JP2018023203A - Motor controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To achieve highly precise compensation of a detection error included in a detection angle of a rotor without necessity of connection of an external apparatus for driving a motor at a constant speed.SOLUTION: An estimation unit 22 inputs an angle detection value θs and a torque detection value τd into an equation of motion showing a relationship between the angle detection value θs including a detection error θc and the torque detection value τd and sequentially applies a least square method to estimate the detection error θc. A compensation unit 23 compensates for the detection error included in the angle detection value θs on the basis of the detection error estimated by the estimation unit 22.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a control device for an electric motor.

  The motor control device performs drive control of the motor, such as speed control and torque control, using angle information of the rotor detected by an angle detection unit such as a resolver. In general, the angle information detected by the angle detection means includes a detection error due to the characteristics of the angle detection means. When drive control of the electric motor is performed using angle information including this detection error, pulsation may occur in the rotational speed and output torque of the electric motor. Therefore, it is necessary to appropriately correct the detection error included in the angle information detected by the angle detection means.

  Patent Documents 1 and 2 are known as conventional techniques for solving the above problems. Patent Document 1 discloses a technique for compensating for a detection error included in an actual detection angle by setting an ideal rotation angle of an electric motor and obtaining a difference between the ideal rotation angle and the actual detection angle.

Patent Document 2 discloses a technique for compensating for a detection error of a detection signal detected by a magnetic position sensor that detects a magnetic pole position of a motor. Specifically, Patent Document 2 describes the phase difference β ^* of the dq current command value when the dq current command value for setting the output torque of the motor to 0 in the no-load state and the actual dq axis current level. A technique for obtaining a phase shift amount from the phase difference β and compensating for a detection error is disclosed.

JP2013-72686A JP 2007-318894 A

  However, in Patent Document 1, it is necessary to accurately drive a motor to be compensated at a constant rotational speed in order to obtain a detection error with high accuracy. The actual detection angle of the motor to be compensated includes a detection error as described above. Therefore, in Patent Document 1, it is difficult to drive the electric motor accurately at a constant speed.

  In Patent Document 1, it is conceivable that an external device for forcibly driving the motor at a constant speed is connected to the motor and the motor is driven at a constant speed. The problem of increasing the size of the apparatus occurs.

  In Patent Document 2, since the detection error of the magnetic position sensor is compensated using the phase shift amount obtained in the no-load state, there is a possibility that the detection error of the detection signal cannot be accurately compensated in the loaded state. There is.

  It is an object of the present invention to provide a motor control device that accurately compensates for a detection error included in a rotor detection angle without connecting an external device that drives the motor at a constant speed.

An electric motor control device according to an aspect of the present invention includes:
An angle detection unit for detecting an actual rotation angle of the rotor including a detection error periodically changing according to a rotation angle of the rotor of the electric motor and a true value of the rotation angle;
A torque detector for detecting an actual output torque of the electric motor;
An estimation unit that estimates the detection error from a periodic variation of the actual output torque generated by using the actual rotation angle for drive control;
Based on the estimated detection error, the detection error included in the actual rotation angle is compensated.

  According to this aspect, the detection error is estimated based on the periodic fluctuation of the actual output torque generated by using the actual rotation angle including the detection error for the drive control. When an actual rotation angle including a periodically varying detection error is used for drive control of the electric motor, a pulsation corresponding to the magnitude of the detection error occurs in the actual output torque in the same cycle as the detection error. Therefore, if the actual rotation angle and the actual output torque are known at the same time, the detection error included in the actual rotation angle can be accurately estimated. Therefore, this aspect can accurately estimate the detection error included in the actual rotation angle.

  Further, since this aspect focuses on the actual rotation angle and the actual output torque, it is not necessary to make the rotation speed of the electric motor constant. Therefore, this aspect can estimate a detection error without using an external device for keeping the rotation speed constant. Therefore, this aspect can suppress an increase in cost and an increase in apparatus scale.

  In the above aspect, the estimation unit estimates the detection error by inputting the actual rotation angle and the actual output torque into an equation of motion indicating a relationship between the actual rotation angle and the actual output torque. Also good.

  According to this aspect, since the detection error is estimated using the equation of motion indicating the relationship between the actual rotation angle and the actual output torque, the detection error can be estimated with high accuracy.

In the above aspect, the actual rotation angle is represented by a sum of an error function obtained by multiplying a trigonometric function having the true value as a variable by an error gain, and the true value,
The estimation unit may estimate the detection error by inputting a plurality of actual rotation angles and a plurality of actual output torques to the equation of motion, and determining the error gain by a least square method. .

  Since the detection error periodically varies according to the rotation angle, it can be expressed by an error function obtained by multiplying a trigonometric function having a true value of the rotation angle as a variable by an error gain. This mode uses an equation of motion in which the actual rotation angle is expressed by the sum of the error function and the true value. In this aspect, a plurality of actual rotation angles and a plurality of actual output torques are input to this equation of motion, and an error gain is determined by the least square method, thereby estimating a detection error. Therefore, the detection error can be estimated with high accuracy. From the viewpoint of reducing the amount of memory, this aspect may use a sequential least square method.

In the above aspect, the estimation unit generates correspondence information in which a plurality of rotation angles and a plurality of detection errors are associated with each other from the error function represented by the error gain determined by the least square method.
The compensation unit may determine a detection error corresponding to the actual rotation angle from the correspondence information, and compensate the actual rotation angle using the determined detection error.

  According to this aspect, correspondence information in which a plurality of rotation angles and a plurality of detection errors are associated with each other is generated in advance from an error function represented by an error gain determined by the least square method. According to this aspect, the detection error corresponding to the actual rotation angle is determined from the correspondence information, and the actual rotation angle is compensated using the determined detection error. Therefore, the detection error corresponding to the actual rotation angle can be determined quickly.

In the above aspect, further comprising a current detector for detecting a current flowing through the electric motor,
The torque detector may calculate the actual output torque based on the detected current.

  According to this aspect, since the actual output torque is calculated based on the current detected by the current detection unit, the actual output torque can be detected without using a torque detection device having a relatively large scale. Therefore, this aspect can further suppress an increase in the device scale.

  In the above aspect, the angle detection unit may be configured by a resolver.

  Since the detection error of the resolver periodically varies according to the rotation angle of the rotor, the detection error can be accurately estimated by using the above aspect.

  According to the present invention, the detection error included in the detection angle of the rotor can be accurately compensated without connecting an external device.

It is a block diagram which shows the structure of the control apparatus of the electric motor which concerns on Embodiment 1 of this invention. It is a block diagram which shows the flow of a process in an estimation part. It is explanatory drawing of correspondence information. It is a block diagram which shows the structure of the control apparatus which concerns on Embodiment 2 of this invention. It is a block diagram which paid its attention to the torque estimation part.

  Hereinafter, a control device for an electric motor (hereinafter referred to as “motor”) according to an embodiment of the present invention will be described with reference to the drawings. In recent years, vector control is generally used for speed control of a three-phase brushless motor such as a permanent magnet synchronous motor. Therefore, also in this embodiment, vector control is adopted as the motor speed control.

(Embodiment 1)
FIG. 1 is a block diagram showing a configuration of an electric motor control apparatus 1 according to Embodiment 1 of the present invention. The control device 1 is a control device that controls the motor M.

  The control device 1 includes a subtractor 11, a speed controller 12, a current command generator 13, subtractors 14a and 14b, a current controller 15, a dq / uvw converter 16, an inverter 17, a torque detector 19, and an angle detector 20. , Uvw / dq conversion unit 21, estimation unit 22, compensation unit 23, adder 24, multiplier 25, and speed calculation unit 26.

  In FIG. 1, a subtractor 11, a speed control unit 12, a current command generation unit 13, subtractors 14a and 14b, a current control unit 15, a dq / uvw conversion unit 16, a uvw / dq conversion unit 21, an estimation unit 22, and a compensation unit. 23, the adder 24, the multiplier 25, and the speed calculation unit 26 are configured by, for example, a computer including a CPU, a ROM, and the like.

  The subtractor 11 subtracts the speed detection value from the target speed to calculate a speed deviation. The target speed varies depending on the device to which the motor M is applied. For example, if the motor M is used to drive a winch drum, an upper swing body, or a lower traveling body of a construction machine, the target speed is set to a value corresponding to the operation amount of the operation lever. Further, if the motor M is applied to a travel motor of an electric vehicle or a hybrid vehicle, the target speed is set to a value corresponding to the accelerator operation amount.

  The speed control unit 12 receives a speed deviation from the subtractor 11 and calculates a torque command value for setting the speed deviation to zero. Here, the speed control unit 12 may calculate the torque command value using, for example, PID (proportional / integral / derivative) control or PI (proportional / integral) control.

  The current command generator 13 calculates a d-axis target current value Id_ref and a q-axis target current value Iq_ref based on the torque command value. Here, the current command generation unit 13 may generate, for example, values predetermined for the torque command value as the target current values Id_ref and Iq_ref.

  The subtractor 14a subtracts the d-axis actual current value Id from the target current value Id_ref to calculate a d-axis current deviation. The subtractor 14b subtracts the q-axis actual current value Iq from the target current value Iq_ref to calculate a q-axis current deviation.

  The current control unit 15 calculates a d-axis voltage command value vd for setting the d-axis current deviation from the subtractor 14a to zero, and sets the q-axis current deviation from the subtractor 14b to zero. The q-axis voltage command value vq is calculated. The voltage command value vd is a command value for controlling the field component of the voltage output to the motor M, and the voltage command value vq is a command value for controlling the torque component of the voltage output to the motor M. Here, the current control unit 15 may calculate the voltage command values vd and vq using, for example, PID control and PI control.

  The dq / uvw conversion unit 16 performs coordinate conversion of the voltage command values vd and vq using the electrical angle α output from the multiplier 25, generates a PWM signal having three phases of UVW, and outputs the PWM signal to the inverter 17. .

  The inverter 17 is configured by, for example, a three-phase inverter, generates UVW three-phase AC power from the three-phase PWM signal, and outputs the generated power to the motor M.

  The motor M is composed of, for example, a three-phase electric motor and is driven according to UVW three-phase AC power output from the inverter 17. For example, if the control device 1 is applied to a construction machine such as a crane or an excavator, the motor M rotates the upper swing body or rotates the winch drum. Moreover, if the control apparatus 1 is applied to an electric vehicle, the electric vehicle is caused to travel.

  The torque detector 19 is configured by a measuring device that directly detects torque applied to the rotation shaft of the motor M, for example. Then, the torque detection unit 19 outputs the detected torque to the estimation unit 22 as a torque detection value.

  The angle detection unit 20 is configured by, for example, a resolver, detects the rotation angle of the rotor of the motor M, and outputs the detected rotation angle as an angle detection value θs. Here, as the angle detection value θs, for example, the rotation angle of the rotor with respect to the reference rotation position can be adopted. The speed detection value has a positive value when, for example, the rotor rotates in a predetermined first direction (for example, clockwise direction), and the rotor has a second value opposite to the first direction. When rotating in a direction (for example, counterclockwise direction), the rotation direction may be distinguished by using a plus or minus sign so as to have a negative value.

  The uvw / dq conversion unit 21 performs coordinate conversion of the alternating currents Iu, Iv, Iw detected by the current sensors 31, 32, 33 using the electrical angle α calculated by the multiplier 25, and performs d- and q-axis conversions. The actual current values Id and Iq are calculated and output to the subtracters 14a and 14b.

  The multiplier 25 calculates the electrical angle α of the motor M using the detected angle value θs and the number of pole pairs of the motor M. Specifically, the multiplier 25 may calculate the electrical angle α by multiplying the detected angle value θs by 1 / pole pair number.

  The speed calculation unit 26 differentiates the detected angle value θs detected by the angle detection unit 20 to calculate the rotation speed of the motor M, and outputs it to the subtractor 11 and the multiplier 25 as a speed detection value.

  Each of the current sensors 31, 32, and 33 is configured by, for example, a Hall-type current sensor using a Hall element, and detects U, V, and W-phase AC currents Iu, Iv, and Iw.

  The above is the basic configuration of the control device 1, and the motor M is feedback-controlled so that the detected speed value follows the target speed.

  The estimation unit 22 estimates the detection error from the periodic fluctuation of the torque detection value τd (actual output torque) generated by using the angle detection value θs (actual rotation angle) including the detection error for driving control of the motor M. .

  The compensation unit 23 compensates for the detection error included in the detected angle value θs based on the detection error estimated by the estimation unit 22.

  In the first embodiment, the control device 1 operates in an estimation phase for estimating the detection error and a compensation phase for removing the detection error included in the angle detection value θs using the detection error estimated in the estimation phase. Divided. The estimation unit 22 operates in the estimation phase, and the compensation unit 23 operates in the compensation phase.

<Estimation phase>
In the estimation phase, the estimation unit 22 inputs the angle detection value θs and the torque detection value τd to the equation of motion indicating the relationship between the angle detection value θs and the torque detection value τd, and sequentially applies the least square method, Estimate the detection error.

  Equation (1) is an equation of motion used in the estimation phase.

θs: Angle detection value including detection error θ: True value of angle detection value τd: Torque detection value J: Moment of inertia of motor M N: Resolver multiple angle Δc: Error gain of cosine component of detection error Δs: Detection error The error gain of the sine component Here, the detected angle value θs is expressed by the following equation (2).

  In Expression (2), the first term on the right side indicates the true value θ, and the second and third terms on the right side indicate detection errors.

  The torque detector 19 is configured by a resolver. The detection error detected by the resolver has a characteristic that varies periodically according to the rotation angle of the motor M. Specifically, the detection error is expressed by a mathematical expression in which cos (Nθ) and sin (Nθ) are linearly combined, where N is the doubler number of the resolver (N is a natural number). Therefore, the detection error can be expressed by an error function (Δc · cos (Nθ) + Δs · sin (Nθ)) as shown in the second and third terms of the equation (2). As can be seen from this equation, the error function has a characteristic of changing 360 degrees while the motor M rotates 1 / N. As the multiple N, a value set in advance from the resolver specification can be adopted.

  On the other hand, the detected angle value θs and the detected torque value τd can be expressed by the equation of motion of the rotating system: Jθs ″ = τd + Δτ. However, Δτ is a torque difference between the output torque of the motor M and the detected torque value τd based on the detected angle value θs indicated by Jθs ″. If θs ″ does not include a detection error, Δτ = 0. “·” Shown in the formula is represented by “′” in the text due to the limitation of the description. “·” Indicates the first derivative, and “··” indicates the second derivative.

  Expression (1) is obtained by substituting Expression (2) into the equation of motion of this rotating system to express the torque difference Δτ.

  When formula (1) is arranged with respect to error gains Δc and Δs, formula (3) is obtained.

  In the estimation phase, the detected angle value θs and the detected torque value τd detected at the same time are input to the left side of Equation (3), and the error gains Δc and Δs are estimated by sequentially applying the least square method.

The sequential least square method, as shown in Equation (4), and the variable z _n, a variable y _n, by using the variable R _n, by calculating repeatedly estimated parameter P _{n (hat),} error gain Δc , Δs. In addition, “数 式” shown in the mathematical expression is a symbol indicating an estimated value, and is expressed as (hat) in the text due to the limitation of description.

Here, the estimation parameter P _n (hat) is expressed by the equation (5), and the variable θ _n , the variable y _n , and the variable R _n corresponding to the variable z _n and the true value θ are respectively expressed by the equations (6), It is represented by (7), (8), (9).

  The subscript n is an index indicating the nth calculated data, and is an integer of 1 or more.

As shown in Expression (5), the estimation parameter P _n (hat) is a matrix of 2 rows and 1 column indicating the error gains Δc _n and Δs _n to be estimated.

As shown in Expression (6), the variable z _n is a variable called a state variable, and is represented by a matrix of 2 rows and 1 column. In Expression (6), the first line shows an expression in parentheses multiplied by Δc of the first term on the right side in Expression (3), and the second line shows the second term on the right side in Expression (3). A mathematical expression in parentheses multiplied by Δs is shown. However, in Expression (6), θ in Expression (3) is expressed as θ → θ _n (hat).

As shown in Expression (7), the variable θ _n (hat) is expressed by θ in the first term on the right side as θ → θ _n (hat) in Equation (2), and cos in the second and third terms on the right side. And θ in the parentheses of sin are expressed as θ → θ _n-1 (hat), expressed as θs → θs (k + n), and expressed as Δc → Δc _n−1 and Δs → Δs _n−1 . In the successive least squares method, the estimation parameter P _n is estimated by repeating the process of calculating the estimation parameter P _n (hat) using _n data groups as one set by a plurality of cycles. Therefore, k of the detected angle value θs (k + n) represents an index for specifying the number of cycles. For example, the detected angle value θs (k + 1) represents the first detected angle value in the kth cycle. In this case, as the value of the detected angle value θs (k), θs obtained up to the (k−1) th cycle may be used, and (k−1) · n used for the estimation up to the (k−1) th cycle. It is not necessary to store the value of the detected angle value θs in the memory. Therefore, the successive least square method can suppress memory consumption. This also applies to the torque detection value τd (k + n).

The variable _{y n} as shown in Equation (8) represents 'a [theta] s' Equation (3) left side of [theta] s of '' → [theta] s '' and (k + n), the left side of .tau.d the .tau.d → .tau.d of formula (3) ( k + n).

As shown in Expression (9), the variable R _n is a determinant expressed using the variable R _n−1 and the variable z _n . In Expression (9), I represents a unit matrix, and T represents a transposed matrix.

FIG. 2 is a block diagram showing the flow of processing in the estimation unit 22. First, when the detected angle value θs (k + n) and the detected torque value τd (k + n) are input, the estimating unit 22 calculates a variable θ _n (hat) using Expression (7) (block 216). Here, the detected angle value θs (k + n) and the detected torque value τd (k + n) are sequentially input at the sampling period Ts.

In the first process, Δc _n−1 , Δs _n−1 , and θ _n−1 (hat) on the right side of Expression (7) are unknown. Therefore, the estimation unit 22 may input initial values (Δc ₀ , Δs ₀ , θ ₀ (hat)) set in advance in Δc _n−1 , Δs _n−1 , θ _n (hat). As an initial value (Δc ₀ , Δs ₀ , θ ₀ (hat)) set in advance, a value close to that value may be adopted in anticipation of an estimated value. Further, in the second and subsequent blocks 216, the estimation unit 22 sets the estimation parameter P _n (hat) = [Δc _n Δs _n ] calculated in the block 215 to [Δc _n−1 Δs _n−1 ] and the expression ( 7).

Next, the estimation unit 22 calculates θ _{n ′} (hat) by first-order differentiation of θ _n (hat) calculated in block 216 (block 217), and second-order differentiation of θ _n (hat). θ _n ″ (hat) is calculated (block 218), and θ _n (hat) is multiplied by 1 / z to calculate θ _n−1 (hat) (block 219).

In the first processing of Expression (7), the initial value is used as θ _n−1 (hat), and therefore the processing of block 219 is omitted. 2 (z−1) / Ts (z + 1) in the block 217 is a discrete transfer function indicating the first derivative. 2 ² (z−1) ² / Ts ² (z + 1) ² in block 218 is a discrete transfer function of second order differentiation. Ts represents a sampling period.

Then, the estimation unit 22, theta _n '' and (hat), theta _n 'and (hat), and theta _n (hat) into the Formula (6), to calculate the variable _{z n} (block 214) .

In parallel with the above processing, the estimation unit 22 performs a process of calculating the variable y _n using Equation (8) (block 213).

  As a pre-processing of the block 213, the estimation unit 22 calculates the τd (k + n) / J by multiplying the torque detection value τd (k + n) by “1 / J” (block 211), and the angle detection value θs (k + n). ) Is second-order differentiated to calculate θs ″ (k + n) (block 212).

Next, the estimation unit 22 inputs the variable z _n obtained in block 214 and the previously obtained variable R _n−1 into equation (9), and calculates the variable R _n (block 215). In the first process, R _{n−1 in} Equation (9) is unknown. Therefore, the estimation unit 22 may input an initial value (R ₀ ) set in advance to R _n−1 . As an initial value (R ₀ ) set in advance, a value close to that value may be adopted in anticipation of the estimated value.

Next, the estimation unit 22 inputs the variable y _n calculated in the block 213, the variable z _n calculated in the block 214, and the estimation parameter P _n−1 (hat) obtained in the previous time into the equation (4). Then, the estimation parameter P _n (hat) = [Δc _n Δs _n ] is calculated (block 215). In the first process, the estimation parameter P _n−1 is unknown. Therefore, the estimation unit 22 may input an initial value (P ₀ = [Δc ₀ Δs ₀ ]) set in advance as the estimation parameter P _n−1 .

By repeating the process of the successive least squares method, first, the error gains Δc _n and Δs _n given appropriate initial values gradually converge to true values. As the number of times of repeating the process of the successive least square method, the number of times that the error gains Δc _n and Δs _n are expected to converge may be adopted.

When the error gains Δc _n and Δs _n are estimated, the estimation unit 22 inputs the error gains Δc _n and Δs _n into the equation (2), and θ−θs = Δc · cos (Nθ) + Δs · sin (Nθ )) To obtain a detection error θc (= θ−θs). Then, the estimation unit 22 generates an error table 302 (corresponding information) indicating the correspondence between the determined detection error θc and the detected angle value θs.

  FIG. 3 is an explanatory diagram of the error table 302. In FIG. 3, a graph 301 is a graph of an error function indicating the detection error θc, with the vertical axis indicating θc and the horizontal axis indicating θs. In the graph 301, it can be seen that the detection error θc fluctuates periodically with a constant amplitude according to the detected angle value θs, with θc = 0 as the center.

  The error table 302 stores a plurality of detected angle values θs and detected errors θc in association with each other as (θs1, θc1), (θs2, θc2),. The estimation unit 22 inputs a plurality of representative angle detection values θs into an error function shown in the graph 301 indicating the detection error θc, thereby calculating a plurality of representative detection errors θc and generating an error table 302. That's fine. When the estimation unit 22 generates the error table 302, the estimation phase ends.

<Compensation phase>
Returning to FIG. When the estimation phase ends, the control device 1 executes the compensation phase using the error table 302 obtained in the estimation phase.

  In the compensation phase, first, the angle detection value θs is input to the compensation unit 23 at the sampling period Ts. In the compensation phase, since the torque detection value τd is not used, the torque detector 19 does not detect the output torque.

  The compensation unit 23 determines a detection error θc corresponding to the input angle detection value θs with reference to the error table 302 generated by the estimation unit 22, and multiplies the determined detection error θc by minus 1 to obtain an error compensation value ( -Θc) is calculated. If the detected angle value θs input to the error table 302 is not registered, the compensating unit 23 linearly complements the error table 302 to determine the detected error θc corresponding to the input detected angle value θs. do it.

  Next, the adder 24 calculates the true value θ (= θs−θc) by adding the error compensation value (−θc) output from the compensation unit 23 to the detected angle value θs. The speed calculation unit 26 calculates a speed detection value using the calculated true value θ. Thereafter, the control device 1 controls the speed of the motor M using the true value θ.

  Thus, according to the control device 1 of the first embodiment, the angle detection value θs and the torque detection value τd are added to the equation of motion (formula (1)) indicating the relationship between the angle detection value θs and the torque detection value τd. By inputting, the detection error θc is estimated. Here, when an angle detection value θs including a periodically varying detection error θc is used for driving control of the motor M, the torque detection value τd has the same period as the detection error θc and the magnitude of the detection error θc. The pulsation according to is generated. Therefore, if the detected angle value θs and the detected torque value τd are known at the same time, the detection error θc included in the detected angle value θs can be accurately estimated. Further, the resolver detection error θc is represented by a trigonometric function that periodically varies in accordance with the detected angle value θs, as shown in a graph 301 of FIG. Therefore, according to this aspect, it is possible to estimate the error gains Δc and Δs by inputting the angle detection value θs and the torque detection value τd to the equation of motion of the equation (1) and sequentially applying the least square method. The detection error θc can be accurately estimated.

(Embodiment 2)
FIG. 4 is a block diagram showing a configuration of a control device 1A according to Embodiment 2 of the present invention. The control device 1A according to the first embodiment includes a torque estimation unit 27 instead of the torque detection unit 19. In the second embodiment, the same components as those in the first embodiment are denoted by the same reference numerals and description thereof is omitted.

  The torque estimation unit 27 estimates the torque detection value τd using the actual current values Id and Iq calculated by the uvw / dq conversion unit 21.

  FIG. 5 is a block diagram focusing on the torque estimation unit 27. Here, it is assumed that the motor M is a permanent magnet type synchronous motor which is an example of a brushless motor. In this case, the torque estimation unit 27 can estimate the detected torque value τd using the equation (10) as shown in FIG.

τd = Pa (ψaIq + 1/2 (Lq−Ld) IdIq) (10)
Pa: number of pole pairs of motor M ψa: linkage flux of motor M Ld: inductance of d axis of motor M Lq: inductance of q axis of motor M Note that the number of pole pairs Pa, linkage flux ψa, and inductances Ld and Lq are motors It has a known value based on the specification of M. Here, the equation (10) is used, but the torque estimation unit 27 may estimate the detected torque value τd using a different equation depending on the type of the motor M. For example, the torque estimation unit 27 may estimate a value obtained by multiplying the actual current value Id by a predetermined coefficient as the torque detection value τd.

  In the estimation phase, the detection error θc is estimated using the torque detection value τd estimated by the torque estimation unit 27 as in the first embodiment.

  Thus, according to the control device 1A according to the second embodiment, the torque estimation value τd is estimated by the torque estimation unit 27. Here, the torque estimation part 27 is comprised by processors, such as CPU, for example. The torque estimation unit 27 can estimate the torque detection value τd using the existing current sensors 31, 32, and 33. Therefore, the control apparatus 1A does not need to detect the torque detection value τd using a dedicated measuring machine. Therefore, 1 A of control apparatuses can suppress the cost and scale of an apparatus.

(Modification 1)
In the first embodiment, the detection error θc is estimated using the successive least square method, but the present invention is not limited to this, and the detection error θc may be estimated using the least square method.

(Modification 2)
In the first embodiment, the error table 302 is finally generated in the estimation phase. However, the present invention is not limited to this, and an error function may be finally generated. In this case, the compensation unit 23 may determine the detection error θc by inputting the angle detection value θs into the error function.

(Modification 3)
In the first embodiment, the processing is performed separately in the estimation phase and the compensation phase, but the present invention is not limited to this, and the estimation and compensation of the detection error θc is performed in real time without separating both phases. Also good.

  In this case, each time the torque detection value τd and the angle detection value θs are input, the estimation unit 22 calculates the error gains Δc and Δs using the equation of motion of the equation (1), and the detection error θc may be obtained. Then, the compensation unit 23 may compensate the angle detection value θs using the calculated detection error θc. In this case, the compensation accuracy of the detection error θc can be increased as the number of times of control of the motor M increases.

(Modification 4)
In the first embodiment, the resolver is employed as the angle detection unit 20, but the present invention is not limited to this, and the detection error θc has a characteristic that periodically varies according to the rotation angle of the rotor of the motor M. Any angle detector may be adopted as long as it is an angle detector.

1, 1A Controller 11 Subtractor 12 Speed controller 13 Current command generator 14a, 14b Subtractor 15 Current controller 16 dq / uvw converter 17 Inverter 19 Torque detector 20 Angle detector 21 uvw / dq converter 22 Estimation Unit 23 Compensator 24 Adder 25 Multiplier 26 Speed calculator 27 Torque estimator 31, 32, 33 Current sensor 301 Graph 302 Error table

Claims (6)

A control device for an electric motor,
An angle detection unit for detecting an actual rotation angle of the rotor including a detection error periodically changing according to a rotation angle of the rotor of the electric motor and a true value of the rotation angle;
A torque detector for detecting an actual output torque of the electric motor;
An estimation unit that estimates the detection error from a periodic variation of the actual output torque generated by using the actual rotation angle for drive control;
An electric motor control device comprising: a compensation unit that compensates for a detection error included in the actual rotation angle based on the estimated detection error.   The said estimation part estimates the said detection error by inputting the said actual rotation angle and the said actual output torque into the equation of motion which shows the relationship between the said actual rotation angle and the said actual output torque. Electric motor control device. The actual rotation angle is represented by a sum of an error function obtained by multiplying a trigonometric function having the true value as a variable by an error gain, and the true value,
3. The estimation unit estimates the detection error by inputting a plurality of actual rotation angles and a plurality of actual output torques to the equation of motion, and determining the error gain by a least square method. The motor control device described. The estimation unit generates correspondence information in which a plurality of rotation angles and a plurality of detection errors are associated with each other from the error function represented by the error gain determined by the least square method.
The motor control device according to claim 3, wherein the compensation unit determines a detection error corresponding to the actual rotation angle from the correspondence information, and compensates the actual rotation angle using the determined detection error. A current detection unit for detecting a current flowing through the electric motor;
The motor control apparatus according to claim 1, wherein the torque detection unit calculates the actual output torque based on the detected current.   The motor control device according to claim 1, wherein the angle detection unit is configured by a resolver.

Images