Hydraulic control methods and system for construction machinery

The hydraulic control method and system for construction machinery addresses simulation challenges by using machine learning to model and inversely transform hydraulic systems, ensuring real-time control for unmanned operations.

KR103046540B1Active Publication Date: 2026-11-10HD CONSTRUCTIONEQUIPMENTCO LTD +1
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
KR1020220003599
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2022-01-10
Publication Date
2026-11-10
Estimated Expiration
2042-01-10

AI Technical Summary

Technical Problem

Hydraulic systems in construction machinery are difficult to mathematically simulate due to nonlinearity and uncertainty caused by complexity, leading to challenges in controlling unmanned automated operations.

Method used

A hydraulic control method and system that models the hydraulic system using machine learning, generates a system model, and inversely transforms it to create a controller model, utilizing Piecewise Linear, Infinite Impulse Response, and Multi-Layer Perceptron functions to interpret nonlinear characteristics and time delays, enabling real-time control.

Benefits of technology

Enables real-time control of hydraulic systems in construction machinery for unmanned automation by accurately predicting joystick signals, reducing network complexity, and ensuring high control performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure R1020220003599_ABST
    Figure R1020220003599_ABST
Patent Text Reader

Abstract

A hydraulic control method for construction machinery includes modeling the hydraulic system of the construction machinery to perform unmanned automation work to generate a system model, inversely transforming the system model to generate a controller model, outputting a control signal through the controller model based on an input target command signal, and controlling the operation of the hydraulic system through the control signal.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to a hydraulic control method and system for construction machinery. More specifically, it relates to a method for controlling a hydraulic system of construction machinery for performing unmanned automated operations and a hydraulic control system of construction machinery for performing the same. Background Technology

[0002] Mathematical analysis techniques utilizing transfer functions and differential equations exist as methods for modeling automated construction machinery systems. However, in the case of hydraulic systems, there are limitations to mathematically simulating the entire system due to nonlinearity and uncertainty caused by the complexity of hydraulic circuits, valve flow characteristics, and fluid compressibility. Furthermore, resolving the delay characteristics arising from these factors presents a problem in that heavy and complex network structures are required. The problem to be solved

[0003] One objective of the present invention is to provide a hydraulic control method for construction machinery that can derive a model of the hydraulic system and improve control performance during unmanned automated operations.

[0004] Another objective of the present invention is to provide a hydraulic control system for a construction machine for performing the hydraulic control method described above. means of solving the problem

[0005] In exemplary embodiments for achieving one objective of the present invention, to perform unmanned automation work, a hydraulic system of a construction machine is modeled to generate a system model, the system model is inversely transformed to generate a controller model, an input target command signal is output as a control signal through the controller model, and the operation of the hydraulic system is controlled through the control signal.

[0006] In exemplary embodiments, modeling the hydraulic system of the construction machine may include generating the system model by performing machine learning using sensor signals regarding state values ​​of the hydraulic system according to a joystick command signal as learning data.

[0007] In exemplary embodiments, the sensor signal may include data regarding the joint angle of the hydraulic system, the pressure of the hydraulic cylinders, the pressure of the hydraulic pump, and the external force applied to the hydraulic system.

[0008] In exemplary embodiments, modeling the hydraulic system to generate the system model may include generating a Piecewise Linear (PL) function unit for analyzing the dead zone and saturation section of the hydraulic system, generating an Infinite Impulse Response (IIR) filter unit for expressing the time delay of the hydraulic system, and generating a Multi-Layer Perceptron (MLP) function unit for analyzing the nonlinear characteristics of the hydraulic system.

[0009] In exemplary embodiments, the segment linear function part may be generated by arithmetic calculating the change in the dead zone according to the ground environment of the construction machine.

[0010] In exemplary embodiments, generating a controller model by inverse transforming the system model may include generating a Multi-Layer Perceptron (MLP) inverse function unit for interpreting the nonlinear characteristics of the hydraulic system, generating an Infinite Impulse Response (IIR) inverse transform filter unit for expressing the time delay of the hydraulic system, and generating a Piecewise Linear (PL) inverse function unit for interpreting the dead zone and saturation section of the hydraulic system.

[0011] In exemplary embodiments, the segmented linear inverse function may be generated by arithmetic the change in the dead zone according to the ground environment of the construction machine.

[0012] In exemplary embodiments, outputting a control signal through the controller model may include outputting a joystick command signal as the control signal based on an error signal between the input target command signal and the sensor signal for the state values ​​of the hydraulic system.

[0013] A hydraulic control system for a construction machine according to exemplary embodiments for achieving another objective of the present invention may include a hydraulic system for a construction machine for performing unmanned automation work, a measuring unit for measuring sensor signals regarding state values ​​of the hydraulic system, and a control device for controlling the operation of the hydraulic system by modeling the hydraulic system to generate a system model, inversely transforming the system model to generate a controller model, and outputting a control signal through the controller model to input a target command signal.

[0014] In exemplary embodiments, the control device can generate the system model by performing machine learning using sensor signals regarding state values ​​of the hydraulic system according to a joystick command signal as learning data.

[0015] In exemplary embodiments, the sensor signal may include data regarding the joint angle of the hydraulic system, the pressure of the hydraulic cylinders, the pressure of the hydraulic pump, and the external force applied to the hydraulic system.

[0016] In exemplary embodiments, the system model may include a Piecewise Linear (PL) function unit for interpreting the dead zone and saturation section of the hydraulic system, an Infinite Impulse Response (IIR) filter unit for expressing the time delay of the hydraulic system, and a Multi-Layer Perceptron (MLP) function unit for interpreting the nonlinear characteristics of the hydraulic system.

[0017] In exemplary embodiments, the system model may further include an auxiliary network for receiving the sensor signal, calculating the change in the dead zone according to the ground environment, and inputting the change in the dead zone to the piece linear function part.

[0018] In exemplary embodiments, the controller model may include a Multi-Layer Perceptron (MLP) inverse function unit for interpreting the nonlinear characteristics of the hydraulic system, an Infinite Impulse Response (IIR) inverse transform filter unit for expressing the time delay of the hydraulic system, and a Piecewise Linear (PL) inverse function unit for interpreting the dead zone and saturation section of the hydraulic system.

[0019] In exemplary embodiments, the control device may output a joystick command signal as a control signal based on an error signal between the input target command signal and the sensor signal. Effects of the invention

[0020] In exemplary embodiments, a hydraulic control method for a construction machine may include modeling a hydraulic system of a construction machine to perform unmanned automation work to generate a system model, inversely transforming the system model to generate a controller model, outputting a control signal through the controller model based on an input target command signal, and controlling the operation of the hydraulic system through the control signal.

[0021] Accordingly, the hydraulic system can be modeled using machine learning that utilizes data acquired during the process of operating the hydraulic system to perform the aforementioned unmanned automation task as training data to generate the system model, and the controller model can be generated by inversely transforming the system model (by dividing the mathematically interpretable parts from the uninterpretable parts and interpreting each nonlinear characteristic). Based on this, the network structure of the learning model can be lightweightly configured, and a control system capable of guaranteeing real-time performance for the entire hydraulic system model can be designed.

[0022] However, the effects of the present invention are not limited to the effects mentioned above and may be extended in various ways without departing from the spirit and scope of the present invention. Brief explanation of the drawing

[0023] FIG. 1 is a side view showing a construction machine according to exemplary embodiments. Figure 2 is a block diagram showing the hydraulic system of the construction machine of Figure 1. FIG. 3 is a block diagram showing a hydraulic control system of a construction machine according to exemplary embodiments. Figure 4 is a block diagram illustrating the process of generating a system model of the hydraulic system of Figure 3. Figure 5 is a block diagram illustrating the process of generating a controller model by inversely transforming the system model of Figure 4. FIG. 6 is a flowchart illustrating a hydraulic control method for a construction machine according to exemplary embodiments. Figure 7 is a flowchart illustrating the process of creating the system model of Figure 6. Figure 8 is a flowchart illustrating the process of generating a controller model by inversely transforming the system model of Figure 7. Specific details for implementing the invention

[0024] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

[0025] In each drawing of the present invention, the dimensions of the structures are depicted enlarged compared to the actual dimensions for the sake of clarity of the present invention.

[0026] In the present invention, terms such as "first," "second," etc., may be used to describe various components, but said components should not be limited by said terms. These terms are used solely for the purpose of distinguishing one component from another.

[0027] The terms used in this invention are used merely to describe specific embodiments and are not intended to limit the invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "having" are intended to indicate the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0028] With respect to the embodiments of the present invention disclosed in the text, specific structural or functional descriptions are provided merely for the purpose of explaining the embodiments of the present invention, and the embodiments of the present invention may be implemented in various forms and should not be interpreted as being limited to the embodiments described in the text.

[0029] That is, the present invention is capable of various modifications and may take various forms, and specific embodiments are illustrated in the drawings and described in detail in the text. However, this is not intended to limit the present invention to the specific disclosed forms, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the invention.

[0030] FIG. 1 is a side view showing a construction machine according to exemplary embodiments. FIG. 2 is a block diagram showing a hydraulic system of the construction machine of FIG. 1.

[0031] Referring to FIGS. 1 and 2, the construction machine may include equipment capable of performing unmanned automated operations using a hydraulic system, such as an excavator, wheel loader, forklift, agricultural machinery, mining equipment, etc. Below, we will describe the case where the construction machine is an excavator. However, it will be understood that this does not limit the control system according to the exemplary embodiments to controlling an excavator.

[0032] Specifically, the construction machine (10) may include a lower driving body (20), an upper rotating body (30) mounted on the lower driving body (20) so as to be rotatable, and a driver's cabin (50) and a work device (60) installed on the upper rotating body (30).

[0033] The lower driving body (20) supports the upper turning body (30) and can drive a construction machine (10), such as an excavator, using power generated from the engine (110). The lower driving body (20) may be a tracked type driving body including endless tracks. Alternatively, the lower driving body (20) may be a wheeled type driving body including driving wheels. The upper turning body (30) is equipped with an upper frame (32) as a base and can set the working direction by rotating on the lower driving body (20) in a plane parallel to the ground.

[0034] The working device (60) may include a boom (70), an arm (80), and a bucket (90). The working device (60) may be operated by the drive of actuators (62), such as a boom cylinder (72), an arm cylinder (82), and a bucket cylinder (92). As the boom cylinder (72), the arm cylinder (82), and the bucket cylinder (92) extend or retract, the boom (70), the arm (80), and the bucket (90) can perform various movements, and the working device (60) can perform various tasks. At this time, the boom cylinder (72), the arm cylinder (82), and the bucket cylinder (92) may extend or retract by hydraulic fluid supplied from a hydraulic pump (120).

[0035] The construction machine (10) may include a slewing device for rotating the upper slewing body (30). The slewing device may be operated by the drive of an actuator (62), such as a slewing motor. At this time, the slewing motor may rotate clockwise or counterclockwise by hydraulic fluid supplied from a hydraulic pump (120).

[0036] The construction machine (10) may include a driving device for rotating the right track (or right wheel) and the left track (or left wheel) of the lower driving body (20). The driving device may include actuators (62), such as a left driving motor and a right driving motor, for rotating the right track (or right wheel) and the left track (or left wheel), respectively. At this time, the left driving motor and the right driving motor may rotate clockwise or counterclockwise by means of hydraulic fluid supplied from a hydraulic pump (120).

[0037] As illustrated in FIG. 2, the hydraulic system of the construction machine may include an engine (110), a hydraulic pump (120) driven by the engine (110), at least one actuator (62) operable by hydraulic fluid discharged from the hydraulic pump (120), a main control valve (130) installed between the hydraulic pump (120) and the actuator (62) and including at least one control valve for controlling the flow rate of hydraulic fluid supplied from the hydraulic pump (120) to the actuator (62) according to the displacement amount of a spool provided inside, and a spool displacement adjustment unit (140) including a spool displacement adjustment valve that supplies a pilot signal pressure to the spool to control the displacement amount of the spool of the control valve in proportion to an input control signal.

[0038] The hydraulic pump (120) can be connected to the output shaft of the engine (110) as a mechanical drive unit. For example, the hydraulic pump (120) may include a pressure-controlled electro-hydraulic pump. The discharge flow rate of the hydraulic pump (120) can be determined by the swash plate angle. The swash plate angle of the hydraulic pump (120) can be controlled by a pump regulator (122). The pump regulator (122) can control the swash plate angle of the hydraulic pump (120) according to a pump control signal input from the control device (200).

[0039] The hydraulic pump (120) can be connected to the main control valve (130) through a high-pressure hydraulic line (124). Actuators (62), such as the boom cylinder (72), arm cylinder (82), bucket cylinder (92), the swing motor, the left drive motor and the right drive motor, can be connected to the main control valve (130) through a high-pressure hydraulic line.

[0040] A pilot pump (150) is connected to the output shaft of the engine (110) and can be driven to discharge control fluid as the output shaft rotates. For example, the pilot pump may be a gear pump. In this case, the operating fluid and the control fluid may contain substantially the same material.

[0041] The control fluid discharged from the pilot pump (150) can be supplied to the spool of the control valve through the spool displacement adjustment valve of the spool displacement adjustment unit (140). The spool displacement adjustment valve can supply a pilot signal pressure to the spool of the control valve to control the amount of displacement of the spool of the control valve in proportion to the input control signal.

[0042] For example, a pair of spool displacement adjustment valves may be provided on each side of the spool of a corresponding control valve. The control valve may be switched by selectively supplying a pilot signal pressure output from the spool displacement adjustment valve to both sides of the spool within the corresponding control valve. The spool displacement adjustment valve may supply a pilot signal having a magnitude proportional to the input control signal. The movement of the spool within the control valve may be controlled by the pilot signal pressure. That is, the direction of movement of the spool is determined according to the supply direction of the pilot signal pressure, and the amount of displacement of the spool may be determined according to the strength of the pilot signal pressure.

[0043] For example, the main control valve (130) as an assembly having the above-mentioned control valve may be an electro-hydraulic main control valve. The spool displacement adjustment valve may include an electro-proportional reduction valve (EPPRV) that controls the pilot working fluid applied to the spool within the control valve according to an input electrical signal.

[0044] In exemplary embodiments, the control device (200) may output a pressure command signal to the spool displacement adjustment valve as a control signal for operating the work device (60) according to a target command signal for unmanned automation work or a manual command signal from an operator. The electronic proportional pressure reducing valve may control the spool with an electrical control signal by outputting a secondary pressure proportional to the pressure command signal to the corresponding spool.

[0045] As described above, the hydraulic system of the construction machine includes a hydraulic control device such as a main control valve (130) and a spool displacement adjustment unit (140), and the hydraulic control device can control the operation of the hydraulic system of the construction machine (the working device, the slewing device, and the driving device) according to a control signal output from the control device (200).

[0046] Below, the hydraulic control system of the above-mentioned construction machine will be described.

[0047] FIG. 3 is a block diagram showing a hydraulic control system of a construction machine according to exemplary embodiments. FIG. 4 is a block diagram showing the process of generating a system model of the hydraulic system of FIG. 3. FIG. 5 is a block diagram showing the process of generating a controller model by inversely transforming the system model of FIG. 4.

[0048] Referring to FIGS. 3 to 5, the hydraulic control system of a construction machine may include a hydraulic control device for controlling the operation of the hydraulic system (100) of the construction machine according to an input control signal, a measuring unit (300) for measuring sensor signals regarding state values ​​of the hydraulic system (100), and a control device (200) for generating and outputting the control signal to perform unmanned automation work of the construction machine. Additionally, the hydraulic control system of the construction machine may further include an upper control unit (400) for providing terrain information and work information of the work terrain for unmanned automation work of the construction machine.

[0049] In exemplary embodiments, the hydraulic control device can control the hydraulic system (100) according to the control signal received from the control device (200). The hydraulic control device can receive the control signal from the control device (200) which receives a work command from a control unit that controls the site through a higher-level control unit (400), such as a motion planner, for unmanned or automated work. The hydraulic control device can convert the result value generated during the operation of the hydraulic system (100) into a feedback signal and transmit it to the control device (200). The feedback signal may include the joint angle and joint angular velocity described later.

[0050] In exemplary embodiments, the measuring unit (300) can measure sensor signals regarding state values ​​of the hydraulic system (100). The sensor signals may include pressures of hydraulic cylinders, pressures of hydraulic pumps, and external forces applied to the hydraulic system. For example, the measuring unit (300) can measure the state values ​​of the hydraulic system (100) in real time during the operation of the construction machine. The state values ​​of the hydraulic system (100) measured by the measuring unit (300) can be expressed as follows in Equation (1).

[0051] ------ Equation (1)

[0052] Here, The working device (60) is a state value of the boom (70), arm (80), and bucket (90), and is the pressure of the above hydraulic cylinders, and is the pressure of the above hydraulic pump, and is an external force applied to the above hydraulic system.

[0053] For example, the sensor signal may include the angle and angular velocity of each of the joints of the boom (70), arm (80), and bucket (90) of the work device (60). Angle sensors and IMU sensors can measure the angle, angular velocity, and angular acceleration values ​​of each of the joints of the boom (70), arm (80), and bucket (90).

[0054] Additionally, the measuring unit (300) may include a sensor for acquiring surrounding environment perception data necessary for unmanned automation work of the construction machine. The sensor may acquire surrounding environment perception data to identify the work terrain and detect surrounding objects. The sensor may acquire front ground information and object information by scanning the terrain or objects in front.

[0055] For example, the sensor may include a LiDAR sensor, a stereo camera sensor, a radar sensor, an ultrasonic sensor, etc.

[0056] In exemplary embodiments, the measuring unit (300) may transmit the surrounding environment perception data to the control unit (200). The control unit (200) may transmit the surrounding environment perception data to the upper control unit (400) and receive terrain information and work information of the work area for the unmanned automation work from the upper control unit (400). For example, the work area may be defined as an area for an excavator to move and work.

[0057] The measurement unit (300) can be wirelessly connected to the upper control unit (400). The upper control unit (400) can receive sensor signals directly from the measurement unit (300). For example, when there is physical interaction with the surrounding environment, such as when a large load is applied to the construction machine during the operation, the upper control unit (400) can receive feedback of the sensor signals from the measurement unit (300) to consider such load conditions and generate a target command signal.

[0058] The upper control unit (400) may be provided as a subsystem within the construction machine or as a separate system (e.g., a control unit). The upper control unit (400) may determine a target command signal in kinematics by utilizing received commands and surrounding environment perception data. For example, the target command signal may include a trajectory in a three-dimensional coordinate system of an end-effector in the field of robotics or an angular trajectory of a joint part. The target command signal may include angular velocity command signals for each of the joints of the boom (70), arm (80), and bucket (90). The upper control unit (400) may not be directly involved in the driving of the hydraulic system (100) of the construction machine.

[0059] In exemplary embodiments, the control device (200) may output the control signal for controlling the operation of the hydraulic system (100). The control device (200) may include a microcomputer and various interfaces and may control the operation of the hydraulic system (100) according to program and recipe information stored in external memory or internal memory. The control device (200) may output the control signal through the target command signal. As described below, the control device (200) may include a controller configured to generate a control signal based on an error signal between the target command signal and a state value measured by the measurement unit (300). For example, the control signal may include a joystick command signal for controlling the operation of the hydraulic system (100).

[0060] The control device (200) may be connected to the upper control unit (400) wirelessly or via a wired connection. For example, the control device (200) may include a wireless transceiver for wireless communication with the upper control unit (400). The wireless transceiver may use known communication standards such as cellular communication like CDMA and GSM, Wi-Fi, and radio communication. The wireless transceiver may include a Telematics Monitor System (TMS) module.

[0061] The control device (200) may include a controller for compensating for the error between the work trajectory from the upper control unit (400) and the actual work trajectory of the hydraulic system (100). The controller may generate a control signal capable of minimizing the error between the target command signal and the feedback sensor signal. The controller may maintain high control performance in various ground environments by considering the physical interaction between the hydraulic system (100) and the ground. The control device (200) may further include a tracking control unit (202) that receives the target command signal and the feedback signal and generates a joint angular velocity signal modified through error feedback.

[0062] The control device (200) can generate a system model (210) by modeling the hydraulic system (100), and can generate a controller model (220) to be described later by inversely transforming the system model (210). For example, the control device (200) can design the controller model (220) by inversely transforming the system model (210) after learning the system model (210) for the hydraulic system (100) has been performed. The system model (210) may be a simulation of the hydraulic system (100) and may be implemented as a look-up table or mathematically.

[0063] For example, the control device (200) may define the system delay characteristics of the hydraulic system (100) as a delay system to classify and learn the system model (210). Based on the delay system, the object to be considered before the delay occurs may be defined as a pre-delay map, and the object to be considered after the delay occurs may be defined as a post-delay map. For example, the pre-delay map may include characteristics at the input end that must be considered before the delay occurs as the input signal passes through the system. The post-delay map may include arbitrary non-linear characteristics that must be considered for the signal after the delay occurs.

[0064] The control device (200) can output the control signal through a controller having the controller model (220) using the target command signal input from the upper control unit (400). The control device (200) can predict and output the control signal for controlling the hydraulic system (100) through the controller model (220).

[0065] The control device (200) can perform offline learning based on the system model (210) using actual work data of the construction machine. For example, the system model (210) can be generated by acquiring data output from an actual hydraulic system according to a joystick signal and performing modeling of the hydraulic system using machine learning. The data may include sensor signals including joint angles, pressure of cylinders and pumps, and external force (excavation force).

[0066] In exemplary embodiments, the system model (210) generated through the system modeling may include a piecewise linear (PL) function section (212) for interpreting the dead zone and saturation section of the hydraulic system, an infinite impulse response (IIR) filter section (214) for expressing the time delay of the hydraulic system, and a multi-layer perceptron (MLP) function section (216) for interpreting the non-linear characteristics of the hydraulic system. The system model (210) may further include an auxiliary network (230) for receiving the sensor signal and calculating the change in the dead zone according to the ground environment.

[0067] The above-mentioned piece linear function part (212) may include a piece linear function having a monotonically increasing function. Because the piece linear function part (212) includes the monotonically increasing function, the dead zone and the saturation section can be easily expressed, and the inverse function of the dead zone and the saturation section can be easily expressed. The piece linear function can be expressed as shown in the following equation (2).

[0068] ------ Equation (2)

[0069] Here, is the joystick input value, is the above-mentioned piece linear function.

[0070] For example, the auxiliary network (230) can receive the sensor signal and calculate the change in the dead zone according to the ground environment. The auxiliary network (230) can input the dead zone characteristics that change according to the environment or condition of the ground where the construction machine is located into the fragment linear function unit (212). The auxiliary network (230) can learn the interval of the fragment linear function. The fragment linear function unit (212) can receive the interval learned through the environment variable from the auxiliary network (230) and can express the change in the dead zone reflecting the environment variable.

[0071] The infinite impulse response filter unit (214) can represent the time delay of the hydraulic system including past input and output information using an infinite impulse response filter. The infinite impulse response filter may refer to a learnable linear recurrent unit having an infinite impulse response structure. Since the infinite impulse response filter unit (214) does not require future information or optimization like a Recurrent Neural Network (RNN), the computation time is reduced, enabling real-time control. The infinite impulse response filter can be represented as shown in the following equation (3).

[0072] ------ Equation (3)

[0073] Here, is a delay system, and Z is It is the Z-transform of, and is the result of the above piece linear function.

[0074] The above-mentioned multi-neural network model function unit (216) can interpret non-linearities such as input coupling and interaction with the outside of the construction machine using a multi-neural network model. For example, the input of the above-mentioned multi-neural network model function unit (216) may include the output of the infinite impulse response filter, joint angle, pressure, and external force, and the output may include joint angular velocity. The above-mentioned multi-neural network model may be represented as shown in the following equation (4).

[0075] ------ Equation (4)

[0076] Here, is the predicted joint angular velocity, and is the above-mentioned multi-neural network model, and is the result value of the above infinite impulse response.

[0077] In exemplary embodiments, the control device (200) may generate a controller model (220) by inversely transforming the system model (210) that models the hydraulic system (100). The control device (200) may generate the control signal through the controller model (220) based on the target command signal, the sensor signal, and the feedback signal, and transmit it to the hydraulic system (100). The control device (200) may control the operation of the hydraulic system (100) using the control signal generated through the controller model (220). For example, the control signal may include the joystick command signal for controlling the operation of the hydraulic system (100).

[0078] The control device (200) can predict the joystick signal required to correspond to the target joint angular velocity using the target joint angular velocity, joint angle, cylinder and pump pressure, and external force. The target joint angular velocity can take into account an error feedback signal for trajectory tracking.

[0079] The above controller model (220) may include a multi-neural network model inverse function section (222) for interpreting the non-linear characteristics of the hydraulic system (100), an infinite impulse response inverse transform filter section (224) for expressing the time delay of the hydraulic system (100), and a piece linear inverse function section (226) for interpreting the dead zone and saturation section of the hydraulic system (100). The above controller model (220) may further include the auxiliary network (230) for receiving the sensor signal, calculating the change in the dead zone according to the ground environment, and inputting the change in the dead zone to the piece linear inverse function section (226).

[0080] The inverse function unit (222) of the multi-neural network model can obtain the joint angular velocity to be tracked using the current joint angle, the planned joint angle, and the angular velocity. The current joint angle, the planned joint angle, and the angular velocity can be calculated in a composite identity manner through the inverse function of the multi-neural network model. The inverse function unit (222) of the multi-neural network model can consider interaction with the external environment and non-linearity through the inverse function of the multi-neural network model. The inverse function of the multi-neural network model can be expressed as shown in the following equation (5).

[0081] ------ Equation (5)

[0082] Here, is the command joint angular velocity, and is the inverse function of the above-mentioned multi-neural network model.

[0083] The infinite impulse response inverse transform filter unit (224) can compensate for the delay effect of the hydraulic system (100) by utilizing the inverse transform of the infinite impulse response. The infinite impulse response inverse transform filter unit (224) may include a transfer function capable of ramp tracking. The infinite impulse response inverse transform filter unit (224) can calculate the transfer function through offline learning. Since the infinite impulse response inverse transform filter unit (224) requires only the current input, past input, and past output during real-time control of the construction machine, rapid calculation may be possible. The inverse transform of the infinite impulse response can be expressed as shown in the following equation (6).

[0084] ------ Equation (6)

[0085] Here, is a delay tracking system, and Z is It is the Z-transform of, and is the result of the inverse function part of the above-mentioned multi-neural network model.

[0086] The above-mentioned segment linear inverse function unit (226) can compensate for the dead zone and the saturation section using the inverse function of the segment linear function. The above-mentioned segment linear inverse function unit (226) can predict the joystick signal corresponding to the target joint angular velocity using the inverse function of the segment linear function. The inverse function of the segment linear function can be expressed as shown in the following equation (7).

[0087] ------ Equation (7)

[0088] Here, is the predicted joystick signal above, and is the inverse function of a piecewise linear function, and is the result of the inverse impulse response filter section above.

[0089] As described above, the hydraulic control system of the construction machine can generate the system model (210) by modeling the hydraulic system (100) through machine learning using data acquired during the process of operating the hydraulic system (100) as learning data to perform the unmanned automation work, and can generate the controller model (220) by inversely transforming the system model (210) (by dividing the parts that can be mathematically interpreted and those that cannot, and interpreting each non-linear characteristic). Based on this, the network structure of the learning model can be lightly configured, and a control system capable of guaranteeing real-time performance for the entire hydraulic system model can be designed.

[0090] Below, we will explain how to control the hydraulic system of a construction machine using the hydraulic control system of the construction machine of FIG. 3.

[0091] FIG. 6 is a flowchart illustrating a hydraulic control method for a construction machine according to exemplary embodiments. FIG. 7 is a flowchart illustrating the process of generating the system model of FIG. 6. FIG. 8 is a flowchart illustrating the process of generating a controller model by inversely transforming the system model of FIG. 7.

[0092] Referring to FIGS. 1 to 8, first, a system model (210) can be created by modeling the hydraulic system (100) of a construction machine to perform unmanned automation work (S110).

[0093] In exemplary embodiments, the system model (210) may be obtained by performing offline learning using actual work data of the construction machine as learning data. For example, the system model (210) may be generated by obtaining data output from an actual hydraulic system according to a joystick signal and performing modeling of the hydraulic system through machine learning. The data may include sensor signals including joint angles, pressure of cylinders and pumps, and external force (excavation force).

[0094] For example, the above system model can be defined by classifying the delay characteristics of the hydraulic system into a Delay System, the factors to be considered before the delay occurs based on the Delay System into a Pre-Delay Map, and the factors to be considered after the delay occurs into a Post-Delay Map.

[0095] The system model (210) generated through the above system modeling may include a Piecewise Linear (PL) function section (212), an Infinite Impulse Response (IIR) filter section (214), and a Multi-Layer Perceptron (MLP) function section (216). The system model (210) may further include an auxiliary network (230).

[0096] Next, the system model (210) can be inversely transformed to generate a controller model (220) (S120).

[0097] In exemplary embodiments, the controller model (220) may mean a model for outputting a control signal capable of automatically controlling the hydraulic system by inversely converting the system model (210). For example, the controller model (220) may predict a joystick signal required to correspond to the target joint angular velocity using the target joint angular velocity, the joint angle, the cylinder and pump pressure, and the external force.

[0098] The above controller model (220) may include a multi-neural network model inverse function section (222), an infinite impulse response inverse transform filter section (224), and a piece linear inverse function section (226). The above controller model (220) may further include the auxiliary network (230).

[0099] Next, the input target command signal can be output as a control signal through the controller model (220) (S130), and the hydraulic system (100) can be operated through the control signal (S140).

[0100] In exemplary embodiments, the control signal may refer to a command signal learned using the system model (210) and the inverse transformation model. Accordingly, the control signal may refer to a command signal for automatically operating the construction machine from a target command signal input from the upper control unit and a sensor signal measured by the measurement unit. For example, the control signal may include a joystick command signal for controlling the operation of the hydraulic system. Since the control signal can be generated using a network structure of a lightly configured learning model, the hydraulic system can be controlled in real time. The target command signal may include angular velocity command signals for each of the joints of the boom (70), arm (80), and bucket (90).

[0101] As illustrated in FIG. 7, creating a system model (210) by modeling the hydraulic system (100) can first generate a piecewise linear (PL) function part (212) for analyzing the dead zone and saturation section of the hydraulic system (S210).

[0102] In exemplary embodiments, the piece linear function part (212) may include a piece linear function having a monotonically increasing function. Because the piece linear function part (212) includes the monotonically increasing function, the dead zone and the saturation section can be easily expressed, and the inverse function of the dead zone and the saturation section can be easily expressed.

[0103] Next, an infinite impulse response (IIR) filter section (214) for expressing the time delay of the hydraulic system can be generated (S220).

[0104] In exemplary embodiments, the infinite impulse response filter unit (214) can represent the time delay of the hydraulic system including past input and output information using an infinite impulse response filter. The infinite impulse response filter may refer to a learnable linear recurrent unit having an infinite impulse response structure. Since the infinite impulse response filter unit (214) does not require future information or optimization, such as a Recurrent Neural Network (RNN), the computation time is reduced, enabling real-time control.

[0105] Next, a multi-layer perceptron (MLP) function part (216) for interpreting the non-linear characteristics of the hydraulic system (100) can be generated (S230).

[0106] In exemplary embodiments, the multi-neural network model function (216) can interpret non-linearities such as input coupling and interaction with the outside of the construction machine using a multi-neural network model. For example, the input of the multi-neural network model function (216) may include the output of the infinite impulse response filter, joint angle, pressure, and external force, and the output may include joint angular velocity.

[0107] As illustrated in FIG. 8, generating a controller model (220) by inversely transforming the system model (210) can first generate a multi-neural network model inverse function part (222) for interpreting the non-linear characteristics of the hydraulic system (100) (S310).

[0108] In exemplary embodiments, the multi-neural network model inverse function unit (222) can obtain the joint angular velocity to be tracked using the current joint angle and the planned joint angle and angular velocity. The current joint angle and the planned joint angle and angular velocity can be calculated in a composite identity manner through the inverse function of the multi-neural network model. The multi-neural network model inverse function unit (222) can consider interaction with the external environment and non-linearity through the inverse function of the multi-neural network model.

[0109] Next, an infinite impulse response (IIR) inverse transform filter section (224) for expressing the time delay of the hydraulic system can be generated (S320).

[0110] In exemplary embodiments, the infinite impulse response inverse transform filter unit (224) can compensate for the delay effect of the hydraulic system (100) by utilizing the inverse transform of the infinite impulse response. The infinite impulse response inverse transform filter unit (224) may include a transfer function capable of ramp tracking. The infinite impulse response inverse transform filter unit (224) can calculate the transfer function through offline learning. The infinite impulse response inverse transform filter unit (224) may be able to calculate quickly because it requires only the current input, past input, and past output during real-time control of the construction machine.

[0111] Next, a piecewise linear (PL) inverse function (226) for analyzing the dead zone and saturation section of the hydraulic system can be generated (S330).

[0112] In exemplary embodiments, the piece linear inverse function unit (226) can compensate for the dead zone and the saturation section using the inverse function of the piece linear function. The piece linear inverse function unit (226) can predict the joystick signal corresponding to the target joint angular velocity using the inverse function of the piece linear function.

[0113] Although the present invention has been described above with reference to embodiments thereof, those skilled in the art will understand that various modifications and changes can be made to the invention without departing from the spirit and scope of the invention as described in the following claims. Explanation of the symbols

[0114] 10: Construction machinery 20: Undercarriage 30: Upper pivot body 32: Upper frame 40: Counterweight 50: Cab 60: Working device 62: Actuator 70: Boom 72: Boom Cylinder 80: Female 82: Female cylinder 90: Bucket 92: Bucket Cylinder 100: Hydraulic system 110: Engine 120: Hydraulic pump 122: Pump regulator 124: High-pressure hydraulic line 130: Main control valve 140: Spool displacement adjustment unit 150: Pilot pump 200: Control unit 202: Tracking control unit 210: System Model 212: Piece of Linear Function 214: Infinite impulse response filter section 216: Multi-neural network model function section 220: Controller Model 222: Inverse Function of Multi-Neural Network Model 224: Infinite impulse response inverse transform filter section 226: Fragmentary linear inverse function section 230: Auxiliary network 300: Measurement unit 400: Upper control unit

Claims

Claim 1 A hydraulic control method for a construction machine, comprising: modeling a hydraulic system of a construction machine to generate a system model for performing unmanned automation work; inversely transforming the system model to generate a controller model; outputting a control signal through the controller model based on an input target command signal; and controlling the operation of the hydraulic system through the control signal, wherein the system model is a model simulating the hydraulic system of the construction machine, and modeling the hydraulic system of the construction machine includes generating the system model by performing machine learning using sensor signals regarding state values ​​of the hydraulic system according to a joystick command signal as learning data. Claim 2 delete Claim 3 A hydraulic control method for a construction machine according to claim 1, wherein the sensor signal includes data regarding the joint angle of the hydraulic system, the pressure of the hydraulic cylinders, the pressure of the hydraulic pump, and the external force applied to the hydraulic system. Claim 4 A hydraulic control method according to claim 1, wherein modeling the hydraulic system to generate the system model comprises generating a Piecewise Linear (PL) function part for analyzing the dead zone and saturation section of the hydraulic system; generating an Infinite Impulse Response (IIR) filter part for expressing the time delay of the hydraulic system; and generating a Multi-Layer Perceptron (MLP) function part for analyzing the non-linear characteristics of the hydraulic system. Claim 5 In claim 4, generating the segmented linear function part is a hydraulic control method of a construction machine that generates the segmented linear function part by arithmetically calculating the change in the dead zone according to the ground environment of the construction machine. Claim 6 A hydraulic control method according to claim 1, wherein generating the controller model by inverse transforming the system model comprises generating a Multi-Layer Perceptron (MLP) inverse function part for analyzing the nonlinear characteristics of the hydraulic system; generating an Infinite Impulse Response (IIR) inverse transform filter part for expressing the time delay of the hydraulic system; and generating a Piecewise Linear (PL) inverse function part for analyzing the dead zone and saturation section of the hydraulic system. Claim 7 In claim 6, the method of generating the segmented linear inverse function part is a hydraulic control method of a construction machine that generates the segmented linear inverse function part by arithmetically calculating the change in the dead zone according to the ground environment of the construction machine. Claim 8 A hydraulic control method for a construction machine according to claim 1, wherein outputting a control signal through the controller model includes outputting a joystick command signal as the control signal based on an error signal between the input target command signal and the sensor signal for the state values ​​of the hydraulic system. Claim 9 A hydraulic control system for a construction machine for performing unmanned automated operations; a measuring unit for measuring sensor signals for state values ​​of the hydraulic system; and a control device for modeling the hydraulic system to generate a system model, inversely transforming the system model to generate a controller model, and outputting a control signal through the controller model to control the operation of the hydraulic system by inputting a target command signal, wherein the system model is a model simulating the hydraulic system of the construction machine, and the control device generates the system model by performing machine learning using sensor signals for state values ​​of the hydraulic system according to a joystick command signal as learning data. Claim 10 delete Claim 11 In claim 9, the sensor signal comprises data regarding the joint angle of the hydraulic system, the pressure of the hydraulic cylinders, the pressure of the hydraulic pump, and the external force applied to the hydraulic system, in a hydraulic control system of a construction machine. Claim 12 In claim 9, the system model comprises: a Piecewise Linear (PL) function section for analyzing the dead zone and saturation section of the hydraulic system; an Infinite Impulse Response (IIR) filter section for expressing the time delay of the hydraulic system; and a Multi-Layer Perceptron (MLP) function section for analyzing the non-linear characteristics of the hydraulic system, thereby forming a hydraulic control system for construction machinery. Claim 13 A hydraulic control system for construction machinery according to claim 12, wherein the system model further comprises an auxiliary network for receiving the sensor signal, calculating the change in the dead zone according to the ground environment, and inputting the change in the dead zone to the segmented linear function part. Claim 14 In claim 9, the controller model comprises: a Multi-Layer Perceptron (MLP) inverse function section for interpreting the nonlinear characteristics of the hydraulic system; an Infinite Impulse Response (IIR) inverse transform filter section for expressing the time delay of the hydraulic system; and a Piecewise Linear (PL) inverse function section for interpreting the dead zone and saturation section of the hydraulic system, thereby forming a hydraulic control system for construction machinery. Claim 15 A hydraulic control system for construction machinery according to claim 14, wherein the controller model further comprises an auxiliary network for receiving the sensor signal, calculating the change in the dead zone according to the ground environment, and inputting the change in the dead zone to the segmented linear inverse function part. Claim 16 A hydraulic control system of a construction machine according to claim 9, wherein the control device outputs a joystick command signal as a control signal based on an error signal between the input target command signal and the sensor signal.

Citation Information

Patent Citations

  • Hydraulic control method and system

    KR1020210123023A

  • Coordinated joint motion control system with position error correction

    US20150322647A1