WO2022186382A1 - 故障予兆検出システムおよび作業車 - Google Patents

故障予兆検出システムおよび作業車 Download PDF

Info

Publication number
WO2022186382A1
WO2022186382A1 PCT/JP2022/009436 JP2022009436W WO2022186382A1 WO 2022186382 A1 WO2022186382 A1 WO 2022186382A1 JP 2022009436 W JP2022009436 W JP 2022009436W WO 2022186382 A1 WO2022186382 A1 WO 2022186382A1
Authority
WO
WIPO (PCT)
Prior art keywords
failure
actuator
detection system
unit
signal
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2022/009436
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
佳成 南
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tadano Ltd
Original Assignee
Tadano Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tadano Ltd filed Critical Tadano Ltd
Priority to JP2023503973A priority Critical patent/JP7501777B2/ja
Publication of WO2022186382A1 publication Critical patent/WO2022186382A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66C—CRANES; LOAD-ENGAGING ELEMENTS OR DEVICES FOR CRANES, CAPSTANS, WINCHES, OR TACKLES
    • B66C15/00—Safety gear
    • E—FIXED CONSTRUCTIONS
    • E02—HYDRAULIC ENGINEERING; FOUNDATIONS; SOIL SHIFTING
    • E02F—DREDGING; SOIL-SHIFTING
    • E02F9/00—Component parts of dredgers or soil-shifting machines, not restricted to one of the kinds covered by groups E02F3/00 - E02F7/00
    • E02F9/26—Indicating devices

Definitions

  • the present invention relates to a failure sign detection system and a working vehicle. More specifically, the present invention relates to a failure sign detection system for a work vehicle having a boom that can be raised and lowered through a swivel on a traveling body, and the work vehicle.
  • the controller of the construction machine described in Patent Document 1 in the failure diagnosis mode of the variable displacement pump, determines the variable displacement in the failure diagnosis mode from the control command for failure diagnosis of the variable displacement pump and the engine speed for failure diagnosis. Calculate the theoretical pump flow rate of the model pump. Furthermore, the controller determines whether the variable displacement pump is faulty based on the actual pump flow rate and the theoretical pump flow rate in the fault diagnosis mode. With this configuration, the controller allows the optimum pump flow rate to flow from the variable displacement pump to the oil passage for judging failure of the variable displacement pump. Accurate determination is possible.
  • the technique of Patent Document 1 cannot detect actuator failure unless the operator performs the failure diagnosis mode.
  • the failure diagnosis mode is a technique for judging the presence or absence of a failure from the operating state of the actuator at the time when the failure diagnosis mode is executed, and does not detect signs of failure.
  • the failure diagnosis mode performs failure determination under specific operating conditions, there is a possibility that failures that occur under conditions other than the specific operating conditions cannot be detected.
  • An object of the present invention is to provide a failure sign detection system and a work vehicle that can detect a sign of failure without being affected by the operating state of the work vehicle.
  • the inventors studied a failure sign detection system and a work vehicle that can detect signs of failure without being affected by the operating state of the work vehicle. As a result of intensive studies, the present inventors came up with the following configuration.
  • a failure sign detection system includes: A failure sign detection system for detecting a sign of failure of an actuator mounted on a work vehicle having a boom, a physical model unit having a physical model that mathematically expresses the normal characteristics of the actuator; It has a learning model that learns the characteristics of the actuator by adjusting the weighting factor based on the difference between the output of the physical model unit and the output of the actuator, and corrects the input to the physical model unit based on the output of the learning model. a learning model unit; a failure prediction unit that detects a sign of failure based on the weighting factor.
  • the present invention it is possible to provide a failure sign detection system and a work vehicle that can detect a sign of failure without being affected by the operating state of the work vehicle.
  • FIG. 1 is a side view showing the overall configuration of the crane.
  • FIG. 2 is a block diagram showing the control configuration of the crane.
  • FIG. 3 is a block diagram showing the control configuration of the control device of this embodiment.
  • FIG. 4 is a diagram showing an inverse dynamics model of a crane.
  • FIG. 5 is a block diagram showing the control configuration of the control system according to the first embodiment of the invention.
  • FIG. 6 is a block diagram showing another control configuration of the control system according to the first embodiment of the invention.
  • FIG. 7 is a block diagram showing the control configuration of the control system according to the second embodiment of the invention.
  • FIG. 8 is a block diagram showing the control configuration of the control system according to the third embodiment of the invention.
  • FIG. 1 is a side view showing the overall configuration of the crane 1.
  • FIG. 2 is a block diagram showing the control configuration of the crane 1.
  • the crane 1 (rough terrain crane) will be described as a work vehicle, but an all-terrain crane, a truck crane, a loading truck crane, an aerial work vehicle, or the like may also be used. That is, the present invention can be applied to a working vehicle having a boom that can be raised and lowered through a swivel base on a traveling body.
  • “(n), (n+1), (n+2)” are the n-th, n+1-th, and n+2-th information acquired per unit time t (e.g., wire rope payout amount).
  • t e.g., wire rope payout amount
  • (n) means information acquired after n ⁇ t time has elapsed from the start of acquisition of information.
  • (n+1) means information acquired after (n+1) ⁇ t time elapses from the start of information acquisition.
  • “(n+2)” means information acquired after (n+2) ⁇ t time elapses from the start of information acquisition.
  • the crane 1 is a mobile crane that can be moved to an unspecified location.
  • the crane 1 has a vehicle 2, a crane device 6 that is a working device, and a load moving operation tool 32 (see FIG. 2) that can operate the crane device 6 on the basis of the load W.
  • a load moving operation tool 32 see FIG. 2
  • the vehicle 2 is a traveling body that transports the crane device 6.
  • a vehicle 2 has a plurality of wheels 3 and runs using an engine 4 as a power source.
  • the vehicle 2 is provided with outriggers 5 .
  • the outriggers 5 are composed of overhang beams that can be hydraulically extended on both sides in the width direction of the vehicle 2 and hydraulic jack cylinders that can be extended in a direction perpendicular to the ground.
  • the crane device 6 is a working device that lifts the load W with a wire rope.
  • the crane device 6 includes a swivel base 7, a boom 9, a jib 9a, a main hook block 10, a sub-hook block 11, a hoisting hydraulic cylinder 12, a main winch 13, a main wire rope 14, a sub winch 15, a sub wire rope 16 and a cabin. 17 and the like.
  • the crane device 6 is provided with at least one actuator.
  • the swivel base 7 is a driving device that allows the crane device 6 to swivel.
  • the swivel base 7 is provided on the frame of the vehicle 2 via an annular bearing.
  • the swivel base 7 is rotatable around the center of an annular bearing.
  • the swivel base 7 is provided with a hydraulic swiveling hydraulic motor 8 as an actuator.
  • the turning hydraulic motor 8 is an actuator that is rotated by a turning valve 23 (see FIG. 2), which is an electromagnetic proportional switching valve.
  • the swivel base 7 is provided with a swivel sensor 27 (see FIG. 2) which is a swivel angle detector for detecting the swivel angle ⁇ z (see FIG. 4) and the swivel speed of the swivel base 7 .
  • the swivel base cameras 7b are provided on both left and right sides in front of the swivel base 7 and on both left and right sides behind the swivel base 7.
  • the swivel base camera 7b is configured to be usable as a set of stereo cameras. That is, the swivel base camera 7b can detect the position information of the suspended load W by using it as a set of stereo cameras. It should be noted that detection of the position information of the load W may be performed by any means capable of detecting the position information of the load W, such as a millimeter wave radar, an acceleration sensor, or a GNSS.
  • the boom 9 is a movable strut that supports the main wire rope 14 and the sub wire rope 16 so that the load W can be lifted.
  • the boom 9 is composed of a plurality of boom members.
  • the base end of the base boom member of the boom 9 is swingably provided substantially at the center of the swivel base 7 .
  • the boom 9 is configured to be telescopic in the axial direction by moving each boom member by a telescopic hydraulic cylinder (not shown), which is an actuator.
  • the telescopic hydraulic cylinder is telescopically operated by an telescopic valve 24 (see FIG. 2), which is an electromagnetic proportional switching valve.
  • the boom 9 is provided with a jib 9a and a boom camera 9b (see FIG. 2).
  • the boom 9 also has a telescopic sensor 28 for detecting the telescopic length lb(n) of the boom 9, an azimuth sensor 29 (see FIG. 2) for detecting the azimuth centered on the tip of the boom 9, and an hoisting angle.
  • An undulation sensor 30 (see FIG. 2) is provided to detect .theta.x (see FIG. 4).
  • the main hook block 10 and the sub hook block 11 are hangers for hanging the load W.
  • the main hook block 10 is provided with a plurality of hook sheaves around which the main wire rope 14 is wound, and a main hook 10a for hanging the load W.
  • the sub-hook block 11 is provided with a sub-hook 11a for hanging the load W.
  • the hoisting hydraulic cylinder 12 is an actuator that raises and lowers the boom 9 and holds the posture of the boom 9 .
  • the hoisting hydraulic cylinder 12 has an end of the cylinder portion swingably connected to the swivel base 7 and an end of the rod portion swingably connected to the base boom member of the boom 9 .
  • the hoisting hydraulic cylinder 12 is expanded and contracted by a hoisting valve 25 (see FIG. 2), which is an electromagnetic proportional switching valve.
  • the main winch 13 and the sub winch 15 are winding devices that carry in (wind up) and let out (lower) the main wire rope 14 and the sub wire rope 16 .
  • the main winch 13 has a main drum around which the main wire rope 14 is wound.
  • the main drum rotates based on the power of a main hydraulic motor (not shown), which is an actuator.
  • the sub winch 15 has a sub drum around which the sub wire rope 16 is wound.
  • the sub-drum rotates based on the power of a sub-hydraulic motor (not shown), which is an actuator.
  • the main hydraulic motor is rotated by a main valve 26m (see Fig. 2), which is an electromagnetic proportional switching valve.
  • the sub hydraulic motor is rotationally operated by a sub valve 26s (see FIG. 2), which is an electromagnetic proportional switching valve.
  • the main winch 13 and the sub winch 15 are provided with a winding sensor 33 (see FIG. 2) for detecting the let-out amount l(n) (see FIG. 4) of the main wire rope 14 and the sub wire rope 16, respectively.
  • the cabin 17 is the cockpit covered by the housing.
  • the cabin 17 is mounted on the swivel base 7 .
  • the cabin 17 includes an operating tool for operating the vehicle 2, a turning operating tool 18 for operating the crane device 6, a hoisting operating tool 19, an extending operating tool 20, a main drum operating tool 21m, and a sub drum operating tool 21s. etc. are provided (see FIG. 2).
  • the cabin 17 is provided with a load moving operation tool 32 (see FIG. 2), which is a load moving operation unit for inputting the moving direction and moving speed of the load W.
  • the load movement operation tool 32 is an operation tool for inputting instructions regarding the movement direction and speed of the load W on the horizontal plane.
  • the load moving operation tool 32 is configured to transmit an operation signal regarding the tilting direction and tilting amount of the operation stick to the control device 31 (see FIG. 2).
  • the load movement operation tool 32 transmits the operation signal to the control device 31 as a target movement speed signal Vd indicating the target movement speed of the load W.
  • the control device 31 is a control device 31 that controls the actuators of the crane device 6 via each operation valve.
  • the control device 31 may have a configuration in which a CPU, a ROM, a RAM, an HDD, and the like are connected via a bus, or may have a configuration including a one-chip LSI or the like.
  • the controller 31 stores various programs and data for controlling the operations of actuators, switching valves, sensors, and the like.
  • the control device 31 is connected to the swivel base camera 7b, the boom camera 9b, the swivel base camera 7b, the boom camera 9b, the swivel manipulator 18, the raising and lowering manipulator 19, the telescopic manipulator 20, the main drum manipulator 21m and the sub drum manipulator 21s. , an image from the boom camera 9b, and the amount of operation of each operation tool can be obtained.
  • the control device 31 is connected to the swing valve 23, the expansion/contraction valve 24, the hoisting valve 25, the main valve 26m and the sub valve 26s, and is an actuation signal for operating each actuator to a target actuation amount.
  • a certain target actuation signal Md (see FIG. 4) can be transmitted.
  • the control device 31 is connected to the turning sensor 27, the extension/retraction sensor 28, the GNSS type azimuth sensor 29, the undulation sensor 30, and the winding sensor 33.
  • the angle ⁇ x, the payout amount l(n) and the orientation of the main wire rope 14 or the sub wire rope 16 (hereinafter simply referred to as "wire rope") can be obtained.
  • the turning sensor 27 may be used as the orientation sensor 29 .
  • the control device 31 generates a target actuation signal Md corresponding to each operating tool based on the amount of operation of the turning operating tool 18, the hoisting operating tool 19, the main drum operating tool 21m and the sub drum operating tool 21s.
  • the control device 31 Based on the orientation of the tip of the boom 9 acquired by the orientation sensor 29, the control device 31 calculates a target trajectory signal Pd ⁇ (see FIG. 3), which is an input signal for moving the load W to the target operation amount. Further, the control device 31 calculates the target position coordinates p(n+1) of the load W, which is the target position of the load W, from the target trajectory signal Pd ⁇ . The control device 31 generates a target actuation signal Md for the swing valve 23, the expansion/contraction valve 24, the hoisting valve 25, the main valve 26m, and the sub valve 26s, which move the load W to the target position coordinate p(n+1). (See Figure 3).
  • the crane 1 configured in this way can move the crane device 6 to any position by running the vehicle 2 .
  • the crane 1 can increase the lifting height and working radius of the crane device 6 by setting the boom 9 to an arbitrary hoisting angle and expansion/contraction length using each operating tool.
  • the crane 1 can convey the load W by operating the swivel base 7 and the boom 9 .
  • the crane 1 moves the cargo W in the tilting direction of the cargo moving operation tool 32 at a speed corresponding to the tilting amount.
  • FIG. 3 is a block diagram showing the control configuration of the control device 31 according to the present invention.
  • FIG. 4 shows an inverse dynamics model of the crane 1 according to the invention.
  • the control device 31 has a target trajectory calculator 31a, a boom position calculator 31b, and an actuation signal generator 31c.
  • the control device 31 can acquire the current position information of the load W by a pair of swivel base cameras 7 b on both left and right sides in front of the swivel base 7 . (See Figure 2).
  • the target trajectory calculation unit 31a is a part of the control device 31, and converts a target movement speed signal Vd indicating the target movement speed of the load W into a target trajectory signal Pd ⁇ of the load W.
  • the target trajectory signal Pd ⁇ of the load W includes information on the movement direction and speed of the load W.
  • the target trajectory calculator 31a can acquire the target moving speed signal Vd indicated by the load W from the load moving operation tool 32 every unit time t. Further, the target trajectory calculator 31a can integrate the acquired target moving speed signal Vd to calculate the target trajectory signal Pd ⁇ of the load W in the x-axis direction, the y-axis direction, and the z-axis direction per unit time t.
  • the suffix ⁇ is a code representing any one of the x-axis direction, the y-axis direction, and the z-axis direction.
  • the boom position calculation unit 31b is a part of the control device 31, and calculates the position coordinates of the tip of the boom 9 from the posture information of the boom 9 and the target trajectory signal Pd ⁇ of the load W.
  • the boom position calculator 31b can acquire the target trajectory signal Pd ⁇ from the target trajectory calculator 31a.
  • the boom position calculator 31 b acquires the turning angle ⁇ z(n) of the turning base 7 from the turning sensor 27 .
  • the boom position calculator 31b acquires the extension/retraction length lb(n) from the extension/retraction sensor 28 .
  • the boom position calculator 31 b acquires the hoisting angle ⁇ x(n) from the hoisting sensor 30 .
  • the boom position calculator 31 b acquires the feed-out amount l(n) of the main wire rope 14 or the sub wire rope 16 (hereinafter simply referred to as “wire rope”) from the winding sensor 33 .
  • the boom position calculator 31b acquires the current position information of the load W from the image of the load W captured by the swivel base camera 7b (see FIG. 2).
  • the boom position calculation unit 31b calculates the current position coordinates p(n) of the load W from the acquired current position information of the load W.
  • the boom position calculator 31b calculates the current position of the tip of the boom 9 (wire rope extension position) (hereinafter simply referred to as "current position coordinates q(n) of boom 9").
  • the boom position calculator 31b also calculates the wire rope feedout amount l(n) from the current position coordinates p(n) of the load W and the current position coordinates q(n) of the boom 9 .
  • the boom position calculator 31b also calculates target position coordinates p(n+1) of the load W, which is the position of the load W after the unit time t has elapsed, from the target trajectory signal Pd ⁇ .
  • the boom position calculator 31b calculates the direction vector e(n+1) of the wire rope from which the load W is suspended from the current position coordinates p(n) of the load W and the target position coordinates p(n+1) of the load W. calculate.
  • the boom position calculator 31b uses an inverse dynamics model to determine the position of the tip of the boom 9 after the lapse of unit time t from the target position coordinate p(n+1) of the load W and the direction vector e(n+1) of the wire rope.
  • a target position coordinate q(n+1) of a certain boom 9 is calculated (see FIG. 4).
  • the actuation signal generation unit 31c is a part of the control device 31, and generates a target actuation signal Md and the like for each actuator for moving the boom 9 to the target position coordinate q(n+1) after the unit time t has elapsed.
  • the actuation signal generation unit 31c receives the current position coordinates q(n) of the boom 9, the target position coordinates p(n+1) of the load W after the unit time t has elapsed, and the target position coordinates q(n+1) of the boom 9 from the boom position calculation unit 31b. ).
  • the actuation signal generation unit 31c generates the turning valve 23, the expansion/contraction valve 24, the expansion/contraction valve 24, and the rotation valve 23 from the current position coordinate q(n) of the boom 9, the target position coordinate q(n+1) of the boom 9, and the target position coordinate p(n+1) of the load W. It is configured to generate a target actuation signal Md for the hoisting valve 25, the main valve 26m, or the sub valve 26s. The actuation signal generator 31c transmits the generated target actuation signal Md to each actuator of the crane 1 (see FIG. 4).
  • the control device 31 controls the current position coordinate p(n) of the load W, the current position coordinate q(n) of the boom 9, and the target position coordinate p(n+1) of the load W after the unit time t has elapsed.
  • the inverse dynamics model is defined in an XYZ coordinate system, with the origin O as the center of rotation of the crane 1 .
  • Controller 31 defines q, p, lb, ⁇ x, ⁇ z, l, f and e in the inverse dynamics model, respectively.
  • q indicates the current position coordinate q(n) of the tip of the boom 9
  • p indicates the current position coordinate p(n) of the load W.
  • lb indicates the telescopic length lb(n) of the boom 9
  • .theta.x indicates the hoisting angle .theta.x(n)
  • .theta.z indicates the turning angle .theta.z(n).
  • l indicates the wire rope payout amount l(n)
  • f indicates the wire rope tension f
  • e indicates the wire rope direction vector e(n).
  • the relationship between the target position q of the tip of the boom 9 and the target position p of the load W can be obtained from the target position p of the load W, the mass m of the load W, and the spring constant kf of the wire rope.
  • the target position q of the tip of the boom 9 is calculated by the equation (3), which is expressed by the equation (2) and is a function of the time of the load W.
  • f tension of the wire rope
  • kf spring constant
  • m mass of the load W
  • q current position or target position of the tip of the boom 9
  • p current position or target position of the load W
  • l feed amount of the wire rope.
  • e direction vector
  • g gravitational acceleration
  • the wire rope payout amount l(n) is calculated from the following equation (4).
  • the wire rope feedout amount l(n) is defined by the distance between the current position coordinates q(n) of the boom 9, which is the tip position of the boom 9, and the current position coordinates p(n) of the load W, which is the position of the load W. be.
  • the direction vector e(n) of the wire rope is calculated from the following equation (5).
  • the wire rope direction vector e(n) is the unit length vector of the wire rope tension f (see equation (2)).
  • the tension f of the wire rope is calculated by subtracting the gravitational acceleration from the acceleration of the load W calculated from the current position coordinates p(n) of the load W and the target position coordinates p(n+1) of the load W after the unit time t has elapsed. be done.
  • the target position coordinate q(n+1) of the boom 9, which is the target position of the tip of the boom 9 after the elapse of the unit time t, is calculated from Equation (6), which expresses Equation (2) as a function of n.
  • ⁇ indicates the turning angle ⁇ z(n) of the boom 9 .
  • the target position coordinate q(n+1) of the boom 9 is calculated from the wire rope payout amount l(n), the target position coordinate p(n+1) of the load W, and the direction vector e(n+1) using inverse dynamics. .
  • control device 31 uses an inverse dynamics model to calculate the target trajectory signal Pd ⁇ of the load W and the target position coordinate q(n+1) of the tip of the boom 9 as one of preferred embodiments. I am using However, the control device 31 may perform control based on forward dynamics in which the input amount of each actuator is calculated from the operation signal of the load moving operation tool 32 .
  • FIG. 5 is a block diagram showing the control configuration of the control system according to the first embodiment of the invention.
  • FIG. 6 is a block diagram showing another control configuration of the control system according to the first embodiment of the invention.
  • the control system 34 includes a load moving operation tool 32, a turning sensor 27, an undulating sensor 30, an extending/contracting sensor 28, a turntable camera 7b (see FIG. 2), a target value filter 35 included in the control device 31, and a target operation amount. It includes a calculator 36 and a failure sign detection system 37 .
  • the control system 34 generates the target actuation signal Md through the cooperation of each sensor, the target trajectory calculator 31a, the boom position calculator 31b, and the actuation signal generator 31c of the control device 31 .
  • the target value filter 35 configured in the target trajectory calculation unit 31a of the control device 31 attenuates frequencies equal to or higher than a predetermined frequency.
  • a target movement position signal Pd which is a control signal for moving the load W to the target position, is input to the target value filter 35 .
  • the target movement position signal Pd is obtained by converting the target movement speed signal Vd from the load movement operation tool 32 by an integrator 32a included in the target trajectory calculation unit 31a.
  • a target value filter 35 calculates a target trajectory signal Pd ⁇ of the load W from the target movement position signal Pd.
  • the target value filter 35 consists of the transfer function G(s) of equation (1).
  • the transfer function G(s) is expressed in the form of partial fraction decomposition with T 1 , T 2 , T 3 , T 4 , C 1 , C 2 , C 3 and C 4 as coefficients and s as a differential element.
  • the transfer function G(s) of Equation (1) is set for each x-axis, y-axis and z-axis. In this way, the transfer function G(s) can be expressed as a superposition of first-order lag transfer functions.
  • the target value filter 35 multiplies the target movement position signal Pd of the load W by the transfer function G(s) to convert the target movement position signal Pd into a target trajectory signal Pd ⁇ .
  • a target operation amount calculation unit 36 configured by the boom position calculation unit 31b and the operation signal generation unit 31c of the control device 31 calculates each actuator from the attitude information of the crane 1, the current position information of the load W, and the target trajectory signal Pd ⁇ of the load W. to the target actuation amount.
  • the target actuation amount calculator 36 has the inverse dynamics model described above.
  • the target actuation amount calculator 36 is connected in series with the target value filter 35 .
  • the target operation amount calculation unit 36 calculates the target trajectory signal PD ⁇ calculated by the target value filter 35 using the inverse dynamics model, the current position coordinates p(n) of the load W calculated from the current position information of the load W, and from each sensor After a unit time t, using an inverse dynamics model based on the obtained attitude information of the crane 1 (turning angle ⁇ z(n), telescopic length lb(n), hoisting angle ⁇ x(n), extension amount l(n)) , the target position coordinate q(n+1) of the boom 9 is calculated (see FIG. 4). Next, the target actuation amount calculator 36 generates a target actuation signal Md for each actuator from the calculated target position coordinates q(n+1). The target actuation amount calculator 36 transmits the generated target actuation signal Md to each actuator.
  • the failure sign detection system 37 is a system that detects a sign of failure of each actuator of the crane 1 .
  • the failure sign detection system 37 is configured in the actuation signal generator 31 c of the control device 31 .
  • the failure sign detection system 37 is composed of a physical model section 38 , a learning model section 40 and a failure prediction section 42 .
  • the physical model unit 38 generates an ideal output signal Mdi that is the actuation amount of each actuator with respect to the target actuation signal Md in the crane 1 in an ideal state.
  • the crane 1 in an ideal state means the crane 1 in a state in which all actuators, movable parts, structures, electrical systems, control devices, etc. of the crane 1 are free from defects and can operate normally.
  • the physical model unit 38 has an inverse dynamics model 39, which is a physical model in which n subsystems represent a plurality of characteristics including at least one model transfer function that indicates the characteristics of at least one actuator of the crane 1, for example. is doing.
  • the inverse dynamics model 39 is constructed as a physical model having properties of the crane 1 in an ideal state.
  • the inverse dynamics model 39 is a physical model that mathematically represents the characteristics of the actuator of the crane 1 during normal operation.
  • the physical model unit 38 uses the inverse dynamics model 39 to generate the ideal output signal Mdi of each actuator from the target actuation signal Md of the load W.
  • the characteristics of the crane 1 include the characteristics of at least one actuator that constitutes the crane 1 .
  • the characteristics of the crane 1 may include not only the characteristics of at least one actuator that constitutes the crane 1, but also the characteristics of the structure that constitutes the crane 1, such as rigidity and friction.
  • the physical model unit 38 is connected in parallel downstream of the target actuation amount calculation unit 36 .
  • the physical model unit 38 can acquire the target actuation signal Md generated by the target actuation amount calculation unit 36 . That is, the physical model unit 38 can obtain the target actuation signal Md, which is the input signal generated by the control device 31 .
  • the physical model unit 38 is connected to the learning model unit 40 .
  • the physical model unit 38 can acquire a correction signal Mdc, which will be described later, from the learning model unit 40 .
  • the inverse dynamics model 39 of the physical model unit 38 includes a plurality of first subsystem SM1, second subsystem SM2, third subsystem SM3, . SMn are coupled in parallel.
  • the inverse dynamics model 39 represents the characteristics of the crane 1 in an ideal state.
  • the physical model unit 38 generates an ideal output signal Mdi, which is the output signal of the inverse dynamics model 39 for the signal obtained by adding the target actuation signal Md and the correction signal Mdc.
  • the model representing the characteristics of the crane 1 in an ideal state is composed of the inverse dynamics model 39, but may be represented by a secondary transfer function or the like.
  • the physical model unit 38 calculates the target operation signal Md obtained from the target operation amount calculation unit 36, the current position coordinates p(n) of the load W calculated from the current position information of the load W, and the attitude information of the crane 1 obtained from each sensor. Based on (turning angle ⁇ z(n), telescopic length lb(n), hoisting angle ⁇ x(n), extension amount l(n)), using the inverse dynamics model 39, the target position of the boom 9 after the unit time t has elapsed. A coordinate q(n+1) is calculated (see FIGS. 3 and 4). Next, the physical model unit 38 generates an ideal output signal Mdi for each actuator from the calculated target position coordinates q(n+1). The physical model unit 38 transmits the generated ideal output signal Mdi to the learning model unit 40 .
  • the learning model unit 40 has a learning model (learning type inverse dynamics model 41 described later) that learns the characteristics of the actuator by adjusting the weighting factor based on the difference between the output of the physical model unit 38 and the output of the actuator. .
  • the learning model unit 40 also corrects the input to the physical model unit 38 based on the output of the learning model (learning inverse dynamics model 41).
  • the learning model unit 40 detects the current state of the crane 1 with respect to the crane 1 in the ideal state.
  • the learning model unit 40 has, for example, a learning inverse dynamics model 41 that is a learning model in which n subsystems represent a plurality of characteristics including at least one model transfer function that indicates the characteristics of at least one actuator of the crane 1. have.
  • a weighting factor which will be described later, is set for each of the model transfer functions.
  • the learning inverse dynamics model 41 is configured as a learning model capable of approximating the characteristics of at least one actuator of the current crane 1 .
  • the learning model unit 40 uses the learning inverse dynamics model 41 to learn the learning inverse dynamics model 41 so as to reduce the difference between the actual output signal Mdr and the ideal output signal Mdi of the crane 1 .
  • the learning model unit 40 also uses a learning inverse dynamics model 41 to calculate the output of each actuator.
  • the learning model unit 40 converts the output of each actuator into an input signal of each actuator by an inverse transfer function 40a, and outputs a correction signal Mdc for correcting the target actuation signal Md.
  • the learning model unit 40 functions as an inverse system that converts output to input.
  • the learning model unit 40 is connected in parallel downstream of the target actuation amount calculation unit 36 .
  • the learning model unit 40 acquires the target actuation signal Md generated by the target actuation amount calculation unit 36 . That is, the learning model unit 40 acquires the target actuation signal Md, which is the input signal generated by the control device 31 .
  • the learning model unit 40 is also connected to the physical model unit 38 .
  • the learning model section 40 acquires the ideal output signal Mdi from the physical model section 38 .
  • the learning model unit 40 is connected to each sensor of the crane 1 .
  • the learning model unit 40 acquires the actual output signal Mdr from each sensor of the crane 1 .
  • the learning inverse dynamics model 41 a plurality of first subsystem SM1, second subsystem SM2, third subsystem SM3, . . . n-th subsystem SMn are connected in parallel.
  • the learning inverse dynamics model 41 has a weighting factor w 1 for the first subsystem SM1, a weighting factor w 2 for the second subsystem SM2, a weighting factor w 3 for the third subsystem SM3 . . . and the nth subsystem SMn. is set with a weighting factor wn .
  • the learning inverse dynamics model 41 receives the target actuation signal Md, which is the input signal of the learning inverse dynamics model 41, from the first subsystem SM1, the second subsystem SM2, the third subsystem SM3 , . . . Enter System SMn. Furthermore, the learning type inverse dynamics model 41 applies the first subsystem SM1, the first subsystem SM1 , Multiply the weighting factors w 1 , w 2 , w 3 . . . and wn for each of the second subsystem SM2 , the third subsystem SM3 .
  • the learning model unit 40 adjusts the weighting coefficients w 1 , w 2 , w 3 .
  • the learning model unit 40 adjusts the weighting coefficients w 1 , w 2 , w 3 . .
  • the learning model unit 40 can learn the learning type inverse dynamics model 41 having at least one actuator characteristic of the crane 1 that has changed due to secular change or the like. Further, the learning model unit 40 uses weighting factors w 1 , w 2 , w 3 , . and wn change amount.
  • the learning model unit 40 acquires the target actuation signal Md from the target actuation amount calculation unit 36 .
  • the learning model section 40 acquires the ideal output signal Mdi from the physical model section 38 .
  • the learning model unit 40 acquires the actual output signal Mdr from each sensor of the crane 1 .
  • the learning model unit 40 adjusts the weighting coefficients w 1 , w 2 , w 3 . . . and wn of each subsystem based on the difference between the ideal output signal Mdi and the actual output signal Mdr .
  • the learning inverse dynamics model 41 is composed of a plurality of subsystems with clear physical characteristics.
  • the learning inverse dynamics model 41 can be regarded as a one-layer neural network by multiplying the outputs from a plurality of subsystems by respective weighting factors.
  • the learning inverse dynamics model 41 independently adjusts the weighting coefficients w 1 , w 2 , w 3 , .
  • the control system 34 since the control system 34 generates the target trajectory signal PD ⁇ input to the target actuation amount calculator 36 through the target value filter 35, which is a fourth-order low-pass filter, a singular point occurs in the target trajectory signal Pd ⁇ . is suppressed. Therefore, the control system 34 inputs the target actuation signal Md generated by the target actuation amount calculation unit 36 from the target trajectory signal Pd ⁇ with the suppressed singularity to the learning inverse dynamics model 41, so that the learning inverse dynamics The learning convergence of the model 41 is facilitated. As a result, when the control system 34 controls each actuator with the load W as a reference, it is possible to move the load W in a manner intended by the operator while suppressing the swing of the load W.
  • the target value filter 35 which is a fourth-order low-pass filter
  • control system 34 generates the target trajectory signal Pd ⁇ input to the learning inverse dynamics model 41 through the target value filter 35, which is a fourth-order low-pass filter.
  • the control system 34 only needs to have a low-pass filter in which the differential element included in the learning inverse dynamics model 41 diverges and the learning by the learning inverse dynamics model 41 does not destabilize.
  • the control system 34 may, for example, have a transfer function configured as a first order or higher low pass filter.
  • the learning model unit 40 calculates the target operation signal Md obtained from the target operation amount calculation unit 36, the current position coordinates p(n) of the load W calculated from the current position information of the load W, and the attitude information of the crane 1 obtained from each sensor. Using the learning inverse dynamics model 41 from (swing angle ⁇ z(n), telescopic length lb(n), hoisting angle ⁇ x(n), extension amount l(n)), the boom 9 after the unit time t has passed. A target position coordinate q(n+1) is calculated (see FIGS. 3 and 4). Next, the learning model unit 40 generates a correction signal Mdc for correcting the target actuation signal Md from the calculated target position coordinate q(n+1). The learning model unit 40 adds the generated correction signal Mdc to the target actuation amount calculation unit 36 acquired by the physical model unit 38 .
  • the failure prediction unit 42 detects a sign of failure of each actuator of the crane 1 .
  • the failure prediction unit 42 acquires the operating time of the crane 1, the operating time of each actuator, and the like.
  • the failure prediction section 42 is connected to the learning model section 40 .
  • the failure prediction unit 42 acquires the weight coefficients w 1 , w 2 , w 3 .
  • the failure prediction unit 42 records the acquired weighting coefficients w 1 , w 2 , w 3 . . . and wn of each subsystem. In other words, the control system 34 continuously records the weight coefficients w 1 , w 2 , w 3 .
  • the failure prediction unit 42 predicts the failure location, failure cause, and failure time based on the weighting coefficients w 1 , w 2 , w 3 . It has a database D for The database D stores failure determination information, which is information serving as a reference for determining signs of actuator failure. The failure prediction unit 42 determines whether or not a sign of failure is included based on the values of the acquired weighting coefficients w 1 , w 2 , w 3 . to judge whether That is, the failure prediction unit 42 compares the weighting coefficients acquired from the learning model unit 40 with the failure determination information stored in the database D to determine whether there is a sign of actuator failure. Further, the failure prediction unit 42 predicts the failure location, failure cause, and failure timing.
  • the control system 34 configured in this way converts the target movement speed signal Vd acquired from the load moving operation tool 32 into the target movement position signal Pd of the load W by the integrator 32a.
  • the control system 34 converts the converted target movement position signal Pd into a target trajectory signal Pd ⁇ of the load W by a target value filter 35 .
  • the control system 34 generates the target actuation signal Md by the target actuation amount calculation unit 36 from the converted target trajectory signal Pd ⁇ .
  • the control system 34 sends the generated target actuation signal Md to each actuator.
  • the control system 34 also transmits the converted target actuation signal Md to the physical model section 38 and the learning model section 40 of the failure sign detection system 37 .
  • the physical model unit 38 generates an ideal output signal Mdi for each actuator from the obtained target actuation signal Md.
  • the learning model unit 40 acquires the ideal output signal Mdi and the real output signal Mdr, and based on the difference between the ideal output signal Mdi and the real output signal Mdr, weight coefficients w 1 , w 2 , w 3 ⁇ of each subsystem. . . . and w n are adjusted. Furthermore, the learning model unit 40 generates a correction signal Mdc that corrects the target actuation signal Md.
  • the learning model unit 40 adds the generated correction signal Mdc to the target actuation signal Md sent to the physical model unit 38 . That is, the learning model unit 40 corrects the target actuation signal Md using the correction signal Mdc generated by the learning inverse dynamics model 41 so that the difference between the ideal output signal Mdi and the actual output signal Mdr becomes zero.
  • the learning model unit 40 sets the weight coefficients w 1 , w 2 , w 3 ⁇ . . . and w n are adjusted.
  • the learning type inverse dynamics model 41 changes due to secular change or the like by changing the weighting coefficients w 1 , w 2 , w 3 . It learns the characteristics of the actual crane 1 that has been tested.
  • the failure prediction unit 42 of the failure sign detection system 37 acquires and records the weighting coefficients w 1 , w 2 , w 3 . . . and wn of each subsystem from the learning model unit 40 every unit time t.
  • the failure prediction unit 42 compares the weight coefficients w 1 , w 2 , w 3 . Based on the data base D, it is determined whether or not a sign of failure is included based on the value of wn , the degree of change in the value, and the tendency of change.
  • the failure prediction unit 42 predicts a failure part, a failure cause, a failure timing, etc., when determining that a sign of failure is included. In addition, the failure prediction unit 42 determines the maintenance timing for parts having signs of failure.
  • the failure prediction unit 42 predicts that the n-th subsystem SMn It is determined that the indicated actuator contains a sign of failure. Further, the failure prediction unit 42 predicts the cause of failure and the time of failure from the degree of change in the time series of the weighting factor wn and the trend of change.
  • a failure prediction unit 42 determines a time when the operating state of the crane 1 is not affected and the time when the maintenance cost is the lowest as the maintenance time based on the normal degree of deterioration due to secular change and the predicted time of failure.
  • the failure prediction unit 42 may store information in the database D that serves as a reference for making such determination. The failure prediction unit 42 notifies the operator or the like of information on failure and information on maintenance.
  • the failure sign detection system 37 configured in this manner detects a sign of failure of each actuator from changes in each weighting factor w 1 , w 2 , w 3 . . . and wn by the failure prediction unit 42 .
  • the failure sign detection system 37 detects a sign of failure from the transition of each weighting factor. That is, the failure sign detection system 37 obtains each weighting factor w 1 , w 2 , w 3 , . and wn . As a result, the failure sign detection system 37 can detect a failure sign without being affected by the operating state of the crane 1 .
  • the failure sign detection system 37 also has a learning inverse dynamics model 41 in which weighting factors w 1 , w 2 , w 3 . . . and wn are set for each sub-model.
  • the failure sign detection system 37 detects the learning type reverse power of the crane 1 in which each weighting factor w 1 , w 2 , w 3 . . .
  • a correction signal Mdc for each actuator is calculated by the mathematical model 41 .
  • the failure sign detection system 37 learns the characteristics of each actuator of the crane 1 while calculating the operation amount of each actuator by the learning type inverse dynamics model 41 with the crane 1 in the ideal state as a reference. As a result, the failure sign detection system 37 can detect a failure sign without being affected by the operating state of the crane 1 .
  • control system 34 including the failure sign detection system 37 can perform highly robust control by learning the characteristics of the crane 1 .
  • the control system 34 can move the load W in a manner intended by the operator while suppressing the swinging of the load W due to changes in the characteristics of the crane 1 .
  • the failure sign detection system 37 records each weighting factor w 1 , w 2 , w 3 . Changes in the operating state of each actuator under operating conditions are continuously accumulated as changes in each weighting factor w 1 , w 2 , w 3 . . . and wn .
  • the failure sign detection system 37 indicates a tendency of change, a degree of change, etc. of each weighting factor w 1 , w 2 , w 3 . . . If the degree of change exceeds the range of the degree of change, it is determined that a sign of failure is included, and the failure site, failure cause, failure time, etc. are predicted. As a result, the failure sign detection system 37 detects a sign of failure from the information of the accumulated weighting factors w 1 , w 2 , w 3 . . . be able to.
  • the crane 1 equipped with the failure sign detection system 37 continues changes in the operating state of each actuator under various operating conditions through each weighting factor w 1 , w 2 , w 3 . . . detected. Thereby, the crane 1 can detect a sign of failure by the failure sign detection system 37 without being affected by the operating state.
  • control system 34 controls each actuator by the target actuation signal Md generated by the target actuation amount calculator 36 from the target trajectory signal Pd ⁇ .
  • control system 34 may be configured to correct the target actuation signal Md by the feedback control section 43, the feedforward control section 45, and the like.
  • control system 34 further includes a feedback control section 43 and a feedforward control section 45.
  • the feedback control section 43 generates a feedback correction signal Md1 for each actuator based on the difference between the target actuation signal Md and the actual output signal Mdr.
  • the feedback controller 43 has a feedback controller 44 that corrects the target actuation signal Md.
  • the feedback controller 44 is connected in series with the target actuation amount calculator 36 .
  • the feedback control unit 43 acquires the target actuation signal Md of the load W from the target actuation amount calculation unit 36 . Also, the feedback control unit 43 acquires the actual output signal Mdr from each sensor of the crane 1 . The feedback control unit 43 feeds back (negatively feeds back) the acquired actual output signal Mdr to the acquired target actuation signal Md. The feedback control unit 43 corrects the target actuation signal Md by the feedback controller 44 based on the difference between the actual output signal Mdr and the target actuation signal Md, and calculates the feedback correction signal Md1.
  • the feedforward control unit 45 generates a feedforward correction signal Md2 for each actuator from the target trajectory signal Pd ⁇ .
  • the feedforward controller 45 has a feedforward controller 46 that corrects the target actuation signal Md. .
  • Feedforward controller 46 is connected in parallel with feedback controller 44 .
  • the feedforward control unit 45 uses the feedforward controller 46 to generate the feedforward correction signal Md2 for each actuator from the target trajectory signal Pd ⁇ obtained from the target value filter 35 .
  • the feedforward control unit 45 adds the generated feedforward correction signal Md2 to the feedback correction signal Md1.
  • the control system 34 transmits the feedback correction signal Md1 calculated by the feedback control section 43 and the feedforward correction signal Md2 calculated by the feedforward control section 45 to each actuator of the crane 1 . After transmitting the feedback correction signal Md1 and the feedforward correction signal Md2 to each actuator, the control system 34 uses the feedback control unit 43 to subtract the actual output signal Mdr detected by each sensor of the crane 1 from the target actuation signal Md.
  • FIG. 7 is a block diagram showing the control configuration of the control system according to the second embodiment of the invention.
  • the crane 1A according to the following embodiment refers to the same thing as the crane 1 shown in FIGS. 1 to 6 by using the names, figure numbers, and symbols used in the description. Therefore, in the following embodiments, specific descriptions of the same points as those of the already described embodiments will be omitted, and differences will be mainly described.
  • a control system 47 including a failure sign detection system 48 of the crane 1A includes a load moving operation tool 32, a turning sensor 27, an undulating sensor 30, an extension sensor 28, a swivel base camera 7b (FIG. 2). ), a target value filter 35 included in the control device 31 , a target actuation amount calculator 36 and a failure sign detection system 48 .
  • the control system 47 generates the target actuation signal Md through the cooperation of the target trajectory calculation section 31a, the boom position calculation section 31b, and the actuation signal generation section 31c of the control device 31 .
  • the learning model unit 40 of the failure sign detection system 48 is connected downstream of the target operation amount calculation unit 36 .
  • the learning model unit 40 can add the generated correction signal Mdc to the target actuation signal Md acquired by the physical model unit 38 . Further, the learning model unit 40 can add the correction signal Mdc to the target actuation signal Md input from the target actuation amount calculation unit 36 to each actuator. That is, the learning model unit 40 controls the crane 1A by adding the generated correction signal Mdc not only to the physical model unit 38 but also to the target actuation signal Md, which is the input signal of the crane 1A.
  • the failure sign detection system 48 converts the correction signal Mdc calculated to reduce the difference between the ideal output signal Mdi and the actual output signal Mdr into the target actuation signal Md, which is the input signal input to each actuator. Add together.
  • the failure sign detection system 48 functions as a control device that controls each actuator of the crane 1A. That is, the failure sign detection system 48 corrects the target actuation signal Md with the calculated correction signal Mdc, thereby approximating the actual output signal Mdr of the crane 1A to the ideal output signal Mdi of the crane 1A in the ideal state, while the learning model unit A sign of failure of each actuator is detected from changes in each weighting factor w 1 , w 2 , w 3 . . . Thereby, the failure portent detection system 48 can detect a failure portent while maintaining the normal operating state of the crane 1A.
  • FIG. 8 is a block diagram showing the control configuration of the control system according to the third embodiment of the invention.
  • the control system 49 including the failure sign detection system 50 of the crane 1B includes the load moving operation tool 32, the turning sensor 27, the hoisting sensor 30, the extension sensor 28, the swivel base camera 7b (FIG. 2). ), a target value filter 35 , a target actuation amount calculator 36 , a feedback controller 43 , a feedforward controller 45 and a failure sign detection system 50 included in the controller 31 .
  • the target trajectory calculator 31a, the boom position calculator 31b, and the actuating signal generator 31c of the control device 31 cooperate to generate the target actuating signal Md.
  • the control system 49 converts the feedforward controller 46 of the feedforward controller 45 into a learning inverse having a plurality of n-th subsystems SMn to which weighting factors w 1 , w 2 , w 3 . . . and wn are assigned respectively. It is structured as a dynamics model. In this case, in the control system 49 , the feedforward control unit 45 also functions as a learning model unit for the failure sign detection system 48 .
  • the feedforward control unit 45 obtains the feedback correction signal Md1 obtained from the feedback controller 44 .
  • the feedforward control section 45 acquires the ideal output signal Mdi from the physical model section 38 . Also, the feedforward control unit 45 acquires the actual output signal Mdr from each sensor of the crane 1 .
  • the feedforward control unit 45 adjusts the weighting factors w 1 , w 2 , w 3 . . . and wn of each subsystem based on the difference between the ideal output signal Mdi and the actual output signal Mdr .
  • the feedforward control unit 45 transmits the weighting coefficients w 1 , w 2 , w 3 .
  • the feedforward control unit 45 uses the feedforward controller 46 to calculate the output of each actuator. Further, the feedforward control section 45 converts the output of each actuator into an input signal of each actuator by an inverse transfer function 45a, and outputs a feedforward correction signal Md2 as a correction signal for correcting the feedback correction signal Md1. Also, the feedforward control unit 45 transmits a feedforward correction signal Md2 to the physical model unit 38 . Thereby, the failure sign detection system 48 can detect a sign of failure using the feedforward control unit 45 .
  • the failure prediction units 42 of the failure sign detection systems 37, 48, 50 record the acquired weighting factors w 1 , w 2 , w 3 .
  • the failure sign detection systems 37, 48, 50 may transmit to an external server via radio or the like.
  • the failure prediction unit 42 is configured to collectively transmit the acquired weighting coefficients w 1 , w 2 , w 3 . . . there is Further, the failure prediction unit 42 may be configured to acquire information on signs of failure from the server. With this configuration, the failure prediction unit 42 determines whether there is a sign of failure based on the weighting factors w 1 , w 2 , w 3 . . . and wn of the plurality of cranes 1 stored in the server.
  • the failure sign detection system 48 detects failures based on the accumulated weight coefficients w 1 , w 2 , w 3 . can be detected with higher accuracy.
  • the failure sign detection systems 37/48/50 are included in the control devices 31 of the cranes 1, 1A, and 1B. However, the failure sign detection systems 37, 48, 50 may be provided in portable terminals capable of remotely controlling the cranes 1, 1A, 1B.
  • the failure sign detection system is a failure sign detection system that detects a sign of failure of at least one actuator in a working vehicle provided with a boom that can be raised and lowered via a swivel on a traveling body. be.
  • One aspect of the failure sign detection system is a physical model unit that includes a physical model having characteristics of a work vehicle in an ideal state, acquires an actuator input signal generated by a work vehicle control device, and calculates an ideal output signal for the input signal in the physical model;
  • a learning model that includes at least one model transfer function that represents characteristics of an actuator in the work vehicle, has a weighting factor set in the model transfer function, and multiplies an output signal for an input signal calculated by the model transfer function by the weighting factor. , obtain the input signal, the ideal output signal, and the actual output signal of the actuator operated by the input signal, adjust the weighting factor based on the difference between the ideal output signal and the actual output signal, and obtain the ideal output signal and the actual output signal.
  • a learning model unit that generates a correction signal that reduces the difference from the output signal and adds the correction signal to the input signal that is input to the physical model unit; a failure prediction unit that acquires the weighting coefficients adjusted by the learning model unit and detects a sign of failure of the actuator from changes in the weighting coefficients.
  • one aspect of the failure sign detection system is correction for reducing the difference between the ideal output signal for the actuator input signal calculated by the physical model unit and the actual output signal for the actuator input signal.
  • a signal is generated by the learning model part.
  • the learning model unit generates a correction signal using the weighting factor as a variable.
  • the failure sign detection system detects a sign of failure of each actuator from changes in weighting coefficients by the failure prediction unit.
  • the failure predictor system configured in this way detects a predictor of failure from the transition of the weighting factor changed so as to bring the actual output signal closer to the ideal output signal.
  • the failure sign detection system continuously detects changes in the operating state of the actuator under various operating conditions during operation of the work vehicle through weighting factors. As a result, the failure sign detection system can detect a failure sign without being affected by the operating state of the work vehicle.
  • one aspect of the failure sign detection system preferably includes the following configuration.
  • the learning model unit controls the work vehicle by adding the correction signal to the input signal input to the actuator.
  • one aspect of the failure sign detection system adds the correction signal calculated to reduce the difference between the ideal output signal and the actual output signal to the input signal input to the actuator.
  • the failure sign detection system functions as a control device that controls the actuators of the work vehicle.
  • the failure sign detection system corrects the input signal using the calculated correction signal to approximate the output of the actual work vehicle to the output of the work vehicle in an ideal state, and detects the sign of failure of the actuator from the change in the weighting factor. To detect.
  • the failure sign detection system can detect a sign of failure while maintaining a normal operating state of the work vehicle.
  • one aspect of the failure sign detection system preferably includes the following configuration.
  • the learning model section is configured as a feedforward control section that inputs the correction signal to the actuator as a feedforward correction signal.
  • the failure sign detection system of the present invention adds the correction signal calculated to reduce the difference between the ideal output signal and the actual output signal to the input signal input to the actuator as the feedforward correction signal.
  • the failure sign detection system functions as a feedforward control section that controls the actuators of the work vehicle.
  • the failure sign detection system corrects the input signal using the calculated correction signal to approximate the output of the actual work vehicle to the output of the work vehicle in an ideal state, and detects the sign of failure of the actuator from the change in the weighting factor. To detect.
  • the failure sign detection system can detect a sign of failure while maintaining a normal operating state of the work vehicle.
  • the correction signal is an actuator input signal obtained by transforming the actuator output signal multiplied by the weighting factor using an inverse transfer function.
  • the failure sign detection system of the present invention has a learning inverse system with weighting factors set.
  • the failure sign detection system generates a correction signal for the actuator by an inverse system in which the weighting factor is adjusted according to the difference between the ideal output signal and the actual output signal. That is, the failure sign detection system learns the characteristics of the work vehicle from the input signal of the actuator, and inputs the output signal of the actuator based on the learning model to the actuator as the correction signal of the actuator.
  • the failure sign detection system can be incorporated into the control device of the work vehicle with a simple configuration.
  • one aspect of the failure sign detection system preferably includes the following configuration.
  • the failure prediction unit records or transmits the acquired weighting factors to an external server every unit time.
  • the weighting coefficient is recorded or transmitted to an external server for each unit time, so that the actuator can be operated under various operating conditions while the work vehicle is in operation. State changes are continuously accumulated as weighting factor changes. As a result, the failure sign detection system can detect a failure sign from the accumulated weighting coefficient information without being affected by the operating state of the work vehicle.
  • one aspect of the failure sign detection system preferably includes the following configuration.
  • the failure prediction unit predicts the maintenance time or failure time of the corresponding actuator from the change in the acquired weighting factor.
  • the tendency of change, the degree of change, or the like of the weighting coefficient exhibits a tendency of change that differs from the normal range of tendency, or If the degree of change exceeds the range, it is determined to include a sign of failure.
  • the failure sign detection system detects the presence or absence of a sign of failure, the location of the failure, and the timing of the failure based on the accumulated weighting data.
  • the failure sign detection system determines, as the maintenance timing, the timing when the operational state of the work vehicle is not affected and the maintenance cost is lowest based on the normal deterioration due to aging and the predicted failure timing. A sign of failure can be detected without being affected by the operating state of the work vehicle.
  • One aspect of the work vehicle according to the present invention is a work vehicle in which a traveling body is provided with a boom that can be raised and lowered via a swivel base, and is equipped with a failure sign detection system.
  • the work vehicle configured in this manner continuously detects changes in the operating state of the actuator under various operating conditions through the weighting coefficients of the learning model section in the failure sign detection system. As a result, the work vehicle can detect a sign of failure without being affected by the operating state by the failure sign detection system.
  • the present invention is applicable not only to mobile cranes but also to various work vehicles.

Landscapes

  • Engineering & Computer Science (AREA)
  • Mechanical Engineering (AREA)
  • Mining & Mineral Resources (AREA)
  • Civil Engineering (AREA)
  • General Engineering & Computer Science (AREA)
  • Structural Engineering (AREA)
  • Control And Safety Of Cranes (AREA)
PCT/JP2022/009436 2021-03-05 2022-03-04 故障予兆検出システムおよび作業車 Ceased WO2022186382A1 (ja)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2023503973A JP7501777B2 (ja) 2021-03-05 2022-03-04 故障予兆検出システムおよび作業車

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2021-035622 2021-03-05
JP2021035622 2021-03-05

Publications (1)

Publication Number Publication Date
WO2022186382A1 true WO2022186382A1 (ja) 2022-09-09

Family

ID=83154499

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2022/009436 Ceased WO2022186382A1 (ja) 2021-03-05 2022-03-04 故障予兆検出システムおよび作業車

Country Status (2)

Country Link
JP (1) JP7501777B2 (https=)
WO (1) WO2022186382A1 (https=)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000513097A (ja) * 1996-06-24 2000-10-03 アルチェリク・アノニム・シルケト 電気モータ用のモデル・ベースの故障検出システム
JP2004260891A (ja) * 2003-02-25 2004-09-16 Yaskawa Electric Corp モータの制御装置及び制御方法
WO2018220751A1 (ja) * 2017-05-31 2018-12-06 株式会社日立製作所 状態監視装置、並びに機器システム

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000513097A (ja) * 1996-06-24 2000-10-03 アルチェリク・アノニム・シルケト 電気モータ用のモデル・ベースの故障検出システム
JP2004260891A (ja) * 2003-02-25 2004-09-16 Yaskawa Electric Corp モータの制御装置及び制御方法
WO2018220751A1 (ja) * 2017-05-31 2018-12-06 株式会社日立製作所 状態監視装置、並びに機器システム

Also Published As

Publication number Publication date
JPWO2022186382A1 (https=) 2022-09-09
JP7501777B2 (ja) 2024-06-18

Similar Documents

Publication Publication Date Title
JP7172243B2 (ja) クレーンおよびクレーンの制御システム
JP7069888B2 (ja) クレーンおよびクレーンの制御方法
JP7119674B2 (ja) クレーン
EP3822221B1 (en) Crane and crane control method
KR20110004776A (ko) 로드 케이블에 매달린 하물을 취급하기 위한 크레인
US7831333B2 (en) Method for the automatic transfer of a load hanging at a load rope of a crane or excavator with a load oscillation damping and a trajectory planner
JP7176645B2 (ja) 制御システムおよびクレーン
WO2020196808A1 (ja) クレーンの制御方法およびクレーン
WO2020196809A1 (ja) クレーンの制御方法およびクレーン
US12600603B2 (en) Crane, crane characteristic change determination device, and crane characteristic change determination system
JP7172256B2 (ja) クレーン
WO2022186382A1 (ja) 故障予兆検出システムおよび作業車
EP4082957B1 (en) Work machine control system and crane
JP2023081639A (ja) クレーン、サーバ、及び出力復元システム
EP4606756A1 (en) Crane
JP5642326B2 (ja) クレーン又はバガーで吊り荷ロープに吊り下げられている吊り荷を自動的に積み替える方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22763432

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2023503973

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22763432

Country of ref document: EP

Kind code of ref document: A1