WO2016174796A1

WO2016174796A1 - Elevator control device, elevator device, and method for determining rotation angle error of rotation detection unit of electric motor for elevator

Info

Publication number: WO2016174796A1
Application number: PCT/JP2015/085540
Authority: WO
Inventors: 大塚　康司; 酒井　雅也
Original assignee: 三菱電機株式会社
Priority date: 2015-04-30
Filing date: 2015-12-18
Publication date: 2016-11-03
Also published as: CN106464195B; JP6165352B2; BR112016025232A2; CN106464195A; JPWO2016174796A1

Abstract

An elevator control device provided with: a frequency analysis unit for outputting a specific frequency component which is obtained through frequency analysis of a current detected by a current detector for detecting a current of an electric motor used for moving a car up and down in a hoistway; and an angle error estimation unit for estimating, by using the specific frequency component, the amplitude and phase of a periodic angle error which is uniquely determined in accordance with a rotation angle from a rotation detection unit for detecting a rotation angle of the electric motor, and for outputting the results as angle error estimation values. The angle error estimation unit performs control so as to execute a learning operation for causing the car to operate over a specific range, continuously acquires a plurality of the specific frequency components which is obtained by inputting the current detected during the learning operation into the frequency analysis unit, calculates evaluation values each of which is a geometric quantity in a coordinate plane formed by a preset number of continuous specific frequency components out of the acquired specific frequency components, calculates the angle error estimation values, and selects an angle error estimation value at which the evaluation value is the smallest.

Description

Method for obtaining rotation angle error of rotation control unit of elevator control device, elevator device, and elevator motor

The present invention relates to an estimation of an angle error in an elevator control device, particularly a control device having a periodic torque pulsation and a speed pulsation due to a periodic angle error of a rotation sensor attached to an electric motor constituting a hoisting machine. It is.

In a resolver device as a conventional rotation sensor, the error waveform of the resolver is composed of frequency components specific to the resolver and is reproducible. Therefore, the position error is calculated with reference to the detected angle signal, and the position error is calculated. The speed error signal is calculated by differentiating the error, and the magnitude of the detection error for each component obtained by dividing the speed error signal by Fourier transform is calculated. By combining the calculated detection errors, an error waveform signal in which the detection errors included in the angle signal detected by the resolver are restored is generated. The generated error waveform signal is used to correct the resolver angle detection signal including the detection error. By performing Fourier transform on the velocity error obtained from the position error, the detection signal of each resolver can be accurately calculated, and by correcting the detected angle signal using the obtained detection error, an accurate angle signal is obtained. be able to.

JP 2012-145371 A

Conventionally, when estimating an angle error using a signal including pulsation caused by a periodic angle error, if the frequency of the periodic angle error matches the natural frequency of the mechanical system connected as the load of the motor, Resonance between the angle error and the mechanical system temporarily changes the amplitude and phase of the velocity pulsation caused by the angle error and loses reproducibility. Therefore, it is difficult to estimate the angle error with the conventional device method. There is a possibility of obtaining an incorrect estimation result.

In particular, in systems such as elevators where the natural frequency of the mechanical system changes according to the length of the rope, the frequency at which the resonance occurs changes from moment to moment, so where does the resonance occur? Is not clear, and there is a high possibility that the angle error is estimated incorrectly. If the detailed specifications of the elevator mechanical system are known in advance, it is possible to calculate the natural frequency, gain characteristics, and phase characteristics of the elevator and estimate the angle error at a speed and car position that does not cause resonance. However, the machine specifications of the elevator are often different for each property, and a method that relies on prior information increases the design time. Therefore, when a mechanical system having resonance is connected as a load of the electric motor, when the angle error is estimated based on a signal including pulsation caused by the angle error, it is a problem to determine whether the angle error estimation result is successful.

Further, in the conventional apparatus, the position error is calculated by referring to the detected angle signal, the position error is differentiated to calculate the speed error signal, and the speed error signal is Fourier transformed to estimate the angle error. Yes. Here, when the angle error is estimated using the velocity signal, the angle resolution estimation accuracy is determined by the angle detector or the velocity resolution of the velocity detector. For this reason, an angle detector or a velocity detector with a low velocity resolution has a problem that a quantization error occurs and the angle error estimation accuracy cannot be sufficiently obtained.

The present invention has been made in order to solve the above-described problems. In an elevator control apparatus having a motor rotation detection unit including a periodic angle error, the angle error and the mechanical system resonate. It is another object of the present invention to provide an elevator control device and the like that can obtain a highly reliable angle error estimation result without erroneously estimating the angle error.

The present invention relates to a current detector that detects a current flowing in an electric motor that generates power to raise and lower a car in a hoistway, a rotation detector that detects a rotation angle of the electric motor, and a current detected by the current detector A frequency analysis unit that outputs a component of a specific frequency obtained by frequency analysis, and an amplitude of a cyclic angle error that is uniquely determined according to a rotation angle from the rotation detection unit using the component of the specific frequency. An angle error estimator that estimates the phase and outputs an angle error estimated value, and the angle error estimator controls the car to perform a learning operation that operates the specific section, and during the learning operation A plurality of the specific frequency components obtained by inputting the detected current to the frequency analysis unit are continuously acquired, and a set number of the specific frequency components that are consecutive among the acquired specific frequency components are obtained. An evaluation value that is a geometric amount in a coordinate plane to be created is calculated, the angle error estimation value is calculated and the evaluation value is associated with the angle error estimation value, and the angle error estimation when the evaluation value is minimized The value is in the elevator control unit etc.

According to the present invention, it is possible to provide an elevator control device or the like that can obtain a highly reliable angle error estimation result without erroneously estimating the angle error even if the angle error and the mechanical system resonate.

It is a block diagram which shows an example of the control apparatus of the elevator by this invention. It is a block diagram which shows an example of a structure of the angle error estimation part of FIG. It is a figure which shows an example of the frequency characteristic of the speed controller of FIG. It is a graph which shows an example of the gain characteristic of the mechanical system of an elevator. It is a graph which shows an example of the phase characteristic of the mechanical system of an elevator. It is a graph which shows an example of the coordinate plane of the Fourier coefficient which the frequency analysis part of FIG. 1 calculates. 3 is a flowchart showing an example of learning driving operation in the first embodiment by an angle error estimating unit in FIG. 1. It is a graph which shows an example of the area enclosed by the coordinate of the Fourier coefficient which the angle error estimation part of FIG. 1 calculates. 6 is a flowchart illustrating an example of the learning driving operation in the second embodiment by the angle error estimating unit in FIG. 1. 6 is a flowchart illustrating an example of learning driving operation in the third embodiment by an angle error estimating unit in FIG. 1. 6 is a flowchart illustrating an example of the learning driving operation in the fourth embodiment by the angle error estimation unit in FIG. 1. It is a block diagram which shows an example of a structure of the angle error estimation part of FIG. It is a graph which shows an example of the line segment length between the coordinates of the Fourier coefficient which the angle error estimation part of FIG. 1 calculates. It is a flowchart which shows an example of operation | movement of the learning driving | operation in Embodiment 5 by the angle error estimation part of FIG. It is a flowchart which shows an example of operation | movement of the learning driving | operation in Embodiment 6 by the angle error estimation part of FIG.

In the elevator control device according to the present invention,
The frequency analyzer calculates the amplitude and phase of the specific frequency component by performing frequency analysis on the current detected by the current detector,
The angle error estimation unit estimates an angle error composed of a specific frequency component as an angle error estimated value using the amplitude and phase of the specific frequency component calculated by the frequency analysis unit, and when estimating the angle detection error, the car is specified in a specific section. A learning operation is performed, and during the learning operation, the calculation results of the amplitude and phase of the specific frequency component are stored for a plurality of times, and the evaluation is a geometric quantity at coordinates created by the amplitude and phase of the specific frequency component stored for a plurality of times. A value is calculated, and an estimated angle error value when the evaluation value is minimized is selected.
Therefore, it is possible to obtain a highly reliable estimated value of the angle error without erroneously estimating the angle error due to the influence of resonance.

Hereinafter, an elevator control device and the like according to the present invention will be described with reference to the drawings according to each embodiment. In each embodiment, the same or corresponding parts are denoted by the same reference numerals, and redundant description is omitted.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing an example of an elevator control apparatus according to the present invention. In FIG. 1, an elevator car 4 and a counterweight 5 are connected to each other by a hoisting rope 6 and are suspended on a sheave 3 in a hangar manner. The sheave 3 is connected to an electric motor 1 that is an electric motor for driving the car 4, and the car 4 is raised and lowered by the power of the electric motor 1. The electric motor 1 that raises and lowers the car 4 is, for example, a permanent magnet synchronous motor.

Rotation detector 2 for detecting the rotation angle of motor 1 or sheave 3 is mounted on the same axis of motor 1 and sheave 3. For example, the angle information that is the rotation angle of the electric motor 1 that is output from the rotation detection unit 2 that includes a resolver, an encoder, a magnetic sensor, or the like includes a periodic error that is uniquely determined according to the rotation angle of the electric motor 1. Here, the periodic error uniquely determined according to the rotation angle of the electric motor 1 is, for example, a rotation angle such as a resolver detection error, a missing pulse due to a slit failure in an optical encoder, and an imbalance in the distance between pulses. Refers to errors that are reproducible depending on, i.e. occur at the same rotational angular position of each rotation

Note that the frequency analysis unit 8, the angle error estimation unit 9, the speed calculation unit 10, the speed command calculation unit 11, the speed controller 12, the current controller 13, and each subtractor SU1- shown in the functional blocks described below. Among SU3, at least the frequency analysis unit 8, the angle error estimation unit 9, the speed calculation unit 10, the speed command calculation unit 11, and each subtractor SU1-SU3 are configured by a computer including a processor and a memory, for example. Each process is executed according to the program stored in the memory and various setting information necessary for the process. Similarly, the speed controller 12 and the current controller 13 may be configured by the computer. The portions indicated by the functional blocks can also be configured by digital circuits that execute the respective functions.
This also applies to FIG.

The speed command calculation unit 11 calculates and outputs a speed command value for the electric motor 1. Although not shown, the speed command calculation unit 11 may include a position control system. The present invention can be applied even when the speed command calculation unit 11 includes a position control system.

The speed controller 12 inputs the difference between the speed command value from the speed command calculation unit 11 and the rotation speed of the motor 1 calculated by the speed calculation unit 10 from the subtractor SU1, and calculates the current command value for the motor 1 And output. The speed controller 12 may be configured by any control method such as PI control or PD control.

The speed calculation unit 10 is an angular error estimation value of a periodic error that is uniquely determined according to the rotation angle of the motor 1 that is an output from the rotation detection unit 2 and the rotation angle of the motor 1 estimated by the angle error estimation unit 9. The rotational speed of the electric motor 1 is calculated and output based on the corrected rotation angle corrected from the subtractor SU2, which is the difference between the motor and the motor. The speed calculation unit 10 calculates the rotation speed by the time differentiation of the rotation angle most simply. Further, it may be configured to smooth by a low-pass filter (not shown) in order to remove noise due to time differentiation. Furthermore, the speed calculation unit 10 may calculate the rotation speed of the electric motor 1 every predetermined time set in advance, or includes a configuration for measuring the time, for every predetermined fixed rotation angle. The rotational speed may be calculated.

The current controller 13 includes a current command value from the speed controller 12 and a phase current that is an output from the current detector 7 or a phase current of the motor 1 that is dq-axis converted by coordinate conversion (not shown). The difference from the shaft current is input from the subtractor SU3, and the voltage command of the electric motor 1 is calculated and output. The control method of the current controller 13 is not limited like the speed controller 12.

The current detector 7 detects the current of the electric motor 1. For example, when the motor 1 is a three-phase motor, a two-phase phase current is often measured, but a three-phase phase current may be measured. In FIG. 1, the current detector 7 measures the output current of the power converter 14. However, the current detector 7 calculates the bus current of the power converter 14 as in a current measurement method using a one-shunt resistor. Each phase current may be estimated by measurement. Even in this case, the present invention is not affected at all.

The power converter 14 converts a power supply voltage (not shown) into a preferable variable voltage variable frequency based on a voltage command from the current controller 13. The power converter 14 of the present invention includes a power converter or a matrix that converts an AC voltage to a DC voltage by a converter and then converts the DC voltage to an AC voltage by an inverter, like an inverter device that is generally sold. It refers to a variable voltage variable frequency power converter including a power conversion device that directly converts AC voltage into AC variable voltage variable current, such as a converter.

Further, the power converter 14 according to the present invention may include a coordinate conversion function in addition to the above-described inverter. That is, when the voltage command is a dq axis voltage command value, the dq axis voltage command value is converted into a phase voltage or a line voltage to a voltage according to the commanded voltage command value. It is expressed as a power converter 14 including a coordinate conversion function for conversion. Note that the present invention can be applied even if a device or a correction unit for correcting the dead time of the power converter 14 is provided.

The frequency analysis unit 8 performs frequency analysis on the current composed of the phase current or the shaft current detected by the current detector 7 and outputs the amplitude and phase of a specific frequency. Here, the frequency analysis unit 8 is preferably configured to obtain the amplitude and phase at a specific frequency of the input signal, such as Fourier transform, discrete Fourier transform, Fourier series expansion, and fast Fourier transform. However, a specific frequency signal is extracted like a filter combining a notch filter and a bandpass filter, and an amplitude calculation is performed on the output current of the bandpass filter, for example, by an amplitude detection unit or a phase detection unit (not shown). Further, the configuration may be such that the amplitude and phase of a specific frequency of the input signal are calculated by performing phase calculation. Further, the filter used here may be an electrical one that combines a resistor, a capacitor, a coil, or the like, or may be a process that performs processing in a computer. Hereinafter, the frequency analysis unit 8 will be described as being configured to perform Fourier transform.

The angle error estimation unit 9 estimates a periodic angle error included in the rotation angle that is the output of the rotation detection unit 2 using the Fourier coefficient that is the output from the frequency analysis unit 8. The angle error estimation unit 9 stores in advance a conversion equation for calculating an angle error using a Fourier coefficient obtained by performing a Fourier transform on the current of the current detector 7 using information on the rotation angle of the motor. Then, the estimated value of the angle error is calculated from the current using the conversion formula.
In addition, the angle error estimation unit 9 is based on a Fourier coefficient obtained by performing a Fourier transform on the current of the current detector 7 that is an output from the frequency analysis unit 8 using information on the rotation angle of the electric motor. The area of the region surrounded by the coordinates is calculated as an evaluation value that is a geometric quantity on the coordinate plane. A method for calculating the area surrounded by the coordinates of the Fourier coefficient and its meaning will be described later.
The angle error estimated by the angle error estimator 9 is composed of an error amplitude and an error phase described later.
When the error error estimator 9 calculates the error amplitude and the error phase as an estimated value of the angle error, the error error and the error phase are used to reproduce a periodic angle error, and the sine wave or cosine wave correction signal is used. Is calculated and output.

FIG. 2 is a configuration diagram illustrating an example of the angle error estimation unit 9.
The learning speed calculation unit 91 calculates the rotation speed of the electric motor 1 based on the rotation angle of the electric motor 1 detected by the rotation detection unit 2. The learning speed calculation unit 91 calculates the rotation speed by time differentiation of the rotation angle, most simply. Further, a configuration may be used in which smoothing is performed by a low-pass filter in order to remove noise due to time differentiation. Furthermore, the learning speed calculation unit 91 may calculate the rotation speed of the electric motor 1 every predetermined time set in advance, or includes a configuration for measuring the time, for each fixed rotation angle set in advance. Alternatively, the rotation speed may be calculated.
Since the rotation angle detected by the rotation detection unit 2 includes a periodic angle error, the rotation speed of the electric motor 1 calculated by the learning speed calculation unit 91 includes a periodic speed pulsation. The speed information necessary for the angle error estimation unit 9 determines whether or not the rotational speed of the electric motor 1 has reached a preset speed and has reached a constant speed traveling state at the set speed in an angle error learning operation to be described later. It is for use. Accordingly, even if pulsation is included in the speed information, there is no problem because it can be determined that the vehicle has reached a constant speed traveling state.

The car position calculation unit 92 calculates and outputs the position of the car 4 in the hoistway based on the rotation angle of the electric motor 1 that is an output from the rotation detection unit 2. The reference position in the hoistway may be the lowermost floor or the uppermost floor, or an arbitrary floor may be used as a reference. Since the rotation angle of the electric motor 1 that is an output from the rotation detection unit 2 includes a cyclic angle error, the car position calculated by the car position calculation unit 92 also includes an error. The position information of the car 4 required in the present invention is used to determine whether or not traveling in a specific section is completed in an angle error learning operation described later. Therefore, there is no problem because it can be determined that the vehicle has traveled in the specific section even if the position information of the car 4 includes an error. The car position calculation part 92 does not calculate the position of the car 4 based on the rotation angle. For example, the car position calculation part 92 determines that the vehicle has traveled in a specific section by counting the number of times the door zone plate is detected. Also good. Further, it may be determined that the vehicle has traveled in a specific section by a position switch that informs a reference position such as the uppermost floor or the lowermost floor provided in the hoistway.

The learning determination unit 93 determines whether the car 4 is preliminarily determined based on whether or not the rotation speed of the electric motor 1 output from the learning speed calculation unit 91 has reached a constant speed running state and the position of the car 4 output from the car position calculation unit 92. It is determined whether the vehicle is traveling in the set specific section. The learning determination unit 93 outputs a learning command while the rotational speed is in a constant speed traveling state and traveling in a specific section, and does not output a learning command otherwise. That is, the angle error estimator 9 estimates the angle error when the rotation speed of the electric motor 1 is constant and the frequency of the angle error is constant. Thereby, the frequency of the angle error can be treated as known.

When the angle error calculation unit 94 receives a learning command from the learning determination unit 93, the angle error calculation unit 94 performs Fourier transform on the current of the current detector 7, which is the output of the frequency analysis unit 8, using information on the rotation angle of the electric motor 1. An angle error is calculated using the obtained Fourier coefficient. The angle error calculation unit 94 stores in advance a conversion formula for obtaining an angle error from a Fourier coefficient, which is a result of Fourier transforming the current using information on the rotation angle of the electric motor 1, and calculates the angle error from the Fourier coefficient. To do. The angle error calculated by the angle error calculation unit 94 is an error amplitude and an error phase described later.

The area calculation unit 95 uses the Fourier coefficient coordinates based on the Fourier coefficient obtained by Fourier-transforming the current of the current detector 7 that is the output of the frequency analysis unit 8 using the rotation angle information of the electric motor 1. Calculate the area of the area to be created. The area of the region created by the coordinates of the Fourier coefficient is output to the output determination unit 96 for use in determining whether the angle error is estimated. The area formed by the coordinates of the Fourier coefficient and its meaning will be described later.

The output determination unit 96 selects and outputs an error amplitude and an error phase when the area created by the coordinates of the Fourier coefficient that is the output of the area calculation unit 95 is minimized. When the area created by the coordinates of the Fourier coefficient is minimized, the estimated value of the angle error when the influence of the resonance is small and equal when the influence of the resonance is the smallest can be selected.

The error signal calculation unit 97 calculates and outputs a correction signal (angle error estimated value) for correcting the periodic angle error of the rotation detection unit 2 using the error amplitude and error phase output from the output determination unit 96. To do. The correction signal is a value obtained by multiplying the error angle by the sine value or cosine value of the rotation angle obtained by adding the error phase calculated by the angle error calculation unit 94 to the rotation angle of the electric motor 1 that is the output of the rotation detection unit 2 ( Equation (1) below.

Next, the cyclic angle error included in the angle output from the rotation detector 2 will be described. The periodic angular error of the rotation detector 2 can be approximately expressed using a sine wave as shown in the following equation (1). In addition, since there is no essential difference between the notation by the sine wave and the notation by the cosine wave, the present invention unifies the notation by the sine wave.

θ _e = A ₁ sin (Xθ _m + φ) (1)

θ _e : Periodic angle error of rotation detection unit 2 X: Order of angle error of rotation detection unit 2 with respect to mechanical angle of electric motor 1 (known value)
θ _m : rotation angle of motor 1 A ₁ : error amplitude of angle error of rotation detector 2 φ: phase shift (error phase) of rotation detector 2 relative to mechanical angle of motor 1

X represents the order of the angle error of the rotation detection unit 2 with respect to the mechanical angle of the electric motor 1, and is a known value. Therefore, if the rotation angle θ _m of the electric motor 1, that is, the rotation speed of the electric motor 1 is known, the frequency of the cyclic angle error of the rotation detection unit 2 represented by the equation (1) can be known. A ₁ indicates the error amplitude of the angle error of the rotation detector 2, and φ indicates the phase shift (error phase) of the rotation detector 2 with respect to the mechanical angle of the electric motor 1.

The angle error estimator 9 calculates the correction signal shown in the equation (1) using the estimation result of the error amplitude A _{1 and} the estimation result of the error phase φ.

Next, the periodic angle error shown in the equation (1) is converted into a periodic velocity error by the velocity calculation unit 10 as in the following equation (2).

ω _e = XA ₁ ωcos (Xθ _m + φ) = A ₂ cos (Xθ _m + φ) (2)
ω _e : Periodic speed error of rotation detector 2 A ₂ : Amplitude of speed error due to angle error in equation (1) ω: Motor rotation speed

Therefore, the rotation speed of the electric motor 1 output from the speed calculation unit 10 includes the periodic speed error expressed by the equation (2). The speed output from the speed calculation unit 10 is compared with the speed command value output from the speed command calculation unit 11 and input to the speed controller 12. The speed controller 12 determines a current command from the difference between the speed command value and the detected speed, but the rotational speed output from the speed calculation unit 10 includes a periodic speed error as shown in Equation (2), so that the speed control is performed. The current command calculated by the device 12 includes the pulsation caused by the equation (2), that is, the pulsation caused by the angle error of the rotation detector 2 of the equation (1). The pulsation of the current command calculated by the speed controller 12 is expressed by equation (3) from equation (2).

I _e = A ₃ cos (Xθ _m + φ + φ _c ) (3)
I _e : Current command pulsation A ₃ : Current pulsation amplitude due to angle error φ _c : Phase delay by speed controller 12

The current pulsation shown in equation (3) is expressed as the following equation (4) by Fourier series expansion.

I _e = A _n cos (Xθ _m ) + B _n sin (Xθ _m ) (4)

Since the current command output from the speed controller 12 includes the current pulsation represented by the equation (4), the current pulsation represented by the equation (4) is also included in the current detected by the current detector 7. By synthesizing the trigonometric function, the current pulsation of equation (4) can be rewritten as follows.

I _e = √ (A _n ² + B _n ² ) · sin (Xθ _m + γ) (5)
G = √ (A _n ² + B _n ² )
γ = tan ⁻¹ (A _n / B _n )

In equation (5),
G = √ (A _n ² + B _n ² ) indicates the amplitude of the current pulsation caused by the angle error of the rotation detector 2,
G = √ (A _n ² + B _n ² ) = A ₃ .
γ = tan ⁻¹ (A _n / B _n ) represents a phase difference with respect to the mechanical angle of the electric motor 1 of the current pulsation,
γ = tan ⁻¹ (A _n / B _n ) = φ + φ _c
In the following explanation,
G = √ (A _n ² + B _n ² ) is the amplitude of current pulsation,
_Let γ = tan ⁻¹ (A _n / B _n ) be called the current pulsation phase.

Here, a method for obtaining the error amplitude A ₁ and the error phase φ of the angle error from the Fourier coefficients A _n and B _n obtained from the frequency analysis result of the current will be described. First, the error phase φ of the angle error can be obtained as in the following equation (6).

φ = tan ⁻¹ (A _n / B _n ) −φ _c (6)

The phase delay φ _c by the speed controller 12 is determined by the frequency characteristics of the speed controller 12. Since the frequency of the angle error is known, the frequency of the speed error is a known value. FIG. 3 shows the frequency characteristics of the gain and phase of the speed controller 12. For example, when the frequency of the angle error is A, the phase delay by the speed controller 12 is −150 [deg]. When the frequency of the angle error is B, the phase lag by the speed controller 12 is −170 [deg]. The phase lag of the speed controller 12 is uniquely determined by the speed controller 12. Therefore, the phase delay φ _c by the speed controller 12 can be obtained from the frequency characteristics of the speed controller 12, and the error phase φ of the angle error can be obtained from the equation (6).

Similarly, the amplitude can be obtained from the frequency characteristic of the speed controller 12. Considering that the frequency of the angle error is known from the frequency characteristics of the speed controller 12, the gain C1 from the speed to the current command can be obtained. From the frequency characteristic of the speed controller 12 in FIG. 3, for example, when the frequency of the angle error is A, the gain by the speed controller 12 is −10 [dB]. When the frequency of the angle error is B, the gain by the speed controller 12 is −35 [dB]. Therefore, the error amplitude A ₁ of the angle error can be calculated as in the following equation (7).

A ₃ / A ₂ = C 1 → √ (A _n ² + B _n ² ) / XA ₁ ω = C 1
A ₁ = √ (A _n ² + B _n ² ) / XωC1 (7)

The angle error calculation unit 94 stores the equations (6) and (7) in the memory in advance, and calculates the error amplitude A ₁ and the error phase φ of the angle error from the Fourier coefficient obtained from the current frequency analysis result. To do. The angle error calculation unit 94 may store the phase lag and gain due to the frequency characteristics of the speed controller 12 as table data for each of a plurality of frequencies, for example.

Next, calculation of Fourier coefficients A _n and B _{n in} the above equations (4) and (5) will be described. The frequency analysis unit 8 performs a Fourier transform on the current I detected by the current detector 7 using information on the rotation angle of the electric motor 1 and calculates a Fourier coefficient. The Fourier coefficient is calculated by the following well-known formula (8).

A _n = (1 / π) ∫ ₀ ^2π I · cos (Xθ _m ) dθ
B _n = (1 / π) ∫ ₀ ^2π I · sin (Xθ _m ) dθ (8)

The calculation of the Fourier coefficient of equation (8) is performed at regular intervals. For example, the fixed period corresponds to one rotation of the rotation angle θ _m of the electric motor 1 according to the equation (8). In other words, the Fourier coefficient is calculated for each distance traveled by the car 4 for one rotation of the electric motor 1 and for each time required for one rotation of the electric motor 1. The interval for calculating the Fourier coefficient may be two rotations or three rotations instead of one rotation of the electric motor 1. In this case, since an average value for several cycles is obtained, the influence of fluctuations in current pulsation and disturbance can be reduced.

The frequency analysis unit 8 is not configured to calculate the Fourier coefficient shown in Expression (8), but extracts a specific frequency signal like a filter combining a notch filter and a band pass filter, and an amplitude detection unit and a phase detection unit. The unit may calculate the amplitude and phase of a specific frequency of the input signal.
In this case, since tan ⁻¹ (A _n / B _n ) in the equation (6) can be directly detected in the phase, the error phase φ can be obtained by subtracting the phase delay φ _c by the speed controller 12. it can. In the amplitude, since √ (A _n ² + B _n ² ) in the equation (7) can be directly detected, the angle error amplitude A ₁ can be obtained by the same procedure as that in the equation (7).

Next, in the estimation of the angle error represented by the equation (1) according to the present embodiment, when the change in the amplitude and phase of the current pulsation is small, that is, the frequency of the angle error of the rotation detector 2 and the mechanical system of the elevator. A learning operation method for selecting an estimated value when the resonance frequency does not match and the influence of resonance is small will be described.

First, the characteristics of the mechanical system of the elevator will be described. 4 shows an example of the gain characteristic from the angle error included in the rotation angle of the electric motor 1 detected by the rotation detector 2 to the current detected by the current detector 7, and FIG. 5 shows an example of the phase characteristic.
4 and 5, when the frequency of the angle error is A and C, the gain and phase are constant at any position of the car 4 in the hoistway.
On the other hand, when the frequency of the angle error is B, the gain and phase do not change when the position of the car 4 is near the lowermost floor indicated by the broken line or the uppermost floor indicated by the solid line, but resonates near the intermediate floor indicated by the dotted line. The characteristics are shown. The characteristics shown in FIGS. 4 and 5 are an example. However, since elevators have different machine specifications for each property, the gain characteristics and phase characteristics shown in FIGS. 4 and 5 differ for each elevator property. In addition, since the speed of the elevator is different for each property, it is difficult to know in advance at which position and at which speed the angular error and the mechanical system resonate. Therefore, it is difficult to avoid resonance based on prior information, and it is desirable to simultaneously perform learning of angle error and determination of success or failure of estimation.

Next, a method for determining the success or failure of the estimation by the angle error estimation unit 9 will be described. 6 the car 4 of the elevator when brought into a particular section traveling are those two Fourier coefficients A _n calculated by the frequency analysis unit 8, the B _n, the horizontal axis B _n, the vertical axis is plotted as A _n It is. Hereinafter, this plane is referred to as a Fourier coefficient coordinate plane. As can be seen from equation (5), in FIG. 6, the distance G from the origin G = √ (A _n ² + B _n ² ) is equal to the amplitude of the current pulsation, and the angle γ = tan ⁻¹ ( A _n / B _n ) is equal to the phase of the current pulsation.

Next, the relationship between the area surrounded by the Fourier coefficient coordinates on the Fourier coefficient coordinate plane, the amplitude of the current pulsation, and the phase of the current pulsation will be described. In the Fourier coefficient coordinate plane of FIG. 6, when considering the change of the amplitude of current pulsation G = √ (A _n ² + B _n ² ) and the phase of current pulsation γ = tan ⁻¹ (A _n / B _n ), the current pulsation When the change in amplitude G = √ (A _n ² + B _n ² ) and current pulsation phase γ = tan ⁻¹ (A _n / B _n ) is small, the area enclosed by the coordinates of the Fourier coefficients of three or more points is small. Become. In the section A of FIG. 6, since the change in the coordinates of the Fourier coefficient is small, the area surrounded by the coordinates of the Fourier coefficient is small, the current pulsation amplitude G = √ (A _n ² + B _n ² ) and the current pulsation phase γ = The change of tan ⁻¹ (A _n / B _n ) is small. In section B, the area surrounded by the coordinates of the Fourier coefficient is larger than in section A, and the amplitude of current pulsation G = √ (A _n ² + B _n ² ) and the phase of current pulsation γ = tan ⁻¹ (A The change of _n / _Bn ) is also large. Therefore, calculating the area surrounded by the Fourier coefficient coordinates on the Fourier coefficient coordinate plane is equivalent to calculating the current pulsation amplitude and phase variation.

In the Fourier coefficient coordinate plane, the area surrounded by the coordinates of three or more Fourier coefficients can be calculated by the coordinate method. For example, the Fourier coefficients calculated by the frequency analysis unit 8 when traveling in a specific section of the elevator car 4 are calculated in the order of the calculation results (B _n1 , A _n1 ), (B _n2 , A _n2 ), (B _n3 , A _n3 ), the area S surrounded by the coordinates of these three sets of Fourier coefficients can be calculated by the following equation (9).

S =
_{(1/2) | (B n1 A} n2 -B n2 A n1) + (B n2 A n3 -B n3 A n2) + (B n3 A n1 -B n1 A n3) |
(9)

The distance √ (A _n1 ² + B _n1 ² ) from the origin to the coordinates (B _n1 , A _n1 ) is represented by G ₁ ,
The distance √ (A _n2 ² + B _n2 ² ) from the origin to the coordinates (B _n2 , A _n2 ) is G ₂ ,
The distance √ (A _n3 ² + B _n3 ² ) from the origin to the coordinates (B _n3 , A _n3 ) is represented by G ₃ ,
In other words, the amplitude of the current pulsation at each coordinate is G ₁ , G ₂ , G ₃ ,
Coordinates from the origin (B _{_n1,} A _n1) to the distance vector angle ^{_{_{tan -1 (A n1 / B n1}}} ) the gamma _1,
Coordinates from the origin (B _{_n2,} A _n2) distance vector angle to ^{_{_{tan -1 (A n2 / B n2}}} ) a gamma _2,
Coordinates from the origin (B _{_n3,} A _n3) distance vector angle to ^{_{_{tan -1 (A n3 / B n3}}} ) and gamma _3,
That is, if the phase of the current pulsation at each coordinate is γ ₁ , γ ₂ , γ ₃ , the area of equation (9) can be rewritten as follows.

S =
(1/2) | G ₁ G ₂ sin (γ ₂ −γ ₁ ) + G ₂ G ₃ sin (γ ₃ −γ ₂ ) + G ₁ G ₃ sin (γ ₁ −γ ₃ ) |
(10)

The gain characteristic and the phase characteristic from the angle error included in the rotation angle of the electric motor 1 detected by the rotation detector 2 to the current detected by the current detector 7 are the characteristics as shown in FIGS. because when the frequency is a or C of the gain and phase constant values at any position of the car 4 hoistway, i.e., the frequency of the angle error does not resonate with the elevator mechanical system if a and C, G ₁ , G ₂ , G ₃ and γ ₁ , γ ₂ , γ ₃ are constant values, and the area calculated by equation (9) is zero.
On the other hand, when the frequency of the angle error is B, the gain, that is, the amplitude and the phase are constant when the car 4 is near the lowermost floor and the uppermost floor, but the gain and the phase change greatly as they approach the intermediate floor. That is, when the frequency of the angle error is B, resonance occurs near the middle floor and no resonance occurs in other places. In this case, the area calculated by Expression (9) is 0 near the lowermost floor and the uppermost floor, but the area of Expression (9) increases as the intermediate floor is approached. When resonance occurs, the gain and phase change at the same time. Therefore, by calculating the area of Equation (9), the change in the amplitude and phase of the current pulsation can be calculated.

The frequency analysis unit 8 is a filter that combines a notch filter and a bandpass filter, and extracts a specific frequency signal and calculates the amplitude and phase of the specific frequency of the input signal by the amplitude detection unit and the phase detection unit. In this case, the area can be calculated in the same procedure. That is, the frequency analysis unit 8 is a filter that combines a notch filter and a bandpass filter, extracts a specific frequency signal, and calculates the amplitude and phase of the specific frequency of the input signal by the amplitude detection unit and the phase detection unit. Since the amplitude G of the current pulsation and the phase γ of the current pulsation in Equation (10) are obtained, the area can be calculated by Equation (10). Furthermore, if the current pulsation amplitude G and error phase φ are used, since _An = Gsin (φ) and B _n = Gcos (φ), the amplitude and phase of the current pulsation are detected and converted into Fourier coefficients. Thus, the area can also be calculated by the calculation method according to equation (9).

Further, the frequency analysis unit 8 is a filter that combines a notch filter and a band pass filter, and extracts a specific frequency signal, and the amplitude detection unit and the phase detection unit calculate the amplitude and phase of the specific frequency of the input signal. 6, instead of the coordinate plane formed by the Fourier coefficients _An and B _n as shown in FIG. 6, the coordinate plane with the vertical axis as amplitude and the horizontal axis as phase, or the vertical axis as phase and the horizontal axis as amplitude. Even in the coordinate plane, the change in the amplitude and phase of the current pulsation can be calculated by calculating the area by the calculation according to the equation (9). That is, in correspondence with amplitude A _n, the phase B _n corresponding, or is associated with phase A _n, it is sufficient to correspond to amplitude B _n.

From the above, obtaining the area of equation (9) is equivalent to obtaining the amount of change in amplitude and phase of current pulsation, and that the area is large means that the amount of change in amplitude and phase of current pulsation is large. Become. In FIG. 6, the area surrounded by the coordinates of the Fourier coefficients of three points has been described as an example. However, since the area surrounded by the coordinates of the Fourier coefficients can be calculated if the number of points is three or more, it is not necessary to limit to three points.

Next, using the area enclosed by the coordinates of the Fourier coefficient shown in equation (9), the angle when the angle error and the mechanical system of the elevator are not resonating and the amount of change in the amplitude and phase of the current pulsation is small. The learning operation for selecting the estimated error value will be described with reference to the flowchart of FIG.

When the angle error estimation unit 9 sends a learning operation command from the learning determination unit 93 or the output determination unit 96 to the speed command calculation unit 11 in FIG. 1 and starts the learning operation, for example, the learning determination unit 93 starts the learning speed calculation unit 91. It is determined whether or not the rotation speed of the electric motor 1 is constant according to the output (step S71). If the rotation speed of the electric motor 1 is not constant, step S71 is continued until the rotation speed becomes constant. This is because when the rotation speed of the electric motor 1 is constant, the frequency of the angle error of the rotation detector 2 is constant, and the angle detection error is easily estimated.
After the rotation speed of the electric motor 1 becomes constant, the angle error calculation unit 94 performs Fourier transform on the current of the current detector 7, which is an output from the frequency analysis unit 8, using information on the rotation angle of the electric motor 1. The obtained Fourier coefficient calculation result is stored in a memory (step S72).

Next, the angle error calculation unit 94 determines whether the calculation result of the Fourier coefficient is stored three times or three times or more (hereinafter, described as three times or more) (step S73). If three or more Fourier coefficients have not been stored, the process returns to step S72, and the storage of the Fourier coefficient output from the frequency analysis unit 8 is repeated. When the Fourier coefficients are stored three times or more, the area calculation unit 95 calculates the area created by the coordinates of the Fourier coefficients according to the equation (9). Further, the angle error calculation unit 94 uses the Fourier coefficient to calculate an estimated value of the angle error by using the stored Fourier coefficient and angle detection error conversion formula (step S74). The estimated value of the angle error is an error amplitude and an error phase. Then, the output determination unit 96 associates the area and the estimated value of the angle error and stores them in the memory.

When the calculation of the estimated value of the area and angle detection error in step S74 ends, the output determination unit 96 compares the stored area, that is, the area calculated in the previous flow with the area calculated in step S74. (Step S75). The initial value of the areas to be compared is the maximum value that can be stored so that the first calculation result is always saved. The estimated value of the angle detection error may be an arbitrary value.

If the output determination unit 96 compares the stored area with the calculated area in step S75 and determines that the area calculated in step S74 is smaller, the output determination unit 96 calculates the area and angle calculated in step S74. The estimated value of the detection error is stored together. At this time, the already stored area and angle detection error are deleted (step S76).
On the other hand, as a result of comparing the area stored in step S75 with the calculated area, if it is determined that the area calculated in step S74 is greater than or equal to the stored area, no processing is performed. That is, the stored angular error estimation value and area are stored, and the process proceeds to the next step.

Next, the learning determination unit 93 determines whether or not the car has traveled in a specific section from the output of the car position calculation unit 92 (step S77). If it is determined that the vehicle is not traveling in a specific section, the process returns to step S72 and the processes up to step S77 are repeated. If it is determined that the vehicle has traveled in a specific section, the learning operation is terminated.
The fact that the car 4 has traveled in a specific section is determined based on the car position calculated by the car position calculation unit 92. For example, by counting the number of times the door zone plate is detected, the specific section is It may be determined that the vehicle has traveled. Further, it may be determined that the vehicle has traveled in a specific section by a position switch that informs a reference position such as the uppermost floor or the lowermost floor provided in the hoistway.

After completion of the learning operation, the error signal calculation unit 97 calculates a periodic angle error of the rotation detection unit 2 using the estimation result of the error phase and the error amplitude, and outputs it as a correction signal. In SU2, the angle error of the rotation detector 2 is corrected.

In the above operation, it is determined whether or not the car has traveled in a specific section lastly to determine whether or not the learning operation has been performed. First, the output of the learning speed calculation unit 91 is the rotational speed of the motor 1. Indicates that the vehicle is running at a constant speed, and when the output of the car position calculation unit 92 indicates that the car 4 is traveling in a preset specific section, the learning determination unit 93 outputs a learning command, and learning The operation may be started. In this case, step S77 is performed in parallel with step S71.

When the learning operation is performed by the above processing, the area created by the coordinates of the Fourier coefficient obtained by Fourier transforming the current of the current detector 7 using the information of the rotation angle of the electric motor 1 when traveling in a specific section is obtained. It is possible to extract the angle detection error when it becomes the minimum. The minimum area created by the Fourier coefficient coordinates when traveling in a specific section is the same when the amplitude and phase change of the current pulsation are small in the specific section. Specifically, the estimated value of the angle detection error with the smallest estimation error can be extracted.

FIG. 8 shows the relationship between the area calculated using the coordinates of three adjacent Fourier coefficients in the Fourier coefficient coordinate plane of FIG. 6 and the car position. For example, when coordinates of Fourier coefficients as shown in FIG. 6 are obtained in the learning operation of the angle error, the area calculated by the area calculation unit 95 of the angle error estimation unit 9 is as shown in FIG. In section A, the area is small, and in section B, the area is larger than section A. According to the learning operation flow of FIG. 7, the estimated value of the angle error in the section A is extracted. Since a small area is equivalent to non-resonance, it is possible to obtain a highly reliable estimation value of the angle error.

According to the method of the first embodiment, it is possible to simultaneously estimate the angle error and determine the success or failure of the estimation based on the amplitude and phase of the current pulsation, extract the estimated value of the angle error when the influence of resonance is the smallest, In addition, since it is not necessary to check information on resonance, the time required for learning the angle error can be shortened. Further, according to the method of the first embodiment, no matter what mechanical system is connected as the load of the electric motor 1 instead of the elevator, the influence of resonance is the most based on the amount of change in the amplitude and phase of the current pulsation. The estimated value when there are few can be extracted. Although FIG. 6 shows a method performed with four Fourier coefficients, it is sufficient if the area can be calculated at three or more points.

The frequency analysis unit 8 is not configured to calculate a Fourier coefficient, but a specific frequency signal is extracted like a filter combining a notch filter or a band pass filter, and the input signal is detected by an amplitude detection unit or a phase detection unit. Even with a configuration for calculating the amplitude and phase of a specific frequency, a highly reliable estimation error angle can be obtained by the same procedure.

Embodiment 2. FIG.
In the first embodiment, the method of extracting the estimated value when the frequency of the angle error coincides with the frequency of the mechanical system and estimation is difficult and the estimated value when the influence of resonance is the smallest is shown. In the second embodiment, the learning operation of the angle error shown in the first embodiment is performed a plurality of times while changing the speed during the learning operation, and by confirming the consistency of the results of the plurality of learnings, A method for obtaining an estimated value of the angle error with higher reliability than that of the first embodiment will be described.

In the method of the first embodiment, the angle estimation error when the influence of resonance is small can be extracted by extracting the estimated value of the angle error when the area surrounded by the coordinates formed by the Fourier coefficients is minimized. However, depending on the machine specifications of the elevator, there is a speed at which resonance occurs regardless of the position in the hoistway. In such a situation, an estimated value of the angle error when the amplitude and phase change of the current pulsation is the smallest among the resonances is extracted. In this case, since there is a possibility that a highly reliable estimated value of the angle error may not be obtained, it is desirable to change the speed of the learning operation. Therefore, in the second embodiment, the angle error learning is performed a plurality of times while changing the speed during the learning operation, and the consistency of the results of the plurality of learning is confirmed.

FIG. 9 shows a flowchart of the learning operation in the second embodiment. Note that the configuration of the elevator control device and the circuit error estimation unit 9 are basically the same as those shown in FIGS.

When the angle error estimation unit 9 sends a learning operation command from the learning determination unit 93 or the output determination unit 96 to the speed command calculation unit 11 in FIG. 1 and starts the learning operation, for example, the learning determination unit 93 starts the learning speed calculation unit 91. It is determined whether or not the rotation speed of the electric motor 1 is constant according to the output (step S901). If the rotation speed of the electric motor 1 is not constant, step S901 is continued until the rotation speed becomes constant. This is because when the rotation speed of the electric motor 1 is constant, the frequency of the angle error of the rotation detector 2 is constant, so that it is easy to estimate the angle detection error.
After the rotation speed of the electric motor 1 becomes constant, the angle error calculation unit 94 performs Fourier transform on the current of the current detector 7, which is an output from the frequency analysis unit 8, using information on the rotation angle of the electric motor 1. The obtained Fourier coefficient calculation result is stored in a memory (step S902).

Next, the angle error calculation unit 94 determines whether or not the calculation result of the Fourier coefficient is stored three times or more (step S903). If three or more Fourier coefficients have not been stored, the process returns to step S902, and the storage of the Fourier coefficient output from the frequency analysis unit 8 is repeated. When the Fourier coefficients are stored three times or more, the area calculation unit 95 calculates the area created by the coordinates of the Fourier coefficients according to the equation (9). In addition, the angle error calculation unit 94 uses the Fourier coefficient to calculate an estimated value of the angle error by using the stored Fourier coefficient and angle detection error conversion formula (step S904). The estimated value of the angle error is an error amplitude and an error phase. Then, the output determination unit 96 associates the area and the estimated value of the angle error and stores them in the memory.

When the calculation of the estimated value of the area and the angle detection error in step S904 is finished, the output determination unit 96 compares the stored area, that is, the area calculated in the previous flow with the area calculated in step S904. (Step S905). The initial value of the areas to be compared is the maximum value that can be stored so that the first calculation result is always saved. The estimated value of the angle detection error may be an arbitrary value.

When the output determination unit 96 compares the stored area with the calculated area in step S905 and determines that the area calculated in step S904 is smaller, the output determination unit 96 calculates the area and angle calculated in step S904. The estimated value of the detection error is stored together. At this time, the already stored area and angle detection error are erased (step S906).
On the other hand, as a result of comparing the area stored in step S905 with the calculated area, if it is determined that the area calculated in step S904 is greater than or equal to the stored area, no processing is performed. That is, the stored angular error estimation value and area are stored, and the process proceeds to the next step.

Next, the learning determination unit 93 determines from the output of the car position calculation unit 92 whether the car has traveled in a specific section (step S907). If it is determined that the vehicle is not traveling in a specific section, the process returns to step S902 and the processes up to step S907 are repeated. When it is determined that the vehicle has traveled in the specific section, the output determination unit 96 stores the number of learning times and the stored angle error estimated value together in the memory (step S908). The number of learnings can be calculated by counting the number of times of traveling in a specific section using the position of the car 4 calculated by the car position calculation unit 92.
The fact that the car 4 has traveled in a specific section is determined based on the car position calculated by the car position calculation unit 92. For example, by counting the number of times the door zone plate is detected, the specific section is It may be determined that the vehicle has traveled. Further, it may be determined that the vehicle has traveled in a specific section by a position switch that informs a reference position such as the uppermost floor or the lowermost floor provided in the hoistway.

Next, the learning determination unit 93 determines whether or not the angle error has been learned twice or more, that is, a plurality of times, based on the stored number of learning (step S909). If it is determined in step S909 that the learning has not been performed twice or more, a learning operation command is sent from the learning determination unit 93 or the output determination unit 96 to the speed command calculation unit 11 in FIG. 1 to change the speed of the learning operation. Then, the learning operation is performed again (step S914). By changing the learning speed, the frequency of the periodic angular error changes. For example, in FIG. As a result, the tendency of current pulsation due to the cyclic angle error of the rotation detector 2 changes. For example, resonance occurs near the middle floor at frequency B, but resonance does not occur at any level in frequencies A and C. Thereby, the angle error can be estimated under different conditions from the first learning. In addition, there is no particular restriction on the speed setting for learning driving, but the tendency of current pulsation changes greatly if the speed is changed extremely, such as the slowest speed or the fastest speed at which the car can travel. The angle error can be learned.

As for the driving direction of the learning driving, it returns to the same direction as the first learning, that is, when the first learning is finished, returns to the position where the learning is started, and performs again in the same driving direction, or continues from the position where the first learning is completed. You may drive in the same direction, and you may drive in the direction opposite to the driving direction of the first learning from the position where the first learning is completed. The driving direction is not limited as long as the learning speed is changed. The learning method when the learning driving speed is changed is the same as the first learning.

If the learning determination unit 93 determines that the angle error has been learned twice or more in step S909, the consistency determination of the angle error estimation results is performed multiple times (step S910). The estimated value of the angle error is composed of the error amplitude and error phase of equation (1). First, the consistency of the error amplitude estimation result is confirmed. That is, the difference between the estimated values of the error amplitude obtained by a plurality of learning operations is calculated, and it is confirmed whether or not the difference is within a preset value. For example, when the learning operation is performed three times, the consistency check is performed by all combinations of the estimation results obtained as in the first time, the second time, the second time, the third time, the first time, and the third time. The set value of the difference in error amplitude for determining consistency may be stored in advance in a memory or may be input from the outside.
When the error amplitude consistency check is completed, the error phase consistency check is then performed. The error phase consistency is also checked in the same manner as the error amplitude. That is, the difference of the estimated value of the error phase obtained by a plurality of learning operations is calculated, and it is confirmed whether or not the difference is within the set value. The set value of the difference in error amplitude for determining consistency may be stored in advance in a memory or may be input from the outside.

In the error amplitude consistency check and error phase consistency check, if it is determined that either of the learning results does not match multiple times (step S911), whether or not the number of learning is within the maximum number of learning times. Is determined (step S912).
If the number of times of learning is within the maximum number of times of learning, the learning operation is performed again by changing the learning speed (step S914).
In the confirmation of the consistency of the estimation results, when the error amplitude is matched and the error phase is not matched, in the learning operation in which the speed is changed, only the error phase may be re-learned. On the other hand, in checking the consistency of the estimation result, if the error phase is matched and the error amplitude is not matched, only the error amplitude may be re-learned. Further, both the error amplitude and the error phase may be relearned.

If it is determined in step S912 that the number of learnings exceeds the maximum number of learnings, the learning determining unit 93 outputs a learning impossibility notification to a host control device (not shown) (step S913). Note that the maximum number of learning times may be stored in advance in a memory or may be commanded from the outside. In addition, it is desirable that the learning impossible notification output in step S913 be displayed on an elevator control panel (not shown) so as to notify the abnormality. Further, when a learning impossible notification is output, it is desirable to ensure safety by pausing the elevator.

When the consistency check in step S911 is completed or a learning impossible notification is output in step S913, the learning determination unit 93 completes the learning operation. If the consistency of the results of the plurality of learning operations can be confirmed in step S911, the error signal calculation unit 97 uses the error phase and error amplitude estimation results after completion of the learning operation. 1 is output as a correction signal, and the angle error of the rotation detector 2 is corrected by the subtractor SU2 in FIG.

When the learning operation is performed by the above processing, it is possible to extract an angle detection error when the area formed by the coordinates of the Fourier coefficient is minimized when traveling in a specific section. In addition, it is possible to obtain an estimated value of the angle error with higher reliability than the method of the first embodiment by confirming the consistency of the learning results a plurality of times. Furthermore, by limiting the number of learnings, it becomes possible to ensure safety when angle error learning cannot be performed normally.

Embodiment 3 FIG.
In the third embodiment, when the area surrounded by the coordinates formed by the Fourier coefficient corresponding to the estimated value of the angle error obtained by the learning operation is larger than the reference value, the learning operation speed is changed to perform the learning. Thus, a method for obtaining the estimated value of the angle error with higher reliability than that of the first embodiment will be described.

In the method of the first embodiment, the angle estimation error when the influence of resonance is small can be extracted by extracting the estimated value of the angle error when the area surrounded by the coordinates formed by the Fourier coefficients is minimized. However, depending on the machine specifications of the elevator, there is a speed at which resonance occurs regardless of the position in the hoistway. In such a situation, an estimated value of the angle error when the amplitude and phase change of the current pulsation is the smallest among the resonances is extracted. In this case, since there is a possibility that a highly reliable estimated value of the angle error may not be obtained, it is desirable to change the speed of the learning operation.
In the second embodiment, the consistency of the learning results is performed a plurality of times while changing the speed during the learning operation. In the third embodiment, a reference value of the area enclosed by the coordinates formed by the Fourier coefficient is prepared in advance by storing it in a memory, for example, and the Fourier corresponding to the estimated value of the angle error obtained by the learning operation is prepared. When the area surrounded by the coordinates formed by the coefficients is larger than the reference value, the angle error is learned by changing the speed during the learning operation.

FIG. 10 shows a flowchart of the learning operation in the third embodiment. Note that the configuration of the elevator control device and the circuit error estimation unit 9 are basically the same as those shown in FIGS.

When the angle error estimation unit 9 sends a learning operation command from the learning determination unit 93 or the output determination unit 96 to the speed command calculation unit 11 in FIG. 1 and starts the learning operation, for example, the learning determination unit 93 starts the learning speed calculation unit 91. It is determined whether or not the rotation speed of the electric motor 1 is constant according to the output (step S101). If the rotation speed of the electric motor 1 is not constant, step S101 is continued until the rotation speed becomes constant. This is because when the rotation speed of the electric motor 1 is constant, the frequency of the angle error of the rotation detector 2 is constant, so that it is easy to estimate the angle detection error.
After the rotation speed of the electric motor 1 becomes constant, the angle error calculation unit 94 performs Fourier transform on the current of the current detector 7, which is an output from the frequency analysis unit 8, using information on the rotation angle of the electric motor 1. The obtained Fourier coefficient calculation result is stored in a memory (step S102).

Next, the angle error calculation unit 94 determines whether or not the calculation result of the Fourier coefficient is stored three times or three times or more (hereinafter described as three times or more) (step S103). If three or more Fourier coefficients have not been stored, the process returns to step S102, and the storage of the Fourier coefficient output from the frequency analysis unit 8 is repeated. When the Fourier coefficients are stored three times or more, the area calculation unit 95 calculates the area created by the coordinates of the Fourier coefficients according to the equation (9). Further, the angle error calculation unit 94 uses the Fourier coefficient to calculate an estimated value of the angle error based on the stored Fourier coefficient and angle detection error conversion formula (step S104). The estimated value of the angle error is an error amplitude and an error phase. Then, the output determination unit 96 associates the area and the estimated value of the angle error and stores them in the memory.

When the calculation of the estimated value of the area and the angle detection error in step S104 is completed, the output determination unit 96 compares the stored area, that is, the area calculated in the previous flow with the area calculated in step S104. (Step S105). The initial value of the areas to be compared is the maximum value that can be stored so that the first calculation result is always saved. The estimated value of the angle detection error may be an arbitrary value.

When the output determination unit 96 compares the stored area with the calculated area in step S105 and determines that the area calculated in step S104 is smaller, the output determination unit 96 calculates the area and angle calculated in step S104. The estimated value of the detection error is stored together. At this time, the already stored area and angle detection error are deleted (step S106).
On the other hand, as a result of comparing the area stored in step S105 with the calculated area, if it is determined that the area calculated in step S104 is greater than or equal to the stored area, no processing is performed. That is, the stored angular error estimation value and area are stored, and the process proceeds to the next step.

Next, the learning determination unit 93 determines from the output of the car position calculation unit 92 whether the car has traveled in a specific section (step S107). If it is determined that the vehicle is not traveling in a specific section, the process returns to step S102 and the processes up to step S107 are repeated. If it is determined that the vehicle has traveled in a specific section, the learning operation is terminated.
The fact that the car 4 has traveled in a specific section is determined based on the car position calculated by the car position calculation unit 92. For example, by counting the number of times the door zone plate is detected, the specific section is It may be determined that the vehicle has traveled. Further, it may be determined that the vehicle has traveled in a specific section by a position switch that informs a reference position such as the uppermost floor or the lowermost floor provided in the hoistway.

Next, the learning determination unit 93 determines that the area enclosed by the coordinates created by the Fourier coefficient corresponding to the estimated value of the angle error obtained by the learning operation is the reference value of the area enclosed by the coordinates created by the Fourier coefficient stored in advance. It is determined that the following is true (step S108). In step 108, it is determined that the area enclosed by the coordinates created by the Fourier coefficient corresponding to the estimated value of the angle error obtained in the learning operation is larger than the reference value of the area enclosed by the coordinates created by the Fourier coefficient stored in advance. In this case, a learning operation command is sent from the learning determination unit 93 or the output determination unit 96 to the speed command calculation unit 11 in FIG. 1, and the learning operation is performed again by changing the speed of the learning operation (step S109).
By changing the learning speed, the frequency of the periodic angular error changes. For example, in FIG. As a result, the tendency of current pulsation due to the cyclic angle error of the rotation detector 2 changes. For example, resonance occurs near the middle floor at frequency B, but resonance does not occur at any level in frequencies A and C. Thereby, the angle error can be estimated under different conditions from the first learning. In addition, there is no particular restriction on the speed setting for learning driving, but the tendency of current pulsation changes greatly if the speed is changed extremely, such as the slowest speed or the fastest speed at which the car can travel. The angle error can be learned.

When the confirmation in step S108 is completed, the learning determination unit 93 completes the learning operation. In step S108, it is confirmed that the area enclosed by the coordinates created by the Fourier coefficient corresponding to the estimated value of the angle error obtained by the learning operation is equal to or less than the reference value of the area enclosed by the coordinates created by the Fourier coefficient stored in advance. In such a case, after completion of the learning operation, the error signal calculation unit 97 calculates the cyclic angle error of the rotation detection unit 2 using the estimation result of the error phase and the error amplitude, and outputs it as a correction signal. The angle error of the rotation detector 2 is corrected by the subtractor SU2 in FIG.

When the learning operation is performed by the above processing, it is possible to extract an angle detection error when the area formed by the coordinates of the Fourier coefficient is minimized when traveling in a specific section. Further, by comparing and confirming the area surrounded by the coordinates formed by the Fourier coefficients with the reference value, it is possible to obtain an estimated value of the angle error with higher reliability than the method of the first embodiment.

Embodiment 4 FIG.
In the fourth embodiment, a method of determining a resonance based on the length of a line segment between coordinates formed by a Fourier coefficient as an evaluation value that is a geometric amount on a coordinate plane and obtaining a highly reliable estimated value will be described.
First, on the Fourier coefficient coordinate plane, the relationship between the length of the line segment between the coordinates formed by the Fourier coefficient, the current pulsation amplitude, and the current pulsation phase will be described. In the Fourier coefficient coordinate plane of FIG. 6, when considering the change of the amplitude of current pulsation G = √ (A _n ² + B _n ² ) and the phase of current pulsation γ = tan ⁻¹ (A _n / B _n ), the current pulsation When the change in amplitude G = √ (A _n ² + B _n ² ) and current pulsation phase γ = tan ⁻¹ (A _n / B _n ) is small, the length of the line segment between the coordinates formed by the two points of the Fourier coefficient Becomes shorter. In the section A of FIG. 6, since the change in the coordinates of the Fourier coefficient is small, the length of the line segment between the coordinates formed by the Fourier coefficient is short, the current pulsation amplitude G = √ (A _n ² + B _n ² ) and the current pulsation. The change in phase γ = tan ⁻¹ (A _n / B _n ) is small. In section B, the length of the line segment between the coordinates formed by the Fourier coefficient is longer than in section A, and the current pulsation amplitude G = √ (A _n ² + B _n ² ) and the current pulsation phase γ = tan ^{The change of -1} (A _n / B _n ) is also large. Therefore, calculating the length of the line segment between coordinates created by the Fourier coefficient in the Fourier coefficient coordinate plane is equivalent to calculating the amplitude and phase change amount of the current pulsation.

In the Fourier coefficient coordinate plane, the length of the line segment formed by the coordinates of the two Fourier coefficients can be calculated by equation (11). For example, when the Fourier coefficients calculated by the frequency analysis unit 8 when traveling in a specific section of the elevator car 4 are (B _n1 , A _n1 ), (B _n2 , A _n2 ) in order of the calculation results, these 2 The length L of the line segment created by the coordinates of the set of Fourier coefficients is given by the following equation.

L = √ ((A _n2 −A _n1 ) ² + (B _n2 −B _n1 ) ² ) (11)

The distance √ (A _n1 ² + B _n1 ² ) from the origin to the coordinates (B _n1 , A _n1 ) is represented by G ₁ ,
The distance √ (A _n2 ² + B _n2 ² ) from the origin to the coordinates (B _n2 , A _n2 ) is G ₂ ,
That is, the amplitude of the current pulsation at each coordinate is G ₁ and G ₂ ,
Coordinates from the origin (B _{_n1,} A _n1) to the distance vector angle ^{_{_{tan -1 (A n1 / B n1}}} ) the gamma _1,
Coordinates from the origin (B _{_n2,} A _n2) distance vector angle to ^{_{_{tan -1 (A n2 / B n2}}} ) a gamma _2,
In other words, if the phase of the current pulsation at each coordinate is γ ₁ and γ ₂ , the length of the line segment in the equation (11) can be rewritten as follows.

_{L = √ ((G 2 sinγ} 2 -G 2 sinγ 1) 2 + (G 2 cosγ 2 -G 2 cosγ 1) 2) (12)

The gain characteristic and the phase characteristic from the angle error included in the rotation angle of the electric motor 1 detected by the rotation detector 2 to the current detected by the current detector 7 are the characteristics as shown in FIGS. because when the frequency is a or C of the gain and phase constant values at any position of the car 4 hoistway, i.e., the frequency of the angle error does not resonate with the elevator mechanical system if a and C, G ₁ , G ₂ , γ ₁ and γ ₂ are constant values, and the length of the line segment calculated by the equation (12) is zero.
On the other hand, when the frequency of the angle error is B, the gain, that is, the amplitude and the phase are constant when the car 4 is near the lowermost floor and the uppermost floor, but the gain and the phase change greatly as they approach the intermediate floor. That is, when the frequency of the angle error is B, resonance occurs near the middle floor and no resonance occurs in other places. In this case, the length of the line segment calculated by Expression (11) is 0 near the lowermost floor and the uppermost floor, but the length of the line segment of Expression (11) becomes longer as the intermediate floor is approached. . When resonance occurs, the gain and phase change at the same time. Therefore, by calculating the length of the line segment of Equation (11), the change in amplitude and phase of the current pulsation can be calculated.

The frequency analysis unit 8 is a filter that combines a notch filter and a bandpass filter, and extracts a specific frequency signal and calculates the amplitude and phase of the specific frequency of the input signal by the amplitude detection unit and the phase detection unit. In this case, the length of the line segment can be calculated by the same procedure. That is, the frequency analysis unit 8 is a filter that combines a notch filter and a bandpass filter, extracts a specific frequency signal, and calculates the amplitude and phase of the specific frequency of the input signal by the amplitude detection unit and the phase detection unit. Since the amplitude G of the current pulsation and the phase γ of the current pulsation in Equation (12) are obtained, the length of the line segment can be calculated from Equation (12). Furthermore, if the current pulsation amplitude G and error phase φ are used, since _An = Gsin (φ) and B _n = Gcos (φ), the amplitude and phase of the current pulsation are detected and converted into Fourier coefficients. Thus, the length of the line segment can also be calculated by the calculation method according to equation (11).

Further, the frequency analysis unit 8 is a filter that combines a notch filter and a band pass filter, and extracts a specific frequency signal, and the amplitude detection unit and the phase detection unit calculate the amplitude and phase of the specific frequency of the input signal. 6, instead of the coordinate plane formed by the Fourier coefficients _An and B _n as shown in FIG. 6, the coordinate plane with the vertical axis as amplitude and the horizontal axis as phase, or the vertical axis as phase and the horizontal axis as amplitude. Even in the coordinate plane, the change in amplitude and phase of the current pulsation can be calculated by calculating the length of the line segment by the calculation according to the equation (11). That is, in correspondence with amplitude A _n, B _n corresponding to phase or is associated with phase A _n, it is sufficient to correspond to amplitude B _n.

From the above, obtaining the length of the line segment of the equation (11) is equivalent to obtaining the amount of change in the amplitude and phase of the current pulsation, and that the length of the line segment is a change in the amplitude and phase of the current pulsation. The amount will be great.

Next, using the length of the line segment between the coordinates created by the Fourier coefficient expressed by Equation (11), the angle error and the mechanical system of the elevator are not resonating, and the amount of change in the amplitude and phase of the current pulsation is The operation of the learning operation for selecting the estimated value of the angle error when it is small will be described with reference to the flowchart of FIG. 11 and the configuration diagram of the angle error estimating unit 8 of FIG.

FIG. 12 is a diagram illustrating an example of the configuration of the angle error estimation unit 9 according to the fourth embodiment. The basic configuration is the same as that in FIG. 2, and a line segment length calculation unit 98 is provided instead of the area calculation unit 95 in FIG. 2. Therefore, detailed description is omitted.

When the angle error estimation unit 9 sends a learning operation command from the learning determination unit 93 or the output determination unit 96 to the speed command calculation unit 11 in FIG. 1 and starts the learning operation, for example, the learning determination unit 93 starts the learning speed calculation unit 91. It is determined whether or not the rotation speed of the electric motor 1 is constant according to the output (step S111). If the rotation speed of the electric motor 1 is not constant, step S111 is continued until the rotation speed becomes constant. This is because when the rotation speed of the electric motor 1 is constant, the frequency of the angle error of the rotation detector 2 is constant, and the angle detection error is easily estimated.
After the rotation speed of the electric motor 1 becomes constant, the angle error calculation unit 94 performs Fourier transform on the current of the current detector 7, which is an output from the frequency analysis unit 8, using information on the rotation angle of the electric motor 1. The obtained Fourier coefficient calculation result is stored in a memory (step S112).

Next, the angle error calculation unit 94 determines whether or not the calculation result of the Fourier coefficient is stored twice (step S113). If the Fourier coefficients for two times are not stored, the process returns to step S112, and the storage of the Fourier coefficient that is the output from the frequency analysis unit 8 is repeated. When the Fourier coefficients are stored twice, the line segment length calculation unit 98 calculates the length of the line segment between the coordinates of the Fourier coefficient by Expression (11). In addition, the angle error calculation unit 94 uses the Fourier coefficient to calculate an estimated value of the angle error using the stored Fourier coefficient and angle detection error conversion formula (step S114). The estimated value of the angle error is an error amplitude and an error phase. Then, the output determination unit 96 associates the length of the line segment with the estimated value of the angle error and stores it in the memory.

When the calculation of the length of the line segment and the estimated value of the angle detection error in step S114 ends, the output determination unit 96 determines the length of the stored line segment, that is, the length of the line segment calculated in the previous flow. And the length of the line segment calculated in step S114 is compared (step S115). Note that the initial value of the length of the line segment to be compared is the maximum value that can be stored so that the first calculation result is always saved. The estimated value of the angle detection error may be an arbitrary value.

As a result of comparing the length of the stored line segment with the calculated length of the line segment in step S115, the output determination unit 96 determines that the length of the line segment calculated in step S114 is shorter. In this case, the length of the line segment calculated in step S114 and the estimated value of the angle detection error are stored together. At this time, the length of the line segment and the angle detection error already stored are deleted (step S116).
On the other hand, as a result of comparing the length of the line segment stored in step S115 with the calculated length of the line segment, it is determined that the length of the line segment calculated in step S114 is greater than or equal to the length of the stored line segment. If so, do nothing. That is, the stored angle error value and line segment length that have already been stored are stored, and the process proceeds to the next step.

Next, the learning determination unit 93 determines whether or not the car has traveled in a specific section from the output of the car position calculation unit 92 (step S117). If it is determined that the vehicle is not traveling in a specific section, the process returns to step S112 and the processes up to step S117 are repeated. If it is determined that the vehicle has traveled in a specific section, the learning operation is terminated.
The fact that the car 4 has traveled in a specific section is determined based on the car position calculated by the car position calculation unit 92. For example, by counting the number of times the door zone plate is detected, the specific section is It may be determined that the vehicle has traveled. Further, it may be determined that the vehicle has traveled in a specific section by a position switch that informs a reference position such as the uppermost floor or the lowermost floor provided in the hoistway.

In the above operation, it is determined whether or not the car has traveled in a specific section lastly to determine whether or not the learning operation has been performed. First, the output of the learning speed calculation unit 91 is the rotational speed of the motor 1. Indicates that the vehicle is running at a constant speed, and when the output of the car position calculation unit 92 indicates that the car 4 is traveling in a preset specific section, the learning determination unit 93 outputs a learning command, and learning The operation may be started. In this case, step S117 is performed in parallel with step 111.

When the learning operation is performed by the above processing, the line segment between the coordinates of the Fourier coefficient obtained by performing Fourier transform on the current of the current detector 7 using information on the rotation angle of the electric motor 1 when traveling in a specific section. It is possible to extract an angle detection error when the length of is minimum. The minimum length of the line segment between the Fourier coefficients when traveling in a specific section is the same when the amplitude and phase change of the current pulsation are small in the specific section, so the influence of resonance with the mechanical system is the least. Finally, an estimated value of the angle detection error with the smallest estimation error can be extracted.

FIG. 13 shows the relationship between the length of the line segment calculated using the coordinates of two adjacent Fourier coefficients on the Fourier coefficient coordinate plane of FIG. 6 and the car position. For example, when the coordinates of the Fourier coefficient as shown in FIG. 6 are obtained in the learning operation of the angle error, the length of the line segment calculated by the line length calculation unit 98 of the angle error estimation unit 9 is as shown in FIG. It becomes like this. In the section A, the length of the line segment is short, and in the section B, the length of the line segment is longer than that in the section A. According to the learning operation flow of FIG. 11, an estimated value of the angle error in the section A is extracted. Since a short line length is equivalent to not resonating, a highly reliable estimation value of the angle error can be obtained.

According to the method of the fourth embodiment, it is possible to simultaneously estimate the angle error and determine the success or failure of the estimation based on the amplitude and phase of the current pulsation, extract the estimated value of the angle error when the influence of resonance is the smallest, In addition, since it is not necessary to check information on resonance, the time required for learning the angle error can be shortened. Further, according to the method of the fourth embodiment, no matter what mechanical system is connected as the load of the electric motor 1 instead of the elevator, the influence of resonance is the most based on the amount of change in the amplitude and phase of the current pulsation. The estimated value when there are few can be extracted.

Embodiment 5 FIG.
In the fourth embodiment, the length of the line segment formed by the coordinates of the Fourier coefficient can be used to eliminate the estimated value when the frequency of the angle error coincides with the frequency of the mechanical system and estimation becomes difficult, and the influence of resonance The method of extracting the estimated value when there is the least is shown.

In the fifth embodiment, the learning operation of the angle error shown in the fourth embodiment is performed a plurality of times while changing the speed during the learning operation, and the consistency check of the results of the plurality of learnings is performed. A method for obtaining an estimated value of the angle error with higher reliability than that of the fourth embodiment will be described.

In the method of the fourth embodiment, the angle estimation error when the influence of resonance is small can be extracted by extracting the estimated value of the angle error when the length of the line segment between the coordinates formed by the Fourier coefficient is minimum. It was. However, depending on the machine specifications of the elevator, there is a speed at which resonance occurs regardless of the position in the hoistway. In such a situation, an estimated value of the angle error when the amplitude and phase change of the current pulsation is the smallest among the resonances is extracted. In this case, since there is a possibility that a highly reliable estimated value of the angle error may not be obtained, it is desirable to change the speed of the learning operation. Therefore, in the fifth embodiment, the angle error learning is performed a plurality of times while changing the speed during the learning operation, and the consistency of the results of the plurality of learnings is confirmed.

FIG. 14 shows a flowchart of the learning operation in the fifth embodiment. The configuration of the elevator control device and the angle error estimating unit 9 is basically the same as the original shown in FIGS. Further, the flowchart of FIG. 14 is basically the same as the flowchart of FIG. 9 shown in the second embodiment. In FIG. 9, the resonance determination is based on the area, but in FIG. It has been judged.
In FIG. 14, the flow denoted by the same reference numerals as those in FIG. 9 is the same as the operation of the second embodiment, and the description thereof is omitted. In the fifth embodiment, the operations from the process of storing the Fourier coefficient (step S1403) to the process of storing the length of the line segment and the estimated angle error (step S1406) are different.

In the fifth embodiment, in step 1403, the angle error calculation unit 94 determines whether the calculation results of Fourier coefficients are stored twice. If the Fourier coefficients for two or more times are not stored, the process returns to step S902, and the storage of the Fourier coefficients output from the frequency analysis unit 8 is repeated. When the Fourier coefficients are stored twice, the line segment length calculation unit 98 calculates the length of the line segment between the coordinates of the Fourier coefficient by Expression (11). Further, the angle error calculation unit 94 uses the Fourier coefficient to calculate an estimated value of the angle error by using the stored Fourier coefficient and angle detection error conversion formula (step S1404). The estimated value of the angle error is an error amplitude and an error phase. Then, the output determination unit 96 associates the length of the line segment with the estimated value of the angle error and stores it in the memory.

When the calculation of the length of the line segment and the estimated value of the angle detection error in step S1404 ends, the output determination unit 96 determines the length of the stored line segment, that is, the length of the line segment calculated in the previous flow. And the length of the line segment calculated in step S1404 is compared (step S1405). Note that the initial value of the length of the line segment to be compared is the maximum value that can be stored so that the first calculation result is always saved. The estimated value of the angle detection error may be an arbitrary value.

As a result of comparing the length of the stored line segment with the calculated line segment length in step S1405, the output determination unit 96 determines that the length of the line segment calculated in step S1404 is shorter. In this case, the length of the line segment calculated in step S1404 and the estimated value of the angle detection error are stored together. At this time, the lengths and angle detection errors that have already been stored are deleted (step S1406).
On the other hand, as a result of comparing the length of the line segment stored in step S1405 with the calculated length of the line segment, it is determined that the length of the line segment calculated in step S1404 is greater than or equal to the length of the stored line segment. If so, do nothing. That is, the stored angle error value and line segment length that have already been stored are stored, and the process proceeds to the next step.

The subsequent processing is the same as the flowchart shown in FIG. When the learning operation is performed by the above processing, it is possible to extract the angle detection error when the length of the line segment between the coordinates of the Fourier coefficient becomes the minimum when traveling in the specific section. In addition, it is possible to obtain an estimated value of the angle error with higher reliability than the method of the fourth embodiment by confirming the consistency of the learning results a plurality of times. Furthermore, by limiting the number of learnings, it becomes possible to ensure safety when angle error learning cannot be performed normally.

Embodiment 6 FIG.
In the sixth embodiment, when the length of the line segment between the coordinates formed by the Fourier coefficient corresponding to the estimated value of the angle error obtained by the learning operation is longer than the reference value, the learning operation speed is changed and learning is performed. A method for obtaining an estimated value of the angle error with higher reliability than that of the fourth embodiment will be described.

In the method of the fourth embodiment, the angle estimation error when the influence of resonance is small can be extracted by extracting the estimated value of the angle error when the length of the line segment between the coordinates formed by the Fourier coefficient is minimum. It was. However, depending on the machine specifications of the elevator, there is a speed at which resonance occurs regardless of the position in the hoistway. In such a situation, an estimated value of the angle error when the amplitude and phase change of the current pulsation is the smallest among the resonances is extracted. In this case, since there is a possibility that a highly reliable estimated value of the angle error may not be obtained, it is desirable to change the speed of the learning operation.
In the fifth embodiment, the consistency check of the learning results is performed a plurality of times while changing the speed during the learning operation. In the sixth embodiment, a reference value of the length of the line segment between coordinates formed by the Fourier coefficient is prepared in advance by storing it in a memory, for example, and an estimated value of the angle error obtained by the learning operation If the length of the line segment between the coordinates generated by the Fourier coefficient corresponding to is longer than the reference value, the angle error is learned by changing the speed during the learning operation.

FIG. 15 shows a flowchart of the learning operation in the sixth embodiment. The configuration of the elevator control device and the angle error estimating unit 9 is basically the same as the original shown in FIGS. Further, the flowchart of FIG. 15 is basically the same as the flowchart of FIG. 10 shown in the third embodiment. In FIG. 10, the resonance determination is based on the area, but in FIG. It has been judged.
In FIG. 15, the flow denoted by the same reference numerals as those in FIG. 10 is the same as the operation of the third embodiment, and thus the description thereof is omitted. In the sixth embodiment, the operation from the process of storing the Fourier coefficient (step S1503) to the process of storing the length of the line segment and the estimated angle error (step S1506) and the determination based on the reference value of the length of the line segment (Step S1508) is different.

In the sixth embodiment, in step 1503, the angle error calculation unit 94 determines whether or not the Fourier coefficient calculation results are stored twice. If two or more Fourier coefficients have not been stored, the process returns to step S102, and the storage of the Fourier coefficient output from the frequency analysis unit 8 is repeated. When the Fourier coefficients are stored twice, the line segment length calculation unit 98 calculates the length of the line segment between the coordinates of the Fourier coefficient by Expression (11). Further, the angle error calculation unit 94 calculates an estimated value of the angle error by using the Fourier coefficient and the conversion formula between the stored Fourier coefficient and the angle detection error (Step S1504). The estimated value of the angle error is an error amplitude and an error phase. Then, the output determination unit 96 associates the length of the line segment with the estimated value of the angle error and stores it in the memory.

When the calculation of the length of the line segment and the estimated value of the angle detection error in step S1504 ends, the output determination unit 96 determines the length of the stored line segment, that is, the length of the line segment calculated in the previous flow. And the length of the line segment calculated in step S1504 is compared (step S1505). Note that the initial value of the length of the line segment to be compared is the maximum value that can be stored so that the first calculation result is always saved. The estimated value of the angle detection error may be an arbitrary value.

As a result of comparing the length of the stored line segment with the calculated length of the line segment in step S1505, the output determination unit 96 determines that the length of the line segment calculated in step S1504 is shorter. In this case, the length of the line segment calculated in step S1504 and the estimated value of the angle detection error are stored together. At this time, the lengths and angle detection errors that have already been stored are deleted (step S1506).
On the other hand, as a result of comparing the length of the line segment stored in step S1505 with the calculated length of the line segment, it is determined that the length of the line segment calculated in step S1504 is greater than or equal to the length of the stored line segment. If so, do nothing. That is, the stored angle error value and line segment length that have already been stored are stored, and the process proceeds to the next step.

Next, the learning determination unit 93 determines from the output of the car position calculation unit 92 whether the car has traveled in a specific section (step S107). This operation is the same as in the third embodiment.
Next, the learning determination unit 93 determines that the length of the line segment between the coordinates created by the Fourier coefficient corresponding to the estimated value of the angle error obtained by the learning operation is between the coordinates created by the Fourier coefficient stored in advance. It is determined that the length is equal to or less than the reference value of the length of the line segment (step S1508). In step S1508, the length of the line segment between the coordinates created by the Fourier coefficient corresponding to the estimated value of the angle error obtained by the learning operation is the length of the line segment between the coordinates created by the Fourier coefficient stored in advance. 1 is sent from the learning determination unit 93 or the output determination unit 96 to the speed command calculation unit 11 in FIG. 1, and the learning operation is performed again by changing the speed of the learning operation. (Step S109).
By changing the learning speed, the frequency of the periodic angular error changes. For example, in FIG. As a result, the tendency of current pulsation due to the cyclic angle error of the rotation detector 2 changes. For example, resonance occurs near the middle floor at frequency B, but resonance does not occur at any level in frequencies A and C. Thereby, the angle error can be estimated under different conditions from the first learning. In addition, there is no particular restriction on the speed setting for learning driving, but the tendency of current pulsation changes greatly if the speed is changed extremely, such as the slowest speed or the fastest speed at which the car can travel. The angle error can be learned.

When the confirmation in step S1508 is completed, the learning determination unit 93 completes the learning operation. In step S1508, the length of the line segment between the coordinates created by the Fourier coefficient corresponding to the estimated value of the angle error obtained by the learning operation is the length of the line segment between the coordinates created by the Fourier coefficient stored in advance. When it is confirmed that the value is equal to or less than the reference value, the error signal calculation unit 97 uses the estimation result of the error phase and the error amplitude after the learning operation is completed, to calculate the periodic angle error of the rotation detection unit 2. The calculated value is output as a correction signal, and the angle error of the rotation detector 2 is corrected by the subtractor SU2 in FIG.

When the learning operation is performed by the above processing, it is possible to extract the angle detection error when the length of the line segment between the coordinates of the Fourier coefficient becomes the minimum when traveling in a specific section. Further, by comparing and confirming the length of the line segment between the coordinates formed by the Fourier coefficient with the reference value, it is possible to obtain an estimated value of the angle error with higher reliability than the method of the fourth embodiment.

Note that the layout and roping method of the entire elevator are not limited to the example in FIG. For example, the present invention can be applied to a 2: 1 roping elevator. Further, for example, the position of the hoisting machine including the electric motor 1 is not limited to the example of FIG.

The present invention is not limited to the above-described embodiments, and it goes without saying that all of these possible combinations are included.

Industrial applicability

The present invention can be applied to various types of elevators such as machine room-less elevators, double deck elevators, one-shaft multi-car elevators, and skew elevators.

Claims

A current detector for detecting a current flowing in an electric motor that generates power to raise and lower the car in the hoistway;
A rotation detector for detecting a rotation angle of the electric motor;
A frequency analysis unit that outputs a component of a specific frequency obtained by frequency analysis of the current detected by the current detector;
An angle error estimation unit that estimates the amplitude and phase of a cyclic angle error that is uniquely determined according to the rotation angle from the rotation detection unit using the specific frequency component, and outputs the angle error estimated value;
With
The angle error estimator controls the car to perform a learning operation that operates the specific section, and inputs the current detected during the learning operation to the frequency analysis unit to obtain the component of the specific frequency obtained. A plurality of consecutive acquisitions are performed, and an evaluation value, which is a geometric amount in a coordinate plane created by a set number of the specific frequency components that are consecutive among the acquired specific frequency components, is calculated, and the angular error estimation value is calculated. Calculating and associating the evaluation value with the angle error estimate, and selecting the angle error estimate when the evaluation value is minimized;
Elevator control device.
2. The elevator control according to claim 1, wherein the set number is 2 and the geometric amount is a length of a line segment between coordinates formed by an amplitude and a phase of the component of the specific frequency in the coordinate plane. apparatus.
2. The elevator control according to claim 1, wherein the set number is 3 or more, and the geometric amount is an area of a region surrounded by coordinates formed by an amplitude and a phase of the component of the specific frequency in the coordinate plane. apparatus.
The angle error estimation unit selects the angle error estimated value when the amount of change in the component of the specific frequency corresponding to the angle error of the current detected by the current detector is minimized based on the evaluation value. The elevator control device according to any one of claims 1 to 3.
When the evaluation value corresponding to the selected estimated angle error value is greater than or equal to a reference value, the angle error estimation unit changes the operation speed of the learning operation and performs the learning operation again to perform the angle error. The elevator control device according to any one of claims 1 to 4, which estimates an estimated value.
The angle error estimator performs the learning operation a plurality of times at different driving speeds, compares the angle error estimated values obtained in the learning operation a plurality of times, and confirms the consistency, thereby obtaining consistency. The elevator control device according to any one of claims 1 to 5, wherein when there is not, the operation speed is further changed to perform the learning operation.
The elevator control device according to any one of claims 1 to 6, wherein the frequency analysis unit performs a Fourier transform as a frequency analysis to calculate a Fourier coefficient.
The frequency analysis unit includes a band-pass filter that passes the specific frequency component, and performs an amplitude calculation and a phase calculation on an output current of the band-pass filter. Elevator control device.
The elevator control device according to any one of claims 1 to 8, wherein the rotation detection unit includes a resolver, an encoder, or a magnetic sensor.
Analyzing the frequency of the current flowing in the drive motor that raises and lowers the car in the hoistway, and calculating a specific frequency component;
By using the component of the specific frequency, the amplitude and phase of a cyclic angle error that is uniquely determined according to the rotation angle included in the rotation detection unit that detects the rotation angle of the electric motor is estimated as an angle error estimated value. Estimating process;
With
In the step of estimating the angle error as an angle error estimated value,
Control the car to perform a learning operation that operates the specific section,
Continuously acquiring a plurality of components of the specific frequency obtained in the step of calculating the component of the specific frequency of the current flowing through the electric motor detected during the learning operation,
While calculating an evaluation value that is a geometric amount in a coordinate plane created by a specific number of components of the specific frequency among the acquired components of the specific frequency, the angle error estimate is calculated to calculate the evaluation value and the Associating an angle error estimate and selecting the angle error estimate when the evaluation value is minimal;
A method for obtaining an angle error of a rotation angle of a rotation detection unit for an electric motor for driving an elevator.
Whether to go up and down in the hoistway,
An electric motor that generates power to raise and lower the car;
A rotation detector for detecting a rotation angle of the electric motor;
A current detector for detecting a current flowing through the electric motor;
A frequency analysis unit that outputs a component of a specific frequency obtained by frequency analysis of the current detected by the current detector;
An angle error estimation unit that estimates the amplitude and phase of a cyclic angle error that is uniquely determined according to the rotation angle from the rotation detection unit using the specific frequency component, and outputs the angle error estimated value;
With
The angle error estimator controls the car to perform a learning operation that operates the specific section, and inputs the current detected during the learning operation to the frequency analysis unit to obtain the component of the specific frequency obtained. A plurality of consecutive acquisitions are performed, and an evaluation value, which is a geometric amount in a coordinate plane created by a set number of the specific frequency components that are consecutive among the acquired specific frequency components, is calculated, and the angular error estimation value is calculated. Calculating and associating the evaluation value with the angle error estimate, and selecting the angle error estimate when the evaluation value is minimized;
Elevator device.