Control devices, control systems, and machine learning devices

Abstract

The present invention provides a control device, a control system, and a machine learning device. The control device comprises: a data acquisition unit that acquires at least data related to the operating state of an industrial machine; a learning model storage unit that stores a learning model that associates the value of the setting behavior of the base speed of a peak-cutting motor with the operating state of the industrial machine; and a decision unit that uses the learning model stored in the learning model storage unit to determine the setting behavior of the base speed of the peak-cutting motor based on the data related to the operating state of the industrial machine (2) acquired by the data acquisition unit.

Description

Technical Field

[0001] The present invention relates to a control device, a control system and a machine learning device. Background Art

[0002] A plurality of industrial machines such as machine tools, injection molding machines, robots, etc. installed at a manufacturing site such as a factory operate according to instructions from individual control devices that control the industrial machines. These multiple industrial machines are connected to a predetermined power supply device and consume the power provided by the power supply device to operate (for example, Japanese Patent Publication No. 2017-162300). At this time, if the industrial machine executes an instruction that consumes a large amount of power, a large amount of power will be consumed by the industrial machine. In order to reduce the power consumption in this case, for example, Figure 7 As shown, a separate motor is pre-driven from the industrial machinery's drive motor, and the motor is decelerated at the timing when the industrial machinery consumes power, generating regenerative energy and thereby suppressing power consumption. The separate motor provided for this purpose is referred to in this specification as a peak-shaving motor.

[0003] As mentioned above, peak-cutting motors need to decelerate to generate regenerative energy when industrial machinery consumes power. Therefore, they must normally rotate at a predetermined speed (base speed). This base speed is determined by on-site operators based on experience, taking into account the operation of the industrial machinery, such as the operating pattern of the industrial machinery's drive motor. The base speed is typically set slightly higher to prevent the peak-cutting motor's speed from dropping excessively when it decelerates to generate regenerative energy.

[0004] While the base speed of the peak-shaving motor is determined when the industrial machinery consumes the most power, for example, when the output of the industrial machinery's drive motor is at its highest, the peak-shaving motor should be rotated at a high speed even when the output of the industrial machinery's drive motor is low. However, setting the base speed of the peak-shaving motor uniformly high can increase power loss in the peak-shaving motor due to increased iron loss, shorten the life of the peak-shaving motor's bearings, and other issues.

[0005] Therefore, a method of learning and setting an appropriate base speed of a peak-cutting motor according to the operating state of the industrial machine is desired. Summary of the Invention

[0006] A control device according to one embodiment of the present invention solves the above-mentioned problem by learning the appropriate base speed of the peak-cutting motor in each state of the industrial machine by trial and error based on data related to the output of the industrial machine and adjusting the base speed based on the learning result.

[0007] In addition, one embodiment of the present invention is a control device that controls the peak shaving action of a peak shaving motor connected to the same power supply path as at least one industrial machine, and comprises: a data acquisition unit that acquires at least data related to the operating state of the above-mentioned industrial machine; a learning model storage unit that stores a learning model that is associated with the value of the setting behavior of the base speed of the above-mentioned peak shaving motor with respect to the operating state of the above-mentioned industrial machine; and a decision unit that uses the learning model stored in the above-mentioned learning model storage unit to determine the setting behavior of the base speed of the above-mentioned peak shaving motor based on the data related to the operating state of the above-mentioned industrial machine acquired by the above-mentioned data acquisition unit.

[0008] Another aspect of the present invention is a control system including a plurality of control devices connected to each other via a network, and capable of sharing the learning results of the learning unit among the plurality of control devices.

[0009] Another embodiment of the present invention is a machine learning device that learns the setting behavior of the base speed of the peak shaving motor in the control of the peak shaving action of the peak shaving motor connected to the same power supply path as at least one industrial machine, and the machine learning device comprises: a learning model storage unit that stores a learning model of the value of the setting behavior of the base speed of the peak shaving motor associated with the action state of the industrial machine; and a decision unit that uses the learning model stored in the learning model storage unit to determine the setting behavior of the base speed of the peak shaving motor based on the data related to the action state of the industrial machine acquired by the data acquisition unit.

[0010] According to one embodiment of the present invention, since the base speed of the peak cutting motor does not exceed necessary, the loss of the peak cutting motor can be reduced and the bearing life of the peak cutting motor can be extended. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The above and other objects and features of the present invention will be made clear by the following embodiments described with reference to the accompanying drawings. In these drawings:

[0012] Figure 1 This is a schematic hardware configuration diagram of a control device according to one embodiment.

[0013] Figure 2 This is a schematic functional block diagram of the control device according to the first embodiment.

[0014] Figure 3 Explain the power consumption and power regeneration of industrial machinery.

[0015] Figure 4 An operating environment of a control system according to one embodiment is schematically shown.

[0016] Figure 5This is a schematic functional block diagram of a control system according to the second embodiment.

[0017] Figure 6 A system in which multiple control systems operate is schematically shown.

[0018] Figure 7 A structure for reducing power consumption using a conventional peak-cutting motor will be described. DETAILED DESCRIPTION

[0019] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0020] Figure 1 This is a schematic hardware configuration diagram showing the main components of a control device according to one embodiment of the present invention. The control device 1 of this embodiment can be implemented as a control device for controlling a peak-shaving motor. Furthermore, the control device 1 can be implemented as a computer such as a personal computer installed in a factory, a cell computer connected to a factory network, a fog computer, or a cloud server. This embodiment illustrates an example in which the control device 1 is implemented as a control device for controlling a peak-shaving motor.

[0021] The CPU 11 included in the numerical controller 1 of this embodiment is a processor that controls the entire controller 1. The CPU 11 reads a system program stored in the ROM 12 connected via the bus 20 and controls the entire controller 1 according to the system program. The RAM 13 stores temporary calculation data, display data for display on the display device 70, and various data input by the operator via the input device 71.

[0022] The non-volatile memory 14 is composed of, for example, a memory backed up by a battery (not shown), an SSD (Solid State Drive), etc., and maintains the storage state even if the power to the control device 1 is cut off. The non-volatile memory 14 stores a setting area for storing setting information related to the operation of the control device 1, data input from the input device 71 via the interface 18, and control programs. In addition, the non-volatile memory 14 also stores various data obtained from the industrial machines 2 via the network 5 (control programs executed in each industrial machine 2, output calculation values, output measurement values, and motor speed of the driving motor of the industrial machine 2, etc.), data read via an external storage device (not shown), control programs, etc. The programs and various data stored in the non-volatile memory 14 can be expanded into the RAM 13 when executed / used. In addition, the ROM 12 is pre-written with a system program, which includes a well-known analysis program for analyzing various data.

[0023] The control device 1 is connected to a wired / wireless network 5 via an interface 16. At least one industrial machine 2 connected to a power supply path for supplying power to the control device 1 is connected to the network 5, and data can be exchanged with the control device 1.

[0024] Various data read into the memory, data obtained as a result of executing a program, etc., and data output from the machine learning device 100 (described later) are output to and displayed on the display device 70 via the interface 17. Furthermore, the input device 71, which is composed of a keyboard, a pointing device, etc., transmits instructions and data based on operator operations to the CPU 11 via the interface 18.

[0025] The axis control circuit 30 receives an axis movement command from the CPU 11 and outputs the axis command to the servo amplifier 40. The servo amplifier 40 receives the command and drives the peak-shaving servo motor 50. The peak-shaving servo motor 50 has a built-in position / speed detector and feeds the position / speed feedback signal from the position / speed detector back to the axis control circuit 30, performing position / speed feedback control.

[0026] The interface 21 connects the CPU 11 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, etc., a RAM 103 for temporarily storing various processes related to machine learning, and a nonvolatile memory 104 for storing learning models, etc. The machine learning device 100 can observe various information that can be obtained by the control device 1 (such as the control program executed by each industrial machine 2, the calculated output value, measured output value, and motor speed of the drive motor of the industrial machine 2) via the interface 21.

[0027] Furthermore, the control device 1 receives the result output from the machine learning device 100 and performs display on the display device 70 , outputs a command to the peak-cut servo motor 50 to be controlled, and the like.

[0028] Figure 2 This is a schematic functional block diagram of the control device 1 and the machine learning device 100 according to the first embodiment. The control device 1 of this embodiment includes the configuration required for the machine learning device 100 to perform learning and the configuration required for processing based on the decision of the machine learning device 100. Figure 2 The functional blocks shown are composed of Figure 1 The CPU 11 included in the control device 1 and the processor 101 of the machine learning device 100 execute their respective system programs to control the operations of the respective components of the control device 1 and the machine learning device 100 .

[0029] The control device 1 of this embodiment includes a control unit 32, a data acquisition unit 34, and a pre-processing unit 36. The machine learning device 100 included in the control device 1 includes a learning unit 110 and a decision unit 120. Figure 1 The nonvolatile memory 14 shown is provided with an acquired data storage unit 54 for storing data acquired from the industrial machine 2 and the like. Figure 1 The nonvolatile memory 104 of the illustrated machine learning device 100 is provided with a learning model storage unit 130 , which stores a learning model constructed by machine learning of the learning unit 110 .

[0030] The control unit 32 is configured as follows: Figure 1 The CPU 11 of the control device 1 shown in FIG. 1 executes a system program read from a ROM 12, and mainly performs arithmetic processing using the RAM 13 and the nonvolatile memory 14 of the CPU 11 and a control process of the axis control circuit 30 of the peak-cutting servo motor 50. The control unit 32 performs a control process based on the system program stored in the ROM 12. Figure 1 The control program 52 stored in the nonvolatile memory 14 controls the operation of the peak-cut servo motor 50. The control unit 32 includes general control functions required to control the operation of various components of the industrial machine 2. The control program 52 outputs movement commands, etc., to the peak-cut servo motor 50 in each control cycle.

[0031] The control unit 32 controls the peak-shaving servo motor 50 to be driven at a base speed set in the control program 52 or in a setting area in the nonvolatile memory 14 of the control device 1. Furthermore, the control unit 32 controls the peak-shaving servo motor 50 to be driven at a base speed set in the control program 52 or in a setting area in the nonvolatile memory 14 of the control device 1. Furthermore, the control unit 32 obtains the operating status of the industrial machine 2 via the network 5 and, when the power consumption of the industrial machine 2 increases, decelerates the peak-shaving servo motor 50 in accordance with the increase, thereby generating regenerative energy.

[0032] Figure 3 The relationship between the change in the power consumed by the industrial machine 2 and the speed control of the servo motor 50 for peak shaving is schematically shown. Figure 3 As shown, the control unit 32 normally drives the peak-shaving servo motor 50 at a set base speed. When the power consumed by the industrial machine 2 increases, the control unit 32 decelerates the peak-shaving servo motor 50 to generate regenerative power. Furthermore, the speed of the peak-shaving servo motor 50 is accelerated to the base speed based on the timing of regenerative power generation in the industrial machine 2.

[0033] The power consumption of the industrial machine 2 mainly increases when the speed of the driving motor of the industrial machine 2 increases due to the instructions executed by the industrial machine 2. However, the amount of power consumption increase in this case can be calculated based on the acceleration of the driving motor, etc. Therefore, by sequentially obtaining the operating state of the industrial machine 2, the power consumption of the industrial machine 2 can be estimated, and the peak shaving servo motor 50 can be appropriately decelerated to generate appropriate regenerative energy. In addition, the increase in the power consumption of the industrial machine 2 can also be directly measured by installing a sensor such as a power meter on the industrial machine 2. The technology for generating regenerative energy by the control unit 32 to reduce the power peak is already known, for example, from Japanese Patent Application Laid-Open No. 2018-153041, and therefore a detailed description thereof is omitted in this specification.

[0034] When the setting of the base speed of the peak-cut servo motor 50 is output from the machine learning device 100 , the control unit 32 sets the base speed of the peak-cut servo motor 50 to the base speed output from the machine learning device 100 .

[0035] The data acquisition unit 34 is configured as follows: Figure 1 The CPU 11 included in the control device 1 shown executes a system program read from the ROM 12. The CPU 11 primarily performs computational processing using the RAM 13 and nonvolatile memory 14, as well as input and output processing using interfaces 16 and 18. A data acquisition unit 34 acquires various data from the industrial machines 2, the input device 71, and the like. For example, the data acquisition unit 34 acquires information such as the control program executed by each industrial machine 2, the currently executed block within the control program, and the calculated and measured output values and motor speeds of the drive motors of the industrial machines 2, and stores these information in the acquired data storage unit 54. The data acquisition unit 34 may also acquire data from another computer (not shown) via an external storage device (not shown) or the network 5.

[0036] The pre-processing unit 36 is composed of Figure 1 The CPU 11 included in the control device 1 shown executes a system program read from the ROM 12, and the CPU 11 primarily performs computational processing using the RAM 13 and nonvolatile memory 14. The preprocessing unit 36 generates state data for use in learning by the machine learning device 100 based on the data acquired by the data acquisition unit 34. The preprocessing unit 36 generates state data by converting the data acquired by the data acquisition unit 34 into a unified format (such as digitization and sampling) for processing in the machine learning device 100. For example, when the machine learning device 100 performs reinforcement learning, the preprocessing unit 36 generates a set of state data S and judgment data D in a predetermined format for the learning process.

[0037] The state data S generated by the pre-processing unit 36 includes drive motor data S1 , which is data related to the drive motor in the industrial machine 2 , and base speed data S2 , which indicates a base speed setting value in the operating state of the industrial machine 2 indicated by the drive motor data S1 .

[0038] The driving motor data S1 is defined as data indicating the operating state of the driving motor in the industrial machine 2. The driving motor data S1 may also include data indicating the current operating state of the driving motor of the industrial machine 2. Furthermore, the driving motor data S1 may also include data indicating the operating state of the driving motor of the industrial machine 2 within a predetermined period after the acquisition of data obtained by analyzing the control program of the industrial machine 2. Examples of data indicating the operating state of the driving motor of the industrial machine 2 include the speed and displacement of the driving motor of the industrial machine 2, the output value and displacement of the driving motor calculated based on the speed and displacement, and the measured value of the power consumed by the driving motor measured by a power meter or the like. The driving motor data S1 may also include time series data of speed, output value, and power consumption value.

[0039] Base speed data S2 is data indicating the set value of the base speed of peak-shaving servo motor 50 in the operating state of industrial machine 2 indicated by drive motor data S1. Base speed data S2 can use the base speed set by control program 52 or the base speed set in the setting area of nonvolatile memory 14 at the start of operation of control device 1. Alternatively, after the base speed is set based on the set value of the base speed output from machine learning device 100, the set base speed value can be used as is as base speed data S2.

[0040] The determination data D generated by the pre-processing unit 36 is used to determine (evaluate) the appropriateness of the base speed setting indicated by the base speed data S2, given the power consumption indicated by the drive motor data S1. The determination data D includes at least peak-cutting operation determination data D1 indicating the adequacy of the peak-cutting operation for the power consumption of the industrial machine, and operation cost determination data D2 regarding the operation cost of the peak-cutting servo motor 50.

[0041] The peak-cutting operation determination data D1 may, for example, be a value indicating the extent to which peak cutting can be achieved by decelerating the peak-cutting servo motor 50 when power consumption increases in the industrial machine 2. For example, the peak-cutting operation determination data D1 may be the ratio of the amount of regenerative power generated by the peak-cutting servo motor 50 to the amount of power consumed by the industrial machine 2. Alternatively, the peak-cutting operation determination data D1 may be the amount of power supplied from the public power supply to the power supply path when the power consumption of the industrial machine 2 increases. In other words, the peak-cutting operation determination data D1 may be data that can determine how to appropriately perform the peak-cutting operation.

[0042] The operation cost determination data D2 may be, for example, a value indicating the base speed of the peak-cut servo motor 50. That is, the operation cost determination data D2 may be data capable of determining at what base speed the peak-cut servo motor 50 can be operated.

[0043] The learning unit 110 is configured as follows: Figure 1 The processor 101 included in the control device 1 shown executes a system program read from the ROM 102. The processor 101 primarily performs computational processing using the RAM 103 and non-volatile memory 104. The learning unit 110 performs machine learning using data generated by the preprocessing unit 36. Using a known reinforcement learning method, the learning unit 110 generates a learning model that learns the behavior for setting the base speed of the peak-shaving servo motor 50 in response to the operating state of the drive motor in the industrial machine 2, and stores the generated learning model in the learning model storage unit 130. Reinforcement learning is a method that repeats a trial-and-error cycle of observing the current state of the environment (i.e., input) while executing a predetermined behavior (i.e., output) in the current state and providing a certain reward for that behavior, thereby maximizing the total reward (in this case, the behavior for setting the base speed of the peak-shaving servo motor 50) to learn a more appropriate solution. Examples of reinforcement learning techniques performed by the learning unit 110 include Q-learning.

[0044] Regarding the feedback R in the Q learning performed by the learning unit 110, for example, if the regenerative power generated by the peak-cut servo motor 50 can be sufficiently provided relative to the power consumption of the industrial machine 2 (the ratio is 1.0 or greater), the result is determined to be "good," and a positive (+) feedback R is given. Furthermore, if the regenerative power generated by the peak-cut servo motor 50 cannot be sufficiently provided relative to the power consumption of the industrial machine 2 (the ratio is less than 1.0), the result is determined to be "no," and a negative (-) feedback R is given. Furthermore, if the base speed of the peak-cut servo motor 50 can be set lower than a predetermined threshold, the result can be determined to be "good," and a positive (+) feedback R is given. Furthermore, if the base speed of the peak-cut servo motor 50 can be set higher than a predetermined threshold, the result can be determined to be "no," and a negative (-) feedback R is given. The value of the feedback R can also vary depending on the magnitude of the difference between the ratio and the threshold.

[0045] In Q-learning by the learning unit 110, a behavior value table, obtained by associating state variables S, decision data D, and rewards R with behavior values (e.g., numerical values) represented by a function Q, can be used as a learning model. In this case, learning by the learning unit 110 means that the learning unit 110 updates the behavior value table. At the beginning of Q-learning, the correlation between the current state of the environment and the base speed setting of the peak-shaving servo motor 50 is unknown. Therefore, the behavior value table is prepared in such a way that various state variables S, decision data D, and rewards R are associated with randomly determined behavior value values (function Q). Furthermore, as learning progresses, the behavior value values (function Q) are rewritten based on the state variables S, decision data D, and calculated rewards R, and the behavior value table is updated. By repeating this update, the behavior value values (function Q) displayed in the behavior value table are rewritten to larger values for more appropriate behaviors. Furthermore, when learning has progressed sufficiently, a more appropriate behavior for setting the base speed of the peak-shaving servo motor 50 can be selected based on the current state by referring to the behavior value table.

[0046] The learning unit 110 can be configured to use a neural network as the value function Q (learning model), taking state data S and behavior a as inputs to the neural network and outputting the value of behavior a in the state (result y). In such a configuration, a neural network having three layers: an input layer, an intermediate layer, and an output layer can be used as the learning model. As another method, a so-called deep learning method that utilizes a neural network with three or more layers can be used to perform more efficient learning and inference. The learning model generated by the learning unit 110 is stored in the learning model storage unit 130 provided in the non-volatile memory 104 and is used in the decision-making process of the decision unit 120 to determine the setting behavior of the base speed of the peak-shaving servo motor 50.

[0047] The learning unit 110 is essential during the learning phase, but it is no longer necessary after the learning unit 110 has learned the behavior for setting the base speed of the peak-shaving servo motor 50. For example, when shipping the machine learning device 100 to a customer after learning, the learning unit 110 may be removed for shipment.

[0048] The decision-making unit 120 is Figure 1 The control device 1 shown is provided with a processor 101 that executes a system program read from a ROM 102, and is implemented primarily by the processor 101 performing arithmetic processing using a RAM 103 and a nonvolatile memory 104. The decision unit 120 uses the learning model stored in the learning model storage unit 130 based on the state data S input from the preprocessing unit 36 to determine a more appropriate solution for the setting behavior of the base speed of the peak-cutting servo motor 50, and outputs the determined setting behavior of the base speed of the peak-cutting servo motor 50. In the present embodiment, the decision unit 120 inputs the state data S (such as the drive motor data S1) and the base speed setting behavior of the peak-shaving servo motor 50 from the preprocessing unit 36 as input data to a learning model (with parameters determined) generated by reinforcement learning in the learning unit 110. This model calculates the reward for taking that behavior in the current state. This reward is calculated for each of the currently available base speed setting behaviors for the peak-shaving servo motor 50. The calculated rewards are compared, and the base speed setting behavior for the peak-shaving servo motor 50 that yields the maximum reward is determined as the most appropriate solution. The most appropriate solution for the base speed setting behavior of the peak-shaving servo motor 50 determined by the decision unit 120 is output to the control unit 32 for use in setting the base speed. Furthermore, the solution may be displayed on the display device 70 or transmitted to a host computer or cloud computer via a wired or wireless network (not shown).

[0049] In the control device 1 having the above-described structure, the base speed setting of the peak-shaving servo motor 50 is changed while monitoring the operating state of the industrial machine 2. Consequently, when the power consumption of each industrial machine increases, regenerative power can be appropriately generated by decelerating the peak-shaving servo motor 50. Furthermore, if the future operating state of the industrial machine 2 requires less regenerative power, the base speed of the peak-shaving servo motor 50 can be dynamically set to a lower level. Consequently, compared to conventional methods of fixedly setting the base speed, power loss in the peak-shaving servo motor 50 can be reduced, and the life of the peak-shaving servo motor 50 can be extended.

[0050] Next, a control system according to a second embodiment will be described, in which the control device 1 is installed as a computer such as a fog computer and a cloud server.

[0051] Figure 4The environment in which the control system of the second embodiment operates is shown in FIG. Figure 4 As shown, the control system of this embodiment operates in a system environment in which a plurality of devices including a cloud server 6, a fog computer 7, and an edge computer 8 are connected to a wired / wireless network. Figure 4 The system shown is configured to be logically divided into three layers, namely, a layer including a cloud server 6, etc., a layer including a fog computer 7, etc., and a layer including an edge computer 8 (a robot controller that controls the robot included in the cell 9, a control device that controls a machine tool, peripheral machinery such as a handling machine, a power supply device, etc.). In such a system, the control system of this embodiment is configured by installing the functions of the controller 1 described in the first embodiment on computers such as a cloud server 6 and a fog computer 7. In the control system of the second embodiment, data is shared with each of a plurality of devices via a network, or various data obtained by the edge computer 8 are collected to the fog computer 7 and the cloud server 6 for large-scale analysis. In addition, the control system can control the action of each edge computer 8 based on its analysis results. Figure 4 In the system shown, multiple units 9 are installed in factories across different locations (for example, one unit 9 is installed on each floor of the factory), and each unit 9 is managed by a higher-level fog computer 7 in predetermined units (factory units, multiple factories of the same manufacturer, etc.). Furthermore, the data collected and analyzed by the fog computer 7 can be collected and analyzed in a higher-level cloud server 6, and the resulting information can be flexibly used for control purposes in each edge computer 8.

[0052] Figure 5 This is a schematic diagram of the control system of this embodiment. The control system 300 of this embodiment includes a control device 1' installed on a computer such as a cloud server 6 or a fog computer 7, a plurality of industrial machines 2 as edge computers connected to the control device 1' via a network 5, and a motor drive device 3 for controlling the operation of a servo motor 50 for peak shaving. The control device 1' of the control system 300 of this embodiment has the same functions as the control unit 32 except that it does not have a control unit 32. Figure 2 The control device 1 described above has the same structure.

[0053] The motor drive device 3 that controls the operation of the peak-shaving servo motor 50 can control the operation of the servo motor and can be a general motor drive device capable of exchanging data with other devices via a network. The motor drive device 3 performs known peak-shaving operation control on the peak-shaving servo motor 50 based on the power consumption status of the industrial machine 2 obtained via the network. Furthermore, when a base speed setting is instructed by the control device 1' via the network 5, the motor drive device 3 changes the base speed during the peak-shaving operation.

[0054] In the control system 300 of this embodiment, the control device 1' sets the base speed for the motor drive device 3. The machine learning device 100 included in the control device 1' learns how to set the base speed of the peak-shaving servo motor 50 for appropriate peak-shaving operation, depending on the operating state of the managed industrial machine 2. Furthermore, the machine learning device 100 can output and set a more appropriate base speed setting solution based on the operating state of the industrial machine 2.

[0055] Figure 6 Indicates that Figure 4 The illustrated system environment introduces multiple examples of control systems 300 . Figure 6 The illustrated system includes multiple control systems 300 (not shown) each equipped with a control device 1' mounted on a fog computer 7. Each control device 1' manages multiple edge computers 8 and learns based on data collected from the managed industrial machines (edge computers 8). Furthermore, each control device 1' is configured to exchange learning models, the results of each learning process, with other control devices 1' either directly or via a cloud server 6.

[0056] In a control system 300 having such a configuration, for example, a learning model can be obtained and used from another control system 300 equipped with similar industrial machinery. Therefore, in the case of a new factory, for example, a learning model can be transferred (or reused) from a control system 300 operating with a similar combination of industrial machinery. This can significantly reduce the time required for trial operation to construct the learning model when establishing the factory.

[0057] As mentioned above, although embodiment of this invention was described, this invention is not limited to the example of the said embodiment, It is possible to implement it in various forms by making appropriate changes.

[0058] For example, in the above embodiment, the control device 1 and the machine learning device 100 are described as devices having different CPUs (processors), but the machine learning device 100 can be implemented by the CPU 11 of the control device 1 and the system program stored in the ROM 12.

Claims

1. A control device for controlling a peak-cutting operation of a peak-cutting electric motor connected to the same power supply path as at least one industrial machine, characterized in that: The control device has: a data acquisition unit configured to acquire at least data related to an operating state of the industrial machine; a learning model storage unit storing a learning model in which a value of a setting behavior of a base speed of the peak-cutting electric motor is associated with an operating state of the industrial machine; and a decision unit that uses the learning model stored in the learning model storage unit to determine a setting behavior for the base speed of the peak-cutting motor based on the data related to the operating state of the industrial machine acquired by the data acquisition unit; For at least one industrial machine, when power consumption in the at least one industrial machine increases, the peak-cutting electric motor is decelerated corresponding to the operating time, thereby generating regenerative power. When regenerative power is generated in the industrial machine after deceleration, the peak-cutting electric motor is accelerated to the base speed.

2. The control device according to claim 1, characterized in that The data acquisition unit further acquires data related to the base speed of the peak shaving motor. The control device includes a learning unit that generates a learning model in which a setting behavior of a base speed of the peak-cutting electric motor is associated with an operating state of the industrial machine.

3. The control device according to claim 2, characterized in that The learning unit provides a positive feedback when sufficient regenerative power can be supplied from the peak shaving motor relative to the power consumed by the industrial machine, or when the base speed of the peak shaving motor can be set low. The learning unit provides a negative feedback when sufficient regenerative power cannot be supplied from the peak shaving motor relative to the power consumed by the industrial machine, or when the base speed of the peak shaving motor can be set high. The learning unit generates the learning model based on the value of the reward.

4. The control device according to any one of claims 1 to 3, characterized in that: The learning model is a behavior value table that stores the value of the setting behavior of the base speed of the peak-cutting electric motor in association with the operating state of the industrial machine.

5. The control device according to any one of claims 1 to 3, characterized in that: The above learning model is a neural network formed by a multi-layer structure.

6. A control system comprising a plurality of control devices according to claim 2 connected to each other via a network, characterized in that: The learning result of the learning unit can be shared among the plurality of control devices.

7. A machine learning device for learning a base speed setting behavior of a peak-cutting motor in controlling a peak-cutting operation of the peak-cutting motor, wherein the peak-cutting motor is connected to the same power supply path as at least one industrial machine, characterized in that: The machine learning device has: a learning model storage unit storing a learning model in which a value of a setting behavior of a base speed of the peak-cutting electric motor is associated with an operating state of the industrial machine; and a decision unit that uses the learning model stored in the learning model storage unit to determine a setting behavior of the base speed of the peak-cutting motor based on data related to the operating state of the industrial machine; For at least one industrial machine, when power consumption in the at least one industrial machine increases, the peak-cutting electric motor is decelerated corresponding to the operating time, thereby generating regenerative power. When regenerative power is generated in the industrial machine after deceleration, the peak-cutting electric motor is accelerated to the base speed.

8. The machine learning device according to claim 7, wherein: The machine learning device further includes a learning unit that generates a learning model in which a setting behavior of a base speed of the peak-cutting electric motor is associated with an operating state of the industrial machine.

Images