阅读说明：本技术 一种k接近优化估计的电机转子初始位置检测方法和装置 (Method and device for detecting initial position of motor rotor based on K-approach optimization estimation ) 是由 栗伟周 张赞 马军磊 李瑞华 吕存杰 吴鑫鑫 田尚永 于 2019-11-12 设计创作，主要内容包括：本发明提供一种K接近优化估计的电机转子初始位置检测方法和装置,方法包括如下步骤：对待测电机进行标定实验,得到原始标定信号；向待测电机的主电机定子注入设定高频信号,检测主励磁机定子的电流,得到测试信号；计算各原始标定信号的聚类中心,得到中心标定信号；计算测试信号与各中心标定信号之间的欧氏距离,选取与测试信号欧氏距离最近的k个中心标定信号,将k个中心标定信号所对应的标定位置中数量最多的标定位置作为待测主电机转子所在的位置,k为正整数。本发明提供的技术方案主励磁机定子的电流信号直接作为标定信号和测试信号使用,无需使用高频滤波器进行滤波,减少了信号采集的延时,能够提高对电机转子初始位置检测的准确性。(The invention provides a method and a device for detecting the initial position of a motor rotor by K-approach optimization estimation, wherein the method comprises the following steps: carrying out calibration experiment on the motor to be measured to obtain an original calibration signal; injecting a set high-frequency signal into a main motor stator of a motor to be tested, and detecting the current of the main exciter stator to obtain a test signal; calculating the clustering center of each original calibration signal to obtain a center calibration signal; calculating Euclidean distances between the test signals and the central calibration signals, selecting k central calibration signals closest to the Euclidean distances of the test signals, taking the most-numbered calibration positions in the calibration positions corresponding to the k central calibration signals as the positions of the main motor rotor to be tested, wherein k is a positive integer. According to the technical scheme provided by the invention, the current signal of the stator of the main exciter is directly used as a calibration signal and a test signal, a high-frequency filter is not required for filtering, the time delay of signal acquisition is reduced, and the accuracy of detecting the initial position of the motor rotor can be improved.)

1. A method for detecting the initial position of a motor rotor by K-approach optimization estimation is characterized by comprising the following steps:

the method comprises the following steps: and (3) carrying out a calibration experiment on the motor to be tested:

firstly, establishing a calibration position set comprising a plurality of calibration positions of a main motor, then injecting a set high-frequency signal into the main motor stator and detecting the current of the main exciter stator when the position of the main motor stator is at each calibration position, and taking the detected current of the main exciter stator as an original calibration signal to obtain an original calibration signal set;

step two: injecting a set high-frequency signal into a main motor stator of a motor to be tested, detecting the current of the main exciter stator, and taking the obtained current as a test signal;

step three: calculating the clustering center of each original signal in the original calibration signal set to obtain a center calibration signal of each calibration position; calculating Euclidean distances between the test signals and the central calibration signals; selecting k central calibration signals closest to the Euclidean distance of the test signals, and taking the calibration position with the largest quantity in the calibration positions corresponding to the k central calibration signals as the position of the rotor of the main motor to be tested; k is a positive integer.

2. The method for detecting the initial position of the rotor of the motor for K-near optimized estimation according to claim 1, wherein in the step one, when the position of the rotor of the main motor is at each calibration position, the current of the main exciter stator is detected for m periods after injecting a set high frequency signal into the main motor stator; in the third step, each original calibration signal calculates n clustering centers; m and n are both positive integers, and m > n.

3. The method for detecting the initial position of the rotor of the motor based on the K-proximity optimization estimation as claimed in claim 1, wherein the Euclidean distance between the test signal and each center calibration signal is calculated by using the following formula:

wherein d is a test signal q_iWith a central calibration signal p_ZEuclidean distance of q_iFor the ith test signal, p_ZjFor the central calibration signal p_ZThe jth element of (1).

4. The method of claim 1, wherein the Euclidean distance between the test signal and each of the central calibration signals is calculated by normalizing each of the central calibration signals and the test signal.

5. An apparatus for detecting an initial position of a rotor of an electric machine for K-approach optimization estimation, comprising a memory and a processor, the memory having stored thereon a computer program for execution on the processor; wherein the processor implements the following steps when executing the computer program:

the method comprises the following steps: and (3) carrying out a calibration experiment on the motor to be tested:

firstly, establishing a calibration position set comprising a plurality of calibration positions of a main motor, then injecting a set high-frequency signal into the main motor stator and detecting the current of the main exciter stator when the position of the main motor stator is at each calibration position, and taking the detected current of the main exciter stator as an original calibration signal to obtain an original calibration signal set;

step two: injecting a set high-frequency signal into a main motor stator of a motor to be tested, detecting the current of the main exciter stator, and taking the obtained current as a test signal;

step three: calculating the clustering center of each original signal in the original calibration signal set to obtain a center calibration signal of each calibration position; calculating Euclidean distances between the test signals and the central calibration signals; selecting k central calibration signals closest to the Euclidean distance of the test signals, and taking the calibration position with the largest quantity in the calibration positions corresponding to the k central calibration signals as the position of the rotor of the main motor to be tested; k is a positive integer.

6. The apparatus for detecting the initial position of the rotor of an electric motor according to claim 5, wherein in the first step, when the position of the rotor of the main electric motor is at each calibration position, the current of the main exciter stator is detected for m cycles after injecting a set high frequency signal into the main motor stator; in the third step, each original calibration signal calculates n clustering centers; m and n are both positive integers, and m > n.

7. The apparatus for detecting the initial position of the rotor of the electric motor according to claim 5, wherein the Euclidean distance between the test signal and each of the center calibration signals is calculated by the following formula:

wherein d is a test signal q_iWith a central calibration signal p_ZEuclidean distance of q_iFor the ith test signal, p_ZjFor the central calibration signal p_ZThe jth element of (1).

8. The apparatus of claim 5, wherein the Euclidean distance between the test signal and each of the central calibration signals is calculated by normalizing each of the central calibration signals and the test signal.

Technical Field

The invention belongs to the technical field of motor rotor initial position detection, and particularly relates to a motor rotor initial position detection method and device based on K-proximity optimization estimation.

Background

The three-level alternating current starting/generating system plays an important role in the current aviation motor, and the starting/generating integrated function of the system can reduce one set of starting device, reduce the weight of an airplane and improve the running stability of the system. With the gradual application of three-stage ac starting/generating system in multi-electric aircraft, rotor position estimation without position sensor becomes one of the hot spots of the current aviation motor research.

The development of all-electric aircrafts provides higher requirements for a three-level alternating current starting/generating system, a novel three-level synchronous motor system represented by a rotor without a position sensor has obvious advantages, the rotor position can be obtained under the condition that the position sensor is not needed, the system cost is reduced, the system risk caused by the fault of the position sensor is eliminated, the research on obtaining the initial position information of the rotor under the condition that the position sensor is not available has higher value and significance.

At present, the detection methods for the initial position of the stator of the three-stage alternating current motor mainly comprise the following steps:

the first is to regard the structure of the electrically excited rotor synchronous start/generator as a rotary transformer, inject a rotating voltage signal into the stator of the main generator, detect the field current of the main exciter, and then process the current signal to estimate the rotor position of the main generator;

the second method is that firstly, high-frequency square wave voltage is injected into an estimated d axis to obtain high-frequency current response through a current sensor; then decomposing the estimated q-axis current response, multiplying the q-axis current response by a fixed frequency cosine modulation wave, obtaining a rotor position error through a low-pass filter, and obtaining a rotor position initial value through a position tracker; finally, identifying the polarity of the magnetic pole by an impressed current bias method based on the magnetic circuit saturation effect;

the third method is a high-frequency signal injection method, in which a high-frequency signal such as a pulse vibration signal and a rotation signal is injected into the stator of the main motor, and then the high-frequency signal output from the stator of the main exciter is detected, and the initial position of the rotor of the main motor is determined based on the detected high-frequency signal.

Compared with the other two methods, the high-frequency signal injection method in the three methods has the advantages of simplicity in operation and high efficiency, but the accuracy of the detection result of the existing high-frequency injection method on the initial position of the main motor rotor is low.

Disclosure of Invention

The invention aims to provide a motor rotor initial position detection method and device based on K-approach optimization estimation, which are used for solving the problem of low detection result accuracy when the existing high-frequency signal injection method is adopted to detect the initial position of a main motor rotor.

In order to solve the technical problems, the technical scheme of the invention is as follows:

a method for detecting the initial position of a motor rotor by K-approach optimization estimation comprises the following steps:

the method comprises the following steps: and (3) carrying out a calibration experiment on the motor to be tested:

firstly, establishing a calibration position set comprising a plurality of calibration positions of a main motor, then injecting a set high-frequency signal into the main motor stator and detecting the current of the main exciter stator when the position of the main motor stator is at each calibration position, and taking the detected current of the main exciter stator as an original calibration signal to obtain an original calibration signal set;

step two: injecting a set high-frequency signal into a main motor stator of a motor to be tested, detecting the current of the main exciter stator, and taking the obtained current as a test signal;

step three: calculating the clustering center of each original signal in the original calibration signal set to obtain a center calibration signal of each calibration position; calculating Euclidean distances between the test signals and the central calibration signals; selecting k central calibration signals closest to the Euclidean distance of the test signals, and taking the calibration position with the largest quantity in the calibration positions corresponding to the k central calibration signals as the position of the rotor of the main motor to be tested; k is a positive integer.

Further, when the position of the rotor of the main motor is at each calibration position in the first step, after injecting a set high-frequency signal into the stator of the main motor, detecting the current of the stator of the main exciter for m periods; in the third step, each original calibration signal calculates n clustering centers; m and n are both positive integers, and m > n.

Further, the formula adopted when calculating the euclidean distance between the test signal and each central calibration signal is as follows:

wherein d is a test signal q_iWith a central calibration signal p_ZEuclidean distance of q_iFor the ith test signal, p_ZjFor the central calibration signal p_ZThe jth element of (1).

Further, when calculating the euclidean distance between the test signal and each of the central calibration signals, normalization processing is performed on each of the central calibration signals and the test signal.

An apparatus for detecting an initial position of a rotor of an electric machine for K-approach optimization estimation, comprising a memory and a processor, the memory having stored thereon a computer program for execution on the processor; the processor, when executing the computer program, implements the steps of:

the method comprises the following steps: and (3) carrying out a calibration experiment on the motor to be tested:

firstly, establishing a calibration position set comprising a plurality of calibration positions of a main motor, then injecting a set high-frequency signal into the main motor stator and detecting the current of the main exciter stator when the position of the main motor stator is at each calibration position, and taking the detected current of the main exciter stator as an original calibration signal to obtain an original calibration signal set;

step two: injecting a set high-frequency signal into a main motor stator of a motor to be tested, detecting the current of the main exciter stator, and taking the obtained current as a test signal;

step three: calculating the clustering center of each original signal in the original calibration signal set to obtain a center calibration signal of each calibration position; calculating Euclidean distances between the test signals and the central calibration signals; selecting k central calibration signals closest to the Euclidean distance of the test signals, and taking the calibration position with the largest quantity in the calibration positions corresponding to the k central calibration signals as the position of the rotor of the main motor to be tested; k is a positive integer.

Further, when the position of the rotor of the main motor is at each calibration position in the first step, after injecting a set high-frequency signal into the stator of the main motor, detecting the current of the stator of the main exciter for m periods; in the third step, each original calibration signal calculates n clustering centers; m and n are both positive integers, and m > n.

Further, the formula adopted when calculating the euclidean distance between the test signal and each central calibration signal is as follows:

wherein d is a test signal q_iWith a central calibration signal p_ZEuclidean distance of q_iFor the ith test signal, p_ZjFor the central calibration signal p_ZThe jth element of (1).

Furthermore, when calculating the euclidean distance between the test signal and each central calibration signal, normalization processing is performed on each central calibration signal and the test signal first

According to the technical scheme provided by the invention, firstly, a set high-frequency signal is injected into the main motor stator based on a high-frequency signal injection method, a current signal of the main exciter stator is detected, and the initial position of the main motor rotor is obtained by combining the current signal of the main exciter stator. According to the technical scheme provided by the invention, the initial position of the main motor rotor is calculated by adopting a K-proximity optimization estimation algorithm, the current signal of the main exciter stator is directly used as a calibration signal and a test signal, a high-frequency filter is not required for filtering, and the delay of signal acquisition is reduced, so that the obtained initial position of the main motor rotor is more accurate, and the problem of lower accuracy of a detection result when the initial position of the main motor rotor is detected by adopting the conventional high-frequency signal injection method can be solved.

Drawings

FIG. 1 is a schematic structural diagram of a three-stage synchronous motor system in an embodiment of the method of the present invention;

FIG. 2 is a flow chart of a method for detecting the initial position of the rotor of the motor with the K approximate optimization estimation in the embodiment of the method.

Detailed Description

The method comprises the following steps:

the embodiment provides a method for detecting the initial position of a motor rotor based on K-approach optimization estimation, which is used for detecting the initial position of a three-level synchronous motor rotor and solving the problem that the detection result of the initial position of the three-level synchronous motor rotor in the prior art is inaccurate.

The method for detecting the initial position of the rotor of the motor with the K-approach optimized estimation provided by the embodiment is used for detecting a three-stage synchronous motor system as shown in fig. 1, and comprises a main generator, a main exciter and a negative exciter, wherein the detection process is shown in fig. 2, and the method comprises the following steps:

the method comprises the following steps: and carrying out test calibration on the synchronous motor.

The method for testing and calibrating the synchronous motor in the embodiment comprises the following steps:

firstly, a calibration position set is established, the calibration position set comprises the calibration positions of the main motor rotor, and the calibration positions are within the value range of the set position. In this embodiment, the calibration positions in the calibration position set take values from 0 to 90 degrees, and one calibration position is set at intervals of 0.5 degrees, that is, the calibration position set is {0,0.5,1, … … 89.5.5, 90 }.

Then setting the rotor of the main motor to one of the calibration positions, injecting a set high-frequency signal into the stator of the main motor, detecting the currents of m main exciter stators, and taking the current of the main exciter stators as an original calibration signal of the calibration position;

and finally, adjusting the main motor rotor to other calibration positions, and sequentially obtaining the original calibration signals of each calibration position according to the method, thereby obtaining an original calibration signal set corresponding to the calibration position set.

The current signal of the stator of the main exciter is directly used as a calibration signal and a test signal without filtering by using a high-frequency filter.

Step two: injecting a set high-frequency signal into a stator of a main exciter to be tested, detecting the current of the stator of the main exciter to obtain the current of the stator of the main exciter, and taking the current as a test signal;

step three: and establishing a main motor rotor initial position resolving model by adopting a K nearest neighbor optimization estimation method, and judging the main motor rotor initial position.

The method for judging the initial position of the rotor of the main motor in the embodiment comprises the following steps:

firstly, clustering is carried out on each original calibration signal in the original calibration signal set to obtain a clustering center of each original calibration signal. In this embodiment, each original calibration signal includes m periods of excitation current, so that when calculating the cluster center of the original calibration signal, each original calibration signal calculates n cluster centers, the cluster center of each original calibration signal is the center calibration signal of the corresponding calibration position, thereby obtaining the center calibration signal of each calibration position, and each calibration position corresponds to n center calibration signals.

In this embodiment, m is a positive integer greater than 1, and n is a positive integer less than m.

Then calculating Euclidean distances between the test signals and each central calibration signal, and selecting k central calibration signals closest to the Euclidean distances between the test signals from the central calibration signal set; and acquiring the calibration positions corresponding to the k central calibration signals, and taking the calibration position with the largest occurrence frequency as the position of the rotor of the main motor to be tested. In this embodiment, k takes a value of 3, and as another embodiment, k may take other positive integers.

The calculation formula adopted when calculating the Euclidean distance between the test signal and each central calibration signal is as follows:

wherein d is a test signal q_iWith a central calibration signal p_ZEuclidean distance of q_iFor the ith test signal, p_ZjFor the central calibration signal p_ZThe jth element of (1).

In order to reduce the calculation amount, when calculating the euclidean distance between the position of the rotor of the main motor to be measured and each calibration position in the calibration position set, firstly, normalizing each central calibration signal in the central calibration signal set and the test signal, wherein a formula adopted in the normalization processing is as follows:

wherein X is the signal to be normalized_normNormalizing the processed signal X_mixFor the smallest signal among the signals to be processed, X_maxThe largest signal among the signals to be processed.

In this embodiment, a Kmeans cluster is used to calculate the cluster centers of the original calibration signals, and taking one of the original calibration signals as an example, the method for calculating the cluster centers of the original calibration signals includes the following steps:

(1) randomly selecting n clustering centers from m periods of original calibration signals of the calibration position;

(2) respectively calculating Euclidean distances from the original calibration signals of m periods to n clustering center points, determining the original calibration signal closest to each randomly selected clustering center point in the original calibration signals of m periods, and dividing the m original calibration signals into n groups;

(3) calculating new clustering centers of n groups;

(4) and (5) repeating the step (2) and the step (3) until the n clustering center points are not changed any more, and thus obtaining the clustering center of the original calibration signal.

When the Kmeans is adopted to calculate the clustering center of the original calibration signal, the mode of calculating the Euclidean distance is consistent with the Euclidean distance formula of calculating the position of the rotor of the main motor to be measured and each calibration position in the calibration position set.

The embodiment of the system is as follows:

the embodiment provides a device for detecting the initial position of a motor rotor for K-approach optimization estimation, which comprises a memory and a processor, wherein the memory is stored with a computer program for being executed on the processor; the processor, when executing the computer program, implements the method for detecting an initial position of a rotor of an electric machine for K-proximity optimized estimation as provided in the above method embodiments.