KR20050096318A - Approximaion of estimated parameters in the sensorless stroke control for linear compressors - Google Patents

Abstract

The present invention relates to sensorless control of a linear compressor, and more particularly, to propose a method that can significantly reduce the memory capacity in the controller by estimating the parameters of the linear motor and approximating it to a secondary curve. This facilitates sensorless control of the linear compressor by one-chip microcomputer by reducing the memory capacity to store the estimated parameters, although the control error is slightly larger. According to the present invention, by approximating a database of inductance or torque constant as a function of current and stroke to a two-dimensional curved surface, the memory capacity is greatly reduced and replaced by a simple operation.

Description

Approximaion of Estimated Parameters in the Sensorless Stroke Control for Linear Compressors}

In the conventional reciprocating compressor as shown in FIG. 1, the motion of the rotating rotary motor is changed to a straight line by the crank shaft, whereas the linear compressor is a piston type compressor in which a piston is directly driven by a linear motor as shown in FIG. . In the linear compressor, since all driving forces are applied in the linear motion direction, no lateral pushing force is generated by the piston. As a result, friction loss is reduced and noise during operation is reduced compared to conventional reciprocating compressors.

The principle of operation of the linear motor in the linear compressor is shown in FIG. As shown in ⓐ of FIG. 3, when the magnitude of the alternating current is increased, the magnetic field in the counterclockwise direction becomes larger, and the permanent magnet in the center is pushed to the left, and as shown in ⓑ, when the alternating current decreases to 0, the permanent magnet is biased to the leftmost side. As time passes, AC current flows in the opposite direction. As in ⓒ, the magnetic field also changes in direction and occurs clockwise. As a result, the permanent magnet is forced to the right and the permanent magnet is biased to the right side at ⓓ. When the AC current is 60Hz, the permanent magnet vibrates left and right 60 times a second. If the frequency of AC current is kept constant at 60Hz and the amplitude is increased, the left and right oscillation widths of the permanent magnets, i.e., the stroke, become larger. As the stroke of the permanent magnet increases during the unit time, the linear speed of the piston connected to the permanent magnet increases, increasing the refrigerant flow rate of the linear compressor, and consequently obtaining greater cooling.

As shown in FIG. 1, a conventional reciprocating compressor uses a crank shaft to change the rotational motion of the motor into a linear motion. This reduces energy efficiency, but the piston is constrained by the crankshaft to maintain safe operation without leaving the top and bottom of the structurally designed cylinder. However, in the linear compressor as shown in Fig. 2, since it is not mechanically constrained, it is necessary to control the stroke of the piston so that the piston vibrates safely within a certain area and does not hit the cylinder head. In addition, the stroke control of the piston is required to adjust the outflow rate of the refrigerant for cooling.

To do this, of course, it is necessary to know the position of the piston correctly. Although position sensors such as LVDTs can be used to measure the exact position, these sensors are not only inexpensive, they are also difficult to mount, and inconvenient to have several strands coming out of the compressor vessel. Therefore, there is a need for an efficient way to estimate this indirectly.

The linear motor in the linear compressor may be represented by an electric circuit equivalent model as shown in FIG. 4, and the circuit equation may be represented by a linear differential equation as shown in Equation (1) below. In addition, the thrust Fe of the linear motor can also be expressed linearly as in Equation (2).

<23> (One)

<24> (2)

<25> In formulas (1) and (2) Is a constant representing the relationship between the thrust and current of the linear motor. Because the magnetic flux density varies depending on the position of the piston Should be expressed as a function of piston position, but for this motor, The change of is small and its effect is also small, so it is assumed to be a constant. Effective Inductance And effective resistance In the case of, the value varies depending on the position of the piston, but it is assumed to be a constant. Is the current flowing through the winding, and Is the counter electromotive force generated in the winding by the movement of the linear motor. Equation (1) can be summarized as Equation (3) with respect to x (t).

<26> (3)

The estimated value of the piston position integrating Eq. (3) is shown in Eq. (4).

<28>

(4)

Equation (4): Digitally calculated position estimate of the piston The nth value of can be expressed as Equation (5).

<30> ,

(5)

Where T is the sampling period.

5 shows a closed loop control block diagram for sensorless stroke control of a linear compressor piston. The voltage applied to the linear compressor measures the voltage at both ends of the linear compressor, and the current value is measured by attaching a sensing resistor to the inlet of the linear compressor, sampling them with the A / D converter, and inputting them to the DSP processor. The measured voltage and current of the linear motor, together with the motor constant, are used to estimate the position of the piston connected to the linear motor by equation (5). Command values of strokes depending on the load are input and strokes obtained from the estimated piston position are compared to determine the amplitude of the voltage input to the linear motor via a suitable controller. The frequency of the voltage is determined and input at a frequency that meets the design specifications of the linear motor.

In the case of controlling the sensorless stroke as shown in FIG. 5 under the assumption of a constant motor constant, the position estimation error of the piston becomes large, which affects the control performance of the system. Thus, the most influential We estimate the position error of piston by estimating and inducting the effective inductance as a function of piston position. Reordering Eq. (3) gives Eq. (6).

<34> (6)

<35> in equation (4) Is the position value of the piston estimated by Eq. (4) from the motor constant, motor voltage, and motor current. Is a direct position measurement using a sensor such as LVDT. When the piston makes a linear movement from side to side, let one cycle be tn in steady state and divide each n into 0, t1, t2, ..., tn-1, tn. Then n equations as shown in equation (7) can be obtained.

<36>

                       . . . (7)

If Equation (7) is rearranged, it can be expressed as Equation (8).

<38> (8)

Here, n x 2 matrix A and n x 1 vector b are given by the following equation (9).

<40> ,

(9)

Using pseudo inverse, Equation (8) can be summarized as Equation (10) below.

<42> 10

The parameters and of the linear motor thus obtained are shown in Figs. 6 and 7, respectively. As shown in the figure, the estimated parameters and Changes badly.

6 and 7 estimated parameters according to changes in stroke or current Wow If you estimate the stroke by storing the database in memory, the error will be greatly reduced. However, this takes up a lot of memory capacity, which is a major obstacle to the commercialization of the sensorless controller of the linear compressor. Therefore, in order to reduce the memory capacity and to facilitate commercialization, the present invention intends to significantly reduce the memory capacity by formulating FIGS. 6 and 7 into two-dimensional curved surfaces and storing them in a controller instead of a database.

If the estimated inductance of FIG. 7 is approximated as a quadratic surface, the following Equation (11) is obtained.

<46> (11)

Where is the stroke, is the current, and is the estimated inductance. a, b, c, d, e, and f are constants of the quadratic surfaces, respectively. Given n data sets It can be summarized as in the following equation (12).

<48> (12)

In this case, when the matrix n x 1 of the left side is referred to as the matrix n x 6 of the right side, the coefficient of each term can be obtained by the following equation (13).

<50> (13)

6 and 7 are approximated to the secondary curved surface through the processes of equations (11) to (13), respectively, and are shown in FIGS. 8 and 9, respectively.

In controlling the stroke of the linear motor inside the linear compressor, it is necessary to know the correct stroke value in order to accurately control the stroke. It is expensive to measure this, and thus indirectly estimates and uses this value for stroke control. The estimation requires the motor parameters such as resistance, inductance and torque constant. However, inductance, torque constant, and the like do not have a constant value and have a characteristic that varies greatly with current or stroke. Therefore, in order to accurately estimate the stroke, an inductance or torque constant, which is a function of current and stroke, is stored in a database and stored in a memory, and when the stroke is estimated, the error can be greatly reduced. However, this requires a lot of memory. Therefore, in the present invention, the database is approximated to a two-dimensional curved surface, and the memory capacity is drastically reduced even though a slight error is large, which greatly benefits the commercialization.

<1> Figure 1 is a reciprocating compressor

<2> Figure 2 is a linear compressor

3 shows the operation principle of the linear motor.

4 is an electric circuit equivalent model of the linear motor.

5 shows stroke control of a linear compressor.

6 shows a database of motor torque constants.

7 is a database of motor inductance

8 shows two-dimensional surface functionalization of the motor torque constant

9 shows the two-dimensional surface functionalization of the motor inductance

<10> <Definitions of terms and abbreviations used in formulas>

<11> : Torque constant of motor

<12> : Effective inductance of motor

<13> : Effective resistance of motor

<14> : Power supply voltage

<15> : Current flowing in winding

<16> : Stroke value

<17> : Stroke estimate

Claims (1)

In the sensorless control of the linear compressor, the motor parameters required for estimating the stroke of the linear motor are expressed as a quadratic surface function rather than a database, which greatly reduces the memory capacity.