EP1574469A1 - Design method for the controller design of an active elevator car shock attenuation system - Google Patents

Abstract

Design of regulator for vibration damping at elevator car, where the regulator design is based on a model of the elevator car, comprises determining an overall model of the elevator car with model parameters, which are at least known and/or estimated; identifying the parameters for the elevator car by comparison of transfer functions and/or frequency responses; changing the model parameters; and designing an optimum regulator. Design of regulator for vibration damping at elevator car (1), where the regulator design is based on a model of the elevator car, comprises determining an overall model of the elevator car with model parameters, which are at least known and/or estimated; identifying the parameters for the elevator car by comparison of transfer functions and/or frequency responses of the model with respective measured transfer functions and measured frequency responses; changing the model parameters to achieve a greatest possible correspondence with the measured frequency responses; and designing an optimum regulator for active vibration damping of the elevator car, where the model together with the identified parameters serves as a basis for the design.

Description

The invention relates to a method for the design of a Regulator for vibration damping in an elevator car, wherein the regulator design is based on a model of the elevator cabin.

Patent EP 0 731 051 B1 discloses a device and a method for vibration damping on a Elevator car became known. Transverse to the direction of travel occurring vibrations or accelerations are through a quick scheme reduced, so they in the Elevator car are no longer noticeable. To capture the Measurements are arranged on the cabin frame inertial sensors. In addition, a slow position controller leads the Elevator cabin with one-sided imbalance to the Guide rails automatically in a central position, wherein Position sensors that supply measured values to position controllers.

To reduce the vibrations or accelerations The elevator car is a multi-size controller and another Multi-size controller for maintaining the game to the Guide rollers or the upright position of the elevator car intended. The control signals of the two controllers are added and control one actuator per roller guide and per Horizontal direction.

The controller design is based on a model of the elevator car, which takes into account the essential structural resonances.

The disadvantage is that the overall model is too complex tends, despite sophisticated methods for reducing the number Poles. As a result, the model-based controller becomes also complex.

The invention aims to remedy this situation. The invention, how it is characterized in claim 1, solves the problem, the Disadvantages of the known method to avoid and a simple procedure for the design of a controller propose.

Advantageous developments of the invention are in the specified dependent claims.

Advantageously, in the inventive method a Overall model of the elevator car with known structure specified. It is a so-called Multibody (MBS) model, which has several rigid bodies includes. The MKS model describes the essential elastic Structure of the elevator car with the guide rollers and the Actuators and the power coupling with the guide rails. The model parameters are more or less well known or there are estimates, the parameters for the used to identify or determine the elevator car used are. The transfer functions or Frequency responses of the model with the measured Transmission functions or frequency responses compared. With Help of an algorithm to optimize functions with several variables are the estimated model parameters changed in order to achieve the greatest possible agreement.

Next advantageous is that as a measuring device for the measuring transmission functions or frequency responses the active vibration damping system of the elevator car itself is usable. With the actuators is the elevator car stimulated and with the acceleration sensors or with the Position sensors are measured the answers.

This model-based design method of the controller ensures the best possible active vibration damping for the individual elevator cabins with very different ones Parameters.

With above-mentioned identification method is ensured that as a result the simplest and most most consistent model of the elevator car is present. Advantageously, based on this model Controller better quality and better control quality. In addition, the method is systematically writable and is largely automated and essential perform a shorter time.

Based on the MKS model with identified parameters a robust multi-size controller is designed for reduction acceleration and a position regulator for Maintaining the game on the leadership roles.

The acceleration controller has the behavior of a Bandpass filters and the best effect in a medium Frequency range from about 1 Hz to 4 Hz. Below and above this frequency band is the gain and thus reduces the effectiveness of the acceleration regulator.

In the low frequency range, the effect of the Acceleration controller through the available game to the Guide rollers and the position controller to be interpreted thereon limited. The position controller causes the elevator car an average of the rail profiles follows while the Accelerator causes a rectilinear motion. This conflict of objectives is resolved by turning the two knobs in different frequency ranges are effective. The Gain of the positioner is at low frequencies big and then decreases. That is, he has the property a low-pass filter. Conversely, the Acceleration controller at low frequencies a small Gain.

In the high frequency range, the effect of the Acceleration controller by the elasticity of the Elevator car limited. The first structural resonance can For example, at 12 Hz, this value is strong depends on the construction of the elevator car and clearly can be lower. Above the first structural resonance can the controller no longer accelerates the car body to reduce. There is even a risk that Structural resonances are stimulated, or that instability can occur. With knowledge of the dynamic system model the controller, the controller can be designed so that this can be avoided.

With reference to the accompanying figures, the present Invention explained in more detail:

Fig. 1
ein Mehrkörpersystem (MKS) Modell einer Aufzugskabine,

Fig. 2
eine Führungsrolle mit Rollenkräften,

Fig. 3
ein Stellglied mit Führungsrolle, Aktuator und Sensoren,

Fig. 4
eine schematische Darstellung der geregelten Achsen,

Fig. 5
die Verstärkung der gemessenen Beschleunigung und des identifizierten Modells,

Fig. 6 und Fig. 7
einen optimierten Regler mit den identifizierten Parametern zur aktiven Schwingungsdämpfung,

Fig. 8
Signalflussschema für den Entwurf eines H _∞-Reglers mit Regler und Regelstrecke,

Fig. 9
den Verlauf der Singularwerte eines Positionsreglers in y-Richtung,

Fig. 10
den Verlauf der Singularwerte eines Beschleunigungsreglers in y-Richtung und

Fig. 11 ein Kraftsignal zur Anregung der Aktuatoren.

Show it:

Fig. 1
a multi-body system (MBS) model of an elevator car,

Fig. 2
a leadership role with role forces,

Fig. 3
an actuator with guide roller, actuator and sensors,

Fig. 4
a schematic representation of the controlled axes,

Fig. 5
the gain of the measured acceleration and the identified model,

Fig. 6 and Fig. 7
an optimized controller with the identified parameters for active vibration damping,

Fig. 8
Signal flow diagram for the design of H _∞ knob with controller and control loop,

Fig. 9
the course of the singular values of a position controller in the y-direction,

Fig. 10
the course of the singular values of an acceleration controller in the y-direction and

Fig. 11 is a force signal for excitation of the actuators.

The MKS model must have the essential characteristics of Regulate elevator car regarding ride comfort. Since at the Identification of parameters only with linear models All nonlinear effects must be able to be worked be ignored. The first natural frequencies of the elastic elevator cabins are so deep that they are with the so-called rigid body natural frequencies of the entire cabin can overlap.

As shown in Fig. 1 are for modeling the elastic Elevator car 1 at least two rigid bodies required namely cabin body 2 and cabin frame 3. Cabin body 2 and cabin frame 3 are by means of elastomer springs 4.1 to 4.6, the so-called cabin insulation 4 connected. This is reduced the transmission of structure-borne noise from the frame to the Cab body. For modeling a rigid elevator car 1 It is sufficient, cabin body and cabin frame as a whole to look at a body.

The transverse rigidity of cabin body 2 and cab frame 3 is much lower than the stiffness in Vertical direction. This can be done with the division in each at least two rigid bodies, namely cabin body 2.1 and 2.2 and cabin frames 3.1 and 3.2 are modeled. The at least two sub-bodies are horizontal by springs 5, 6.1 and 6.2 coupled and can be connected vertically as rigid to be viewed as.

The guide rollers 7.1 to 7.8 with the proportionate masses of Levers and actuators can work with at least 8 rigid bodies modeled or neglected. This is dependent from the associated natural frequencies of the guide rollers and considered from the upper limit of the frequency range of becomes. Since the natural frequency of the actuator-roller system in Regulated condition can lead to instability, the Modeling with rigid bodies is preferred. These are only perpendicular to the footprint on the rail opposite the Sliding frame and with the roller guide springs 8.1 bis 8.8 coupled. In the other directions they are rigid with connected to the frame.

As shown in Fig. 2, the leadership behavior or the Force coupling between guide rollers and guide rails important. To model are essentially only the two horizontal force components necessary. The vertical one Force component that results from the rolling resistance can be ignored. The normal force results from the elastic compression of the roller covering 9.1 to 9.8. The Axial or lateral force results from the angle between the Straight lines perpendicular to the roll axis and parallel to the rail and the actual direction of movement of the Roll center.

F_RA

: Rollenkraft in Achsrichtung in [N]

α

: Schräglaufwinkel in [rad]

F_RN

: Rollenkraft normal zur Aufstandsfläche [N]

K

: Konstante ohne Dimension, wird durch Messung bestimmt

Mathematically, the following relationships are relevant: F RA = - tan (α) * F RN * K

F _RA

: Rolling force in axial direction in [N]

α

: Slip angle in [rad]

F _RN

: Roller Force Normal to Contact Area [N]

K

: Constant without dimension, determined by measurement

The force law {1} becomes invalid at the latest when the Limits of static friction can be achieved as well large slip angle α. This one becomes smaller Driving speed quickly greater and is at a standstill about 90 degrees. The law of force {1} only applies to the moving cabin.

v_K :

Vertikalgeschwindigkeit der Kabine [m/s]

v_A :

Geschwindigkeit der Kabine in Achsrichtung [m/s]

For the roller force in the axial direction when the car is moving then approximately: F RA = - v A / v K * F RN * K F RA = - v A * (F RN * K / v K )

v _K :

Vertical speed of the cabin [m / s]

v _A :

Speed of the cabin in the axial direction [m / s]

K is a constant, v _K and F _RN can be considered constant if the preload force is significantly greater than the dynamic part of the normal force. This means that the roller force in the axial direction is proportional and opposite to the speed in the axial direction and inversely proportional to the driving speed of the elevator car.

Transverse vibrations of the cabin are thus due to the roles steamed as from a viscous damper, with the effect becomes smaller with increasing travel speed.

As shown in Fig. 3, the guide rollers 7 with a order an axle 10 'rotatable lever 10 with the cab frame 3 connected, wherein the roller guide spring 8 a force generated between the lever and cab frame. An actuator 11 generates a force parallel to the roller guide spring acts. A position sensor 12 measures the position of the lever 10 and the guide roller 7. An acceleration sensor 13th measures the acceleration of the cabin frame 3 perpendicular to Footprint of the roller lining 9 on the guide rail 14. The reference number of the respective element applies as in Fig. 1 shown (for example, on the elevator car 1 below right: 7.1.8.1.9.1,10.1,11.1,12.1,13.1).

At the elevator car 1 four lower guide rollers 7.1 to 7.4 provided with actuators and position sensors. In addition, four upper guide rollers 7.5 to 7.8 be provided with actuators and position sensors. The Number of required acceleration sensors 13 corresponds the number of controlled axles, with at least three and at most six acceleration sensors are provided.

As shown in Fig. 4, for the active vibration damping of the elevator car 1, the number of axles is reduced from eight to six, or from four to three axles, if only down active is controlled. For each axis An _i is a triple of signals Fn _i , Pn _i , _i for actuator force, position and acceleration. The index i is the consecutive numbering in the respective axis system and n stands for number of axes of the system.

Das Kraftsignal F6₁ für die Aktuatoren 11.1 und 11.3 bzw. das Kraftsignal F6₄ für die Aktuatoren 11.5 und 11.7 wird in eine positive und eine negative Hälfte aufgeteilt. Jeder Aktuator wird nur von einer Hälfte angesteuert und kann nur Druckkraft im Rollenbelag erzeugen. Von den Signalen der Positionssensoren 12.1 und 12.3 wird ein Mittelwert gebildet und das gleiche gilt für die Positionssensoren 12.5 und 12.7. Von den Signalen der Beschleunigungssensoren 13.1 und 13.3, bzw. 13.5 und 13.7 wird ebenfalls ein Mittelwert gebildet. Da die Beschleunigungssensoren 13.1 und 13.3, bzw. 13.5 und 13.7 auf einer Achse liegen und durch den unteren oder oberen Kabinenrahmen starr verbunden sind, messen sie prinzipiell das Gleiche und es kann je ein Sensor des jeweiligen Paares weggelassen werden.

The signals of the lower and the upper roller pair between the guide rails 14.1 and 14.2 are summarized as follows:

The force signal F6 ₁ for the actuators 11.1 and 11.3 and the force signal F6 ₄ for the actuators 11.5 and 11.7 is divided into a positive and a negative half. Each actuator is controlled by only one half and can only generate compressive force in the roller lining. Of the signals of the position sensors 12.1 and 12.3 an average value is formed and the same applies to the position sensors 12.5 and 12.7. Of the signals of the acceleration sensors 13.1 and 13.3, or 13.5 and 13.7, an average value is also formed. Since the acceleration sensors 13.1 and 13.3, or 13.5 and 13.7 lie on one axis and are rigidly connected by the lower or upper cabin frame, they measure in principle the same and it can ever be omitted a sensor of the respective pair.

During the measuring runs, one or more actuators are actuated with a force signal as shown in FIG. 11, and the elevator car 1 is excited to vibrations transverse to the direction of travel in such a way that clearly measurable signals are produced in the position sensors 12 and in the acceleration sensors 13. So that the correlation of the measurements with the force signals can be reliably determined, usually only one actuator or actuator pair is driven. As shown in Table 1, at least as many test drives are necessary as active axles are provided.

The frequency spectrum of the force signals and the measured position signals and acceleration signals are determined by Fourier transformation. The transfer functions in the frequency domain or frequency responses G _{i , j} (ω) with the angular frequency ω as an argument are determined by dividing the spectra of the measurements by the associated spectrum of the force signal. Where i is the index of the measurement and j is the index of the force. G P i . j (ω) = P i (Ω) F j (Ω) G a i . j (ω) = a i (Ω) F j (Ω)

G ^P _{i , j} (ω) are the individual frequency responses from force to position and G ^a _{i , j} (ω) are the individual frequency responses from force to acceleration. The matrix G ^P (ω) contains all frequency responses force to position and matrix G ^a (ω) all frequency responses force to acceleration. Matrix G (ω) arises from the vertical composition of G ^P (ω) and G ^a (ω).

For a 6-axis system, this results in 2 x 6 x 6 = 72 Transfer functions and for a 3-axis system 2 x 3 x 3 = 18 transfer functions. In cabins their focus on the axis lies between the guide rails 14.1 and 14.2, are the couplings and the correlation between the two horizontal directions x and y weak. That's why only about half of the transfer functions continue to use, the rest are eliminated, because of insufficient Correlation.

The MKS model of the cabin is generally a linear system. If this contains nonlinear parts, a fully linearized model is generated in a suitable operating state by numerical differentiation. In the linear state space, the MKS model is described with the following equations: x = Ax + Bu y = cx + You

Positionen und Geschwindigkeiten der Schwerpunkte im Starrkörpermodell, sowie Drehwinkel und Drehgeschwindigkeiten. Ableitungen der Zustände sind Geschwindigkeiten und Beschleunigungen. Die Geschwindigkeit ist damit Zustand und Ableitung zugleich.

x is the vector of the states of the system, which are generally not visible from the outside. States of the system in the present case are:

Positions and speeds of the centers of gravity in the rigid body model, as well as rotation angles and rotational speeds. Derivatives of the states are speeds and accelerations. Speed is thus state and derivative at the same time.

The vector x ˙ contains the derivatives of x after the time. y is a vector containing the measured quantities, ie positions and accelerations. The vector u contains the inputs (actuator forces) of the system. A, B, C and D are matrices which together form the so-called Jacobian matrix by which a linear system is completely described. The frequency response of the system is given by G ^ (ω) = D + C ( j ω I - A ) -1 B ,

G ^ (ω) is a matrix with the same number of rows as measurements in vector y and the same number of columns as inputs in vector u and contains all frequency responses of the MKS model of the cabin.

A Jacobian matrix contains all partial derivatives of a System of equations. In a linear system of Coupled differential equations of 1st order are the constant coefficients of the A, B, C and D matrices.

w(ω) ist eine von der Frequenz abhängige Gewichtung. Sie sorgt dafür, dass nur wichtige Teile der gemessenen Frequenzgänge im Modell nachgebildet werden.The model contains a number of well known parameters such as dimensions and mass and a number of poorly known parameters such as spring rates and damping constants. It is important to identify these poorly known parameters. The identification is performed by comparing the frequency responses of the model with the measured frequency responses. With an optimization algorithm, the poorly known model parameters are changed until the minimum of the sum e of all deviations of the frequency responses of the model from the measured frequency responses is found.

w (ω) is a frequency-dependent weighting. It ensures that only important parts of the measured frequency responses are modeled in the model.

An optimization algorithm can briefly rewrite as follows Given is a function with multiple variables. We are looking for a minimum or maximum of this function. One Optimization algorithm searches for these extremes. There are many different algorithms, e.g. the method of the fastest Descent seeks the greatest gradient with the help of partial derivatives and finds local minima quickly, can but others overlook it. Optimization is in many Subject areas applied mathematics and an important area scientific research.

Fig. 5 shows the frequency-dependent gains of the acceleration measured and of the identified model. | G ^a _1,1 | means magnitude or amplitude of the transfer function or the frequency response force to acceleration with output acceleration of axis 1 and with input force of axis 1. Dimension: 1 mg / N = 1 milli-g / N = 0.0981 m / s ^ 2 / N ~ 1 cm / s ^ 2 / N.

11 shows the force signal for excitation of the actuators 11. The excitation takes place with a so-called Random Binary Signal generated by a random generator, the Amplitude of the signal fixed, for example to ± 300 N. can be adjusted and the spectrum wide and is evenly distributed.

The model with the identified parameters forms the Basis for the design of an optimal controller for active Vibration damping. Controller structure and parameters are depending on the characteristics of the track to be regulated, in this case of the elevator car. The elevator car has a static and dynamic behavior caused by the Model is described. Important parameters are: masses and Moments of inertia, geometry such as height (s), Width (s), depth (s), track gauge, etc., spring rates and Attenuation values. If the parameters change, so does that Influence on the behavior of the elevator car and thus on the settings of the vibration damping controller. at a classic PID controller (proportional, integral and Differential control) are three reinforcements set, which can be managed manually well. The regulator for the present case has well over a hundred parameters, with a Manual adjustment is practically impossible. The Parameters must therefore be determined automatically. This is only with the help of a model that is the essential Describes characteristics of the elevator car, possible.

Ein Positionsregler 15 und ein Beschleunigungsregler 16. Es sind auch andere Strukturen der Regelung möglich, insbesondere eine Kaskadenschaltung von Positions- und Beschleunigungsregler wie in Fig. 7 gezeigt. Die Regler sind linear, zeitinvariant, zeitdiskret und sie regeln mehrere Achsen gleichzeitig, daher kommt die Bezeichnung MIMO für Multi Input, Multi Output. n ist der fortlaufende Index des Zeitschrittes in einem zeitdiskreten oder "digitalen" Regler.

The control as shown in Fig. 6 is divided into two regulators connected in parallel:

A position controller 15 and an acceleration controller 16. Other structures of the control are possible, in particular a cascade connection of position and acceleration controller as shown in Fig. 7. The controls are linear, time-invariant, time-discrete, and they control multiple axes simultaneously, hence the term MIMO for multi-input, multi-output. n is the continuous index of the time step in a time-discrete or "digital" controller.

The updated states x (n + 1) for the next time step are calculated to be available there.

A dynamic system is time-invariant when the descriptive parameters remain constant. A linear one Controller is time-invariant when system matrices A, B, C and D not change. Controller on a digital computer are realized are always time-discrete. That means, you make the inputs, calculations and outputs in fixed Intervals.

In the controller design, the so-called H _∞ method is used. Fig. 8 shows the signal flow scheme of the closed-loop H _∞ drafting process. The main advantage of the H _∞ design process is that it can be automated. Standard minimizes the system to be controlled with a closed loop - The H _∞ is. The H _∞ - norm of a matrix A with m × n elements is given by:

• w_v modelliert die Störungen im Frequenzbereich am Eingang des Systems
• w_r ist ein kleiner konstanter Wert
• w_u limitiert den Reglerausgang
• w_y hat den Wert eins

The system to be controlled is the identified model of the elevator car 1 designated P for Plant as shown in FIG. The desired behavior of the regulator K with reference numeral 17 is generated by means of additional weighting functions at the input and the output of the system.

• w _v models the disturbances in the frequency domain at the input of the system
• w _r is a small constant value
• w _u limits the controller output
• w _y has the value one

Fig. 8 is a diagram for designing the regulator with the H _∞ method. w is the vector signal at the input and is composed of v and r. z is the vector signal at the output, where z = T * w. T consists of controller, control section and weighting functions. P6 or a6 constitute the closed-loop feedback, the position controller or the accelerometer separately designed. F6 is the output or the control signal of the controller.
The H _∞ -norm of ∥z∥ _∞ / ∥w∥ _∞ = ∥T∥ _∞ is minimized. Again, an optimization algorithm is needed that changes the parameters of the controller until a minimum is found.

9 shows the course of the singular values of a Position controller in y direction. This one has predominantly one integrating behavior.

10 shows the course of the singular values of a Acceleration controller in y-direction. This one has one Bandpass characteristic.

Singular values are a measure of the overall gain of a Matrix. An n x n matrix has n singular values. Dimension: 1 N / mg = 1 N / milli-g = N / (0.0981 m / s ^ 2) ~ 1 N / (cm / s ^ 2).

Claims (10)

Method for the design of a vibration damping controller on an elevator car (1), the controller design being based on a model of the elevator car (1),
characterized in that an overall model of the elevator car (1) with more or less well known or estimated model parameters is used, wherein the parameters for the elevator car used are identified by comparing the transmission functions or the frequency responses of the model with the measured transmission functions or the measured ones Frequency response and the model parameters are changed in order to achieve the greatest possible match with the measured frequency responses, with the model with the identified parameters as the basis for the design of an optimal controller for active vibration damping is used. Method according to claim 1,
characterized in that the active vibration damping system of the elevator car (1) itself is provided as the measuring device for the transmission functions or frequency responses to be measured, wherein the elevator car (1) is excited by means of actuators (11) and by means of acceleration sensors (13) or by means of position sensors ( 12) the answers are measured. Process according to claims 1 or 2,
characterized in that the model parameters are changed by means of an optimization algorithm until the minimum of the sum (e) of all deviations of the frequency responses of the model from the measured frequency responses is found. Method according to claim 3,
characterized in that the deviations between the frequency responses of the model and the measured frequency responses in the calculation of the sum (e) are weighted by a frequency-dependent value w (w). Method according to one of the preceding claims,
characterized; is designed that the controller (17) using the H _∞ method. Method according to claim 5,
characterized in that the controller (17) has a position controller (15) which controls the actuators (11) in dependence on the position of the elevator car (1), wherein the Führungslemente (7) assume a predetermined position and
in that the controller (17) has an acceleration regulator (16) which actuates the actuators (11) as a function of the acceleration of the elevator car (1), vibrations occurring on the elevator car (1) being suppressed. Method according to claim 6,
characterized in that the position controller (15) and the acceleration controller (16) are connected in parallel, wherein the control signals of the position controller (15) and the acceleration controller (16) added and the actuators (11) are supplied as a sum signal. Method according to claim 6,
characterized in that the position controller (15) and the acceleration controller (16) are connected in series, wherein the control signal of the position controller (15) is supplied to the acceleration controller (16) as an input signal. Method according to one of claims 6 to 8,
characterized in that the position controller (15) and the acceleration controller (16) are effective in substantially different frequency ranges. Method according to one of the preceding claims,
characterized in that the multibody system (MBS) model for an elastic elevator car comprises at least two bodies describing the cabin body (2) and the cabin frame (3) or, as a whole, for a rigid elevator car (1) cabin body (2) and cabin frame (3) Body includes.

Images