Method for vibration damping for hydraulic lifting mechanisms of mobile working machines and hydraulic lifting mechanism having vibration damping

Abstract

There are disclosed a method and a lifting mechanism for actively damping in mobile working machines vibrations which may occur as a result of raised attachments during travel. The method or the lifting mechanism has a prediction or a predictor for estimating the future vibration. The prediction is preferably carried out with a recursive least squares algorithm. The mobile working machine may, for example, be a tractor.

Claims

1. A method for active vibration damping for a hydraulic lifting mechanism of a mobile working machine, an attachment being coupled to the hydraulic lifting mechanism, the method comprising: measuring a current vibration; predicting a future vibration with a predictor using the current vibration multiplied by coefficients determined using a Kalman amplification vector and an a priori error between the current vibration and an estimate of the current vibration; generating an output signal with the predictor based upon the predicted future vibration; summing with a summation member the output signal of the predictor and an output signal of an operating element; and controlling the hydraulic lifting mechanism based on the summed output signals of the predictor and of the operating element.

2. The method according to claim 1, the predicting further comprising: predicting the future vibration based on a recursive least squares algorithm.

3. The method according to claim 1 further comprising: determining one of (i) a dominant frequency of the current vibration and (ii) a natural frequency of the current vibration.

4. The method according to claim 1, wherein the method is carried out with a closed control circuit.

5. The method according to claim 4 further comprising: calculating, offline, a transfer function of the closed control circuit.

6. The method according to claim 4 further comprising: determining a phase shift of the closed control circuit.

7. The method according to claim 1, further comprising: providing the summed output signals of the predictor and of the operating element to a controller, wherein controlling the hydraulic lifting mechanism based on the summed output signals comprises: controlling the hydraulic lifting mechanism based on the summed output signals of the predictor and of the operating element with the controller.

8. A lifting mechanism of a mobile working machine, the lifting mechanism comprising: a sensor configured to measure a current vibration of an attachment coupled to the lifting mechanism; a predictor configured to predict a future vibration of the attachment using the current vibration multiplied by coefficients determined using a Kalman amplification vector and an a priori error between the current vibration and an estimate of the current vibration; and a summation member configured to sum an output signal of the predictor and an output signal of an operating element.

9. The lifting mechanism according to claim 8, wherein the lifting mechanism is configured to use the summed output signals of the predictor and of the operating element to control the lifting mechanism.

10. The lifting mechanism of claim 8, further comprising: a controller configured to receive the summed output signals of the predictor and of the operating element and control the lifting mechanism based on the summed output signals of the predictor and of the operating element.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) A plurality of embodiments of the method according to the disclosure or the lifting mechanism according to the disclosure for active vibration damping in mobile working machines with a lifting mechanism are illustrated in the Figures.

(2) In the drawings:

(3) FIG. 1 shows a mobile working machine and the vibration of an attachment with respect to the mobile working machine during the vibration damping according to the disclosure,

(4) FIG. 2 shows the temporal offset of a control of the lifting mechanism based on the predicted force path,

(5) FIG. 3 is a circuit diagram of a first embodiment of the method according to the disclosure,

(6) FIG. 4 is a circuit diagram of a second embodiment of the method according to the disclosure, and

(7) FIG. 5 is a flow chart of a method according to the disclosure.

DETAILED DESCRIPTION

(8) FIG. 1 shows a mobile working machine 1, in this example a tractor, having a control according to the disclosure for vibration damping by means of the lifting mechanism 2 thereof. An attachment 4, in this example a plow, is coupled or articulated to the lifting mechanism 2.

(9) When the attachment 4 is raised and the mobile working machine 1 travels over bumps, the attachment is excited to an undesirable vibration h.sub.t which is, for example, a pivot movement about a rear axle 6 of the mobile working machine 1.

(10) The predictive vibration damping according to the disclosure may be implemented as a pure software module in the existing lifting mechanism 2. Therefore, the use of a predictive vibration damping in the lifting mechanism 2 in order to increase the performance and to further reduce the vibrations h.sub.t which occur during travel operation is possible without expanding the hardware. The subsequent expansion of existing systems for vibration damping by means of a software update is thereby possible.

(11) In the alternative control concept of the existing lifting mechanism 2 in question here, which dispenses with the direct measurement of the force in the lifting mechanism bearing, the phase shift in the effect chain is in principle in most cases greater. Resultant performance losses with the vibration damping according to the prior art are compensated for by the predictive vibration damping according to the disclosure. As a result of the greater phase shift in the effect chain, the potential of the predictive vibration damping according to the disclosure is high.

(12) In order to predict the vibration h.sub.t of the attachment 4, a recursive least squares algorithm is used to estimate the future vibration h.sub.t+1. There can be used as an input signal the signals of different sensors which enable a conclusion relating to the vibration movement. In the embodiment shown, this sensor is a camera 8 and/or pressure sensors in the hydraulic cylinders (both not shown) of the lifting mechanism 2.

(13) Based on a direct or indirect measurement of the force in the lifting mechanism 2 and the dynamic change thereof brought about by the vibration h.sub.t, the active precontrol for compensation for the phase shift in the effect chain is explained below.

(14) With reference to the superimposed vibrations of the measured force path in the lifting mechanism, the learning algorithm learns the vibration properties (for example, natural frequency, spectrum) of the overall system in the current system configuration. Based on the estimated vibration properties, the force path in the lifting mechanism 2 can be predicted a short time Δt. The requirement for this is a periodic path of the vibrations. For control of the lifting cylinders, the predicted force path is used as a basis. There is thereby also produced a time displacement of the control by Δt. The control runs ahead of the actual force path. The optimal prediction time Δt must in this instance correspond to the duration of the effect chain from the control to the actual movement of the lifting mechanism 2. The phase shift of the active chain can be determined from the established transfer function and the natural frequency at which the overall system vibrates. The reaction of the lifting mechanism 2 can consequently be carried out at the optimum time for the vibration damping. The greater the phase shift in the effect chain between the control and lifting mechanism movement is, the greater is the potential for a precontrol based on the prediction according to the disclosure. The phase shift is influenced by all the components involved in the control circuit, such as the sensor, the data transmission, the controller and the mechanical components.

(15) FIG. 2 shows the temporal offset of the control of the lifting mechanism 2 based on the predicted force path F. Both values are to this end plotted against time t.

(16) According to FIG. 3, the vibration h.sub.t of the overall system at the time t is measured by a measurement device 10 and used as an input value for estimating the future vibration h.sub.t+1. The estimation is based on a recursive least squares (RLS) algorithm.

(17) For estimation, a linear model θ
h.sub.t+1={right arrow over (h.sub.t.sup.T)}.Math.{right arrow over (θ)}.sub.t
is used, wherein the measurement data vector {right arrow over (h.sub.t.sup.T)} comprises N measurements up to the time t, and where {right arrow over (θ)} are the coefficients.

(18) First, the so-called Kalman amplification vector

(19) ${\stackrel{.fwdarw.}{K}}_t=\frac{P_{i-1}.Math.\stackrel{.fwdarw.}{h_{t-1}}}{\lambda +\stackrel{.fwdarw.}{h_{t-1}^T}{P}_{i-1}\stackrel{.fwdarw.}{h_{t-1}}}$
is calculated, where λ(0≤λ≤1) is the so-called forgetting factor and P.sub.i is the inverse correlation matrix of the measurement data.

(20) The a priori error ∈ between the actual measurement value h.sub.t and the estimated value h.sub.t+1 is subsequently determined as
ε.sub.i=h.sub.t−{right arrow over (h.sub.t−1.sup.r)}{right arrow over (θ.sub.t−1)}

(21) The coefficients θ.sub.t are updated with reference to the equation
{right arrow over (θ.sub.t)}={right arrow over (θ.sub.t−1)}+{right arrow over (K.sub.t)}ε.sub.t
and the inverse correlation matrix P.sub.t with reference to

(22) $P_t=\frac{P_{i-1}-{\stackrel{.fwdarw.}{K}}_i\stackrel{.fwdarw.}{h_{t-1}^T}{P}_{i-1}}{\lambda }$

(23) Subsequently, an estimation of the vibration h.sub.t+1 or the measurement values is calculated at a time t+n. To this end, a new measurement data vector of corresponding length is iteratively produced and multiplied by the coefficient vector {right arrow over (θ)}.
h.sub.t+n=(h.sub.t−N+n−1,h.sub.t−N÷, . . . ,h.sub.t+n−1).Math.{right arrow over (θ.sub.t)}

(24) An existing controller 16, which is in the embodiment shown constructed as a PID controller, enables an improved performance by incorporating future information relating to the vibration. Accordingly, the expanded structure is divided into a “prediction of the vibration” and an “active precontrol”.

(25) The performance of the active vibration damping is determined significantly by the delay of the closed control circuit 14, which is influenced by the controller 16, the characteristic of the mechanical components which are installed in each case and the signal delays as a result of the measurement device 10 (bus delays, update times, etcetera).

(26) By using a predictor 12 in the precontrol, a damping of these delays of the closed control circuit 14 can be carried out by means of a suitable prediction horizon.

(27) The property of the predictor 12 brings about a periodic path of the input variable or vibration h.sub.t in order to be able to make the most precise prediction possible h.sub.t+1. This is determined by the characteristic of the vibration h.sub.t.

(28) In order to establish the optimum prediction time, an offline transfer function of the closed control circuit 14 is carried out. Furthermore, at the known dominant frequency or natural frequency of the vibration h.sub.t, the phase shift of the closed control circuit 14 is determined. Based on the established phase shift, the prediction time is determined in accordance with the update rate.

(29) The sum of the predicted future vibration h.sub.t+1 formed by a summation member 23 and a user command of an operating element 24 constitutes the input of the system for active vibration damping according to the prior art. The inner structure of the lifting mechanism of the prior art does not have to be changed.

(30) In addition to the above-described method, the performance can be additionally improved using a feed forward inversion of the control path.

(31) A disadvantage of the above-described method involves the requirement for an offline phase delay analysis which is solved by an expansion according to FIG. 4.

(32) An adaptive controller 22 modifies its properties in accordance with the process dynamic and the characteristic of the vibration h.sub.t. The adaptation process requires an additional online parameter estimation using a parameter estimator 26 of the closed control circuit 14 in real time. The parameter estimator 26 is based on a prediction algorithm which carries out a stable inversion of the estimated model. The predicted vibration information is also included in the calculation of the correcting variable of the adaptive controller.

(33) There are disclosed a method and a lifting mechanism 2 for actively damping in mobile working machines 1 vibrations which may occur as a result of raised attachments 4 during travel. The method or the lifting mechanism 2 has a prediction or a predictor 12 for estimating the future vibration h.sub.t+1. The prediction is preferably carried out with a recursive least squares algorithm. The mobile working machine may, for example, be a tractor, a wheel loader or an excavator.

(34) A method is carried out in one embodiment in accordance with the process 50 shown in FIG. 5. At block 52 a current vibration is measured. At block 54, a known dominant frequency or natural frequency of the vibration h.sub.t is used to determine the phase shift of the closed control circuit 14 and at block 56 an offline transfer function of the closed control circuit 14 is used to determine an optimum prediction time. At block 58 a present vibration is used in the prediction of an estimated vibration. In one embodiment the prediction is based upon a recursive least squares algorithm. The summation member 23 sums the predicted future vibration h.sub.t+1 and a user command of an operating element 24 at block 60 and provides the sum to the adaptive controller 22. At block 62 a stable feed forward inversion of an estimated model is carried out by the parameter estimator 26 using a prediction algorithm to estimate an online parameter. The estimate of the online parameter is done in real time and used along with the sum to modify properties of the adaptive controller 22 at block 64. Output from the adaptive controller is then used by a controller to control the system at block 66.