Method and apparatus for adaptive control of motor, and storage medium

Patent metadata

Field	Value
Publication number	US-11290048-B2
Application number	US-202016941573-A
Country	US
Kind code	B2
Filing date	Jul 29, 2020
Priority date	Aug 9, 2019
Publication date	Mar 29, 2022
Grant date	Mar 29, 2022

Abstract

A method for adaptive motor control includes acquiring current parameters in an operation process of the motor at a current moment; determining a type of a region in which the motor operates at the current moment according to the current parameters; triggering a corresponding motor model according to the type of the region in which the motor operates at the current moment; and inputting the current parameters into the corresponding motor model, generating control parameters for motor operation according to the current parameters, and controlling the operation of the motor according to the control parameters for motor operation. An apparatus and a computer-readable storage medium are also disclosed. In comparison with the conventional motor control which uses the single nonlinear model, the motor control method disclosed herein can greatly improve the reliability of the control.

First claim

What is claimed is:

1. A method for adaptive control of a motor, comprising:
a data acquisition step of acquiring current parameters in an operation process of the motor at a current moment;
a determination step of determining a type of a region in which the motor operates at the current moment according to the current parameters;
a triggering step of triggering a corresponding motor model according to the type of the region in which the motor operates at the current moment; and
a controlling step of inputting the current parameters into the corresponding motor model, generating control parameters for motor operation according to the current parameters, and controlling the operation of the motor according to the control parameters for motor operation,
wherein the type of the region in which the motor operates at the current moment comprises a linear region and a nonlinear region, and
when it is determined that the type of the region in which the motor operates at the current moment is the linear region, a linear model is triggered and the method proceeds to a linear control step; and
when it is determined that the type of the region in which the motor operates at the current moment is the nonlinear region, a neural network model is triggered and the method proceeds to a nonlinear control step.

2. The method for adaptive control of a motor according to claim 1 , wherein the linear control step comprises inputting the current parameters into the linear model to obtain the control parameters for motor operation, and controlling the operation of the motor according to the control parameters for motor operation; and
the nonlinear control step comprises inputting the current parameters into the neural network model to obtain the control parameters for motor operation, and controlling the operation of the motor according to the control parameters for motor operation.

3. The method for adaptive control of a motor according to claim 2 , wherein the neural network model comprises a time delay neural network model.

4. The method for adaptive control of a motor according to claim 3 , wherein the method comprises a step of training the time delay neural network model, the step of training comprising:
acquiring individual parameters in the operation process of the motor from historical data, wherein the individual parameters are taken as parameters of input layer nodes of the time delay neural network model;
acquiring control parameters in the operation process of the motor from historical data, wherein the control parameters are taken as parameters of output layer nodes of the time delay neural network model; and
determining the coefficient of each hidden layer node of the time delay neural network model using a back propagation algorithm, and training the time delay neural network model.

5. The method for adaptive control of a motor according to claim 2 , wherein the current parameters in the operation process of the motor at the current moment comprise the displacement of a motor vibrator at the current moment, and the determination step comprises:
comparing the displacement of the motor vibrator at the current moment with a corresponding displacement threshold set by a system; and
determining the type of the region in which the motor operates at the current moment according to a result of the comparison between the displacement of the motor vibrator at the current moment and the displacement threshold.

6. The method for adaptive control of a motor according to claim 5 , wherein the data acquisition step comprises:
predicting the displacement of the motor vibrator at the current moment by adopting the linear model upon initial control by the system or when the motor operates in the linear region; and
predicting the displacement of the motor vibrator at the current moment by adopting the neural network model when the motor operates in the nonlinear region.

7. The method for adaptive control of a motor according to claim 6 , wherein the linear model adopts a second-order physical model, a differential equation of which is as follows:

∑

b
⁢

(
x
)

⁢
i

=

m
⁢

d
2

⁢
x

d
⁢

t
2

+

R
m

⁢

d
⁢
x

d
⁢
t

+

k
⁡

(
x
)

⁢
x

-

L
x

⁡

(
x
)

⁢

i
2

2

u
=

R
e

⁢
i

+

d
⁡

(

L
⁡

(
x
)

⁢
i

)

d
⁢
t

+

Assignees

• Aac Technologies Pte Ltd

Inventors

• Lu Xiang
• Xiang Zheng
• Zheng Yajun

Classifications

• H02P23/0022

  Model reference adaptation, e.g. MRAS or MRAC, useful for control or parameter estimation · CPC title

• H02P23/0077

  Characterised by the use of a particular software algorithm · CPC title

• H02P23/0018 Primary

  using neural networks · CPC title

• H02P23/14 Primary

  Estimation or adaptation of motor parameters, e.g. rotor time constant, flux, speed, current or voltage · CPC title

• G05B13/042

  in which a parameter or coefficient is automatically adjusted to optimise the performance · CPC title