Work machinery and control devices

Abstract

This disclosure provides a work machine capable of predicting the precise behavior of each part of an attachment affected by nonlinear elements. [Solution] The shovel 100 is a work machine comprising an upper rotating body as the main body, an attachment provided on the upper rotating body, and a controller 30 as a control device. The controller 30 stores a dynamics model 301 of the attachment and a learning model 302 that outputs the nonlinear elements of the dynamics model 301. The controller 30 predicts the behavior of the attachment using the dynamics model 301, which includes the nonlinear elements output by the learning model 302.

Description

Technical Field

[0001] The present disclosure relates to a work machine and a control device.

Background Art

[0002] Conventionally, an attitude control system for various construction machines, particularly construction machines such as cranes that have a high center of gravity and are at risk of tipping over, is known (see Patent Document 1 below). The attitude control system described in Patent Document 1 includes sensors that detect the attitude, movement, and working load of the main body.

[0003] Based on the detection values of the sensors, the attitude control system constructs a model representing the current and future mechanical behavior of the construction machine main body with respect to the attitude while referring to a database, and determines whether the main body will tip over. Further, when tipping over is predicted, the attitude control system stops the ongoing work operation and starts an operation to avoid tipping over, thereby preventing tipping over. Also, when tipping over is predicted, the system notifies the operator to that effect.

[0004] According to the attitude control system, based on the values detected by the sensors and referring to a database, it is possible to predict tipping over of the main body in advance and prevent tipping over.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] The conventional attitude control system described in Patent Document 1 above stores knowledge and parameters regarding the mechanical behavior of the construction machine body, consisting of the posture, movement, direction, magnitude, and application position of the work load, and combinations of their temporal changes, in the above-mentioned database. However, since the behavior of the work machine attachment is affected by nonlinear elements such as friction, it is difficult to accurately predict the behavior even by referring to the above-mentioned database based on values ​​detected by sensors. Similarly, since the work machine attachment is affected by nonlinear elements such as friction, it is difficult to improve the calculation accuracy of the load applied to the attachment.

[0007] This disclosure provides a work machine and control device capable of predicting the precise behavior of each part of an attachment affected by nonlinear elements. Furthermore, this disclosure provides a work machine and control device capable of improving the calculation accuracy of loads applied to an attachment affected by nonlinear elements. [Means for solving the problem]

[0008] One aspect of the present disclosure provides a work machine comprising a main body, an attachment provided on the main body, and a control device that stores a dynamic model of the attachment and a learning model that outputs nonlinear elements of the dynamic model, wherein the control device predicts the behavior of the attachment using the dynamic model including the nonlinear elements output by the learning model.

[0009] Another aspect of the present disclosure provides a work machine comprising a main body, an attachment provided on the main body, and a control device storing a dynamic model of the attachment and a learning model that outputs nonlinear elements of the dynamic model, wherein the control device calculates the load applied to the attachment using the dynamic model including the nonlinear elements output by the learning model.

[0010] Another aspect of the present disclosure provides a control device for predicting the behavior of an attachment of a work machine comprising a main body and an attachment provided on the main body, wherein the control device stores a dynamic model of the attachment and a learning model that outputs nonlinear elements of the dynamic model, and predicts the behavior of the attachment using the dynamic model including the nonlinear elements output by the learning model.

[0011] Another aspect of the present disclosure provides a control device for calculating a load applied to an attachment of a work machine comprising a main body and an attachment provided on the main body, wherein the control device stores a dynamic model of the attachment and a learning model that outputs nonlinear elements of the dynamic model, and calculates the load applied to the attachment using the dynamic model including the nonlinear elements output by the learning model. [Effects of the Invention]

[0012] According to each of the above embodiments of this disclosure, it is possible to provide a work machine and control device that can predict the precise behavior of each part of an attachment affected by nonlinear elements, and a work machine and control device that can improve the calculation accuracy of the load applied to the attachment. [Brief explanation of the drawing]

[0013] [Figure 1] This is a side view showing a shovel, which is an embodiment of the work machine related to this disclosure. [Figure 2] Figure 1 is a block diagram showing an example of the configuration of an excavator. [Figure 3] Figure 1 is a diagram showing the configuration of the hydraulic system installed in the excavator. [Figure 4] Figure 1 shows an example of a dynamics model stored in the control device of the excavator. [Figure 5] Figure 1 shows an example of an NN model stored in the control device of the excavator. [Figure 6] Figure 1 is a functional block diagram showing the processing flow by the control device of the excavator. [Figure 7]Figure 1 is a functional block diagram showing the processing flow by the control device of the excavator. [Figure 8] This is a schematic diagram showing an embodiment of the control device relating to this disclosure. [Modes for carrying out the invention]

[0014] Embodiments of this disclosure will be described below with reference to the drawings. The embodiments described below are illustrative and do not limit the invention. Not all features and combinations thereof in the embodiments of this disclosure are necessarily essential to the invention. In each drawing, the same or corresponding components are denoted by the same or corresponding reference numerals, and redundant descriptions may be omitted.

[0015] First, with reference to Figures 1 to 3, an excavator 100, which is an embodiment of the work machine according to this disclosure, will be described. Figure 1 is a side view showing the excavator 100 according to this embodiment. As shown in Figure 1, the excavator 100 comprises a lower traveling body 1, an upper rotating body 3, an attachment AT, and a controller 30.

[0016] The lower travel unit 1 includes, for example, a crawler that is driven by a travel hydraulic motor 2M to move the shovel 100. Specifically, the lower travel unit 1 includes a left crawler driven by a left travel hydraulic motor 2ML shown in Figure 2, and a right crawler driven by a right travel hydraulic motor 2MR shown in Figure 2. The shovel 100 may be a wheeled hydraulic shovel with tires on the lower travel unit 1.

[0017] The upper revolving body 3 is the main body of the excavator 100 according to the present embodiment. The upper revolving body 3 is provided rotatably on the lower traveling body 1. Specifically, the upper revolving body 3 is rotatably attached onto the lower traveling body 1 via a slewing mechanism 2. The slewing mechanism 2 is driven by a slewing hydraulic motor 2A shown in FIG. 2 to slew the upper revolving body 3 on the lower traveling body 1. The slewing hydraulic motor 2A is a slewing actuator mounted on the upper revolving body 3 to drive the slewing mechanism 2. Note that the slewing actuator may be an electric actuator.

[0018] On the front left side of the upper revolving body 3, a cab 10 as an operator's cab of the excavator 100 is provided. Inside the cab 10, in addition to a controller 30, an operating device 26 shown in FIG. 2 and the like are provided. Further, on the upper revolving body 3, in addition to the slewing hydraulic motor 2A, a power source such as an engine 11 is mounted. Further, an external recognition device 70 is attached to the upper revolving body 3.

[0019] The external recognition device 70 is configured to recognize objects around the excavator 100. The external recognition device 70 may be configured to calculate the distance to the recognized objects. The external recognition device 70 includes, for example, an imaging device, LiDAR, millimeter wave radar, ultrasonic sensor, infrared sensor, or any combination thereof. The external recognition device 70 includes, for example, a front recognition device 70F, a rear recognition device 70B, a left recognition device 70L, and a right recognition device (not shown).

[0020] The forward recognition device 70F is mounted, for example, on the upper front of the cabin 10 to recognize objects in front of the shovel 100. The rear recognition device 70B is mounted on the rear of the upper rotating body 3 to recognize objects behind the shovel 100. The left recognition device 70L is mounted on the left side of the upper rotating body 3 to recognize objects to the left of the shovel 100. The right recognition device is mounted on the right side of the upper rotating body 3 to recognize objects to the right of the shovel 100. These external environment recognition devices 70 are arranged so that the object detection ranges of adjacent external environment recognition devices 70 partially overlap, and are configured to recognize objects present in a 360-degree radius around the shovel 100.

[0021] Note that the forward, backward, left, and right directions of the shovel 100 are the directions as viewed from the operator sitting in the cabin 10. Figure 1 shows a three-dimensional Cartesian coordinate system in which the forward, backward, left, right, up, and down directions of the shovel 100 are defined as the positive X-axis, negative X-axis, positive Y-axis, negative Y-axis, positive Z-axis, and negative Z-axis, respectively.

[0022] The attachment AT is located in the front center of the upper slewing body 3, adjacent to the right side of the cabin 10. The attachment AT includes a boom 4, an arm 5, and a bucket 6. The boom 4 is rotatably supported on the upper slewing body 3 via a boom foot pin. The arm 5 is rotatably supported on the boom 4 via a boom top pin. The bucket 6 is rotatably supported on the arm 5 via a bucket pin.

[0023] The boom 4 rotates up and down, driven by the boom cylinder 7. Specifically, the boom cylinder 7 performs a boom raising operation by extending the piston rod to rotate the boom 4 upward, and a boom lowering operation by retracting the piston rod to rotate the boom 4 downward.

[0024] The arm 5 is driven by the arm cylinder 8 to open and close relative to the boom 4. Specifically, the arm cylinder 8 extends the piston rod to perform an arm closing operation, closing the arm 5 relative to the boom 4, and retracts the piston rod to perform an arm opening operation, opening the arm 5 relative to the boom 4.

[0025] The bucket 6 is driven by a bucket cylinder 9 connected via a linkage mechanism to open and close relative to the arm 5. Specifically, the bucket cylinder 9 performs a bucket closing operation by extending its piston rod to close the bucket 6 relative to the arm 5, and a bucket opening operation by retracting its piston rod to open the bucket 6 relative to the arm 5.

[0026] Bucket 6 is an end attachment for excavation that is attached to the tip of attachment AT. The end attachment is not limited to bucket 6, and may be other types of buckets, such as a large bucket, a slope bucket, or a dredging bucket. Bucket 6 may also be provided with a bucket tilt mechanism.

[0027] The controller 30 (an example of a control device) is installed, for example, inside the cabin 10 and controls the drive of the shovel 100. The functions of the controller 30 may be realized by any hardware, software, or a combination thereof. For example, the controller 30 is mainly composed of a microcomputer including a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), a non-volatile auxiliary storage medium, and various input / output interfaces. The controller 30 realizes various functions by executing various programs stored in the ROM or non-volatile auxiliary storage medium with the CPU.

[0028] Figure 2 is a block diagram showing an example of the configuration of excavator 100 in Figure 1. In Figure 2, double lines indicate the transmission of mechanical power, and solid lines indicate the high-pressure hydraulic path. Dashed lines indicate the transmission path of pilot pressure, and dotted lines indicate the transmission paths of electrical signals and control signals.

[0029] Excavator 100 is equipped with a hydraulic drive system that drives hydraulic actuators including a boom cylinder 7, an arm cylinder 8, a bucket cylinder 9, a slewing hydraulic motor 2A, a left travel hydraulic motor 2ML, and a right travel hydraulic motor 2MR. The hydraulic drive system of excavator 100 includes, for example, an engine 11, a regulator 13, a main pump 14, a pilot pump 15, and a control valve unit 17.

[0030] The engine 11 is the main power source in the hydraulic drive system and is mounted, for example, at the rear of the upper slewing body 3. Specifically, the engine 11 rotates at a constant speed at a preset target rotational speed under direct or indirect control by the controller 30 (described later) and drives the main pump 14 and the pilot pump 15. The power source of the shovel 100 may be a combination of a power source such as a battery or fuel cell and an electric motor. The engine 11 is, for example, a diesel engine that uses light oil as fuel. The engine 11 may also be a gasoline engine or a hydrogen engine, etc.

[0031] The regulator 13 controls the discharge rate of the main pump 14. For example, the regulator 13 adjusts the angle (tilt angle) of the swash plate of the main pump 14 in response to a control command from the controller 30.

[0032] The main pump 14 (an example of a hydraulic pump) is mounted at the rear of the upper slewing body 3, similar to the engine 11, and supplies hydraulic fluid to the control valve unit 17 through the high-pressure hydraulic line 16. The main pump 14 is driven by the engine 11, as described above. The main pump 14 is, for example, a variable displacement hydraulic pump. As described above, under the control of the controller 30, the piston stroke length of the main pump 14 can be adjusted by adjusting the tilt angle of the swash plate by the regulator 13, thereby controlling the discharge flow rate (discharge pressure).

[0033] The pilot pump 15 is an example of a pilot pressure generating device and is configured to supply hydraulic fluid to hydraulic control equipment via a pilot line. In this embodiment, the pilot pump 15 is a fixed-displacement hydraulic pump. The pilot pressure generating device may also be implemented by the main pump 14. That is, the main pump 14 may have the function of supplying hydraulic fluid to the control valve unit 17 via a hydraulic fluid line, as well as the function of supplying hydraulic fluid to various hydraulic control equipment via a pilot line. In this case, the pilot pump 15 may be omitted.

[0034] The control valve unit 17 is a hydraulic control device that controls the hydraulic system in the excavator 100. In this embodiment, the control valve unit 17 includes control valves 171 to 176. The control valve unit 17 is configured to selectively supply the hydraulic fluid discharged by the main pump 14 to one or more hydraulic actuators through the control valves 171 to 176.

[0035] Control valves 171 to 176 control, for example, the flow rate of hydraulic fluid from the main pump 14 to the hydraulic actuator, and the flow rate of hydraulic fluid from the hydraulic actuator to the hydraulic fluid tank. More specifically, control valve 171 corresponds to the left travel hydraulic motor 2ML, control valve 172 to the right travel hydraulic motor 2MR, and control valve 173 to the swing hydraulic motor 2A. In addition, control valve 174 corresponds to the bucket cylinder 9, control valve 175 to the boom cylinder 7, and control valve 176 to the arm cylinder 8.

[0036] The operating system of the shovel 100 according to this embodiment includes, for example, an operating device 26, a discharge pressure sensor 28, an operating sensor 29, and a controller 30.

[0037] The operating device 26 is a device used by the operator to operate the actuator. The operating device 26 includes, for example, an operating lever and an operating pedal. The actuator includes at least one of a hydraulic actuator and an electric actuator.

[0038] The discharge pressure sensor 28 is configured to detect the discharge pressure of the main pump 14. In this embodiment, the discharge pressure sensor 28 outputs the detected value to the controller 30.

[0039] The operation sensor 29 is configured to detect the operator's actions using the operation device 26. In this embodiment, the operation sensor 29 detects the operating direction and amount of operation of the operation device 26 corresponding to each actuator, and outputs the detected values ​​to the controller 30.

[0040] The controller 30 controls the opening area of ​​the proportional valve 31 according to the output of the operation sensor 29. The controller 30 then supplies the hydraulic fluid discharged by the pilot pump 15 to the pilot port of the corresponding control valve in the control valve unit 17. The pressure of the hydraulic fluid supplied to each pilot port (pilot pressure) is, in principle, the pressure corresponding to the operating direction and amount of the operating device 26 corresponding to each hydraulic actuator. In this way, the operating device 26 is configured to supply the hydraulic fluid discharged by the pilot pump 15 to the pilot port of the corresponding control valve in the control valve unit 17.

[0041] The control system of the excavator 100 according to this embodiment includes a controller 30, a display device D1, an input device D2, and a communication device T1. The control system of the excavator 100 also includes, as a configuration related to the semi-automatic operation function, a proportional valve 31, a boom angle sensor S1, an arm angle sensor S2, a bucket angle sensor S3, a machine body tilt sensor S4, a slewing angle sensor S5, an imaging device S6, and a positioning device PS.

[0042] The controller 30 sets a target rotational speed based on the operator's actions, for example, and performs drive control to keep the engine 11 rotating at a constant speed. The controller 30 also outputs control commands to the regulator 13 as needed to change the discharge amount of the main pump 14. The controller 30 controls the regulator 13 based on the detected pilot pressure values ​​corresponding to the operating states of various operating elements (i.e., various hydraulic actuators) in the operating device 26, which are input from the operation sensor 29, and adjusts the discharge amount of the main pump 14. The controller 30 also performs control related to a machine guidance function that guides the operator's manual operation of the shovel 100 through the operating device 26. The controller 30 also performs control related to a machine control function that automatically assists the operator's manual operation of the shovel 100 through the operating device 26.

[0043] Furthermore, the controller 30 includes, for example, a dynamics model 301, a learning model 302, and a behavior prediction unit 303. The controller 30 may also include an attachment control unit 304, a payload calculation unit 305, and a learning unit 306. Each of these parts of the controller 30 represents a function of the controller 30, which is realized by executing various programs stored in ROM or a non-volatile auxiliary storage medium that constitutes the controller 30 on the CPU. Furthermore, each of the above parts of the controller 30 may be realized by, for example, any hardware, software, or a combination thereof.

[0044] The display device D1 is installed in a location easily visible to a seated operator inside the cabin 10 and displays various information images under the control of the controller 30. The display device D1 may be connected to the controller 30 via an in-vehicle communication network such as CAN (Controller Area Network), or it may be connected to the controller 30 via a one-to-one dedicated line. Furthermore, the display device D1 is not limited to a device pre-installed in the cabin 10, but may be a separately installed monitor. Moreover, the display device D1 can be any device capable of displaying information; for example, a tablet terminal capable of communicating with the communication device T1 may be used.

[0045] The input device D2 is located within reach of a seated operator in the cabin 10 and receives various operation inputs from the operator, outputting signals corresponding to the operation inputs to the controller 30. The input device D2 includes a touch panel mounted on the display of the display device D1 which displays various information images, a knob switch located at the tip of the lever device of the operation device 26, and button switches, levers, toggles, rotary dials, etc., installed around the display device D1. Signals corresponding to the operations performed on the input device D2 are received by the controller 30.

[0046] The communication device T1 communicates with external devices through a predetermined network, including a mobile communication network with a base station as its endpoint, a satellite communication network, and the Internet network. The communication device T1 is, for example, a mobile communication module that supports mobile communication standards such as LTE (Long Term Evolution), 4G (4th Generation), and 5G (5th Generation), or a satellite communication module for connecting to a satellite communication network.

[0047] The proportional valve 31 functions as a control valve for machine control. The proportional valve 31 is located in the pipeline connecting the pilot pump 15 and the pilot port of the control valve in the control valve unit 17, and is configured to change the flow area of ​​the pipeline. In this embodiment, the proportional valve 31 operates in response to control commands output by the controller 30. Therefore, the controller 30 can supply the hydraulic fluid discharged by the pilot pump 15 to the pilot port of the control valve in the control valve unit 17 via the proportional valve 31, independently of the operator's operation of the operating device 26.

[0048] This configuration allows the controller 30 to operate the hydraulic actuator corresponding to a specific operating device 26 even when no operation is being performed on that specific operating device 26. Furthermore, if the excavator 100 does not have machine control or remote control functions, the excavator 100 does not need to have a proportional valve 31.

[0049] The boom angle sensor S1 detects the rotation angle of the boom 4. In this embodiment, the boom angle sensor S1 is an acceleration sensor attached to the boom 4 and can detect the boom angle, which is the rotation angle of the boom 4 relative to the upper slewing body 3. The boom angle is smallest when the boom 4 is lowered to its lowest position, and increases as the boom 4 is raised. The detection signal corresponding to the boom angle from the boom angle sensor S1 is input to the controller 30.

[0050] The arm angle sensor S2 detects the rotation angle of the arm 5. In this embodiment, the arm angle sensor S2 is an acceleration sensor attached to the arm 5 and can detect the arm angle, which is the rotation angle of the arm 5 relative to the boom 4. The arm angle is smallest when the arm 5 is closed to its shortest extent, and increases as the arm 5 is opened. The detection signal corresponding to the arm angle from the arm angle sensor S2 is input to the controller 30.

[0051] The bucket angle sensor S3 detects the rotation angle of the bucket 6. In this embodiment, the bucket angle sensor S3 is an acceleration sensor attached to the link mechanism that drives the bucket 6, and can detect the bucket angle, which is the rotation angle of the bucket 6 relative to the arm 5. The bucket angle is smallest when the bucket 6 is closed to its fullest extent, and increases as the bucket 6 is opened. The detection signal corresponding to the bucket angle from the bucket angle sensor S3 is input to the controller 30.

[0052] The boom angle sensor S1, arm angle sensor S2, and bucket angle sensor S3 may be a potentiometer using a variable resistor, a stroke sensor that detects the stroke amount of the corresponding hydraulic cylinder, or a rotary encoder that detects the rotation angle around the connecting pin. The boom angle sensor S1, arm angle sensor S2, and bucket angle sensor S3 constitute an attitude sensor that detects the attitude of the excavation attachment.

[0053] The machine body tilt sensor S4 is attached to the upper rotating body 3 and is configured to detect the tilt of the upper rotating body 3 with respect to a predetermined plane. In this embodiment, the machine body tilt sensor S4 is an acceleration sensor that detects the tilt angle of the upper rotating body 3 around the longitudinal axis and the tilt angle around the left-right axis with respect to the horizontal plane. The longitudinal axis and left-right axis of the upper rotating body 3 are, for example, orthogonal to each other and pass through the shovel center point, which is a point on the rotation axis of the shovel 100.

[0054] The rotation angle sensor S5 is attached to the upper rotating body 3 and is configured to detect the rotational angular velocity of the upper rotating body 3. In this embodiment, the rotation angle sensor S5 is a gyro sensor. The rotation angle sensor S5 may also be a resolver or a rotary encoder, etc. The rotation angle sensor S5 may also detect the rotational speed. The rotational speed may be calculated from the rotational angular velocity.

[0055] Furthermore, if the aircraft tilt sensor S4 includes a gyro sensor, a 6-axis sensor, an IMU, etc., capable of detecting angular velocity around three axes, the rotation state of the upper rotating body 3 (for example, rotational angular velocity) may be detected based on the detection signal from the aircraft tilt sensor S4. In this case, the rotational angle sensor S5 may be omitted.

[0056] The imaging device S6 is an example of the external environment recognition device 70 shown in Figure 1. The imaging device S6 recognizes objects around the shovel 100 by capturing images of objects present around the shovel 100. The imaging device S6 includes, for example, a front camera as a front recognition device 70F, a rear camera as a rear recognition device 70B, a left camera as a left recognition device 70L, and a right camera as a right recognition device.

[0057] The positioning device PS is configured to acquire information regarding the position of the shovel 100. In this embodiment, the positioning device PS is configured to measure the position and orientation of the shovel 100. Specifically, the positioning device PS is a GNSS receiver incorporating an electronic compass, which measures the latitude, longitude, and altitude of the current position of the shovel 100, as well as the orientation of the shovel 100.

[0058] The shovel 100 operates actuators (e.g., hydraulic actuators) in response to the operation of the operator sitting in the cabin 10, driving the moving elements (hereinafter referred to as "driven elements") such as the lower traveling body 1, the upper slewing body 3, the boom 4, the arm 5, and the bucket 6.

[0059] Furthermore, instead of being configured to be operable by the operator in the cabin 10, or in addition to being configured to be operable by the operator in the cabin 10, the shovel 100 may also be configured to be remotely operated from outside the shovel 100. When the shovel 100 is remotely operated, the inside of the cabin 10 may be unoccupied.

[0060] Furthermore, the shovel 100 may automatically operate its actuators regardless of the operator's actions. As a result, the controller 30 of the shovel 100 has the function of automatically operating at least some of the multiple actuators that operate each of the driven elements such as the lower traveling body 1, the upper slewing body 3, the boom 4, the arm 5, and the bucket 6, that is, a so-called "automatic driving function" or "machine control function".

[0061] The automatic driving function may include a function that automatically operates driven elements (actuators) other than the target driven element (actuator) in response to the operator's operation of the control device 26 or remote control, i.e., a so-called "semi-automatic driving function" or "operation-assist type machine control function". The automatic driving function may also include a function that automatically operates at least some of the multiple driven elements (hydraulic actuators) on the premise that there is no operation of the operator's control device 26 or remote control, i.e., a so-called "fully automatic driving function" or "fully automatic machine control function". In the case of the excavator 100, when the fully automatic driving function is enabled, the interior of the cabin 10 may be unoccupied. Furthermore, the semi-automatic driving function and fully automatic driving function may include a mode in which the operation content of the driven elements (actuators) that are the target of automatic driving is automatically determined according to predetermined rules. Furthermore, semi-autonomous driving functions and fully autonomous driving functions may include a mode in which the shovel 100 autonomously makes various decisions, and the operation of the driven elements (hydraulic actuators) that are subject to autonomous driving is determined autonomously in accordance with the results of those decisions (so-called "autonomous driving function").

[0062] Specifically, when the arm 5 is operated by the operator via the operating device 26, the controller 30 may automatically operate at least one of the boom 4 and the bucket 6 so that the tip position of the bucket 6 coincides with a predetermined target construction surface. In addition, the controller 30 may also automatically operate the arm 5 regardless of the operating state of the operating device 26 that operates the arm 5. In other words, the controller 30 may trigger the operation of the operating device 26 by the operator to perform predetermined operations on the attachment. Hereinafter, the function of the controller 30 that operates not only the arm 5 but also at least one of the boom 4 and the bucket 6 in response to the operation of the operating device 26 corresponding to the arm 5 will be referred to as the "semi-automatic operation function". The semi-automatic operation function may be executed, for example, by operating a predetermined switch (hereinafter referred to as the "MC (Machine Control) switch") located at the tip of any of the lever devices included in the operating device 26. In this embodiment, a paddle switch may be used as the MC switch, in which the machine control function is executed while it is pressed.

[0063] Furthermore, some of the functions of controller 30 may be implemented by other controllers (control devices). In other words, the functions of controller 30 may be implemented in a manner distributed among multiple controllers. For example, machine guidance functions and machine control functions may be implemented by dedicated controllers (control devices).

[0064] Next, with reference to Figure 3, an example of the configuration of the hydraulic system mounted on the excavator 100 according to this embodiment will be described. Figure 3 is a diagram showing an example of the configuration of the hydraulic system mounted on the excavator 100 according to this embodiment. In Figure 3, the mechanical power transmission system, hydraulic fluid line, pilot line, and electrical control system are shown by double lines, solid lines, dashed lines, and dotted lines, respectively.

[0065] The hydraulic system of the Shovel 100 mainly includes an engine 11, a regulator 13, a main pump 14, a pilot pump 15, a control valve unit 17, an operating device 26, a discharge pressure sensor 28, an operating sensor 29, and a controller 30, etc.

[0066] In Figure 3, the hydraulic system is configured to circulate hydraulic fluid from the main pump 14, driven by the engine 11, to the hydraulic fluid tank via the center bypass pipeline 40 or the parallel pipeline 42.

[0067] The engine 11 is the power source for the shovel 100. In this embodiment, the engine 11 is, for example, a diesel engine that operates to maintain a predetermined rotational speed. The output shaft of the engine 11 is connected to the input shafts of the main pump 14 and the pilot pump 15, respectively.

[0068] The main pump 14 is configured to supply hydraulic fluid to the control valve unit 17 via a hydraulic fluid line. In this embodiment, the main pump 14 is a swashplate type variable displacement hydraulic pump.

[0069] The regulator 13 is configured to control the discharge rate of the main pump 14. In this embodiment, the regulator 13 controls the discharge rate of the main pump 14 by adjusting the swash plate tilt angle of the main pump 14 in response to a control command from the controller 30.

[0070] As described above, the pilot pump 15 is configured to supply hydraulic fluid to the hydraulic control equipment via the pilot line.

[0071] As described above, the control valve unit 17 includes control valves 171 to 176. Control valve 175 includes control valves 175L and 175R, and control valve 176 includes control valves 176L and 176R. As described above, the control valve unit 17 is configured to selectively supply the hydraulic fluid discharged by the main pump 14 to one or more hydraulic actuators through control valves 171 to 176. As described above, control valves 171 to 176 control the flow rate of hydraulic fluid flowing from the main pump 14 to the hydraulic actuators, and the flow rate of hydraulic fluid flowing from the hydraulic actuators to the hydraulic fluid tank. As described above, the hydraulic actuators include a boom cylinder 7, an arm cylinder 8, a bucket cylinder 9, a left travel hydraulic motor 2ML, a right travel hydraulic motor 2MR, and a slewing hydraulic motor 2A.

[0072] The operating device 26 is configured to allow an operator to operate the actuator. In this embodiment, the operating device 26 includes a hydraulic actuator operating device configured to allow an operator to operate a hydraulic actuator. Specifically, the hydraulic actuator operating device is configured to supply hydraulic fluid discharged by the pilot pump 15 to the pilot port of the corresponding control valve in the control valve unit 17 via a pilot line. The pressure of the hydraulic fluid supplied to each pilot port (pilot pressure) is a pressure corresponding to the operating direction and amount of the operating device 26 corresponding to each hydraulic actuator.

[0073] As described above, the discharge pressure sensor 28 detects the discharge pressure of the main pump 14 and outputs the detected value to the controller 30. As described above, the operation sensor 29 detects the operation of the operation device 26 by the operator, detects the operating direction and amount of the operation device 26 corresponding to each actuator, and outputs the detected value to the controller 30.

[0074] The main pump 14 includes a left main pump 14L and a right main pump 14R. The left main pump 14L circulates the hydraulic fluid to the hydraulic fluid tank via the left center bypass pipeline 40L or the left parallel pipeline 42L, while the right main pump 14R circulates the hydraulic fluid to the hydraulic fluid tank via the right center bypass pipeline 40R or the right parallel pipeline 42R.

[0075] The left center bypass pipeline 40L is a hydraulic fluid line that passes through control valves 171, 173, 175L, and 176L located within the control valve unit 17. The right center bypass pipeline 40R is a hydraulic fluid line that passes through control valves 172, 174, 175R, and 176R located within the control valve unit 17.

[0076] The control valve 171 is a spool valve that supplies the hydraulic fluid discharged by the left main pump 14L to the left travel hydraulic motor 2ML, and switches the flow of hydraulic fluid to discharge the hydraulic fluid discharged by the left travel hydraulic motor 2ML to the hydraulic fluid tank.

[0077] The control valve 172 is a spool valve that supplies the hydraulic fluid discharged by the right main pump 14R to the right travel hydraulic motor 2MR, and switches the flow of hydraulic fluid to discharge the hydraulic fluid discharged by the right travel hydraulic motor 2MR to the hydraulic fluid tank.

[0078] The control valve 173 is a spool valve that supplies the hydraulic fluid discharged by the left main pump 14L to the swivel hydraulic motor 2A, and switches the flow of hydraulic fluid to discharge the hydraulic fluid discharged by the swivel hydraulic motor 2A to the hydraulic fluid tank.

[0079] The control valve 174 is a spool valve that supplies the hydraulic fluid discharged by the right main pump 14R to the bucket cylinder 9 and switches the flow of the hydraulic fluid in order to discharge the hydraulic fluid in the bucket cylinder 9 to the hydraulic fluid tank.

[0080] Control valve 175L is a spool valve that switches the flow of hydraulic fluid to supply the hydraulic fluid discharged by the left main pump 14L to the boom cylinder 7. Control valve 175R is a spool valve that supplies the hydraulic fluid discharged by the right main pump 14R to the boom cylinder 7 and also switches the flow of hydraulic fluid to discharge the hydraulic fluid inside the boom cylinder 7 to the hydraulic fluid tank.

[0081] Control valve 176L is a spool valve that supplies the hydraulic fluid discharged by the left main pump 14L to the arm cylinder 8 and switches the flow of hydraulic fluid to discharge the hydraulic fluid in the arm cylinder 8 to the hydraulic fluid tank. Control valve 176R is a spool valve that supplies the hydraulic fluid discharged by the right main pump 14R to the arm cylinder 8 and switches the flow of hydraulic fluid to discharge the hydraulic fluid in the arm cylinder 8 to the hydraulic fluid tank.

[0082] The left parallel pipeline 42L is a hydraulic fluid line running parallel to the left center bypass pipeline 40L. The left parallel pipeline 42L can supply hydraulic fluid to a control valve further downstream if the flow of hydraulic fluid through the left center bypass pipeline 40L is restricted or blocked by any of the control valves 171, 173, and 175L. The right parallel pipeline 42R is a hydraulic fluid line running parallel to the right center bypass pipeline 40R. The right parallel pipeline 42R can supply hydraulic fluid to a control valve further downstream if the flow of hydraulic fluid through the right center bypass pipeline 40R is restricted or blocked by any of the control valves 172, 174, and 175R.

[0083] The regulator 13 includes a left regulator 13L and a right regulator 13R. The left regulator 13L controls the discharge volume of the left main pump 14L by adjusting the swash plate tilt angle of the left main pump 14L in accordance with the discharge pressure of the left main pump 14L. Specifically, the left regulator 13L reduces the discharge volume by adjusting the swash plate tilt angle of the left main pump 14L in accordance with an increase in the discharge pressure of the left main pump 14L. The same applies to the right regulator 13R. This is to ensure that the absorption power (absorption horsepower) of the main pump 14, which is expressed as the product of the discharge pressure and the discharge volume, does not exceed the output power (output horsepower) of the engine 11.

[0084] The operating device 26 includes a left operating lever 26L, a right operating lever 26R, and a travel lever 26D. The travel lever 26D includes a left travel lever 26DL and a right travel lever 26DR.

[0085] The left operating lever 26L is used for slewing and operating the arm 5. When the left operating lever 26L is operated in the forward / backward direction, it uses the hydraulic fluid discharged by the pilot pump 15 to introduce a control pressure corresponding to the lever operation amount into the pilot port of the control valve 176. When it is operated in the left / right direction, it uses the hydraulic fluid discharged by the pilot pump 15 to introduce a control pressure corresponding to the lever operation amount into the pilot port of the control valve 173.

[0086] Specifically, when the left operating lever 26L is operated in the arm closing direction, it introduces hydraulic fluid into the right pilot port of control valve 176L and into the left pilot port of control valve 176R. When the left operating lever 26L is operated in the arm opening direction, it introduces hydraulic fluid into the left pilot port of control valve 176L and into the right pilot port of control valve 176R. Furthermore, when the left operating lever 26L is operated in the left rotation direction, it introduces hydraulic fluid into the left pilot port of control valve 173, and when operated in the right rotation direction, it introduces hydraulic fluid into the right pilot port of control valve 173.

[0087] In the example shown in Figure 3, the left control lever 26L functions as an arm control lever when operated in the forward / backward direction and as a swivel control lever when operated in the left / right direction.

[0088] The right operating lever 26R is used to operate the boom 4 and the bucket 6. When the right operating lever 26R is operated in the forward / backward direction, it uses the hydraulic fluid discharged by the pilot pump 15 to introduce a control pressure corresponding to the lever operation amount into the pilot port of the control valve 175. When it is operated in the left / right direction, it uses the hydraulic fluid discharged by the pilot pump 15 to introduce a control pressure corresponding to the lever operation amount into the pilot port of the control valve 174.

[0089] Specifically, when the right operating lever 26R is operated in the boom lowering direction, it introduces hydraulic fluid into the left pilot port of the control valve 175R. When the right operating lever 26R is operated in the boom raising direction, it introduces hydraulic fluid into the right pilot port of the control valve 175L and into the left pilot port of the control valve 175R. Furthermore, when the right operating lever 26R is operated in the bucket closing direction, it introduces hydraulic fluid into the right pilot port of the control valve 174, and when it is operated in the bucket opening direction, it introduces hydraulic fluid into the left pilot port of the control valve 174.

[0090] In the example shown in Figure 3, the right operating lever 26R functions as a boom operating lever when operated in the forward / backward direction and as a bucket operating lever when operated in the left / right direction.

[0091] The travel lever 26D is used to operate the crawler. Specifically, the left travel lever 26DL is used to operate the left crawler. The left travel lever 26DL may be configured to be linked with the left travel pedal. When the left travel lever 26DL is operated in the forward or backward direction, it uses the hydraulic fluid discharged by the pilot pump 15 to introduce a control pressure corresponding to the lever operation amount into the pilot port of the control valve 171. The right travel lever 26DR is used to operate the right crawler. The right travel lever 26DR may be configured to be linked with the right travel pedal. When the right travel lever 26DR is operated in the forward or backward direction, it uses the hydraulic fluid discharged by the pilot pump 15 to introduce a control pressure corresponding to the lever operation amount into the pilot port of the control valve 172.

[0092] The discharge pressure sensor 28 includes discharge pressure sensors 28L and 28R. Discharge pressure sensor 28L detects the discharge pressure of the left main pump 14L and outputs the detected value to the controller 30. The same applies to discharge pressure sensor 28R.

[0093] The operation sensor 29 includes operation sensors 29LA, 29LB, 29RA, 29RB, 29DL, and 29DR. Operation sensor 29LA detects the operator's forward and backward movement of the left operation lever 26L and outputs the detected value to the controller 30. The operation details include, for example, the direction of lever operation and the amount of lever operation (lever operation angle).

[0094] Similarly, the operation sensor 29LB detects the operator's left-right operation of the left operation lever 26L and outputs the detected value to the controller 30. The operation sensor 29RA detects the operator's forward-backward operation of the right operation lever 26R and outputs the detected value to the controller 30. The operation sensor 29RB detects the operator's left-right operation of the right operation lever 26R and outputs the detected value to the controller 30. The operation sensor 29DL detects the operator's forward-backward operation of the left travel lever 26DL and outputs the detected value to the controller 30. The operation sensor 29DR detects the operator's forward-backward operation of the right travel lever 26DR and outputs the detected value to the controller 30.

[0095] In this embodiment, the description of the operating device 26 is based on a hydraulic operating lever equipped with a hydraulic pilot circuit. However, an electric operating lever equipped with an electric pilot circuit may be used instead of a hydraulic operating lever. In this case, the amount of lever operation of the electric operating lever is input to the controller 30 as an electrical signal. A solenoid valve is also placed between the pilot pump 15 and the pilot port of each control valve. The solenoid valve is configured to operate in response to an electrical signal from the controller 30. With this configuration, when manual operation is performed using the electric operating lever, the controller 30 can move each control valve by controlling the solenoid valve with an electrical signal corresponding to the amount of lever operation to increase or decrease the pilot pressure. Note that each control valve may be composed of an electromagnetic spool valve. In this case, the electromagnetic spool valve operates in response to an electrical signal from the controller 30 corresponding to the amount of lever operation of the electric operating lever.

[0096] The controller 30 receives the output of the operation sensor 29 and, if necessary, outputs a control command to the regulator 13 to change the discharge amount of the main pump 14. The controller 30 also receives the output of the control pressure sensor 19 located upstream of the throttle 18 and, if necessary, outputs a control command to the regulator 13 to change the discharge amount of the main pump 14. The throttle 18 includes a left throttle 18L and a right throttle 18R, and the control pressure sensor 19 includes a left control pressure sensor 19L and a right control pressure sensor 19R.

[0097] In the left center bypass pipeline 40L, a left throttle 18L is located between the control valve 176L, the downstreammost control valve, and the hydraulic fluid tank. Therefore, the flow of hydraulic fluid discharged by the left main pump 14L is restricted by the left throttle 18L. The left throttle 18L then generates a control pressure to control the left regulator 13L. The left control pressure sensor 19L is a sensor for detecting this control pressure and outputs the detected value to the controller 30. The controller 30 controls the discharge amount of the left main pump 14L by adjusting the swash plate tilt angle of the left main pump 14L in accordance with this control pressure. The controller 30 decreases the discharge amount of the left main pump 14L as the control pressure increases, and increases the discharge amount of the left main pump 14L as the control pressure decreases. The discharge amount of the right main pump 14R is controlled in the same way.

[0098] Specifically, as shown in Figure 3, when the hydraulic actuators in the shovel 100 are in a standby state and not being operated, the hydraulic fluid discharged from the left main pump 14L passes through the left center bypass pipe 40L to the left constriction 18L. The flow of hydraulic fluid discharged from the left main pump 14L increases the control pressure generated upstream of the left constriction 18L. As a result, the controller 30 reduces the discharge volume of the left main pump 14L to the minimum allowable discharge volume, suppressing the pressure loss (pumping loss) as the discharged hydraulic fluid passes through the left center bypass pipe 40L. On the other hand, when any of the hydraulic actuators are operated, the hydraulic fluid discharged from the left main pump 14L flows into the operated hydraulic actuator via the control valve corresponding to the operated hydraulic actuator. The flow of hydraulic fluid discharged from the left main pump 14L reduces or eliminates the amount reaching the left constriction 18L, lowering the control pressure generated upstream of the left constriction 18L. As a result, the controller 30 increases the discharge volume of the left main pump 14L, ensuring sufficient hydraulic fluid circulation to the hydraulic actuator being operated and guaranteeing reliable operation of the hydraulic actuator. The controller 30 also controls the discharge volume of the right main pump 14R in the same manner.

[0099] With the configuration described above, the hydraulic system in Figure 3 can suppress unnecessary energy consumption in the main pump 14 when in standby mode. Unnecessary energy consumption includes pumping losses caused by the hydraulic fluid discharged by the main pump 14 in the center bypass pipeline 40. Furthermore, when operating a hydraulic actuator, the hydraulic system in Figure 3 can reliably supply the necessary and sufficient hydraulic fluid from the main pump 14 to the hydraulic actuator being operated.

[0100] Furthermore, boom cylinder 7 is equipped with boom rod pressure sensor S7R and boom bottom pressure sensor S7B. Arm cylinder 8 is equipped with arm rod pressure sensor S8R and arm bottom pressure sensor S8B. Bucket cylinder 9 is equipped with bucket rod pressure sensor S9R and bucket bottom pressure sensor S9B. The boom rod pressure sensor S7R, boom bottom pressure sensor S7B, arm rod pressure sensor S8R, arm bottom pressure sensor S8B, bucket rod pressure sensor S9R, and bucket bottom pressure sensor S9B are collectively referred to as "cylinder pressure sensors". In addition, the swing hydraulic motor 2A is equipped with left swing pressure sensor S10L and right swing pressure sensor S10R.

[0101] The boom rod pressure sensor S7R detects the pressure in the rod-side oil chamber of the boom cylinder 7 (hereinafter referred to as "boom rod pressure"), and the boom bottom pressure sensor S7B detects the pressure in the bottom-side oil chamber of the boom cylinder 7 (hereinafter referred to as "boom bottom pressure"). The arm rod pressure sensor S8R detects the pressure in the rod-side oil chamber of the arm cylinder 8 (hereinafter referred to as "arm rod pressure"), and the arm bottom pressure sensor S8B detects the pressure in the bottom-side oil chamber of the arm cylinder 8 (hereinafter referred to as "arm bottom pressure"). The bucket rod pressure sensor S9R detects the pressure in the rod-side oil chamber of the bucket cylinder 9 (hereinafter referred to as "bucket rod pressure"), and the bucket bottom pressure sensor S9B detects the pressure in the bottom-side oil chamber of the bucket cylinder 9 (hereinafter referred to as "bucket bottom pressure"). The left slewing pressure sensor S10L detects the hydraulic fluid pressure at the left port of the slewing hydraulic motor 2A. The right slewing pressure sensor S10R detects the hydraulic fluid pressure at the right port of the slewing hydraulic motor 2A. The values ​​detected by each sensor are transmitted to the controller 30.

[0102] Next, with reference to Figures 4 to 7, the controller 30, which acts as a control device for predicting the behavior of the attachment AT, will be described. Figure 4 is a mathematical formula showing an example of the dynamics model 301 of the attachment AT stored in the controller 30. Figure 5 is a diagram showing an example of the learning model 302 stored in the controller 30.

[0103] The controller 30 includes, for example, a dynamics model 301 and a learning model 302, as shown in Figure 2. The dynamics model 301 and the learning model 302 are stored, for example, in a ROM or auxiliary storage medium that constitutes the controller 30.

[0104] The dynamics model 301 includes, for example, the dynamics equation shown as equation (1) in Figure 4. In equation (1), θ = [θ1, θ2, θ3], where θ1 is the boom angle, θ2 is the arm angle, and θ3 is the bucket angle. This dynamics equation (1) is set for each part of the attachment AT, for example. Specifically, for example, if the working machine is a shovel 100, the dynamics equation (1) is set for each of the boom 4, arm 5, and bucket 6 that make up the attachment AT.

[0105] In equation (1) shown in Figure 4, the equivalent mass term on the left side is determined by the orientation of each part of the attachment AT. The acceleration term on the left side is calculated by differential approximation from the velocities of each part of the attachment AT detected by sensors such as the boom angle sensor S1, arm angle sensor S2, and bucket angle sensor S3.

[0106] Furthermore, in equation (1) shown in Figure 4, the thrust in the first term on the right-hand side is calculated, for example, from the pressure in the oil chambers of the boom cylinder 7, arm cylinder 8, and bucket cylinder 9, as detected by the cylinder pressure sensor. The Coriolis force in the second term on the right-hand side is determined by the attitude and velocity of each part of the attachment AT. The gravity term in the third term on the right-hand side is also determined by the attitude of each part of the attachment AT.

[0107] Furthermore, in equation (1) shown in Figure 4, the fourth term on the right-hand side is a nonlinear element. The nonlinear element includes, for example, frictional forces and viscous forces acting on various parts of the attachment AT. The nonlinear element fluctuates due to, for example, temperature, humidity, field environment, and the passage of time. In this embodiment, the nonlinear element is output from the learning model 302.

[0108] Figure 5 illustrates an example of the learning model 302. The machine learning used to generate the learning model 302 is, for example, machine learning using a deep neural network (DNN), and deep learning is applied. In other words, the learning model 302 is, for example, a neural network model. The learning model 302 takes the detection results X_1, X_2, ..., X_D of various sensors as input to the input layer (layer 0) of the neural network, and outputs the nonlinear elements Y_1, Y_2, Y_3 of each part of the attachment AT from the output layer (layer n).

[0109] The detection results X_1, X_2, ..., X_D from various sensors input to the input layer (layer 0) of the neural network are, for example, the displacement, velocity, and oil chamber pressure of the hydraulic actuator that drives the attachment AT. The nonlinear elements Y_1, Y_2, Y_3 output from the output layer (layer n) of the neural network are, for example, the frictional force and viscous force acting on various parts of the attachment AT. Alternatively, the output layer (layer n) of the neural network may also output the Coriolis force, which is the second term on the right-hand side of the dynamic equation represented by equation (1) in Figure 4.

[0110] For example, in the excavator 100, the displacement and velocity of the boom cylinder 7, arm cylinder 8, and bucket cylinder 9, based on the detection results of the stroke sensors constituting the boom angle sensor S1, arm angle sensor S2, and bucket angle sensor S3, are input to the input layer (layer 0). Alternatively, the displacement and velocity of the boom 4, arm 5, and bucket 6, based on the detection results of the acceleration sensors constituting the boom angle sensor S1, arm angle sensor S2, and bucket angle sensor S3, may also be input to the input layer (layer 0).

[0111] In addition, in the excavator 100, the pressure of the oil chambers in the boom cylinder 7, arm cylinder 8, and bucket cylinder 9, which are the detection results of the cylinder pressure sensors, are input to the input layer (layer 0). The pressure of the oil chambers of each cylinder includes the boom bottom pressure, arm bottom pressure, and bucket bottom pressure detected by the boom bottom pressure sensor S7B, arm bottom pressure sensor S8B, and bucket bottom pressure sensor S9B. Furthermore, the pressure of the oil chambers of each cylinder may also include the boom rod pressure, arm rod pressure, and bucket rod pressure detected by the boom rod pressure sensor S7R, arm rod pressure sensor S8R, and bucket rod pressure sensor S9R.

[0112] Furthermore, in the shovel 100, the nonlinear elements Y_1, Y_2, and Y_3 of each part of the attachment AT output from the output layer (n layer) of the neural network include estimated values ​​of frictional and viscous forces acting on the boom 4, arm 5, and bucket 6, respectively.

[0113] Figure 6 is a functional block diagram showing the processing flow of the controller 30 in the excavator 100 of this embodiment. The dynamics model 301 generates dynamic equations for each part of the attachment AT based on inputs from the various sensors mentioned above and outputs them to the behavior prediction unit 303. The learning model 302 outputs the nonlinear terms of the dynamics model 301 corresponding to each part of the attachment AT, i.e., the nonlinear terms of the dynamic equations, to the behavior prediction unit 303 based on inputs from the various sensors mentioned above.

[0114] The behavior prediction unit 303 predicts the behavior of the attachment AT using the dynamic model 301 of each part of the attachment AT, which is a dynamic equation including nonlinear terms output by the learning model 302. Specifically, the behavior prediction unit 303 predicts, for example, the acceleration of each part of the attachment AT.

[0115] The attachment control unit 304 controls the operation of the attachment AT based on the behavior of the attachment AT predicted by the behavior prediction unit 303. Specifically, the attachment control unit 304 generates control commands for the proportional valve 31 using, for example, the outputs of the various sensors mentioned above and the behavior of the attachment AT, including the acceleration of each part of the attachment AT, as predicted by the behavior prediction unit 303. As a result, the attachment control unit 304 drives the boom cylinder 7, arm cylinder 8, bucket cylinder 9, etc., and performs machine control, which is automatic control of the attachment AT.

[0116] The payload calculation unit 305 calculates the load applied to the attachment AT based on the behavior of the attachment AT predicted by the behavior prediction unit 303. Specifically, the payload calculation unit 305 calculates the load applied to the attachment AT using, for example, the outputs of the various sensors mentioned above and the behavior of the attachment AT, including the acceleration of each part of the attachment AT, as predicted by the behavior prediction unit 303.

[0117] Next, the learning of the learning model 302 by the controller 30 will be explained with reference to Figure 7. Figure 7 is a functional block diagram showing the flow of learning the learning model 302 by the learning unit 306 of the controller 30. Note that the learning unit 306 may perform the learning of the learning model 302 while the shovel 100 is in operation, or it may complete the learning of the learning model 302 before the shovel 100 is in operation.

[0118] The learning unit 306 acquires detection results from the various sensors mentioned above, for example, which are input to the input layer of the learning model 302. The learning unit 306 also acquires the behavior of each part of the attachment AT predicted from the behavior prediction unit 303 based on the detection results of the various sensors. The learning unit 306 then calculates the difference between the behavior of the attachment AT predicted by the behavior prediction unit 303 and the actual behavior of the attachment AT based on the detection results of the various sensors.

[0119] Furthermore, the learning unit 306 acquires the displacement of each part of the attachment AT and the pressure in the oil chamber of the hydraulic cylinder that drives the attachment AT, based on the detection results of the various sensors mentioned above. The learning unit 306 then uses the difference in the behavior of the attachment AT, the displacement of each part of the attachment AT, and the pressure in the oil chamber of the hydraulic cylinder that drives the attachment AT as training data to train the learning model 302.

[0120] Furthermore, the learning unit 306 may train the learning model 302 even when no load is applied to the attachment AT. In this case, the learning unit 306 determines whether or not a load is applied to the attachment AT based on the recognition result of the external environment recognition device 70, such as the imaging device S6. For example, if a load is applied to the attachment AT, the learning unit 306 does not train the learning model 302, but if no load is applied to the attachment AT, it trains the learning model 302.

[0121] Specifically, for example, in the case of the shovel 100, the learning unit 306 determines whether or not excavation material such as soil is contained in the bucket 6 of the attachment AT, based on the recognition result of the forward recognition device 70F. For example, if excavation material is contained in the bucket 6, the learning unit 306 does not train the learning model 302, and if excavation material is not contained in the bucket 6, it trains the learning model 302.

[0122] The operation of the shovel 100 in this embodiment will be described below.

[0123] The excavator 100 of this embodiment is a work machine comprising an upper rotating body 3 as the main body, an attachment AT provided on the upper rotating body 3, and a controller 30 as a control device. The controller 30 stores a dynamics model 301 of the attachment AT and a learning model 302 that outputs the nonlinear elements of the dynamics model 301. The controller 30 predicts the behavior of the attachment AT using the dynamics model 301, which includes the nonlinear elements output by the learning model 302.

[0124] This configuration makes it possible to accurately estimate the nonlinear elements included in the dynamic equations of the dynamic model 301, which were previously difficult to estimate, using the learning model 302. As a result, in the case of the excavator 100, which is an example of a work machine according to this disclosure, the precise behavior of each part of the attachment AT that is affected by the nonlinear elements can be predicted using the dynamic model 301.

[0125] Furthermore, in the excavator 100, which is an example of a work machine related to this disclosure, the controller 30, which acts as a control device, controls the operation of the attachment AT based on the predicted behavior of the attachment AT.

[0126] With this configuration, the behavior of the attachment AT in response to the operating input can be accurately predicted based on the dynamics model 301, which includes nonlinear terms output by the learning model 302. Therefore, with the excavator 100, which is an example of a work machine according to this disclosure, the accuracy of automatic control, such as machine control of the attachment AT, can be improved.

[0127] Furthermore, in the excavator 100, which is an example of a work machine related to this disclosure, the controller 30, as a control device, trains the learning model 302 using the following items as training data. The training data consists of the difference between the predicted behavior of the attachment AT and the actual behavior of the attachment AT, the displacement of each part of the attachment AT, and the pressure of the hydraulic cylinder that drives the attachment AT.

[0128] This configuration allows the learning model 302 to be trained using training data, thereby improving the estimation accuracy of the nonlinear elements of the dynamics model 301, which is the output of the learning model 302.

[0129] Furthermore, in the excavator 100, which is an example of a work machine related to this disclosure, the controller 30, which is a control device, trains the learning model 302 when no load is applied to the attachment AT.

[0130] This configuration eliminates training data in which a load is applied to attachment AT, which acts as a disturbance during the training of the learning model 302, thereby improving the learning accuracy of the learning model 302.

[0131] Furthermore, in the excavator 100, which is an example of a work machine related to this disclosure, the nonlinear elements of the dynamics model 301 include frictional force.

[0132] This configuration makes it possible to accurately estimate frictional forces that fluctuate due to factors such as temperature, humidity, on-site environment, and the passage of time using the learning model 302.

[0133] Furthermore, as mentioned above, by having the learning model 302 output the Coriolis force, which is the second term on the right-hand side of the dynamic equation shown in equation (1) in Figure 4, it is possible to reduce the computational load on the controller 30 and improve the accuracy of the dynamic model 301.

[0134] Furthermore, the excavator 100 of this embodiment is a work machine comprising an upper rotating body 3 as the main body, an attachment AT provided on the upper rotating body 3, and a controller 30 as a control device. The controller 30 stores a dynamics model 301 of the attachment AT and a learning model 302 that outputs the nonlinear elements of the dynamics model 301. The controller 30 calculates the load applied to the attachment AT using the dynamics model 301, which includes the nonlinear elements output by the learning model 302.

[0135] With this configuration, the loads applied to the attachment AT, such as the payload (the weight of the excavated object contained in the bucket 6), can be calculated based on the dynamics model 301, which includes nonlinear terms, output by the learning model 302. This improves the accuracy of the calculation of the loads applied to the attachment AT.

[0136] Furthermore, the controller 30, which is an embodiment of the control device according to this disclosure, predicts the behavior of the attachment AT of a shovel 100, which is a work machine comprising an upper rotating body 3 as the main body and an attachment AT provided on the upper rotating body 3. The controller 30 stores a dynamics model 301 of the attachment AT and a learning model 302 that outputs the nonlinear elements of the dynamics model 301. The controller 30 predicts the behavior of the attachment AT using the dynamics model 301, which includes the nonlinear elements output by the learning model 302.

[0137] This configuration makes it possible to accurately estimate the nonlinear elements included in the dynamic equations of the dynamic model 301, which were previously difficult to estimate, using the learning model 302. As a result, according to the controller 30, which is an embodiment of the control device according to this disclosure, the precise behavior of each part of the attachment AT that is affected by the nonlinear elements can be predicted using the dynamic model 301.

[0138] Furthermore, the controller 30, which is an embodiment of the control device according to this disclosure, calculates the load applied to the attachment AT of a shovel 100, which is a work machine comprising an upper rotating body 3 as the main body and an attachment AT provided on the upper rotating body 3. The controller 30 stores a dynamics model 301 of the attachment AT and a learning model 302 that outputs the nonlinear elements of the dynamics model 301. The controller 30 calculates the load applied to the attachment AT using the dynamics model 301, which includes the nonlinear elements output by the learning model 302.

[0139] This configuration makes it possible to accurately estimate the nonlinear elements included in the dynamic equations of the dynamic model 301, which were previously difficult to estimate, using the learning model 302. As a result, the controller 30, which is an embodiment of the control device according to this disclosure, can improve the calculation accuracy of the load applied to the attachment AT, which is affected by the nonlinear elements.

[0140] As described above, this embodiment provides a shovel 100 and controller 30 as a work machine and control device that can predict the precise behavior of each part of the attachment AT that is affected by nonlinear elements.

[0141] Next, another embodiment of the control device according to the present disclosure will be described with reference to Figure 8. Figure 8 is a block diagram showing an embodiment of the control device according to the present disclosure. The remote controller 30E, which serves as the control device in this embodiment, constitutes part of the operating system SYS of a shovel 100, which is an example of a work machine.

[0142] The excavator operating system SYS of this embodiment includes, for example, the excavator 100 and the remote control room RC. Note that the detailed configuration of the excavator 100 is omitted from Figure 8 because the excavator 100 shown in Figure 8 has the same configuration as the excavator 100 shown in Figure 1.

[0143] The shovel 100 and the remote control room RC are connected to each other so that data can be sent and received via the remote communication device 60E and the communication line NW. Alternatively, the shovel 100 and the remote control room RC may be connected to each other so that data can be sent and received directly without using the communication line NW. In the illustrated example, the shovel 100 transmits information about the work site and the detection results of each sensor to the remote control room RC. This allows the remote operator RO in the remote control room RC to understand the situation at the work site and the status of the shovel 100 based on the information from the shovel 100.

[0144] The shovel 100 is equipped with sensors capable of recognizing the position and shape of objects present at the work site in three dimensions. For example, the shovel 100 is equipped with an external environment recognition device 70. Therefore, the shovel 100 can transmit the results of three-dimensional measurements of the work site to the remote control room RC.

[0145] The external environment recognition device 70 is a device for recognizing the space surrounding the shovel 100. In the illustrated example, the spatial recognition device is a LiDAR. The LiDAR measures, for example, the distance between the LiDAR and each of more than one million points within the monitoring range. The external environment recognition device 70 can be any device capable of measuring the distance to an object. For example, the external environment recognition device 70 may be a stereo camera, or a combination of an imaging device S6 and a ranging device such as a millimeter-wave radar.

[0146] The operating system SYS may include one or more excavators 100. If it includes multiple excavators 100, the remote operator RO operating a specific excavator 100 can obtain information about the work sites obtained by that specific excavator 100, as well as information about the work sites obtained by one or more other excavators 100.

[0147] The remote control room RC is equipped with a remote communication device 60E, a remote controller 30E, a remote control device 26E, an operation sensor 29E, and a remote output device 50E including a display device. The remote control room RC also has an operator's seat DS where the remote operator RO sits to remotely control the shovel 100.

[0148] The remote communication device 60E is configured to communicate with the communication device T1 attached to the shovel 100.

[0149] The remote controller 30E is a computing device that performs various calculations. In this embodiment, the remote controller 30E is composed of a microcomputer including a CPU and memory. The various functions of the remote controller 30E are realized by the CPU executing a program stored in memory.

[0150] The remote controller 30E, like the controller 30 mounted on the excavator 100, is a control device that stores the dynamics model 301 of the attachment AT and a learning model 302 that outputs the nonlinear elements of the dynamics model 301. In other words, the remote controller 30E predicts the behavior of the attachment AT using the dynamics model 301, which includes the nonlinear elements output by the learning model 302.

[0151] The display device included in the remote output device 50E is a device capable of displaying various types of information. The display device displays images based on information transmitted from the shovel 100 so that the remote operator RO in the remote control room RC can visually inspect the area around the shovel 100. In the illustrated example, the display device is a liquid crystal display that displays images captured by the imaging device S6 mounted on the shovel 100. The display device may also be a display or projector that enables naked-eye stereoscopic viewing, or it may be a VR goggle or the like.

[0152] The remote control device 26E is equipped with an operation sensor 29E for detecting the operation of the remote control device 26E. The operation sensor 29E is, for example, a tilt sensor that detects the tilt angle of the operating lever, or an angle sensor that detects the oscillation angle of the operating lever around its pivot axis. The operation sensor 29E may also consist of other sensors such as a pressure sensor, a current sensor, a voltage sensor, or a distance sensor. The operation sensor 29E outputs information regarding the operation of the remote control device 26E that it has detected to the remote controller 30E. The remote controller 30E generates an operation signal based on the received information and transmits the generated operation signal to the shovel 100. The operation sensor 29E may be configured to generate the operation signal. In this case, the operation sensor 29E may output the operation signal to the remote communication device 60E without going through the remote controller 30E. With this configuration, the remote operator RO can remotely operate the shovel 100 from the remote control room RC.

[0153] As described above, the remote controller 30E, which serves as the control device in this embodiment, predicts the behavior of the attachment AT of the shovel 100, which is a work machine comprising an upper rotating body 3 as the main body and an attachment AT provided on the upper rotating body 3. The remote controller 30E stores a dynamics model 301 of the attachment AT and a learning model 302 that outputs the nonlinear elements of the dynamics model 301. The remote controller 30E predicts the behavior of the attachment AT using the dynamics model 301, which includes the nonlinear elements output by the learning model 302.

[0154] This configuration makes it possible to accurately estimate the nonlinear elements included in the dynamic equations of the dynamic model 301, which were previously difficult to estimate, using the learning model 302. As a result, with the remote controller 30E, which is an embodiment of the control device according to this disclosure, the precise behavior of each part of the attachment AT affected by the nonlinear elements can be predicted using the dynamic model 301.

[0155] Furthermore, the remote controller 30E, which serves as the control device in this embodiment, calculates the load applied to the attachment AT of the shovel 100, which is a work machine comprising an upper rotating body 3 as the main body and an attachment AT provided on the upper rotating body 3. The remote controller 30E stores a dynamics model 301 of the attachment AT and a learning model 302 that outputs the nonlinear elements of the dynamics model 301. The remote controller 30E calculates the load applied to the attachment AT using the dynamics model 301, which includes the nonlinear elements output by the learning model 302.

[0156] This configuration makes it possible to accurately estimate the nonlinear elements included in the dynamic equations of the dynamic model 301, which were previously difficult to estimate, using the learning model 302. As a result, the remote controller 30E, which is an embodiment of the control device according to this disclosure, can improve the calculation accuracy of the load applied to the attachment AT, which is affected by the nonlinear elements.

[0157] Preferred embodiments of the present disclosure have been described above. However, the inventions of the present disclosure are not limited to the embodiments described above. Various modifications, substitutions, etc., can be applied to the embodiments described above without departing from the scope of the inventions of the present disclosure. Furthermore, each of the features described with reference to the embodiments described above may be combined as appropriate, as long as they do not contradict each other technically.

[0158] For example, the work machine may be a machine other than an excavator, such as a crawler crane, wheel crane, stationary crane, wheel loader, or forklift, which has a main body and attachments provided on that main body. Also, in the case of an excavator, which is an example of a work machine, all or part of the driven parts such as the lower traveling body, upper rotating body, boom, arm, and bucket may be electrically driven. That is, the excavator may be a hybrid excavator or electric excavator in which all or part of the driven parts are driven by electric actuators. Specifically, the excavator may be an electric excavator that drives all of the driven parts using only an electric motor as the power source. A hybrid excavator is typically an excavator that uses a combination of an internal combustion engine such as a diesel engine and an electric motor driven by a battery mounted on the upper rotating body as the power source, while an electric excavator is typically an excavator that uses only an electric motor driven by a battery mounted on the upper rotating body as the power source. However, an electric excavator may also be an excavator that uses only an electric motor connected to an external power source as the power source. [Explanation of Symbols]

[0159] 3. Upper rotating body (main body) 4. Boom (Each part of the attachment) 5. Arms (parts of the attachment) 6. Buckets (parts of the attachment) 7. Boom Cylinder (Hydraulic Cylinder) 8. Arm Cylinder (Hydraulic Cylinder) 9. Bucket Cylinder (Hydraulic Cylinder) 30 Controller (control device) 30E Remote Controller (Control Device) 100 Shovel (working machine) 301 Dynamics Model 302 Learning Models AT attachment

Claims

1. The main body and The attachment provided on the main body, The system includes a control device that stores a dynamics model of the attachment and a learning model that outputs the nonlinear elements of the dynamics model, The control device predicts the behavior of the attachment using the dynamics model which includes the nonlinear elements output by the learning model. Agricultural machinery.

2. The control device controls the operation of the attachment based on the predicted behavior of the attachment. The work machine according to claim 1.

3. The control device trains the learning model using the difference between the predicted behavior of the attachment and the actual behavior of the attachment, the displacement of each part of the attachment, and the pressure of the hydraulic cylinder that drives the attachment as training data. The work machine according to claim 1.

4. The work machine according to claim 3, wherein the control device trains the learning model when no load is applied to the attachment.

5. The aforementioned nonlinear element includes frictional force, The work machine according to claim 1.

6. The main body and The attachment provided on the main body, The system includes a control device that stores a dynamics model of the attachment and a learning model that outputs the nonlinear elements of the dynamics model, The control device calculates the load applied to the attachment using the dynamic model which includes the nonlinear elements output by the learning model. Agricultural machinery.

7. A control device for predicting the behavior of an attachment of a work machine comprising a main body and an attachment provided on the main body, The system stores a dynamics model of the attachment and a learning model that outputs the nonlinear elements of the dynamics model, and uses the dynamics model, including the nonlinear elements output by the learning model, to predict the behavior of the attachment. Control device.

8. A control device for calculating the load applied to an attachment of a work machine, which comprises a main body and an attachment provided on the main body, The system stores a dynamic model of the attachment and a learning model that outputs the nonlinear elements of the dynamic model, and calculates the load applied to the attachment using the dynamic model including the nonlinear elements output by the learning model. Control device.

Images