CN108667370B - Built-in permanent magnet synchronous motor weak magnetic curve tracking method and device based on autonomous learning - Google Patents

Abstract

The built-in permanent magnet synchronous motor maximum torque current ratio curve tracking method based on autonomous learning comprises an angle resolving module, a voltage comparator, an autonomous learning derivative observer, a PI regulator and a current comparator; the angle calculation is carried out according to the angle compensation output by the PI regulator and the current angle to obtain a new angle value; the voltage comparator judges whether the voltage limit is reached or not by comparing the current terminal voltage amplitude with the bus voltage; the self-learning derivative observer calculates the partial derivative of the torque to the angle by a disturbance observation method, the partial derivative is output to the PI regulator, the PI regulator calculates the required angle compensation amount, when the derivative of the error torque to the angle is within a preset range, the judgment and the tracking are finished, and the magnitude of the current vector is updated immediately. The static calibration mode is changed into the dynamic search mode, so that the development efficiency can be improved, the curve correctness is ensured, and the method has better applicability to different motors.

Description

Built-in permanent magnet synchronous motor weak magnetic curve tracking method and device based on autonomous learning

Technical Field

The invention belongs to the technical field of control of permanent magnet synchronous motors, and particularly relates to a built-in permanent magnet synchronous motor weak magnetic curve tracking method and device based on autonomous learning.

Background

With the development of national economy and science and technology, the motor plays more and more important roles in various industries. The permanent magnet synchronous motor benefits from a plurality of advantages in the aspects of design, manufacture and control, and is widely applied to various industrial production and living occasions. In addition, the rare earth resources in China are rich, and the application market of the permanent magnet synchronous motor is particularly large in China. The permanent magnet synchronous motor can be divided into a surface-mounted type and a built-in type according to the difference of quadrature-direct axis inductance, and the built-in type permanent magnet synchronous motor (IPMSM) can have a wider speed regulation interval under the condition of weak magnetism, so that the application is wider.

In the control strategy of IPMSM, in order to maximize efficiency and maximize utilization of current capacity, the motor is controlled to operate on a maximum torque to current ratio (MTPA) curve before field weakening, and as the speed of the motor increases, the short voltage represented by d-axis q-axis currents Id and Iq, a permanent magnet flux linkage and an electrical angular velocity w gradually increases, and the specific expression is as follows:

where Umax is the inverter dc bus voltage that drives the motor. The current limiting relationship limited to the upper voltage limit can be obtained from the above equation. Seen in dq coordinate system is an ellipse which shrinks as w increases (see fig. 1). The dq current coordinate must fall within this ellipse. As shown in the figure, in the field weakening control stage, the dq current limit curve is composed of two parts, namely, the original MTPA curve (which becomes a section a in fig. 1) and the voltage limit curve corresponding to the current w (which becomes a section B in fig. 1), which are called field weakening limit curves.

Before the motor operates in the MTPV (maximum torque voltage ratio) stage, from the torque output point of view, when the dq axis current is positioned on the flux weakening limit curve, the output torque is the maximum value corresponding to the current.

According to the control theory of the motor, when the motor runs on the curve of the section B, the voltage amplitude of the output end of the motor is just equal to the voltage of a direct current bus of the inverter. When the motor runs on the A curve, the derivative of the torque signal to the current angle is recorded as (dTe (beta)) to be 0. In the scheme, dTe (beta) is used as a feedback signal, and a reasonable PI adjusting module is used for searching an angle value with the optimal torque under the current vector magnitude in order to adjust dTe (beta) to be 0 finally.

In general motor driving products, torque sensors are not included, and even with the torque sensors, the derivation of the torque sensors is difficult due to problems such as sampling time filtering parameters.

The invention discloses a method and a system for controlling an MTPA (maximum torque Angle Power) of a built-in permanent magnet synchronous motor, which are disclosed by the invention on 2016 (year 01), 20 (day) and 105262394A (Chinese published patent No. 105262394A), and discloses a method and a system for controlling an MTPA of a built-in permanent magnet synchronous motor. The invention provides an MTPA experimental method based on variable interval search and binary quadratic polynomial fitting, and an MTPA control method with higher efficiency and precision is realized.

The calibration work of the traditional mode on the motor curve can occupy a large time, and the development efficiency and the engineering progress are seriously influenced; because only one or a plurality of sample machines are calibrated in the calibration process, when the consistency is deficient, the calibration curve and the actual curve have larger difference.

The core of the autonomous learning algorithm adopted by the scheme is that on the premise of not adopting a torque sensor, reasonable observed quantity is searched to represent the partial derivative of the torque to the angle, and a reasonable PI adjusting module is utilized to search out the optimal angle value of the torque under the current vector magnitude, while in the existing solution, the weak magnetic curve is usually obtained through artificial repeated trial and error calibration. That is, after the current level is fixed, the angle of the current (beta shown in FIG. 1) is repeatedly adjusted to finally trace the curve shown. The problems of manual calibration are more, such as low precision, long time consumption, low efficiency and the like.

Disclosure of Invention

The invention aims to overcome the defects of long construction period, low efficiency, poor accuracy, poor robustness and the like of the conventional weak magnetic curve manual calibration method and provides a built-in permanent magnet synchronous motor weak magnetic curve tracking algorithm based on autonomous learning and a matching method.

Built-in permanent magnet synchronous motor weak magnetic curve tracking means based on autonomic study includes:

the controlled synchronous motor is electrically connected with the main control unit, the angle resolver and the voltage limit comparison module;

the main control unit is used for storing current and angle values, exchanging data of each unit and controlling the motor to run and is electrically connected with the controlled synchronous motor, the current comparator, the voltage limit comparison module, the autonomous learning derivative observer and the PI regulation module;

the current comparator is used for comparing whether the currents of the d axis and the q axis are both larger than zero, and is electrically connected with the controlled synchronous motor and the main control unit; the angle calculating module is arranged on the controlled synchronous motor, is used for calculating the angle of the synchronous motor, and is electrically connected with the main control unit and the autonomous learning derivative observer;

the voltage limit comparison module is characterized in that a first acquisition end is electrically connected with the direct current bus, a second acquisition end is electrically connected with a synchronous motor terminal, and an output end is electrically connected with the main control unit;

the autonomous learning derivative observer is used for calculating the angle derivative of the virtual torque and is electrically connected with the angle resolving module, the PI adjusting module and the main control unit;

and the PI adjusting module is used for calculating the motor control quantity for compensating the angle from the virtual torque to the angle derivative, and is electrically connected with the autonomous learning derivative observer and the main control unit.

The angle resolving module is used for adding the current angle to the angle compensation output by the PI adjusting module to obtain a new angle value; the autonomous learning derivative observer calculates a partial derivative of the torque to the angle by a disturbance observation method, the partial derivative is output to the PI adjusting module, and the PI adjusting module calculates the required angle compensation amount; when the error judger detects that the derivative of the torque to the angle is in the preset range, the tracking is judged to be finished, and the magnitude of the current vector is updated immediately.

The built-in permanent magnet synchronous motor weak magnetic curve tracking method based on autonomous learning comprises the following steps:

m1, setting a preset voltage limit value, a threshold value Slim and a zero moment angle value;

m2, the angle calculating unit reads d-axis current i of the synchronous motor_dAnd q-axis current i_qAngle calculation unit calculates i_dTo i_qThe ratio is subjected to arc tangent transformation calculation to obtain the current angle;

m3, the high frequency injection unit reads the current angle calculated by the angle solution unit and superposes the current angle with the high frequency component sin _ omega _ h to recalculate the virtual current value i_{d_h}And i_{q_h}；

M4, virtual Torque solving Unit readingTaking the direct-axis voltage V of the synchronous motor_dQuadrature axis voltage V_qRotational speed w_mechWhile reading the virtual current value i calculated by the high-frequency injection unit_{d_h}And i_{q_h}And d-axis current i of synchronous motor_dAnd q-axis current i_qAccording to the calculation formula:

calculating a virtual torque T_{e_h}(ii) a Wherein R is the armature winding of the motor body, and the value of the armature winding is 1/2 of the resistance value between two phases under the condition that the motor is disconnected;

m5, virtual Torque T_{e_h}Inputting the value into a band-pass filter to obtain a signal Te _ h _ fil;

m6, multiplying the signal Te _ h _ fil by the high frequency component sin _ omega _ h and performing band-pass filtering to obtain a derivative value S of the torque to the angle.

Preferably, the following steps are further provided before the step M2:

PM1, Current setting calculator reads d-axis Current i_dAnd judging whether the current value is greater than zero, reading q-axis current i_qJudging whether the current values are larger than zero, if so, judging that the current values pass through or not, if not, ending the process, and if so, executing a step PM 2;

PM2, detecting the rotation angle value of the synchronous motor, judging whether the ratio of the front end voltage amplitude to the bus voltage reaches the voltage limit, if so, recording the current and angle values, otherwise, executing the step M2;

PM3, reading the derivative value S of the torque to the angle in the step M6, judging whether the derivative value S of the torque to the angle is smaller than a preset Slim, if so, jumping to the step PM4, otherwise, jumping to the step PM 5;

PM4 for d-axis current i_dSubtracting the fixed operating value 1 for the q-axis current i_qSubtract the fixed operating value of 1 and perform step PM 1;

PM5, inputting the derivative value S of the torque to the angle to a PI regulator, and calculating a control quantity for compensating the existing angle by the PI regulator;

the PM6 updates the current angle value by the control amount for compensating the current angle, and moves to step PM 2.

The angle derivative value is S which is the derivative of the torque observed by the autonomous learning derivative observer to the angle, the optimal torque can be considered only when the value of S is 0, but a threshold value Slim needs to be set because the convergence speed problem in the solving process is hard to reach 0, and the optimal torque is considered to be searched when the derivative value S is smaller than a secondary threshold; so in the flow, when S < Slim, data storage is carried out, and then the operation of reducing one by the fixed operation value current instruction is carried out; and judging that the front end voltage amplitude is U and the bus voltage is Umax, so that the condition that U is greater than Umax or S < Slim is met, storing data, and subtracting a fixed value from the current operating value.

Preferably, the step M3 includes the following sub-steps:

a1, generating a high frequency component sin _ omega _ h by a waveform generator;

a2, superposing the current angle calculated by the angle calculating unit and the high-frequency component sin _ omega _ h to obtain an intermediate value beta sum;

a3, respectively calculating a sine value and a cosine value of the beta sum to obtain a middle value sin beta sum and a cos beta sum;

a4, reading d-axis current i_dAnd q-axis current i_qAnd executing the following calculation formula:

calculating to obtain a total current Ia;

a5, multiplying the intermediate value cos beta sum with the total current Ia to obtain the virtual current value i_{q_h}；

A6, negating the intermediate value sin β sum and multiplying the result by the total current Ia to obtain a virtual current value i_{d_h}。

Wherein i_dU into formula_[1],i_qU into formula_[2]。

The substantial effects of the invention are as follows: in the traditional calibration table look-up method, the calibration work of the motor curve takes a longer time, and the development efficiency and the engineering progress are seriously influenced; because the motor is calibrated manually, the offset error is large, and even the calibration results of different operators are different, the accuracy of the curve cannot be ensured; the method adopts a static mode, namely a principle of 'one-time calibration and permanent use', and when the parameters of the motor change along with the increase of the use time, the flux weakening curve of the motor can change obviously. The difference between the curve obtained by the static calibration method and the actual curve is increased, so that the overall efficiency of the motor is influenced; because only one or a plurality of sample machines are calibrated in the calibration process, when the consistency is deficient, the calibration curve and the actual curve have larger difference; the method of the invention can accelerate the development progress, the weak magnetic curve changes along with the state change of the motor, the similarity between the weak magnetic curve and the actual curve can be maximized, the operation efficiency of the motor is improved, the high similarity of the motor production line is not needed, the same method can be used along with different batches of motors, and the production cost of the motor and the development cost of the weak magnetic curve are reduced.

Drawings

FIG. 1 is a graph of a prior art limit curve;

FIG. 2 is a schematic diagram of an angle calculation module;

FIG. 3 is a high-frequency injection unit for solving a virtual Id and Iq current value calculation graph;

FIG. 4 is a schematic diagram of a virtual torque resolving unit;

FIG. 5 is a graph of a derivative solution unit solution.

Detailed Description

The technical solution of the present invention is further specifically described below by way of specific examples in conjunction with the accompanying drawings.

Example 1

As shown in fig. 2, fig. 3, fig. 4 and fig. 5, the weak magnetic curve tracking method of the interior permanent magnet synchronous motor based on autonomous learning includes the following steps:

PM1, Current setting calculator reads d-axis Current i_dAnd judging whether the current value is greater than zero, reading q-axis current i_qJudging whether the current values are larger than zero, if so, judging that the current values pass through or not, if not, ending the process, and if so, executing a step PM 2;

PM2, detecting the rotation angle value of the synchronous motor, judging whether the ratio of the front end voltage amplitude to the bus voltage reaches the voltage limit, if so, recording the current and angle values, otherwise, executing the step M2;

PM3, reading the derivative value S of the torque to the angle in the step M6, judging whether the derivative value S of the torque to the angle is smaller than a preset Slim, if so, jumping to the step PM4, otherwise, jumping to the step PM 5;

PM4 for d-axis current i_dSubtracting the fixed operating value 1 for the q-axis current i_qSubtract the fixed operating value of 1 and perform step PM 1;

PM5, inputting the derivative value S of the torque to the angle to a PI regulator, and calculating a control quantity for compensating the existing angle by the PI regulator;

the PM6 updates the current angle value by the control amount for compensating the current angle, and moves to step PM 2.

M1, setting a preset voltage limit value, a threshold value Slim and a zero moment angle value;

m2, the angle calculating unit reads d-axis current i of the synchronous motor_dAnd q-axis current i_qAngle calculation unit calculates i_dTo i_qThe ratio is subjected to arc tangent transformation calculation to obtain the current angle;

m3, the high frequency injection unit reads the current angle calculated by the angle solution unit and superposes the current angle with the high frequency component sin _ omega _ h to recalculate the virtual current value i_{d_h}And i_{q_h}；

M4, virtual torque solution unit reads the direct axis voltage V of the synchronous motor_dQuadrature axis voltage V_qRotational speed w_mechWhile reading the virtual current value i calculated by the high-frequency injection unit_{d_h}And i_{q_h}And synchronizationD-axis current i of motor_dAnd q-axis current i_qAccording to the calculation formula:

calculating a virtual torque T_{e_h}(ii) a Wherein R is the armature winding of the motor body, and the value of the armature winding is 1/2 of the resistance value between two phases under the condition that the motor is disconnected;

m5, virtual Torque T_{e_h}Inputting the value into a band-pass filter to obtain a signal Te _ h _ fil;

m6, multiplying the signal Te _ h _ fil by the high frequency component sin _ omega _ h and performing band-pass filtering to obtain a derivative value S of the torque to the angle.

The angle derivative value is S which is the derivative of the torque observed by the autonomous learning derivative observer to the angle, the optimal torque can be considered only when the value of S is 0, but a threshold value Slim needs to be set because the convergence speed problem in the solving process is hard to reach 0, and the optimal torque is considered to be searched when the derivative value S is smaller than a secondary threshold; so in the process, when S<Slim, storing data, and then performing the operation of reducing one by the current instruction with a fixed operation value; wherein, the front end voltage amplitude is judged to be U, the bus voltage is Umax, so that the condition of U is satisfied>Umax condition, or S<Slim condition, the data is stored and the current operating value is reduced by a fixed value of one, i.e., for the d-axis current i as described in step PM4_dSubtracting the fixed operating value 1 for the q-axis current i_qThe fixed operating value 1 is subtracted.

As shown in fig. 2, the angle calculating module takes a given current value as an input, and obtains a current angle by calculating an arctangent of a ratio of the given current value and the given current value.

As shown in FIG. 3, the high frequency injection unit superposes a high frequency component sin _ omega _ h on the basis of the angle calculation unit, and recalculates the virtual i_d、i_qCurrent value (i)_{d_h}And i_{q_h})。

As shown in fig. 5, the derivative solving unit is composed of a band-pass filter F1, a multiplier M1, and a low-pass filter F2. The signal Te _ h _ fil obtained by the virtual torque Te _ h through the band-pass filter F1 is multiplied by the high-frequency component sin _ omega _ h through the multiplier M1, and then low-pass filtering F2 is carried out to obtain the derivative of the torque to the angle.

The step M3 includes the following sub-steps:

a1, generating a high frequency component sin _ omega _ h by a waveform generator;

a2, superposing the current angle calculated by the angle calculating unit and the high-frequency component sin _ omega _ h to obtain an intermediate value beta sum;

a3, respectively calculating a sine value and a cosine value of the beta sum to obtain a middle value sin beta sum and a cos beta sum;

a4, reading d-axis current i_dAnd q-axis current i_qAnd executing the following calculation formula:

calculating to obtain a total current Ia;

a5, multiplying the intermediate value cos beta sum with the total current Ia to obtain the virtual current value i_{q_h}；

A6, negating the intermediate value sin β sum and multiplying the result by the total current Ia to obtain a virtual current value i_{d_h}。

Built-in permanent magnet synchronous motor weak magnetic curve tracking means based on autonomic study includes: the controlled synchronous motor is electrically connected with the main control unit, the angle resolver and the voltage limit comparison module; the main control unit is used for storing current and angle values, exchanging data of each unit and controlling the motor to run and is electrically connected with the controlled synchronous motor, the current comparator, the voltage limit comparison module, the autonomous learning derivative observer and the PI regulation module; the current comparator is used for comparing whether the currents of the d axis and the q axis are both larger than zero, and is electrically connected with the controlled synchronous motor and the main control unit; the angle calculating module is arranged on the controlled synchronous motor, is used for calculating the angle of the synchronous motor, and is electrically connected with the main control unit and the autonomous learning derivative observer; the voltage limit comparison module is characterized in that a first acquisition end is electrically connected with the direct current bus, a second acquisition end is electrically connected with a synchronous motor terminal, and an output end is electrically connected with the main control unit; the autonomous learning derivative observer is used for calculating the angle derivative of the virtual torque and is electrically connected with the angle resolving module, the PI adjusting module and the main control unit; and the PI adjusting module is used for calculating the motor control quantity for compensating the angle from the virtual torque to the angle derivative, and is electrically connected with the autonomous learning derivative observer and the main control unit.

Claims (3)

1. The built-in permanent magnet synchronous motor weak magnetic curve tracking method based on autonomous learning is characterized by comprising the following steps of:

m1, setting a preset voltage limit value, a threshold value Slim and a zero moment angle value;

m2, the angle calculating unit reads d-axis current i of the synchronous motor_dAnd q-axis current i_qAngle calculation unit calculates i_dTo i_qThe ratio is subjected to arc tangent transformation calculation to obtain the current angle;

m3, the high frequency injection unit reads the current angle calculated by the angle solution unit and superposes the current angle with the high frequency component sin _ omega _ h to recalculate the virtual current value i_{d_h}And i_{q_h}；

M4, virtual torque solution unit reads the direct axis voltage V of the synchronous motor_dQuadrature axis voltage V_qRotational speed w_mechWhile reading the virtual current value i calculated by the high-frequency injection unit_{d_h}And i_{q_h}And d-axis current i of synchronous motor_dAnd q-axis current i_qAccording to the calculation formula:

calculating a virtual torque T_{e_h}Wherein R is the armature winding of the motor body, and the value of R is 1/2 of the resistance value between two phases under the condition of motor open circuit;

m5, virtual Torque T_{e_h}Inputting the value into a band-pass filter to obtain a signal Te _ h _ fil;

m6, multiplying the signal Te _ h _ fil by the high-frequency component sin _ omega _ h and carrying out band-pass filtering to obtain a derivative value S of the torque to the angle;

the following steps are also provided before the step M2:

PM1, Current setting calculator reads d-axis Current i_dAnd judging whether the current value is greater than zero, reading q-axis current i_qJudging whether the current values are larger than zero, if so, judging that the current values pass through or not, if not, ending the process, and if so, executing a step PM 2;

PM2, detecting the rotation angle value of the synchronous motor, judging whether the ratio of the front end voltage amplitude to the bus voltage reaches the voltage limit, if so, recording the current and angle values, otherwise, executing the step M2;

PM3, reading the derivative value S of the torque to the angle in the step M6, judging whether the derivative value S of the torque to the angle is smaller than a preset Slim, if so, jumping to the step PM4, otherwise, jumping to the step PM 5;

PM4 for d-axis current i_dSubtracting the fixed operating value 1 for the q-axis current i_qSubtract the fixed operating value of 1 and perform step PM 1; PM5, inputting the derivative value S of the torque to the angle to a PI regulator, and calculating a control quantity for compensating the existing angle by the PI regulator;

the PM6 updates the current angle value by the control amount for compensating the current angle, and moves to step PM 2.

2. The tracking method for the weak magnetic curve of the interior permanent magnet synchronous motor based on the autonomous learning of claim 1, wherein the step M3 comprises the following substeps:

a1, generating a high frequency component sin _ omega _ h by a waveform generator;

a2, superposing the current angle calculated by the angle calculating unit and the high-frequency component sin _ omega _ h to obtain an intermediate value beta sum;

a3, respectively calculating a sine value and a cosine value of the beta sum to obtain a middle value sin beta sum and a cos beta sum;

a4, reading d-axis current i_dAnd q-axis current i_qAnd executing the following calculation formula:

calculating to obtain a total current Ia, wherein i_dU into formula_[1]，i_qU into formula_[2]；

A5, multiplying the intermediate value cos beta sum with the total current Ia to obtain the virtual current value i_{q_h}；

A6, negating the intermediate value sin β sum and multiplying the result by the total current Ia to obtain a virtual current value i_{d_h}。

3. The built-in permanent magnet synchronous motor weak magnetic curve tracking device based on autonomous learning is suitable for the built-in permanent magnet synchronous motor weak magnetic curve tracking method based on autonomous learning, and is characterized by comprising the following steps:

the controlled synchronous motor is electrically connected with the main control unit, the angle resolver and the voltage limit comparison module;

the main control unit is used for storing current and angle values, exchanging data of each unit and controlling the motor to run and is electrically connected with the controlled synchronous motor, the current comparator, the voltage limit comparison module, the autonomous learning derivative observer and the PI regulation module;

the current comparator is used for comparing whether the currents of the d axis and the q axis are both larger than zero, and is electrically connected with the controlled synchronous motor and the main control unit; the angle calculating module is arranged on the controlled synchronous motor, is used for calculating the angle of the synchronous motor, and is electrically connected with the main control unit and the autonomous learning derivative observer;

the voltage limit comparison module is characterized in that a first acquisition end is electrically connected with the direct current bus, a second acquisition end is electrically connected with a synchronous motor terminal, and an output end is electrically connected with the main control unit;

the autonomous learning derivative observer is used for calculating the angle derivative of the virtual torque and is electrically connected with the angle resolving module, the PI adjusting module and the main control unit;

and the PI adjusting module is used for calculating the motor control quantity for compensating the angle from the virtual torque to the angle derivative, and is electrically connected with the autonomous learning derivative observer and the main control unit.