JP6453075B2 - Method and apparatus for controlling steadying of trolley crane - Google Patents

Description

  The present invention relates to a control method and apparatus for detecting a swing angle of a suspended load generated during operation of a trolley crane and performing steadying control.

  There are various types of cranes used at harbors and construction sites, etc. Some bridge cranes, overhead cranes, some jib cranes, etc. traverse trolleys along horizontal jibs and girders. There is a so-called trolley crane that hangs a suspended load such as a grab bucket or a hook with a rope and transports it.

  In such a trolley crane, a swing along the moving direction of the trolley occurs in the suspended load as the trolley traverses. When the suspended load reaches the target position, if the suspended load is shaken, the transported object cannot be lifted and lowered to an accurate position until the shake is settled, and work efficiency is lowered. In addition, depending on the angle and size of the generated deflection, the suspended load and the rope may interfere with surrounding machines and structures, and it is desirable to suppress the suspended load as much as possible. In order to suppress the swing, it is effective to detect the swing angle of the suspended load and apply appropriate control (stabilization control) to the operation of the trolley or the suspended load based on the swing angle.

  As literature which described the technique for steadying control in such a trolley type crane, there exist the following patent documents 1-3, for example. In addition to detecting the swing angle and swing angular velocity with the swing angle detector, various numerical values related to the swing angle such as the position and speed of the trolley are measured, and based on these data, the swing angle is calculated using an observer and Kalman filter. The estimation is performed and the steadying control is performed based on the estimated deflection angle.

JP 2009-234699 A JP 2000-7274 A JP 2000-318973 A

  However, in actual operation of a trolley crane, it is practically difficult to directly detect the swing angle of a suspended load. For example, in the deflection angle control device described in Patent Document 1, an optical device that uses a trolley-side imaging device to photograph a marker attached to the suspended load side and detects the deflection angle by image processing is used. Such an optical deflection angle detector requires expensive equipment and is costly, and requires much labor to attach and detach the equipment, and also requires frequent maintenance. Also, depending on the weather and the surrounding environment, shooting and image processing may be difficult, and the shake angle may not be detected accurately.

  As another method for detecting the swing angle, for example, a rod that follows the hoisting rope is attached near the fulcrum on the base side of the hoisting rope, and the deflection angle of the hoisting rope is detected as the angle of the rod. is there. However, these mechanical deflection angle detectors still take time to attach and detach the equipment, and because there are many contact parts and sliding parts between the equipment and the hoisting rope, dirt is likely to accumulate, which makes detection possible. There is a concern that the accuracy will decrease.

  In addition to the problems related to the detection of the deflection angle described above, there are also problems associated with modeling. In other words, in order to achieve an accurate prediction value of the deflection angle and high control performance, the model to be controlled is often complicated, and handling such a complex model is costly and troublesome in design. Will increase. In addition, depending on how the model is set up, there is a risk that the robustness of the entire system against disturbances to trolleys and suspended loads, modeling errors, and the like may be reduced.

  In view of such circumstances, the present invention is intended to provide a trolley type crane steady-state control method and apparatus capable of suitably performing steady-state control on a suspended load while reducing cost and maintenance problems. is there.

The present invention relates to a trolley crane including a trolley that traverses a girder or a boom, a suspended load that is suspended from the trolley by a plurality of ropes, and a winch drum that drives the plurality of ropes and traverses the trolley. A steady rest control method, wherein a sheave load detector is attached to each sheave provided in the middle of the plurality of ropes to detect the tension of the rope, and an estimated deflection angle is determined from the tension detected for each sheave. And calculating a predicted swing angle of the suspended load by performing a process using the Kalman filter on the estimated swing angle, and performing feedback control for suppressing the swing of the suspended load on the trolley using the predicted swing angle. In addition to the predicted deflection angle, the feedback control includes the traverse position of the trolley, the traverse direction speed of the trolley, the suspension load of the suspended load. Deriving a gain scheduled H _∞ controller with up length, the gain scheduled H _∞ a controller by steadying the feedback control speed instruction for said trolley obtained, shake linearly combining the traverse operation speed command for the trolley stop the control speed instruction in which according to the steadying control method for a trolley crane, characterized in row Ukoto.

  In this way, the swing angle of the suspended load can be accurately predicted by a simple and inexpensive mechanism.

  In the steady control method for a trolley crane of the present invention, it is preferable to design the Kalman filter with respect to a mathematical model defined assuming that the trolley crane is a one-mass system model of a spring mass damper. A highly accurate deflection angle prediction value can be obtained with a simple mathematical model.

The present invention relates to a trolley crane including a trolley that traverses a girder or a boom, a suspended load that is suspended from the trolley by a plurality of ropes, and a winch drum that drives the plurality of ropes and traverses the trolley. A steady rest control method comprising: a sheave provided in the middle of each of the plurality of ropes; a sheave load detector attached to the sheave for detecting the tension of the rope; and a tension detected for each sheave. An estimated deflection angle is calculated from the estimated deflection angle, a Kalman filter is applied to the estimated deflection angle to calculate a predicted deflection angle of the suspended load, and the deflection of the suspended load relative to the trolley is calculated using the predicted deflection angle. and a controller configured to perform feedback control to suppress, control device detects the hanging rope length of the transverse position and the lifted load of the trolley Hanging with a load current position detector is, transverse position of the trolley, in addition to the deflection angle prediction value, transverse direction velocity of the trolley, derives the gain scheduled H _∞ controller with hanging rope length of the suspended load The trolley is controlled by a steady-state feedback control speed command for the trolley obtained by the gain schedule H _∞ controller and a steady-state control speed command for linearly connecting a transverse operation speed command for the trolley. The present invention relates to a steady rest control device for a trolley crane, which is characterized by

  In the sway control device for the trolley crane of the present invention, the Kalman filter may be designed with respect to a mathematical model defined by assuming that the trolley crane is a one-mass system model of a spring mass damper, and mounted on the control device. preferable.

  According to the trolley-type crane steady-state control method and apparatus of the present invention, it is possible to achieve an excellent effect that the steady-rest control can be suitably performed on a suspended load while reducing cost and maintenance problems.

It is a side view which shows an example of the grab bucket type unloader to which this invention is applied. It is a perspective view which shows an example of the grab bucket type unloader to which this invention is applied. It is a control block diagram in the Example of the steadying control method and apparatus of the trolley type crane of this invention. It is a control block diagram in the Example of the steadying control method and apparatus of the trolley type crane of this invention. It is the schematic which illustrates the tension concerning a sheave. It is explanatory drawing of the numerical formula model of a trolley type crane. It is a block diagram which converts a force input model into a speed input model. It is a flowchart which shows the process of the calculation in a Kalman filter. It is a diagram which shows by comparing the swing angle of the suspended load estimated by the implementation of the present invention and the actual swing angle. It is a block diagram for deriving a linear parameter variation system model. It is a block diagram which shows the generalized plant used for derivation _| leading-out of a _Hinfinity end point controller. It is a diagram which shows the behavior at the time of steadying operation in the experimental machine of the grab bucket type unloader to which the present invention is applied, (a) is the speed command for the trolley and the suspended load, (b) is the traversing position of the trolley, (c) Indicates the swing angle of the suspended load.

  Embodiments of the present invention will be described below with reference to the accompanying drawings.

  FIG. 1 and FIG. 2 show an example of an embodiment for implementing the trolley crane steadying control method and apparatus of the present invention, and illustrates the case where the present invention is applied to a grab bucket unloader.

  The grab bucket type unloader is a type of bridge crane provided on the quay to unload loose objects such as ore and coal loaded on a bulk carrier. The grab bucket type unloader shown in FIG. 1 includes a machine body 4 having a seaside limb 1 and a landside limb 2 and traveling on a rail 3 on a quay, and a land side above the machine body 4. A boom 6 which can be lifted from the girder 5 provided on the sea side around the pin 6 a, a trolley 7 which traverses along the longitudinal direction of the boom 6 and the girder 5, and is suspended from the trolley 7. It has a grab bucket 8 as a suspended load that can be lifted and lowered. Then, the grab bucket 8 opened from the trolley 7 located on the sea side of the boom 6 is suspended from the upper opening 9a of the bulk carrier 9 and placed on the bulk, and the grab bucket 8 is closed. The grab bucket 8 is moved up, and then the trolley 7 is moved to the land side to move the grab bucket 8 to the land side. The grab bucket 8 is provided in the machine body 4 with a hopper. By opening the door 10 when it reaches the top, the rose is put into the hopper 10. The bulk material thrown into the hopper 10 is supplied to the land conveyor 12 by an in-machine conveyor 11 provided in the machine body 4. In FIG. 2, reference numeral 13 denotes a mobile operation room in which an operator who operates the unloader is boarded, and reference numeral 14 denotes a machine room provided at the land side end of the upper part of the machine body 4.

  As the grab bucket type unloader as described above, for example, as shown in FIG. 2, four winch drums are provided, and the trolley 7 is traversed and the grab bucket 8 is moved up and down and opened and closed by driving the winch drum. There is a 4-drum unloader.

  The hoisting rope 16 fed out from the hoisting drum 15 as the winch drum is guided to the sheave 18 on the trolley 7 through the sheave 17 provided at the land side end of the girder 5 (see FIG. 1), and then downward. The lower end is directed and fixed to one side (land side) of the grab bucket 8. Further, the hoisting rope 16 ′ fed out from the hoisting drum 15 ′ as the winch drum is guided to the sheave 18 ′ on the trolley 7 through the sheave 17 ′ provided at the sea end of the boom 6 (see FIG. 1). After that, the lower end directed downward is fixed to the other side (sea side) of the grab bucket 8.

  The open / close rope 20 fed out from the open / close drum 19 as the winch drum is guided to the sheave 22 on the trolley 7 through the sheave 21 provided at the land side end of the girder 5 (see FIG. 1), and then downward. A lower moving sheave 23 that is guided and attached to the connecting portion of the bucket body 8a, 8a of the grab bucket 8, and an upper fixed sheave 24 that is attached to the upper frame 8c that supports the bucket body 8a by pin connection via a tie rod 8b (see FIG. 1), and is fixed to a required portion of the grab bucket 8. On the other hand, the open / close rope 20 ′ fed out from the open / close drum 19 ′ as the winch drum is guided to the sheave 22 ′ on the trolley 7 through the sheave 21 ′ provided at the sea end of the boom 6 (see FIG. 1). After that, it is guided downward and is hung a plurality of times between the lower moving sheave 23 of the grab bucket 8 and the upper fixed sheave 24 (see FIG. 1), and is fixed to a required portion of the grab bucket 8.

  In the four-drum unloader shown in FIG. 2, when the hoisting drums 15 and 15 ′ are stopped and the open and close ropes 20 and 20 ′ are simultaneously fed out by the open and close drums 19 and 19 ′, the lower moving sheave of the grab bucket 8 is obtained. 23 and the upper fixed sheave 24 (see FIG. 1) are opened, the grab bucket 8 is opened, and when the open / close ropes 20 and 20 ′ are simultaneously wound by the open / close drums 19 and 19 ′, the lower moving sheave 23 and the upper fixed sheave 24 The interval (see FIG. 1) becomes narrower and the grab bucket 8 is closed.

  Further, if the operation of feeding the hoisting ropes 16 and 16 ′ by the hoisting drums 15 and 15 ′ and the operation of feeding the opening and closing ropes 20 and 20 ′ by the opening and closing drums 19 and 19 ′ are simultaneously performed, the grab bucket 8 is lowered. Further, when the operation of winding the hoisting ropes 16 and 16 ′ by the hoisting drums 15 and 15 ′ and the operation of winding the opening and closing ropes 20 and 20 ′ by the opening and closing drums 19 and 19 ′ are performed simultaneously, the grab bucket 8 is raised. To do.

  On the other hand, a winding operation of the hoisting drum 15 and the opening / closing drum 19 leading the hoisting rope 16 and the opening / closing rope 20 from the land-side sheaves 17, 21 to the land-side sheaves 18, 22 on the trolley 7, The hoisting drum 15 'and the opening / closing drum 19' are led out simultaneously from the sheaves 17 ', 21' to the seaside sheaves 18 ', 22' of the trolley 7 with the hoisting rope 16 'and the opening / closing rope 20'. Then, the trolley 7 and the grab bucket 8 traverse to the land side. On the contrary, when the winding operation of the hoisting drum 15 and the opening / closing drum 19 and the winding operation of the hoisting drum 15 ′ and the opening / closing drum 19 ′ are performed simultaneously, the trolley 7 and the grab bucket 8 traverse to the sea side. That is, the trolley 7 and the grab bucket 8 can be traversed by operating the hoisting drums 15 and 15 ′ and the opening and closing drums 19 and 19 ′.

  In this embodiment, the swing angle of the suspended load (grab bucket) 8 is determined from the tension of each rope (the hoisting ropes 16, 16 ′ and the opening / closing ropes 20, 20 ′) detected by using the sheave load detector 25. In addition to estimation, the current position of the suspended load 8 is detected by the suspended load current position detector 26.

  The sheave load detector 25 is a load sensor such as a load cell attached to the hoist sheaves 17 and 17 ′ and the open and close sheaves 21 and 21 ′, and is detected in the four sheaves as shown in FIGS. The obtained load is input to the deflection angle calculation unit 28 of the control device 27 as a load signal 25a.

  The suspended load current position detector 26 traverses the trolley 7 and also operates a winch drum (winding ropes 16 and 16 'and opening and closing ropes 20 and 20') for raising and lowering and opening and closing the grab bucket 8. A rotation sensor such as an encoder for measuring the drum rotation speed of the drums 15 and 15 ′ and the open / close drums 19 and 19 ′), and the current load position calculation unit 29 of the control device 27 uses the measured drum rotation speed as a rotation speed signal 26a. To enter.

Further, by the controller 30 provided in the moving cab 13 operated by an operator, the operation signal 30a is input to the controller 27 as a traverse operation speed command V _ref and transverse position command X _ref to the trolley 7 It is like that. The traversing target position command X _ref can be calculated by integrating the traversing operation speed command V _ref .

The control device 27 estimates the swing angle θ of the grab bucket 8 based on the load signal 25 a input from the sheave load detector 25, and the estimated swing angle θ and the rotation input from the suspended load current position detector 26. We stopped to calculate the feedback control speed command u _fb shake on the basis of the transverse position x or the like of the suspension rope length l and trolley 7 of the suspended load 8 is calculated from the number signal 26a, shake the the stop feedback control speed command u _fb The steadying control speed command u linearly combined with the traverse operation speed command _Vref is output to the inverter 32 of the winch drum motor 31 as a control signal 27a. The inverter 32 controls the traversing operation of the winch drum via the motor 31 based on the control signal 27a.

  Next, details of the control in the present embodiment will be described with reference to FIGS.

Control device 27, which is converged to zero deflection angle of the load 8 suspended on the basis of the estimated swing angle theta _T transverse operating speed while traversing the trolley 7 on the basis of a command V _ref, the suspended load 8 to the trolley 7 It is.

First, the suspended load current position calculation unit 29 (see FIG. 4) of the control device 27 performs the traversing position (position detection value) of the trolley 7 based on the rotation speed signal 26a input from the suspended load current position detector 26. x _r , the traversing speed (speed detection value) V _{fb of} the trolley 7, and the distance (hanging rope length detection value) 1 from the trolley 7 to the suspended load (grab bucket) 8 are calculated. At this time, the traverse speed of the trolley 7 can be calculated as a differential value of the position of the trolley 7 in the traverse direction.

Next, the deflection angle calculation unit 28 of the control device 27 estimates the deflection angle θ of the suspended load 8 based on the sheave load signal 25 a input from the sheave load detector 25. Specifically, as shown in FIG. 5, the sheaves (winding sheaves 17, 17 ′, open / close sheaves 21, 21 ′) are each subjected to a tension of T _m1 , T _m2 , T _a1 , T _a2 . these tensions will vary according to the estimated swing angle theta _T of the suspended load 8. That is, the estimated deflection angle θ _T can be calculated as a function using the tensions T _m1 , T _m2 , T _a1 , and T _a2 as independent variables as follows.
θ _T = f (T _m1 , T _m2 , T _a1 , T _a2 )

Next, the controller 27 detects the position detection value x _r , the speed detection value V _fb , the suspension rope length detection value l calculated by the suspended load current position calculation unit 29, and the estimation calculated by the deflection angle calculation unit 28. the deflection angle theta _T input to the Kalman filter 33 (see FIG. 4). The Kalman filter 33 predicts the deflection angle of the load 8 suspended by a crane model 34 that defines inside of the Kalman filter 33, the deflection angles the prediction compared to the estimated swing angle theta _T, and its error variance minimum Thus, the system configuration is such that an optimum estimation result for the deflection angle is returned.

  Hereinafter, a design procedure of the Kalman filter 33 will be described. In order to design the Kalman filter 33, it is necessary to derive a mathematical model (crane model) 34 of a trolley crane (in the case of this embodiment, a grab bucket unloader) that is a control target.

The mathematical model 34 of the trolley crane is derived by the procedure described below. First, the model definition for deriving the mathematical model 34 of the trolley crane is assumed as follows.
a) Rope mass is not considered.
b) The trolley 7 is a one-mass system model of a spring mass damper that takes into account only the movement in the transverse direction (x direction).
c) The control input (stabilization control speed command) u acts directly on the trolley 7.

  Based on the above definition, a mathematical model 34 of a trolley crane is assumed as shown in FIG. The definitions of the main symbols are shown below.

M _t : trolley mass [kg], M _o : suspended load mass [kg], l (t): suspended rope length [m], u (t): control input [N], x _r (t): trolley 7 x-direction position (transverse position) [m], x _o (t): x-direction position [m] of the suspended load 8, y _o (t): y-direction position [m], θ (t) of the suspended load 8 ): Deflection angle [rad] of the suspended load 8; k _t : equivalent spring constant [N / m] of the trolley 7; c _t : equivalent damping coefficient [N · s / m] of the trolley 7; K _pp : inverter Equivalent speed gain, g: gravitational acceleration [m / s ² ], t: time [s]. The subscript t represents a trolley, and the subscript o represents a suspended load.

The kinetic energy T _t , potential energy V _t , and external force P _t applied to the trolley 7 are as follows. Here, the dot symbol (•) written at the top of the variable represents the derivative (the differential value with respect to time) of the variable. When two dot symbols are attached, it represents the second derivative of the variable.

Kinetic energy T _o , potential energy V _o , and external force P _o applied to the suspended load 8 are as follows.

The equation of motion is derived by applying the above [Equation 1] and [Equation 2] to the Lagrangian equation. When q _i is a generalized coordinate, the Lagrangian equation is as follows.

Ｍは２×２の質量行列、Ｃは２×２の減衰行列、Ｋは２×２の剛性行列、Ｆは２×１の入力行列である。 From the above [Equation 3], the following equation of motion is obtained.

M is a 2 × 2 mass matrix, C is a 2 × 2 attenuation matrix, K is a 2 × 2 stiffness matrix, and F is a 2 × 1 input matrix.

として展開すると、状態方程式および出力方程式は以下の通りとなる。

Furthermore, the state quantity x is

, The state equation and output equation are as follows.

Here, although the state equation is a force input, in the actual machine, the trolley 7 is driven by a traverse operation speed command via the inverter 32. Therefore, the speed control gain equivalent of the inverter 32 is set to _Kpp, and the speed control is realized by the block diagram shown in FIG. Therefore, the state equation of [Formula 6] can be expressed by the following state equation of the speed control input model.

The Kalman filter 33 is designed based on the output equation of [Equation 6] and the state equation of [Equation 7]. The trolley crane to be controlled by the Kalman filter 33 is given by the following equation of state and output, taking into account the process noise w and the observation noise v defined by the covariance matrix in [Equation 6] and [Equation 7]. .

The process noise w is an error between the behavior simulated by the mathematical model and the actual behavior. The magnitude of the process noise w is set by the covariance matrix Q. The covariance matrix Q is calculated by setting the standard deviation σ _w of each component of the process noise w.

  As factors of process noise, disturbances such as wind, errors due to modeling, errors due to discretization of motion equations, calculation errors in the control device, and the like can be considered.

The magnitude of the observation noise v is also set by the covariance matrix R. As in the case of the process noise w, the covariance matrix R is calculated by setting the standard deviation σ _v of each component of the observation noise v.

  Note that the cause of the observation noise may be electrical noise or drift inherent to the sensor, rope other than the one-mass pendulum vibration, and other mechanical vibrations.

  A control calculation executed inside the Kalman filter 33 will be described with reference to FIG. In executing the control calculation, first, as step S1, specific numerical values of each component are set for the process noise w, the observation noise v, and the covariance matrices Q and R thereof. These numbers are determined empirically.

の初期値

および初期共分散ＣｏｖＸ_０を以下の通り設定する。ここで、ハット記号（＾）は上記モデルを用いた予測値であることを表す。

Further, the state quantity of the mathematical model 34 inside the Kalman filter 33

Initial value of

And the initial covariance CovX ₀ is set as follows: Here, the hat symbol (^) represents a predicted value using the above model.

Next, as step S2, the coefficient matrix in the crane model 34 (see [Equation 8]) is discretized to derive discretized coefficient matrices A _t , B _t , and C _t . The subscript t in this calculation represents discrete time.

Next, as step S3, the internal state and the observed value after Δt time (t + Δt time) are predicted as follows from the estimation result of the internal state at the present time (t time). Δt represents a control calculation cycle. Here, not the traversing operation speed command V _{ref to} the trolley 7 but the sensor detection value V _fb that is the actual speed of the trolley 7 is used as the control input. The value of the above [Equation 12] is used as an internal state at the time of initial calculation.

Further, as step S4, the covariance between the internal state and the observed value after Δt time (t + Δt time) is predicted. Here, since each covariance includes process noise and observation noise, the predicted value of covariance is as follows.

In step S5, a Kalman gain K _{t + Δt} that is a weight of the correction amount of the internal state is calculated. This Kalman gain is composed of the covariance calculated in [Equation 15] and includes process noise.

As step S6, the estimated value of the internal state is updated by the following [Equation 17] from the observed value that predicted the internal state and the actual observed value. The estimated value updated here is substituted in step S3 of the next calculation cycle.

In step S7, the covariance of the internal state is updated by the following [Equation 18]. The estimated value updated here is substituted in step S4 of the next calculation cycle.

  When the calculation up to step S7 is completed, the process returns to step S2, and the steps from step S2 to S7 are repeated at the above calculation cycle while correcting the internal state and its covariance. Thereby, the deflection angle θ of the suspended load 8 can be accurately estimated in real time.

である。実線はシーブ荷重検出器（ロードセル）２５とは別のセンサにより検出した振れ角を示しており、これが真値であると仮定することができる。破線で示される振れ角予測値は、実線で示される真値とよく適合していることがわかる。このように、本実施例においては、巻上シーブ１７，１７'や開閉シーブ２１，２１'にかかる張力から吊り荷８の振れ角を推定し、さらにカルマンフィルタ３３による処理を行って吊り荷８の振れ角を予測しているので、上記特許文献１、２等に記載されているような複雑な機構や高価な機器を用いることなく、単純で安価な機器により吊り荷８の振れ角を精度良く予測することができる。しかも、カルマンフィルタ３３はトロリ式クレーンをばねマスダンパの１質点系モデルと仮定した数式モデル３４に対して設計されるので、簡単なクレーンモデル３４で精度の高い振れ角予測値を得ることができる。 FIG. 9 is a diagram showing comparison between the swing angle data predicted by the above process and the actual swing angle data in the grab bucket unloader experimental machine. The experimental machine used in this comparative verification test is a scale-down type produced by reducing the size of the actual grab bucket unloader. An estimated swing angle theta _T is the is based on the load signal 25a input from the sheave load detector 25 shown by a dashed line, deflection angle is shown by a broken line is predicted by the Kalman filter process from the estimated swing angle theta _T Predicted value

It is. A solid line indicates a deflection angle detected by a sensor different from the sheave load detector (load cell) 25, and it can be assumed that this is a true value. It can be seen that the predicted deflection angle indicated by the broken line is well matched with the true value indicated by the solid line. As described above, in this embodiment, the swing angle of the suspended load 8 is estimated from the tension applied to the hoist sheaves 17 and 17 ′ and the open and closed sheaves 21 and 21 ′, and further, the process by the Kalman filter 33 is performed. Since the deflection angle is predicted, the swing angle of the suspended load 8 can be accurately determined by a simple and inexpensive device without using a complicated mechanism or an expensive device as described in Patent Documents 1 and 2 above. Can be predicted. In addition, since the Kalman filter 33 is designed with respect to the mathematical model 34 in which the trolley crane is assumed to be a one-mass system model of a spring mass damper, it is possible to obtain a highly accurate deflection angle prediction value with a simple crane model 34.

Using the predicted value (the above [Equation 19]) of the deflection angle θ obtained by the above process using the Kalman filter 33, the gain corresponding to the variation of the suspended rope length l of the suspended load is applied to the crane model 34. The schedule H _∞ controller 35 (see FIG. 4) is derived, and the steady state feedback control speed command u _fb is generated by the gain schedule H _∞ controller 35 (see FIG. 4). The gain schedule _H∞ controller 35 is derived by a formalized method called an endpoint method in which an LTI (linear time invariant) endpoint controller is interpolated to perform scheduling.

The gain schedule H _∞ controller 35 derives an LPV (linear parameter variation system) model and obtains the LPV model using LMI (linear matrix inequality).

尚、

は零行列である。 Derivation of the LPV model will be described below. First, from [Equation 6] and [Equation 7], the state equation and the output equation of the speed control input model of the trolley crane that is the control target can be expressed by the following [Equation 20].

still,

Is a zero matrix.

Here, in designing the gain schedule _H∞ controller 35 based on LMI, there is a restriction that the control input matrix B _pp must be invariant. However, B _pp includes the suspended rope length l as a component, and this suspended rope length l is a variable parameter, and the above [Equation 20] cannot be used as it is in the gain schedule _H∞ controller 35. Therefore, a first-order low-pass filter (LPF) is added to the speed control input model of [Equation 20] as shown in FIG. Thus, the above [Equation 20] can be expressed by the state equation and output equation of the expanded system shown below, and the LPF coefficient matrix D _f = 0.
Thus, the input matrix B _S can be made unchanged.

Above, including the parameter variations of the system matrix A _s matrix B _pp large plants the number 22] and control design model to derive the LPV model for this.

System matrix A _S in the [Equation 22] is a variation component is a multiple of the parameter Z is (4,1) component of the matrix is divided into other constant component, expressed as follows.

Therefore, state equations of [Equation 22] is used system matrix _{A S} [Expression 23] can formula modified as follows Equation 24] as LPV model. Note that A _S0 is a matrix in which only constant terms are removed from the matrix A _S by removing components that vary depending on Z, and A _Z is a matrix that depends only on the coefficients of Z excluding A _S0 from the matrix A _S.

For the LPV model P _s (Z) defined by the above [Equation 24], an H _∞ end point controller K _{i (i = 1, 2)} is derived. In deriving the H _∞ endpoint controller K _i is first defined generalized plant as shown in Figure 11. W _T , W _s , and D _W are design parameters (frequency weights) for reflecting the control specifications in the design, and have the following meanings. W _T is a weighting to compensate for the modeling error is an error of the mathematical model and the actual machine. W _s (W _s1 , W _s2 ) is a weight for positioning of the trolley and steadying of the suspended load. _Dw is introduced to satisfy the assumption of standard _H∞ control. w ₁ and w ₂ indicate disturbance and virtual observation noise, respectively, and z ₁ and z ₂ indicate control amounts.

The derivation of the H _∞ end point controller K _i means that the H _∞ norm from the virtual disturbance w to the controlled variables z ₁ and z ₂ is represented by G (s) as the round trip transfer function of the generalized plant in FIG. It is to solve the following inequality (H _∞ control problem) that is less than 1.

Here, the transfer function from w to z ₁ is as follows.

T _m is called a complementary sensitivity function, and it is assumed that robust stability is satisfied by satisfying the following expression.

The transfer function from w to z ₂ is as follows.

S _m is called a settling function, and the smaller this is, the better the response. The controller is designed so as to satisfy the following equation where γ is the smallest possible constant.

From [Equation 26] to [Equation 29], [Equation 25] can be rewritten as follows.

With respect to the design parameters W _T and W _S , the controller K _i is uniquely derived with the values appropriately determined first, and the final W _T and W _S are determined by numerical simulation and experimental response of the closed loop system. . When W _T and W _S are determined, the continuous time system H _∞ end point controllers K ₁ and K ₂ are determined by the following formula by the design program. Therefore, the controllers K ₁ and K ₂ are mounted on the control device.

The controller K _i of the continuous time system obtained above is mounted on the control device 27, and using the parameter Z that depends on the suspended rope length l of the suspended load 8, linearity by convex interpolation of the following equation for each control period Gain scheduling is performed by interpolation. Thus, the gain schedule _H∞ controller K is obtained in real time.

Since the gain schedule _H∞ controller K obtained here is a continuous time controller, it is necessary to discretize every control cycle Δt of the control device simultaneously with the gain schedule calculation. A discrete time controller is obtained by Padé approximation and a desired discrete time gain schedule _H∞ controller _Kd is derived. By passing the discrete time controller _Kd , the steady-state feedback control speed command u _fb is generated.

In actual design, Matlab (registered trademark) of control design software can be used to derive the above-mentioned [Equation 31] continuous time H _∞ end point controller. Thereby, since the continuous time H _∞ end point controllers K ₁ and K _{2 of} [Equation 31] are obtained in the binary file format, this binary file is mounted on the control device 27, and the above calculation is performed in the control device 27 for each control cycle. As a result, the final discrete time controller K _d is obtained as the gain schedule H _∞ controller 35.

The steady-state feedback control speed command u _fb generated by the derivation of the gain schedule H _∞ controller 35 in this way is the traverse operation speed command V _ref to the trolley 7 in the linear combination controller 36 as shown in FIG. The steady-state control speed command u is generated by linear combination at a predetermined ratio. The steadying control speed command u is output from the control device 27 to the inverter 32 as a control signal 27a, and the inverter 32 that receives the control signal 27a drives the winch drum via the motor 31. Thereby, the deflection angle of the suspended load 8 can be converged to zero while keeping the traverse speed of the trolley 7 at the command speed.

  An example of a test result for verifying the effect of the above-described steadying control is shown in FIG. FIG. 12 is data in which the effect of steady rest control is verified using the above-described grab bucket unloader scale-down experimental machine, and is a suspended load in the grab bucket unloader experimental machine that is a trolley crane. When the grab bucket 8 is reciprocated from the sea side (hatch area, near the top of the carrier ship 9 in FIG. 1) to the land side (hopper area, near the top of the hopper 10 in FIG. 1) by operation using steady rest control. The behavior is shown. (A) is the speed command with respect to the trolley 7 and the grab bucket 8, (b) is the traversing position of the trolley 7, (c) has shown the deflection angle of the grab bucket 8 which is a suspended load, respectively.

  In this test, the trolley 7 is in the hatch region (position near 14 m in (b)) at the time of Time = 0 [sec], and the grab bucket 8 has an initial shake. From this state, the trolley 7 is traversed toward the hopper area on the land side. From Time = 5 [sec] to Time = about 15 [sec], a speed command is given to the trolley 7 as shown in (a), and according to the speed command, the trolley 7 has a hopper region ( (Position near 5 m in (b)).

  Here, in this test, the trolley 7 starts to decelerate from the time point near Time = about 13 [sec] when the trolley 7 approaches the hatch area, and at this deceleration, as shown by the solid line in FIG. The steady rest control is implemented. Then, from Time = about 13 [sec] to Time = about 15 [sec], as shown in (c), the swing angle of the grab bucket 8 quickly converges to zero, and the time when the trolley 7 reaches the hopper region. At the time around 15 [sec], the swing angle of the grab bucket 8 is almost zero.

  After the trolley 7 arrives at the hopper region, the grab bucket 8 is opened from Time = about 15 [sec] to Time = about 20 [sec] in this test. During this time, the swing angle of the grab bucket 8 is kept substantially zero by controlling the swing of the trolley 7.

  Next, from the time point of Time = about 20 [sec], the trolley 7 is caused to traverse from the hopper area to the hatch area. As shown in (c), a swing angle is generated in the grab bucket 8 at the time when Time = about 20 [sec], and the trolley 7 starts to traverse, but from the time near Time = about 24 [sec], the trolley 7 When the steadying control is performed at the same time as the deceleration, the deflection angle of the grab bucket 8 quickly converges to zero, and the deflection angle is approximately 27 [sec] when the trolley 7 reaches the hatch region. Becomes almost zero again.

  In this way, even when the suspended load has an initial runout, the runout of the suspended load can be immediately reduced to the allowable runout range where cargo handling can be performed smoothly upon arrival at the hopper area or hatch area. The sufficient effect of the above-described steadying control can be confirmed. In addition, when there is no initial runout in the suspended load, it is possible to confirm the effect more than when there is an initial runout.

Thus, in the present embodiment, the trolley 7 traversing the girder 5 and the boom 6 and a plurality of ropes (the hoisting ropes 16, 16 ′ and the opening and closing ropes 20, 20 ′) are suspended from the trolley 7. The present invention relates to a steadying control method for a trolley crane comprising a suspended load 8 and a winch drum (winding drums 15, 15 ′, opening and closing drums 19, 19 ′) that drives the plurality of ropes to traverse the trolley 7. A sheave load detector 25 is attached to sheaves (winding sheaves 17, 17 ′, open and closed sheaves 21, 21 ′) provided in the middle of a plurality of ropes to detect the tension of the rope, and an estimated deflection angle is determined from the tension. theta _T is calculated, the calculated deflection angle prediction value of the load 8 suspended by performing the processing of the Kalman filter 33 for the estimated swing angle, deflection of the load 8 the relative trolley 7 hanging with the shake angular predicted value Since the feedback control to suppress, by a simple and inexpensive mechanism, the swing angle of the suspended load 8 can be predicted accurately.

  In this embodiment, the Kalman filter 33 is designed with respect to the mathematical model 34 defined by assuming that the trolley crane is a one-mass system model of the spring mass damper. A predicted value can be obtained.

Further, in this embodiment, in addition to the predicted deflection angle value, the gain schedule H _∞ controller 35 using the traverse position x of the trolley 7, the traverse direction speed V of the trolley 7, and the suspended rope length l of the suspended load 8. And the steady state control speed command u for the trolley 7 is obtained by the gain schedule H _∞ controller 35, so that a robust control system in consideration of disturbance and modeling error can be constructed for the entire system.

Further, in the present embodiment, the trolley 7 is controlled by the anti-sway control speed command u obtained by linearly combining the _anti- sway feedback control speed command u _fb from the gain schedule H _∞ controller 35 and the transverse operation speed command V _ref for the trolley 7. Therefore, the swing angle of the suspended load 8 is converged to zero based on the estimated swing angle θ _T of the suspended load 8 while traveling the trolley 7 based on the transverse operation speed command V _ref to the trolley 7. be able to.

  Therefore, according to the present embodiment, the steadying control can be suitably performed on the suspended load while reducing the cost and maintenance problems.

  The steadying control method and apparatus for a trolley crane according to the present invention is not limited to the above-described embodiment. For example, in the above-described embodiment, a grab bucket unloader has been described as an example. Of course, various modifications can be made without departing from the gist of the present invention, such as being applicable to each type of crane.

5 Girder 6 Boom 7 Trolley 8 Suspended load (grab bucket)
15,15 'winch drum (winding drum)
16, 16 'rope (winding rope)
17, 17 'sheave
19, 19 'winch drum (open / close drum)
20, 20 'rope (opening and closing rope)
21,21 'sheave (open and close sheave)
25 Sheave load detector 26 Suspended load current position detector 27 Controller 33 Kalman filter 34 Formula model 35 Gain schedule H _∞ controller

Claims (4)

A steadying control method for a trolley crane comprising a trolley traversing on a girder or boom, a suspended load suspended from the trolley by a plurality of ropes, and a winch drum that drives the plurality of ropes and traverses the trolley Because
A sheave load detector is attached to each sheave provided in the middle of the plurality of ropes to detect the tension of the rope, and an estimated deflection angle is calculated from the tension detected for each sheave, and the estimated deflection angle is calculated. Kalman filter processing performed by calculating the deflection angle predicted value of the suspended load, have rows feedback control to suppress the vibration of the suspended load with respect to the trolley with the shake angular predicted value for,
The feedback control is
A gain schedule H _∞ controller is derived using the traverse position of the trolley, the traverse direction speed of the trolley, and the suspended rope length of the suspended load in addition to the predicted deflection angle , and the gain schedule H _∞ controller The steady-state feedback control speed command for the obtained trolley,
Steadying control method for a trolley crane, characterized in rows Ukoto by the traverse operation speed command and shake by linearly combining the set control speed instruction for said trolley.   2. The trolley-type crane steady-state control method according to claim 1, wherein the Kalman filter is designed with respect to a mathematical model defined by assuming that the trolley-type crane is a one-mass system model of a spring mass damper. A trolley crane steadying control device comprising: a trolley traversing on a girder or a boom; a suspended load suspended from the trolley by a plurality of ropes; and a winch drum that drives the plurality of ropes and traverses the trolley. Because
A sheave each provided in the middle of the plurality of ropes;
A sheave load detector attached to the sheave for detecting the tension of the rope;
An estimated deflection angle is calculated from each detected tension for each sheave, a predicted swing angle value of the suspended load is calculated by performing a process using the Kalman filter on the estimated deflection angle, and the predicted deflection angle value is used to calculate the predicted deflection angle. A control device configured to perform feedback control to suppress the swing of the suspended load with respect to the trolley ,
The control device
A suspended load current position detector for detecting a traverse position of the trolley and a hanging rope length of the suspended load, in addition to the predicted deflection angle, the traverse position of the trolley, the traverse direction speed of the trolley, the suspended load A gain schedule H _∞ controller is derived using the suspension rope length of the sway, and the steadying feedback control speed command for the trolley obtained by the gain schedule H _∞ controller,
A steady-state control speed command linearly combined with a traverse operation speed command for the trolley;
The trolley type crane steady-state control device is configured to control the trolley. 4. The runout of the trolley crane according to claim 3 , wherein the Kalman filter is designed for a mathematical model defined on the assumption that the trolley crane is a one-mass system model of a spring mass damper, and is mounted on the control device. Stop control device.

Images