the trained model based on an operation instruction from a user. The output unit 180 may output the trained model to, for example, an external device 72. The output unit 180 may also output the trained model to another computer such as the fog computer 6, the cloud server 7, or the abnormal vibration prediction device 2 via the network 5.

[0024] The machine learning device 1 having the above configuration generates a trained model that can be used to estimate information related to the occurrence of abnormal vibration and information related to the cause of the abnormal vibration from the machining program and machining information. The generated trained model can be used to predict the occurrence of abnormal vibration and the cause of the abnormal vibration based on the machining program and machining information prepared by the user before actually performing machining.

[0025] Second Embodiment A machine learning device according to a second embodiment of the present disclosure will be described below. The machine learning device 1 according to this embodiment has the same hardware configuration as the machine learning device 1 according to the first embodiment.

[0026] 4 is a schematic block diagram illustrating functions of the machine learning device 1 according to the second embodiment of the present disclosure. Each function of the machine learning device 1 according to this embodiment is realized by the CPU 11 of the machine learning device 1 shown in FIG. 1 executing a system program and controlling the operation of each unit of the machine learning device 1.

[0027] The machine learning device 1 of this embodiment includes a preprocessing unit 105 in addition to a machining program acquisition unit 100, a machining information acquisition unit 110, an operation information acquisition unit 120, a learning data creation unit 130, a learning unit 140, and an output unit 180. In addition, a model storage unit 160, which is an area for storing the trained model created by the learning unit 140, is prepared in advance in the RAM 13 to the nonvolatile memory 14 of the machine learning device 1.

[0028] The processing information acquisition unit 110, the motion information acquisition unit 120, the learning unit 140, and the output unit 180 included in the machine learning device 1 according to this embodiment have the same functions as those according to the first embodiment.

[0029] The machining program acquisition unit 100 according to this embodiment acquires a machining program to be executed by the control device 3 when the industrial machine 4 is in operation. The machining program acquired by the machining program acquisition unit 100 may include multiple programs that are in a calling relationship. The machining program acquisition unit 100 may acquire the machining program from the user via the input device 71. Also, the machining program may be acquired via the network 5 from the control device 3 that controls the industrial machine 4 in accordance with the user's operation. Furthermore, the machining program recorded in the external device 72 may be acquired in accordance with the user's operation. The machining program acquisition unit 100 outputs the acquired machining program to the pre-processing unit 105.

[0030] The preprocessing unit 105 performs preprocessing on the machining programs acquired by the machining program acquisition unit 100. Preprocessing is, for example, a process of combining multiple machining programs into one machining program. A machining program may call another machining program using a subprogram call command or a macro call command. FIG. 5 is a schematic diagram showing an example of a subprogram call. In the example of FIG. 5, the subprogram "O0002" is called in block N6 of the machining program "O0001." In this case, the preprocessing unit 105 replaces block N6 of the machining program "O0001" with the blocks of the subprogram "O0002," thereby combining the two machining programs into one machining program. FIG. 6 is a schematic diagram showing an example of combining the two machining programs illustrated in FIG. 5 into one machining program. In the example of FIG. 6, the command "M98P2" located at block N6 of the machining program "O0001" has been replaced with a group of blocks included in the subprogram "O0002."

[0031] The preprocessing performed by the preprocessing unit 105 may be, for example, a process of expanding a cycle command into multiple commands. A cycle command allows frequently used machining operations to be instructed in a single block. When a cycle command is executed, multiple commands are executed in a predetermined order. When a cycle command is included in a machining program, the preprocessing unit 105 replaces the cycle command with multiple commands. FIG. 7 is a schematic diagram showing an example of a machining program including a cycle command. In the example of FIG. 7, the N2 block of the machining program "O0003" includes a cycle command "G1450." In this case, the preprocessing unit 105 replaces the cycle command "G1450" in the N2 block of the machining program "O0003" with multiple commands executed by the cycle command. FIG. 8 is a schematic diagram showing an example of replacing a cycle command included in the machining program shown in FIG. 7. In the example of FIG. 8, the cycle command "G1450" located in the N2 block of the machining program "O0003" has been replaced with multiple commands executed by the cycle command.

[0032] The preprocessing unit 105 outputs the preprocessed machining program to the learning data creation unit 130. Then, the learning data creation unit 130 according to this embodiment creates learning data to be used in supervised learning, using the machining program preprocessed by the preprocessing unit 105 and the machining information acquired by the machining information acquisition unit 110 as input data, and the information on the occurrence or non-occurrence of abnormal vibration and the information on the cause of the occurrence of abnormal vibration acquired by the operation information acquisition unit 120 as label data.

[0033] The machine learning device 1 according to this embodiment, which has the above configuration, can convert multiple machining programs into a single machining program and use it to generate a trained model. Even if special cycle commands or manufacturer-dependent cycle commands are included, the machine learning device 1 can generate a trained model that learns by replacing the commands with commonly used commands.

[0034] 9 is a schematic hardware configuration diagram showing the main parts of an abnormal vibration prediction device according to a third embodiment of the present disclosure. The abnormal vibration prediction device 2 according to this embodiment can be implemented, for example, on a control device that controls industrial machinery. The abnormal vibration prediction device 2 can also be implemented, for example, on a personal computer attached to the control device, or on a computer such as a personal computer, cell computer, fog computer 6, or cloud server 7 that is connected to the control device via a wired or wireless network. This embodiment shows an example in which the abnormal vibration prediction device 2 is implemented on a computer that is connected via a network to the control device that controls industrial machinery.

[0035] The CPU 31 provided in the abnormal vibration prediction device 2 according to this embodiment is a processor that controls the entire abnormal vibration prediction device 2. The CPU 31 reads a system program stored in the ROM 32 via the bus 42, and controls the entire abnormal vibration prediction device 2 in accordance with the system program. The RAM 33 temporarily stores temporary calculation data, display data, various data acquired from outside, and the like.

[0036] The nonvolatile memory 34 is composed of, for example, a battery-backed memory (not shown) or an SSD (Solid State Drive), and retains its stored state even when the power to the abnormal vibration prediction device 2 is turned off. The nonvolatile memory 34 stores programs and data read from an external device 82 via the interface 35, programs and data input via the input device 81, and programs and data acquired via the network 5 from the machine learning device 1, the control device 3 that controls the industrial machine 4, and other devices. The programs and data stored in the nonvolatile memory 34 may be expanded into the RAM 33 when executed / used. In addition, various system programs such as known analysis programs are written in the ROM 32 in advance.

[0037] The interface 35 is an interface for connecting the CPU 31 of the abnormal vibration prediction device 2 to an external device 82 such as a USB device. For example, system programs, setting data, etc. are read from the external device 82. In addition, programs, setting data, etc. created or edited within the abnormal vibration prediction device 2 can be stored in external storage means via the external device 82.

[0038] The interface 40 is an interface for connecting the CPU 31 of the abnormal vibration prediction device 2 to a wired or wireless network 5. The network 5 may communicate using technologies such as serial communication such as RS-485, Ethernet (registered trademark), optical communication, wireless LAN, Wi-Fi (registered trademark), Bluetooth (registered trademark), etc. The machine learning device 1, a control device 3 that controls at least one industrial machine 4, a fog computer 6, a cloud server 7, etc. are connected to the network 5, and data is exchanged between the network 5 and the abnormal vibration prediction device 2.

[0039] The display device 80 displays data and the like obtained as a result of executing various data and programs loaded into the memory, output via the interface 37. An input device 81, which is composed of at least one input device such as a keyboard, a pointing device, a voice input device, an imaging device, etc., passes commands, data, etc. based on user operations to the CPU 31 via the interface 38.

[0040] 10 is a schematic block diagram showing functions of the abnormal vibration prediction device 2 according to the third embodiment of the present disclosure. Each function of the abnormal vibration prediction device 2 according to this embodiment is realized by the CPU 31 included in the abnormal vibration prediction device 2 shown in FIG. 9 executing a system program and controlling the operation of each part of the abnormal vibration prediction device 2.

[0041] The abnormal vibration prediction device 2 of this embodiment includes a machining program receiving unit 200, a machining information receiving unit 210, an estimation unit 230, and an output unit 290. The RAM 33 to the nonvolatile memory 34 of the abnormal vibration prediction device 2 also include a model storage unit 240, which is an area for storing a trained model generated by the machine learning device 1 according to the first or second embodiment.

[0042] The machining program receiving unit 200 receives a machining program to be executed by the control device 3 when the industrial machine 4 is in operation. The machining program received by the machining program receiving unit 200 is, for example, a numerical control program written in G-code or M-code as illustrated in FIG. 3 . The machining program receiving unit 200 may receive a machining program from a user via the input device 81. Alternatively, the machining program may be received via the network 5 from the control device 3 that controls the industrial machine 4 in accordance with a user operation. Furthermore, the machining program may be received from a processing unit 82 stored in an external device 82 in accordance with a user operation. The machining program receiving unit 200 outputs the received machining program to the estimation unit 230.

[0043] The machining information receiving unit 210 receives machining information related to machining in the control device 3 and industrial machine 4 that operate the machining program received by the machining program receiving unit 200. The machining information may include, for example, machining conditions set in the control device 3 and the industrial machine 4. It may also include information on the material shape and material quality of the workpiece to be machined in the industrial machine 4 before machining. In this case, the material shape may be indicated, for example, in text, or an image or a three-dimensional shape file (e.g., CAD data) showing the shape of the material before machining may be received. It may also include tool information related to a tool attached to the industrial machine 4. In this case, the tool information may be, for example, the tool model number, or an image or a three-dimensional shape file (e.g., CAD data) showing the shape of the tool. The machining information receiving unit 210 may receive machining information from a user via the input device 81. It may also receive machining information via the network 5 from the control device 3 that controls the industrial machine 4 in accordance with a user's operation. It may also receive machining information recorded in an external device 82 in accordance with a user's operation. The processing information receiving unit 210 outputs the received processing information to the estimation unit 230 .

[0044] The estimation unit 230 performs estimation using a trained model stored in the model storage unit 240 based on the machining program received by the machining program receiving unit 200 and the machining information received by the machining information receiving unit 210. When the control device 3 that controls the industrial machine 4 executes the machining program acquired by the machining program acquisition unit 100 under the machining information acquired by the machining information acquisition unit 110, the estimation unit 230 estimates information related to whether or not abnormal vibration has occurred in the industrial machine 4 and information related to the cause of the abnormal vibration if abnormal vibration has occurred. The model storage unit 240 stores a trained model generated in advance by the machine learning device 1 according to the first or second embodiment. When a set of a machining program and machining information is input, the trained model outputs a set of information related to whether or not abnormal vibration has occurred and information related to the cause of the abnormal vibration. The estimation unit 230 inputs the set of the machining program received by the machining program receiving unit 200 and the machining information received by the machining information receiving unit 210 to the trained model and obtains, as its output, a set of information related to whether or not abnormal vibration has occurred and information related to the cause of the abnormal vibration. The set of information on whether or not abnormal vibration has occurred and information on the cause of the abnormal vibration output by the trained model has the same format as the label data used for training by the machine learning device 1 according to the first or second embodiment. The estimation unit 230 outputs the set of information on whether or not abnormal vibration has occurred and information on the cause of the abnormal vibration obtained as a result of the estimation to the output unit 290.

[0045] The output unit 290 outputs information regarding the occurrence or non-occurrence of abnormal vibration estimated by the estimation unit 230 and information regarding the cause of the abnormal vibration. The output unit 290 may be configured to display and output to the display device 80. Alternatively, the output unit 290 may be configured to print out from a printing device (not shown). Furthermore, the output unit 290 may be configured to transmit and output via the network 5 to other computers such as the control device 3, the fog computer 6, and the cloud server 7. FIG. 11 is a screen diagram showing an example of the display output of the information regarding the occurrence or non-occurrence of abnormal vibration and information regarding the cause of the abnormal vibration by the output unit 290. As exemplified in FIG. 11 , the output unit 290 may be configured to display the occurrence or non-occurrence of abnormal vibration for each block of the machining program. Furthermore, blocks in which abnormal vibration occurs may be highlighted, for example, by using a different color or by highlighting.

[0046] The abnormal vibration prediction device 2 according to this embodiment, which is configured as described above, can predict whether or not abnormal vibration will occur before actually running a machining program. In addition, if abnormal vibration does occur, it can also identify the cause of the occurrence. This allows the user to take effective measures based on the cause of the abnormal vibration.

[0047] Fourth Embodiment An abnormal vibration prediction device according to a fourth embodiment of the present disclosure will be described below. The abnormal vibration prediction device 2 according to this embodiment has the same hardware configuration as the abnormal vibration prediction device 2 according to the third embodiment.

[0048] 12 is a schematic block diagram showing functions of the abnormal vibration prediction device 2 according to the third embodiment of the present disclosure. Each function of the abnormal vibration prediction device 2 according to this embodiment is realized by the CPU 31 included in the abnormal vibration prediction device 2 shown in FIG. 9 executing a system program and controlling the operation of each part of the abnormal vibration prediction device 2.

[0049] The abnormal vibration prediction device 2 of this embodiment further includes a preprocessing unit 205 in addition to the machining program receiving unit 200, the machining information receiving unit 210, the estimation unit 230, and the output unit 290. Furthermore, the RAM 33 to the nonvolatile memory 34 of the abnormal vibration prediction device 2 are provided with a model storage unit 240, which is an area for storing the trained model generated by the machine learning device 1 according to the first or second embodiment.

[0050] The processing information receiving unit 210 and the output unit 290 included in the abnormal vibration prediction device 2 according to this embodiment have the same functions as those according to the third embodiment.

[0051] The machining program receiving unit 200 according to this embodiment receives a machining program to be executed by the control device 3 when the industrial machine 4 is in operation. The machining program received by the machining program receiving unit 200 may include multiple programs that are in a calling relationship. The machining program receiving unit 200 may receive a machining program from a user via the input device 81. Also, the machining program may be received via the network 5 from the control device 3 that controls the industrial machine 4 in accordance with a user operation. Furthermore, the machining program may be received from a machining program recorded in an external device 82 in accordance with a user operation. The machining program receiving unit 200 outputs the received machining program to the pre-processing unit 205.

[0052] The pre-processing unit 205 performs pre-processing on the machining program received by the machining program receiving unit 200. The pre-processing is, for example, a process of combining multiple machining programs into one machining program. The pre-processing performed by the pre-processing unit 205 may also be, for example, a process of expanding a cycle command into multiple commands. The pre-processing unit 205 outputs the pre-processed machining program to the estimating unit 230. Then, the estimating unit 230 according to this embodiment estimates information on whether or not abnormal vibration has occurred and information on the cause of the abnormal vibration, based on the machining program pre-processed by the pre-processing unit 205 and the machining information received by the machining information receiving unit 210.

[0053] The abnormal vibration prediction device 2 according to this embodiment, which is configured as described above, can make estimations related to a plurality of machining programs. Furthermore, even if a special cycle command or a manufacturer-dependent cycle command is included, the estimation is made after replacing it with a commonly used command, so that the accuracy of the estimation can be expected to be improved.

[0054] Fifth Embodiment An abnormal vibration prediction device according to a fifth embodiment of the present disclosure will be described below. The abnormal vibration prediction device 2 according to this embodiment has the same hardware configuration as the abnormal vibration prediction device 2 according to the third embodiment.

[0055] 13 is a schematic block diagram showing functions of the abnormal vibration prediction device 2 according to the fifth embodiment of the present disclosure. Each function of the abnormal vibration prediction device 2 according to this embodiment is realized by the CPU 31 included in the abnormal vibration prediction device 2 shown in FIG. 9 executing a system program and controlling the operation of each part of the abnormal vibration prediction device 2.

[0056] The abnormal vibration prediction device 2 of this embodiment includes a determination unit 270 and a change unit 280 in addition to a machining program reception unit 200, a machining information reception unit 210, an estimation unit 230, and an output unit 290. The RAM 33 to the nonvolatile memory 34 of the abnormal vibration prediction device 2 also include a model storage unit 240, which is an area that stores the trained model generated by the machine learning device 1 according to the first or second embodiment, and a change condition storage unit 250, which is an area that stores change conditions for the machining program in advance.

[0057] The processing information receiving unit 210 included in the abnormal vibration prediction device 2 according to this embodiment has the same functions as those according to the third embodiment.

[0058] The machining program receiving unit 200 according to this embodiment outputs the received machining program to the estimating unit 230 and the changing unit 280. Furthermore, the estimating unit 230 according to this embodiment outputs to the determining unit 270 the estimation results of information relating to the presence or absence of abnormal vibration and information relating to the cause of the occurrence of abnormal vibration.

[0059] The determination unit 270 determines whether or not to change the machining program based on information related to the presence or absence of abnormal vibration included in the estimation result by the estimation unit 230. The information related to the presence or absence of abnormal vibration may include information indicating in which block of the machining program abnormal vibration will occur. If the information related to the estimated presence or absence of abnormal vibration indicates that abnormal vibration will not occur, the determination unit 270 outputs to the modification unit 280 a message indicating that the machining program does not need to be changed. On the other hand, if the information indicates that abnormal vibration will occur, the determination unit 270 outputs to the modification unit 280 a message indicating that the machining program needs to be changed together with information related to the block in which abnormal vibration will occur.

[0060] When the determination unit 270 determines that abnormal vibration will not occur, the change unit 280 outputs this to the output unit 290. On the other hand, when the determination unit 270 determines that abnormal vibration will occur, the change unit 280 refers to the change conditions stored in the change condition storage unit 250 and identifies a method for changing the machining program to eliminate the abnormal vibration. Then, the change unit 280 changes the machining program according to the identified change method. Then, the change unit 280 outputs the changed machining program to the output unit 290.

[0061] FIG. 14 is a schematic diagram illustrating example change conditions stored in the change condition storage unit 250. The change condition storage unit 250 stores, as change conditions, change methods for the machining program to eliminate abnormal vibrations, each of which has a predetermined priority. In the example of FIG. 14 , for example, a change method for the machining program to "reduce the cutting depth in the block where abnormal vibration occurs" is stored as the change condition with the highest priority. This change method may include instructions for making the change and an arithmetic expression for parameters. The change unit 280 acquires unapplied change conditions with high priority stored in the change condition storage unit 250. The change unit 280 then changes the machining program based on the change method for the acquired change conditions. When changing the machining program, for example, it is sufficient to include instructions indicated in the change method for the block where abnormal vibration occurs, or change a predetermined parameter to a value calculated using an arithmetic expression.

[0062] The output unit 290 according to this embodiment outputs the result of the determination made by the determination unit 270 regarding the presence or absence of abnormal vibration, together with the machining program changed by the change unit 280 .

[0063] The abnormal vibration prediction device 2 according to the present embodiment having the above configuration changes the machining program that generates abnormal vibration based on the presence or absence of abnormal vibration and the cause of the abnormal vibration estimated by the estimation unit. Therefore, the operator can obtain a machining program that does not generate abnormal vibration based on the initially input machining program without any effort.

[0064] As a modified example of the abnormal vibration prediction device 2 according to this embodiment, the estimation unit 230 may re-estimate information regarding the occurrence or non-occurrence of abnormal vibration and information regarding the cause of the abnormal vibration for the machining program changed by the change unit 280. Then, such machining program change and estimation processing is repeated until the determination unit 270 determines that abnormal vibration will not occur. If the abnormal vibration is not resolved, the change unit 280 may acquire a change condition other than the applied change condition. With this configuration, the estimation processing and machining program change are automatically performed until an appropriate change is made to the machining program, thereby reducing the effort required for the user to check the estimation results.

[0065] Sixth Embodiment An abnormal vibration prediction device according to a sixth embodiment of the present disclosure will be described below. The abnormal vibration prediction device 2 according to this embodiment has the same hardware configuration as the abnormal vibration prediction device 2 according to the third embodiment.

[0066] 15 is a schematic block diagram showing functions of an abnormal vibration prediction device 2 according to the sixth embodiment of the present disclosure. Each function of the abnormal vibration prediction device 2 according to this embodiment is realized by a CPU 31 included in the abnormal vibration prediction device 2 shown in FIG. 9 executing a system program and controlling the operation of each part of the abnormal vibration prediction device 2.

[0067] The abnormal vibration prediction device 2 of this embodiment further includes a change method selection unit 275 in addition to a machining program reception unit 200, a machining information reception unit 210, an estimation unit 230, a determination unit 270, a change unit 280, and an output unit 290. The RAM 33 to the nonvolatile memory 34 of the abnormal vibration prediction device 2 are provided with a model storage unit 240 which is an area for storing the trained model generated by the machine learning device 1 according to the first or second embodiment, and a change condition storage unit 250 which is an area for storing in advance change conditions for the machining program.

[0068] The abnormal vibration prediction device 2 according to this embodiment includes a machining program receiving unit 200, a machining information receiving unit 210, an estimation unit 230, and an output unit 290, which have the same functions as those of the fifth embodiment.

[0069] The determination unit 270 according to this embodiment outputs the result of the determination regarding the presence or absence of abnormal vibration to the change method selection unit 275. When the determination unit 270 determines that abnormal vibration will occur, the change method selection unit 275 selects how to change the machining program. This selection may be made, for example, by having the user input the details of the changes to the machining program via the input device 81. Alternatively, the change conditions stored in the change condition storage unit 250 may be displayed on the display device 80 to allow the user to select a method of changing the machining program. Alternatively, the change method of the machining program for eliminating the cause of the abnormal vibration may be automatically selected from the change conditions stored in the change condition storage unit 250.

[0070] FIG. 16 is a schematic diagram illustrating example change conditions stored in the change condition storage unit 250 according to this embodiment. The change condition storage unit 250 according to this embodiment stores change conditions that associate causes of abnormal vibration with machining program change methods for eliminating the abnormal vibration. In the example of FIG. 16 , for example, the first change condition associates the cause of abnormal vibration, "large depth of cut," with a machining program change method of "reducing the depth of cut in the block where abnormal vibration occurs." This change method may include commands for making the change and parameter calculation formulas. When automatically selecting a machining program change method, the change method selection unit 275 uses information related to the estimated cause of abnormal vibration to search for the "cause of abnormal vibration" item in the change conditions stored in the change condition storage unit 250. At this time, it is preferable to use well-known techniques such as fuzzy search, concept search, and semantic search. The change method selection unit 275 then selects the machining program change method indicated by the searched change condition.

[0071] The modification method selection unit 275 outputs the selected modification method for the machining program to the modification unit 280. Then, the modification unit 280 modifies the machining program based on the modification method selected by the modification method selection unit 275.

[0072] The abnormal vibration prediction device 2 according to the present embodiment, which is configured as described above, changes the machining program that causes abnormal vibration based on the estimated presence or absence of abnormal vibration and the cause of the abnormal vibration. Because the user can select the method of changing the machining program, it is possible to more reliably eliminate the cause of abnormal vibration.

[0073] [Other Embodiments] In the above-described embodiment, the abnormal vibration prediction device 2 is configured to include the model storage unit 240 and the change condition storage unit 250. However, the model storage unit 240 and the change condition storage unit 250 may be provided on other devices, such as the fog computer 6 or the cloud server 7. In this case, the abnormal vibration prediction device 2 references the model storage unit 240 and the change condition storage unit 250 via the network 5. With such a configuration, it becomes possible to collectively manage trained models and change conditions at a manufacturing site where many control devices 3 and industrial machines 4 are installed.

[0074] Although the embodiments of the present disclosure have been described in detail above, the present disclosure is not limited to the individual embodiments described above. Various additions, substitutions, modifications, partial deletions, etc. are possible in these embodiments without departing from the gist of the invention or the idea and intent of the present disclosure derived from the content described in the claims and their equivalents. For example, in the above-described embodiments, the order of each operation and the order of each process are shown as examples and are not limited to these. The same applies when numerical values ​​or mathematical expressions are used in the description of the above-described embodiments.

[0075] The following are supplementary notes related to embodiments of the present disclosure. (Supplementary Note 1) A machine learning device (1) according to one aspect of the present disclosure includes: a machining program acquisition unit (100) that receives at least one machining program used to operate an industrial machine (4), a machining information acquisition unit (110) that receives machining information including at least one of machining conditions, a material shape before machining, material information, and tool information, an operation information acquisition unit (120) that acquires information on whether or not abnormal vibration occurs in the industrial machine (4) when the machining program is executed under the machining information, and information on the cause of the abnormal vibration if abnormal vibration occurs, a learning unit (140) that performs supervised learning using the machining program and the machining information as input data and the information on whether or not abnormal vibration occurs and the information on the cause of the abnormal vibration as label data, and generates a trained model that outputs information on whether or not abnormal vibration occurs and information on the cause of the abnormal vibration when the machining program and the machining information are input, and an output unit (180) that outputs the trained model.

[0076] (Supplementary Note 2) A machine learning device (1) according to another aspect of the present disclosure further includes a preprocessing unit (105) that performs at least one of a process of combining multiple machining programs that are in a calling relationship into one and a process of converting cycle commands included in the machining programs as preprocessing, and the learning unit (140) performs learning using the machining programs and the machining information preprocessed by the preprocessing unit (105) as input data.

[0077] (Supplementary Note 3) An abnormal vibration prediction device (2) according to one aspect of the present disclosure includes a trained model generated by a machine learning device (1), and includes: a machining program receiving unit (200) that receives at least one machining program to be used in operating an industrial machine (4); a machining information receiving unit (210) that receives machining information including at least any one of machining conditions, a material shape before machining, material information, and tool information; an estimation unit (230) that performs estimation using the trained model based on the machining program and the machining information to estimate information related to the occurrence or non-occurrence of abnormal vibration and information related to the cause of the occurrence of abnormal vibration; and an output unit (290) that outputs the result of the estimation.

[0078] (Supplementary Note 4) An abnormal vibration prediction device (2) according to another aspect of the present disclosure further includes a preprocessing unit (205) that executes, as preprocessing, at least one of a process of combining multiple machining programs that are in a calling relationship into one and a process of converting cycle commands included in the machining programs, and the estimation unit (230) performs estimation based on the machining program preprocessed by the preprocessing unit (205) and the machining information. (Supplementary Note 5) An abnormal vibration prediction device (2) according to another aspect of the present disclosure further includes a determination unit (270) that determines whether to change the machining program based on information related to the presence or absence of abnormal vibration included in the estimation result, a change condition storage unit (250) that stores change conditions for the machining program, and a change unit (280) that changes the machining program based on the result of the determination by the determination unit (270) and the change conditions, and the output unit (290) outputs at least one of the result of the determination and the changed machining program.

[0079] (Supplementary Note 6) In an abnormal vibration prediction device (2) according to another aspect of the present disclosure, the estimation unit (230) estimates information relating to the presence or absence of abnormal vibration during operation of the changed machining program and information relating to the cause of the abnormal vibration, based on the machining program changed by the change unit (280) and the machining information, and repeats the change of the machining program and estimation process until the determination unit (270) determines, based on the estimation result related to the changed machining program, that abnormal vibration will not occur. (Supplementary Note 7) The abnormal vibration prediction device (2) according to another aspect of the present disclosure further includes a change method selection unit (275) that selects a change method for the machining program based on the cause of the abnormal vibration, and the change unit (280) changes the machining program based on the change method selected by the change method selection unit (275).

[0080] (Supplementary Note 8) A computer-readable recording medium according to one aspect of the present disclosure records a program that causes a computer to operate as: a machining program receiving unit (200) that receives at least one machining program to be used for operating an industrial machine (4); a machining information receiving unit (210) that receives machining information including at least any one of machining conditions, pre-machining material shape, material information, and tool information; an estimation unit (230) that performs estimation using a trained model that, when a set of the machining program and the machining information is input, outputs a set of information related to the presence or absence of abnormal vibration and information related to the cause of the abnormal vibration, and estimates the information related to the presence or absence of abnormal vibration and information related to the cause of the abnormal vibration; and an output unit (290) that outputs the result of the estimation.

[0081] LIST OF SYMBOLS 1 Machine learning device 2 Abnormal vibration prediction device 3 Control device 4 Industrial machine 5 Network 6 Fog computer 7 Cloud server 11 CPU 12 ROM 13 RAM 14 Non-volatile memory 15, 17, 18, 20 Interface 22 Bus 70 Display device 71 Input device 72 External device 100 Machining program acquisition unit 105 Pre-processing unit 110 Machining information acquisition unit 120 Operation information acquisition unit 130 Learning data creation unit 140 Learning unit 160 Model storage unit 180 Output unit 31 CPU 32 ROM 33 RAM 34 Non-volatile memory 35, 37, 38, 40 Interface 42 Bus 80 Display device 81 Input device 82 External device 200 Machining program reception unit 205 Pre-processing unit 210 Machining information reception unit 230 Estimation unit 240 Model storage unit 250 Change condition storage unit 270 Determination unit 275 Change method selection unit 280 Change unit 290 Output unit

Claims

1. A machine learning device comprising: a machining program acquisition unit that receives at least one machining program used to operate industrial machinery; a machining information acquisition unit that receives machining information including at least one of machining conditions, material shape before machining, material information, and tool information; an operation information acquisition unit that acquires information on whether or not abnormal vibrations will occur in the industrial machinery when the machining program is executed under the machining information, and information on the cause of the abnormal vibrations if abnormal vibrations occur; a learning unit that performs supervised learning using the machining program and the machining information as input data and the information on whether or not abnormal vibrations will occur and the information on the cause of the abnormal vibrations as label data, and generates a trained model that outputs information on whether or not abnormal vibrations will occur and information on the cause of the abnormal vibrations when the machining program and the machining information are input; and an output unit that outputs the trained model.

2. The machine learning device according to claim 1, further comprising a pre-processing unit that performs at least one of a process of consolidating multiple machining programs that are in a calling relationship into one and a process of converting cycle commands included in the machining programs as pre-processing, and the learning unit performs learning using the machining programs pre-processed by the pre-processing unit and the machining information as input data.

3. An abnormal vibration prediction device comprising a trained model generated by the machine learning device according to claim 1 or 2, and comprising: a machining program receiving unit that receives at least one machining program to be used in operating industrial machinery; a machining information receiving unit that receives machining information including at least any of machining conditions, material shape before machining, material information, and tool information; an estimation unit that performs estimation using the trained model based on the machining program and the machining information, and estimates information related to the occurrence or non-occurrence of abnormal vibration and information related to the cause of the occurrence of abnormal vibration; and an output unit that outputs the results of the estimation.

4. An abnormal vibration prediction device as described in claim 3, further comprising a pre-processing unit that performs at least one of a process of combining multiple machining programs that are in a calling relationship into one and a process of converting cycle commands included in the machining programs as pre-processing, and the estimation unit makes an estimation based on the machining programs pre-processed by the pre-processing unit and the machining information.

5. An abnormal vibration prediction device as described in claim 3, further comprising: a judgment unit that judges whether or not to change the machining program based on information related to the presence or absence of abnormal vibration contained in the estimation result; a change condition storage unit that stores change conditions for the machining program; and a change unit that changes the machining program based on the result of the judgment by the judgment unit and the change conditions, wherein the output unit outputs at least one of the result of the judgment and the changed machining program.

6. The abnormal vibration prediction device according to claim 5, wherein the estimation unit estimates information relating to whether or not abnormal vibration will occur during operation of the modified machining program and information relating to the cause of the abnormal vibration, based on the machining program modified by the modification unit and the machining information, and the determination unit repeats modifying the machining program and the estimation process until it determines that abnormal vibration will not occur based on the estimation result relating to the modified machining program.

7. An abnormal vibration prediction device as described in claim 5, further comprising a change method selection unit that selects a method of changing the machining program based on the cause of the abnormal vibration, and the change unit changes the machining program based on the change method selected by the change method selection unit.

8. A computer-readable recording medium having recorded thereon a program that causes a computer to operate as: a machining program receiving unit that receives at least one machining program used to operate industrial machinery; a machining information receiving unit that receives machining information including at least any one of machining conditions, material shape before machining, material information, and tool information; an estimation unit that performs estimation using a trained model that, when a set of the machining program and the machining information is input, outputs a set of information regarding the occurrence or non-occurrence of abnormal vibration and information regarding the cause of the abnormal vibration, and estimates information regarding the occurrence or non-occurrence of abnormal vibration and information regarding the cause of the abnormal vibration; and an output unit that outputs the results of the estimation.

Images