KR20110077423A - Method for estimating external disturbance and external disturbance estimator - Google Patents

Abstract

PURPOSE: A method and a device for tracking external disturbance are provided to reduce physical volume by observing the disturbance flowing into the controlling system without a sensor. CONSTITUTION: The Input and output values of a control system are measured. The control system is a motor control system. The measured input and output values are applied to a kalman filter. An external disturbance torque coupled with output values is estimated. The output values are the angular velocity and the rotation angle of the motor.

Description

Method for estimating external disturbance and external disturbance estimator

The present invention relates to a method for estimating input stage disturbance or output stage disturbance incident as a disturbance in a control system and an external disturbance estimator. More specifically, when a system model is given, the present invention relates to a method for estimating external disturbances without sensors and removing disturbances, or estimating the amount of disturbances for use in remote operation. The main feature is to reduce hardware installation costs by estimating external torque or force without the sensor and to eliminate errors from the sensor.

Sensorless observation of external disturbances entering the control system is efficient in many ways and is a very important technical issue depending on the application. In the case of observing external disturbance without a sensor, it is not necessary to attach a sensor, so that the physical size of the system can be reduced, and errors caused by noise of the sensor can be eliminated.

Observing the external disturbance allows the surgeon to observe the force or torque generated during the cutting of the muscle without using a sensor, thereby delivering a stable force to the operator. Although there is a well-known method of observing external disturbance without a sensor, this method may be somewhat sensitive to noise and has a disadvantage of observing disturbance incident only at an input terminal.

In addition, the conventional disturbance observer can only observe the disturbance incident on the input terminal, the robustness is inferior because it does not use a model for noise, it can be said that the application field is limited.

The present invention seeks to develop an algorithm for estimating external disturbances without sensors. Specifically, we will design a more systematic and reliable disturbance observer using the noise model through the Kalman filter method, and develop an estimator that can observe the disturbance of the output stage as well as the input stage through the model-based signal processing algorithm.

In the present invention, it is intended to develop an external estimator capable of observing external disturbances more robustly in consideration of noise characteristics, and to develop a new concept of external disturbance observer capable of reliably observing disturbances at an output stage. To solve the first problem, we design a disturbance observer based on the Kalman filter. If the noise characteristics change over time, the Kalman filter can be replaced with an adaptive Kalman filter. To observe the disturbance of the input / output stage, we propose a model-based signal processing algorithm that extracts the disturbance using the system model and re-enters the extracted disturbance into the same system.

The external disturbance estimating method according to an aspect of the present invention is an external disturbance estimating method of a control system, comprising: measuring an input value and an output value according to an input value and an input value of the control system; Applying to a Kalman filter, to estimate an external disturbance torque coupled with the output value.

External disturbance estimation method according to another aspect of the present invention, the external disturbance estimation method of the control system, the output of the control system according to the input reflecting the external disturbance, and the control system according to the input that does not reflect the external disturbance Obtaining an output of the control system according to the input of the external disturbance only by obtaining a difference of outputs of the output signal and estimating the external disturbance of the output of the control system according to the input of the external disturbance only through a PID controller. .

An external disturbance estimator according to another aspect of the present invention is an external disturbance estimator for estimating an external disturbance of a control system to which a plurality of transfer function blocks are connected, the output of the control system according to an input in which external disturbances are reflected, and the external disturbance. A reference signal obtainer for obtaining a difference of an output of the control system according to an input that is not reflected and obtaining an output of the control system according to an input of the external disturbance as a reference signal, and estimating the external disturbance using the reference signal. It includes an estimator.

The present invention provides a method for observing disturbance incident on a control system without a sensor, thus eliminating the need for installing a sensor and reducing physical size. This saves cost and reduces physical size, making it suitable for applications with small workspaces. In a specific embodiment, since it is difficult to install a force measuring sensor on the arm of the surgical robot using the method proposed in the present invention can easily measure the force acting on the remote surgical robot and transmit it to the remote operator, on the contrary, the remote operator The force acting on the master robot can be measured without a force sensor and easily transmitted to the remote slab robot.

The method using the Kalman filter or the adaptive Kalman filter proposed in the present invention enables more reliable observation because the characteristics of the noise are considered. In addition, the model-based signal processing algorithm proposed in the present invention proposes a method for stably observing external disturbance incident on an input terminal or an output terminal. Therefore, the two external disturbance estimators to be achieved in the present invention have more excellent characteristics in terms of noise in the conventional method, and can be referred to as more general estimators because they can observe both input and output stages. Two estimators presented in the present invention enable bidirectional force transmission without a sensor in a remote control system such as a surgical robot. In addition to surgical robots, there are various applications.

An external disturbance estimating method and an external disturbance estimator will be described with reference to FIGS. 1 and 2.

1 is an exemplary diagram for describing a process of measuring an external disturbance without a sensor in a motor control system according to an exemplary embodiment of the present invention, and FIG. 2 is a conceptual diagram illustrating an external disturbance estimation method.

In FIG. 1, the disturbance applied from the outside is measured without a sensor on the rotation axis of the motor model indicated by the dotted line. The motor model is a more realistic and accurate model considering the fluid viscosity friction.

<Kalman Filter>

For the disturbance measurement using the Kalman filter, the system model shown in FIG. 1 was expressed using the state space equation, which is as follows.

[Equation 1]

The continuous time model in Equation 1 is modified to a discrete time form state space equation to use a discrete Kalman filter, and then the disturbance torque T _L is added to the fourth state value to form the final state space equation as shown in Equation 2 below. Get

[Equation 2]

As can be seen from the state equation of Equation 2, the disturbance T _L is assumed to have a constant characteristic for the next time k + 1. Since Δt is the sampling time and Δt is very small, very accurate results can be obtained with this assumption. Variables used in the above equations are defined as shown in the table below.

As shown in the above model, the disturbance T _L has a coupling at the angular velocity Wm, so the Kalman filter can be used to predict the disturbance. The input values of the Kalman filter are the current, angular velocity, and angular values, all of which are easy to measure. The operation of the Kalman filter is shown in FIG.

과 초기 공분산(covariance)

, 모델의 연산 에러 분산(variance)을 가지고 현재 시간 k 의 상태 값

과 공분산

를 예측한다. 여기서 k-1은 이전 시간, k는 현재 시간을 의미한다. 즉 도 2의 예측 부분에서 현재 시간 k의 상태값

을 계산하는 (1)식은 이전(시간)에 계산된 값

과 u_k-1로 새로운 현재 상태값

을 예측하는 것이다. 여기서 u_k-1은 입력신호를 의미한다.As shown in FIG. 2, the operation of the Kalman filter may be divided into a prediction part and a correction part. In the prediction part, the initial prediction value

And initial covariance

, The state value of the current time k with the computational error variance of the model

And covariance

Predict. Where k-1 is the previous time and k is the current time. That is, the state value of the current time k in the prediction part of FIG.

Equation (1), which calculates the previous value (time)

And the new current state value with u _k-1

To predict. Here u _k-1 means an input signal.

와 계측 게인 매트릭스 H, 그리고 계측 값의 에러 분산(variance)을 이용하여 칼만 게인 K를 계산한다. In the next calibration section, the predicted covariance

Calculate the Kalman gain K using, and the measurement gain matrix H and the error variance of the measurement values.

을 예측한 후 칼만 게인 K를 이용해 공분산

를 예측한다. 위의 과정을 매시간 k 마다 반복적으로 계산하여 k 에서의 상태 값을 예측해 나간다. State value of current time k using calculated Kalman gain K

And then covariance using Kalman gain K

Predict. The above process is repeated for every hour k to predict the state value at k.

는 4가지 상태 변수

을 가지며 여기서 마지막 값인 T_L로부터 교란을 얻어 낼 수 있다.Predicted state values

Is a four state variable

Where the perturbation can be derived from the last value, T _L.

<Adaptive Kalman Filter>

The Kalman filter described above can only measure optimal disturbances if you know exactly the variance of process and metrology noise, as well as guarantee the quality of the disturbance prediction if the variances change over time. Can't. However, the adaptive Kalman filter takes a sample of data of the desired size and analytically resolves the similarity between the samples, updating the variance of the optimal process noise and measurement noise every hour (or periodically).

The variables used in the equations representing the measured noise variance and the process variance are defined as follows.

Process noise and measurement noise are predicted using the above method, and the rest of the process is the same as the Kalman filter described earlier, so the description is omitted. In this way, the adaptive Kalman filter achieves continuous self-correction and always maintains optimal disturbance prediction quality.

<Model-based Signal Processing Algorithm>

In the case of the Kalman filter described above, accurate modeling in the state space can be used. However, the model-based signal processing algorithm can be applied using only the block diagram corresponding to a part of the system. Hereinafter, an external disturbance estimation method and an external disturbance estimator through the model-based signal processing algorithm will be described.

An external disturbance estimation method and an external disturbance estimator according to another embodiment of the present invention will be described with reference to FIGS. 3 to 6. 3 to 6 are overall system diagrams for explaining an external disturbance estimating method and an external disturbance estimator according to another embodiment of the present invention.

3 is a block diagram of applying a model-based signal processing algorithm to the system model of FIG. As shown in Fig. 3, the overall system includes a system model S1 for finding a system output value in a disturbed state, a system model S2 for finding a value in a state without disturbances, and a disturbance therefrom. It includes a control loop model (S3) for estimating the.

Hereinafter, the entire system illustrated in FIG. 3 is for explaining a process of estimating a disturbance value 0.04N actually acting on the motor in the model S1 of the motor control system controlling the motor. In FIG. 3, the model S1 is a model of the motor control system shown in FIG. 1. Here, the system model (S1) for finding the system output value of the disturbed state is the same as the state equation used in the Kalman filter described above, and is expressed in a form in which transfer function blocks are connected.

First, the operation of this system model S1 will be briefly described.

First, when current is applied to the input, it is multiplied by K in the K1 block to become torque. The disturbance torque is subtracted from this torque value, and the acceleration value can be obtained by dividing the torque by the inertia (J) value. The integrator behind the acceleration changes the acceleration to velocity 1 and multiplies the velocity by B _m / J _m to form a negative feedback on the acceleration. As the speed increases, the acceleration will drop. Velocity 1 is integrated again by an integrator and eventually becomes a position value. Here, speed 1 is a speed according to the influence of the input current and the disturbance torque.

Input Disturbance Observer

Hereinafter, an external disturbance estimator for estimating input disturbance and an external disturbance estimating method will be described. The input is in the form of a current and by this current value a torque is produced, for example in the voice coil, and the disturbance is expressed as a difference in this torque. In FIG. 3, the disturbance added after the K1 block is defined as an input disturbance hereinafter.

After all, the speed value represented by the output in the actual system can be expressed as the sum of the result of the current input and the result of the disturbance. When the difference between the output speed 2 of the model (S2) without disturbance (disturbance torque) is obtained, This value becomes velocity 3 only by the disturbance torque effect.

In the next step, the lower control loop model S3 includes a PID controller and a transfer function blocks after the position where the disturbance torque is input at S1 at the rear end of the PID controller. However, in addition to the PID controller, various types of controllers that are well known may be used. Hereinafter, the PID controller will be described as an example.

The PID controller outputs the PID output signal using the speed 3 based on the disturbance torque as the reference signal. The transfer function blocks after the position where the disturbance torque is input receive and operate a PID output signal output from the PID controller to feed back a feedback signal (output speed). In this case, the feedback signal includes the transfer function blocks after the position where the disturbance torque is input in the system model S1, and the output of the transfer function blocks is the output speed. The PID controller performs PID control so that the feedback signal (output speed) is equal to the reference signal from the transfer function blocks after the disturbance torque is input. Thus, the PID output signal is equivalent to disturbance. From this, external disturbances are estimated.

This algorithm has a number of advantages: it requires only knowing the signal flow diagrams of all systems, not just measurement, but only a block between before and after measurable input signaling output signals that contain disturbances. In addition, the performance of the control loop is the performance of the disturbance observer, and any control technique can be used for the PID controller block position, and even if the variance of the noise is known or unknown, Adding noise reduction techniques by predicting variance also ensures robustness against noise. Therefore, it can be said that it is very scalable.

Output Disturbance Observer

4 is an overall system diagram illustrating an external disturbance estimating method and an external disturbance estimator capable of observing a disturbance applied to an output terminal. In FIG. 4, the disturbance added after the 1 / Js1 block is defined as output stage disturbance. Input and output disturbances are different in the system where the disturbances are entered.

As described above, only the blocks after the disturbance torque is input in S1 are provided behind the PID controller of the lower control loop model S3.

3 and 4, for example, the speed value is used as the measured value. If the speed measurement is difficult, the integrator may be added one by one to use the position value 1 and the position value 2 as the measured values. This is also one of the great advantages of model-based signal processing algorithms. This is illustrated in FIG. 5.

If disturbances occur at the input and output of the system at the same time, the disturbances occurring in several places can be measured simultaneously using the observers of FIGS. 3 and 4 in parallel.

On the other hand, the external disturbance estimator according to an embodiment of the present invention obtains the difference between the output of the control system according to the input reflecting the external disturbance and the output of the control system according to the input which does not reflect the external disturbance, And a reference signal obtaining unit for obtaining an output of the control system as a reference signal, and an estimating unit for estimating external disturbance using the reference signal.

The reference signal acquisition unit may be the model S2 of FIGS. 3, 4, and 5, and the estimation unit may be the control loop model S3. Alternatively, the reference scene acquisition unit may not be the entire model S2 of FIG. 3, but may be an operation unit that calculates a difference between the speed 1 and the speed 2 or an operation unit that calculates the difference between the position value 1 and the position value 2.

-When input and output disturbances exist at the same time

6 is a state in which the input terminal disturbance and the output terminal disturbance defined above exist at the same time. In this case, the input disturbance is in the form of torque, but the output disturbance is in the form of acceleration. However, the observation result of the disturbance observer calculates the torque type value of the input stage disturbance and the acceleration type disturbance of the output stage in the form of the torque of the input stage and calculates the sum of all disturbances as the equivalent torque value generated at the input stage.

Thus any type of disturbance that occurs anywhere in the motor can be predicted and compensated for by the input current value.

1 is an exemplary view for explaining a process of measuring the external disturbance without a sensor in the motor control system according to an embodiment of the present invention.

2 is a conceptual diagram illustrating an external disturbance estimation method.

3 to 6 are overall system diagrams for explaining an external disturbance estimating method and an external disturbance estimator according to another embodiment of the present invention.

Claims (9)

In the external disturbance estimation method of the control system, Measuring an input value of the control system and an output value according to the input value; And Applying the measured input and output values to a Kalman filter to estimate an external disturbance torque coupled with the output value External disturbance estimation method comprising a. The method of claim 1, The control gear system is a motor control system, The input value is a current, the output value is the angular speed, rotation angle of the motor, The estimating step is to estimate an external disturbance torque coupled with the angular velocity. External disturbance estimation method. The method of claim 2, wherein estimating the external disturbances Predicting a state value and a covariance of the current time having the current, angular velocity, rotation angle, and external disturbance torque as state variables; Calculating a Kalman gain using the predicted covariance, the measurement gain matrix, and the error variance of the measurement value; And Using the Kalman gain, correcting the predicted state value of the present time and covariance of the present time. External disturbance estimation method. The method of claim 3, wherein the predicting step Using the state values and covariance calibrated in the calibration step at a previous time External disturbance estimation method. 5. The method of claim 4, Predicting the state value and covariance of the current time includes estimating the state value of the current time using process noise measured every hour, The calculating of the Kalman gain may include calculating the Kalman gain by further using measurement noise measured every hour. External disturbance estimation method. In the external disturbance estimation method of the control system, Obtaining the difference between the output of the control system according to the input reflecting the external disturbance and the output of the control system according to the input which does not reflect the external disturbance and obtaining the output of the control system according to the input of the external disturbance only; And Estimating the external disturbance from an output of the control system according to the input of the external disturbance only External disturbance estimation method comprising a. The method of claim 6, The control word system is a plurality of transfer function block is connected, Estimating the external disturbance Inputting the output of the control system according to the input of the external disturbance only to the PID controller as a reference signal; Outputting, by the PID controller, a PID output signal such that a feedback signal which is an output of a transfer function block after a position where the external disturbance is input among the plurality of transfer function blocks is equal to the reference signal; Transmitting, by the transfer function block after the position where the external disturbance is input, the PID output signal to feed back the feedback signal to the PID controller; And Estimating, by the external disturbance, a PID output signal that makes the feedback signal equal to the reference signal. External disturbance estimation method. In the external disturbance estimator for estimating the external disturbance of a control system to which a plurality of transfer function blocks are connected, Obtain the difference between the output of the control system according to the input reflecting the external disturbance and the output of the control system according to the input which does not reflect the external disturbance, and use the output of the control system according to the input of the external disturbance only as a reference signal. Obtaining a reference signal; And An estimator estimating the external disturbance using the reference signal External disturbance estimator comprising a. The method of claim 8, wherein the estimating unit A PID controller for outputting a PID output signal to have the reference signal and a feedback signal; And Transfer function blocks after the position where the external disturbance is input, including transfer function blocks after the position where the external disturbance is input to operate by receiving the PID output signal to feed back the feedback signal to the PID controller External disturbance estimator.

Images