CN108183639B - Least square classification speed regulation method for brushless direct current motor - Google Patents

Abstract

The invention discloses a least square classification speed regulation method for a brushless direct current motor, which comprises the following steps: 1) detecting the position of a motor rotor and regulating the speed by using the same motor provided with a position sensor, and detecting a back electromotive force signal; 2) taking the back electromotive force signal as input and the zero crossing point signal as output, and performing off-line training of a least square classifier; 3) and classifying the back electromotive force data of the brushless direct current motor control system by using a trained least square classifier to obtain a zero crossing point moment, so that the rotating speed and a phase change signal of the motor are calculated to realize phase change, the rotating speed of the motor is fed back to a PID (proportion integration differentiation) controller of the brushless direct current motor control system to calculate a control quantity, and the speed regulation is realized. The invention realizes the control of the brushless direct current motor without a position sensor, classifies the back electromotive force by using the least square classifier, is used for estimating the zero crossing point signal, and more accurately detects the unconventional back electromotive force zero crossing point.

Description

A Least Square Classification Speed Regulation Method for Brushless DC Motors

technical field

The invention relates to the technical field of brushless direct current motor control, in particular to a least-square classification speed regulation method of a brushless direct current motor.

Background technique

If the brushless DC motor uses the back-EMF zero-crossing point to determine the commutation moment, there is no need to install a Hall sensor to detect the rotor position. Usually, the non-conducting phase terminal voltage is detected to calculate the back-EMF, and then the commutation is obtained by detecting the back-EMF zero point. However, the back-EMF detection method is due to the closing noise of the hardware circuit, external noise, etc., if the load increases, the signal-to-noise ratio decreases, and the probability of detecting a wrong zero-crossing point is very high. The speed change may also change the characteristics of the winding, causing reverse The electromotive force fluctuates, and the false zero-crossing point caused by this makes it difficult to determine the commutation time of the motor. The inaccurate commutation time causes the torque ripple to make the brushless DC motor jitter. above problem.

In this paper, a back-EMF zero-point classifier is designed to classify all the detected back-EMFs, and the non-zero points are classified into one category, and the zero points are classified into another category, so that the zero-crossing points can be detected by the back-EMF classifier. The least-squares classifier can minimize the amount of calculation while ensuring a high accuracy rate, which can be satisfied when it is applied to embedded systems such as DSP. So far, the position sensorless speed regulation of brushless DC motors using the least squares back-EMF classifier for speed measurement has not yet appeared.

SUMMARY OF THE INVENTION

Aiming at the deficiencies of the prior art, the present invention proposes a least-squares classification speed regulation method for a brushless DC motor, so as to solve the problem of inaccurate detection of the zero-crossing point of the back electromotive force in the prior control method.

In order to achieve the above purpose, the technical scheme of the present invention is: a method for least squares classification speed regulation of a brushless DC motor, comprising:

Step S1: collect offline training data, use the same type of motor installed with a position sensor to detect the rotor position of the motor and adjust the speed, and detect the back electromotive force signal at the same time;

Step S2: take the back electromotive force signal as the input and the zero-crossing point signal as the output, and perform offline training of the least squares classifier;

Step S3: use the sampling circuit to sample the back-EMF data of the brushless DC motor control system without the position sensor, use the trained least squares classifier to classify the online back-EMF data, obtain the zero-crossing time, and calculate the motor speed And the commutation signal, realize commutation, feedback the motor speed to the PID controller of the brushless DC motor control system to calculate the control amount and realize the speed regulation.

Further, the step S2 is specifically:

The least squares classifier is trained offline, and the back EMF signals {e(k-1),...,e(k-p)} of p beats are reserved and displayed on the time axis. There is only one zero-crossing point in each cycle, and the zero-crossing point is marked. , the output of the zero-crossing sample is y(k)=1, and the output of the non-zero-crossing sample is y(k)=0; the back EMF input signal of p beat {e(k),...,e(k-p+1 )} and the output signal of p beat {y(k),...,y(k-p+1)} as training samples, the solution feature vector x(k) is: x(k)=P(y(k-1 ),...,y(k-n),e(k),...,e(k-n)), where e(k),...,e(k-n) and y(k),...,y(k-n) are the input and output of the classifier at the kth moment, respectively, and n represents the number of beats of the input and output; the least squares classifier is constructed based on the least squares method.

Further, the specific method for obtaining the zero-crossing time in the step S3 includes:

The three-stage starting method is used to start the motor, which is divided into three stages: rotor pre-positioning, frequency boosting synchronous operation, and switching self-synchronous operation. After the startup is completed, a stable back EMF is continuously detected;

In the brushless DC motor drive system, the trained least squares classifier is used to detect zero-crossing points, assuming the input is {e(k),...,e(k-n+1)}, the least squares classifier is used The judgment function is expressed as: g(x)=ω·x(k); if g(x)>=0, then y(k)=1; if g(x)<0, then y(k)=0 ; x(k) is the feature vector: x(k)=P(y(k-1),...,y(k-n),e(k),...,e(k-n)), where e( k),...,e(k-n) and y(k),...,y(k-n) are the input and output of the classifier at the kth moment, respectively, ω is the weight of the least squares classifier, and e(k) is the inverse Electromotive force, y(k) is the output of the classifier, that is, the zero-crossing signal;

In the speed regulation stage of the brushless DC motor, the back EMF data is continuously collected and input into the least squares classifier to obtain the zero-crossing time.

Compared with the prior art, the present invention has beneficial effects: the present invention saves the position sensor, only needs to expand three back electromotive force sampling circuits on the inverter circuit, the hardware cost is small, and the signal detection is not affected by environmental factors such as temperature and dust , high reliability;

The invention uses the least squares classifier to classify the back electromotive force, which is used for zero-crossing signal estimation, and has high accuracy. The brushless DC motor control system controlled by DSP can improve the control precision.

Description of drawings

1 is a schematic flowchart of the least squares classification speed regulation method of the brushless DC motor of the present invention;

2 is a schematic diagram of the detection of the zero-crossing point of the back electromotive force according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a flow chart of back-EMF data classification according to an embodiment of the present invention.

Detailed ways

The present invention will be further described below with reference to the accompanying drawings and embodiments.

For the trapezoidal wave brushless DC motor with the back EMF waveform of 120° flat-top width, the three opposite EMFs generate six zero-crossing points in one electrical cycle, and the displacement time of each delay of 30° electrical angle is 6 commutation moments, which is accurate. The commutation moment affects the driving effect of the brushless DC motor. The invention adopts the least squares classification method to measure the rotational speed and only needs the back electromotive force of the power circuit as the input signal to calculate the rotational speed and the commutation time, and can replace the rotor position signal measured by the mechanical sensor to complete the commutation, and the rotational speed is fed back to the input end of the PID controller , to achieve speed regulation. In the brushless DC motor control system with position sensor, the back-EMF signal is extracted as a sample to train the least squares classifier. The trained least squares classifier can realize the back-EMF zero-crossing classification. The back-EMF zero-crossing point detection of the gauge is more accurate, and the output is represented by a Boolean value. The motor speed and commutation time are calculated through the output, and the PID algorithm is used to achieve accurate speed regulation.

As shown in FIG. 1 , a method for least squares classification speed regulation of a brushless DC motor of the present invention includes:

Step S1: collect offline training data, use the same type of motor installed with a position sensor to detect the rotor position of the motor and adjust the speed, and detect the back electromotive force signal at the same time;

Step S2: take the back electromotive force signal as the input and the zero-crossing point signal as the output, and perform offline training of the least squares classifier;

Step S3: use the sampling circuit to sample the back-EMF data of the brushless DC motor control system without the position sensor, use the trained least squares classifier to classify the online back-EMF data, obtain the zero-crossing time, and calculate the motor speed And the commutation signal, realize commutation, feedback the motor speed to the PID controller of the brushless DC motor control system to calculate the control amount and realize the speed regulation.

The step S2 is specifically:

The least squares classifier is trained offline, and the back EMF signals {e(k-1),...,e(k-p)} of p beats are reserved and displayed on the time axis. There is only one zero-crossing point in each cycle, and the zero-crossing point is marked. , the output of the zero-crossing sample is y(k)=1, and the output of the non-zero-crossing sample is y(k)=0; the back EMF input signal of p beat {e(k),...,e(k-p+1 )} and the output signal of p beat {y(k),...,y(k-p+1)} as training samples, the solution feature vector x(k) is: x(k)=P(y(k-1 ),...,y(k-n),e(k),...,e(k-n)), where e(k),...,e(k-n) and y(k),...,y(k-n) are the input and output of the classifier at the kth moment, respectively, and n represents the number of beats of the input and output; the least squares classifier is constructed based on the least squares method.

The specific method for obtaining the zero-crossing time in the step S3 includes:

The three-stage starting method is used to start the motor, which is divided into three stages: rotor pre-positioning, frequency boosting synchronous operation, and switching self-synchronous operation. After the startup is completed, a stable back EMF is continuously detected;

In the brushless DC motor drive system, the trained least squares classifier is used to detect zero-crossing points, assuming the input is {e(k),...,e(k-n+1)}, the least squares classifier is used The judgment function is expressed as: g(x)=ω·x(k); if g(x)>=0, then y(k)=1; if g(x)<0, then y(k)=0 ; x(k) is the feature vector: x(k)=P(y(k-1),...,y(k-n),e(k),...,e(k-n)), where e( k),...,e(k-n) and y(k),...,y(k-n) are the input and output of the classifier at the kth moment, respectively, ω is the weight of the least squares classifier, and e(k) is the inverse Electromotive force, y(k) is the output of the classifier, that is, the zero-crossing signal;

In the speed regulation stage of the brushless DC motor, the back EMF data is continuously collected and input into the least squares classifier to obtain the zero-crossing time.

As shown in Figure 2, for a trapezoidal wave brushless DC motor with a back EMF waveform of 120° flat-top width, in one electrical cycle, the three opposite EMFs generate six zero-crossing points, namely Z1-Z6, and six The commutation points are at times S1-S6, and the six switches are VT1-VT6. The drive circuit adopts the two-way conduction mode. At the beginning, VT5 and VT6 are in the conduction state. After the zero-crossing time Z1 is detected, the commutation time S1 can be obtained by delaying the electrical angle by 30°, and the VT1 and VT6 are controlled to be turned on, and then start to detect Z2. , and alternately perform zero-crossing detection and phase commutation according to the positions shown in the figure, and then the motor can be driven to rotate.

As shown in Figure 3, the A/D conversion result is firstly read to obtain the back EMF data. It is necessary to decide which back EMF to read according to the different switch conduction states. The value is standardized to obtain the eigenvector, and the least two Multiply the classifier to determine the back-EMF zero-crossing point. The output of the back-EMF zero-crossing point is set to y(k)=1. If the zero-crossing point is detected, the commutation is completed according to the back-EMF, and finally the speed is adjusted.

The specific embodiments described above describe in detail the purpose, technical solutions and achievements of the present invention. It should be understood that the above are only specific embodiments of the present invention, and are not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (1)

1. a brushless DC motor least square classification speed regulation method, is characterized in that, comprises: Step S1: collect offline training data, use the same type of motor installed with a position sensor to detect the rotor position of the motor and adjust the speed, and detect the back electromotive force signal at the same time; Step S2: take the back electromotive force signal as the input and the zero-crossing point signal as the output, and perform offline training of the least squares classifier; Step S3: use the sampling circuit to sample the back-EMF data of the brushless DC motor control system without the position sensor, use the trained least squares classifier to classify the online back-EMF data, obtain the zero-crossing time, and calculate the motor speed And the commutation signal, realize the commutation, feedback the motor speed to the PID controller of the brushless DC motor control system to calculate the control amount and realize the speed regulation; The step S2 is specifically: The least squares classifier is trained offline, and the back EMF signals {e(k-1),...,e(k-p)} of p beats are reserved and displayed on the time axis. There is only one zero-crossing point in each cycle, and the zero-crossing point is marked. , the output of the zero-crossing sample is y(k)=1, and the output of the non-zero-crossing sample is y(k)=0; the back EMF input signal of p beat {e(k),...,e(k-p+1 )} and the output signal of p beat {y(k),...,y(k-p+1)} as training samples, the solution feature vector x(k) is: x(k)=P(y(k-1 ),...,y(k-n),e(k),...,e(k-n)), where e(k),...,e(k-n) and y(k),...,y(k-n) are the input and output of the classifier at the k-th moment, respectively, and n represents the number of beats of the input and output; the least squares classifier is constructed based on the least squares method; The specific method for obtaining the zero-crossing time in the step S3 includes: The three-stage starting method is used to start the motor, which is divided into three stages: rotor pre-positioning, frequency boosting synchronous operation, and switching self-synchronous operation. After the startup is completed, a stable back EMF is continuously detected; In the brushless DC motor drive system, the trained least squares classifier is used to detect zero-crossing points, assuming the input is {e(k),...,e(k-n+1)}, the least squares classifier is used The judgment function is expressed as: g(x)=ω·x(k); if g(x)>=0, then y(k)=1; if g(x)<0, then y(k)=0 ;ω is the weight of the least squares classifier, e(k) is the back electromotive force, and y(k) is the output of the classifier, that is, the zero-crossing signal; In the speed regulation stage of the brushless DC motor, the back EMF data is continuously collected and input into the least squares classifier to obtain the zero-crossing time.

Images