CN115065290B - Permanent magnet synchronous motor current harmonic suppression method based on data driving - Google Patents

Abstract

The invention provides a data-driven permanent magnet synchronous motor current harmonic suppression method, which can effectively identify voltage interference items in real time by utilizing online data and effectively overcome the defect that the current control of the existing model prediction is seriously dependent on the accuracy of model parameters, so that the identification accuracy is not influenced by the initial value of system parameters. The identified voltage interference term is brought into a current increment prediction model, the optimal current increment and the corresponding duty ratio at the next moment are obtained through calculation, a novel modulation process and a five-segment modulation mode based on current circle tracking can be finally realized, effective control of motor stator current is completed, and current harmonic waves generated due to single effective vector modulation in traditional model prediction current control are effectively avoided.

Description

Permanent magnet synchronous motor current harmonic suppression method based on data driving

Technical Field

The invention belongs to the technical field of permanent magnet synchronous motor harmonic suppression, and particularly relates to a permanent magnet synchronous motor voltage and current harmonic suppression control technology based on a data driving and current loop tracking modulation method.

Background

In the prior art, model predictive current control (Model Predictive Current Control, MPCC) is widely adopted in the control of permanent magnet synchronous motors by virtue of the advantages of good dynamic performance, capability of handling control problems including constraints and the like. However, there are still some disadvantages of the current MPCC, and the control performance is very susceptible to voltage disturbance caused by motor model parameter mismatch and inherent nonlinear properties of the inverter. Meanwhile, due to the limitations of sensor cost and motor manufacturing technology, motor model parameters (resistance, inductance and flux linkage) are difficult to measure in real time or even cannot be measured at all. For the motor parameters or voltage disturbance, there are usually adopted on-line identification, estimated filters (kalman filter) and observers (dimension-reducing observer, full-order observer, sliding mode observer) and the like, but these methods are limited to processing unknown disturbance caused by a specific reason, and the identification performance is very sensitive to the parameter initial value setting in the identification equation.

Disclosure of Invention

In view of the above, the present invention provides a method for suppressing current harmonics of a permanent magnet synchronous motor based on data driving, which specifically includes the following steps:

Step one, establishing a current differential equation model containing a voltage disturbance term under a dq rotating coordinate system aiming at a permanent magnet synchronous motor; expressing the current differential equation model into a super-local first-order local model, and solving an estimated value expression of the voltage disturbance term in a preset variable time window for storing data based on Laplace positive and negative transformation;

Step two, collecting stator current, rotor electric angular speed, rotor position angle data and inverter output voltage data of the permanent magnet synchronous motor in real time, and forming a data register domain for rolling update in each sampling period based on the preset variable time window; calculating an estimated value of the voltage disturbance item in real time by utilizing the data;

Step three, calculating an optimal incremental current vector of the next sampling period and an optimal predicted current vector under the action of each effective voltage vector by using the acquired stator current, the estimated value of the voltage disturbance item calculated in the step two and the reference current of the next sampling period;

Converting the optimal incremental current vector and the optimal predicted current vector obtained by the calculation in the step three into an alpha beta static coordinate system, and constructing a cost function according to the absolute value of the difference between the optimal incremental current vector and the optimal predicted current vector; and taking the optimal increment current vector which enables the cost function to reach the minimum as a main vector, taking an optimal increment current vector which is closest to the main vector and a negative current increment under the action of a zero voltage vector as auxiliary vectors respectively, solving the optimal solutions of the main vector and the auxiliary vectors and the corresponding duty ratios thereof, and outputting the optimal solutions and the corresponding duty ratios to an inverter for modulation in the next sampling period.

Further, in the first step, a current differential equation model containing a voltage disturbance term under a dq rotation coordinate system is established specifically for the surface-mounted permanent magnet synchronous motor, and the following assumption is firstly based:

(1) The motor stator winding is a three-phase symmetrical winding;

(2) The magnetomotive force of electrons is distributed in a sine form along the air gap;

(3) Neglecting eddy current, hysteresis and core losses of the motor;

(4) The components in the motor stator inductance dq rotational coordinate system are equal, i.e., L _d＝L_q＝L_s.

On the basis of which the differential equation of the form:

Wherein u _d、u_q and i _d、i_q are respectively the d-axis and q-axis stator voltages and the d-axis and q-axis stator currents in the dq coordinate system; phi _r is the rotor flux linkage; r _s is the stator resistance; l _s is a stator inductance; omega _e is the electrical angular velocity of the rotor; u _d,par、U_q,par is a d-axis and q-axis voltage disturbance term generated by parameter mismatch; u _d,inv、U_q,inv is the d-axis and q-axis voltage disturbance term generated by the nonlinear characteristic of the inverter; t is a time variable;

The above description is rewritten as:

wherein X _d、X_q is the sum of polynomials including U _d,par、U_q,par、U_d,inv、U_q,inv in the d _q coordinate system; x _d、x_q is a scaling factor of the dq axis input stator voltage, typically set to (L _s)^-1.

Further, the current differential equation model is expressed in the form of the following superlocal first order local model:

dy/dt＝F+αu

Wherein y and u represent system state variables and control inputs, respectively; f represents a time-varying unknown system state parameter. Due to the undetectable nature of F, it is assumed that a constant function is possible in each control period And (5) approximate estimation.

And carrying out Laplace transformation on the obtained product to obtain the following steps:

Wherein y ₀ is an initial value of a state variable corresponding to a time period [ t- Γ, t ], Γ representing the preset variable time window length; deriving s in the formula eliminates y ₀ to obtain:

to attenuate the influence of the noise domain on the s time domain calculation, the two ends of the above formula are multiplied by s ^-2 at the same time, so as to obtain:

Inverse Laplace transformation is carried out on the above method, and integral calculation is carried out, so that the time period of [0, Γ ] is obtained The calculated expression of (2) is as follows:

Wherein Γ=n _T·T_s,n_T is defined as an integer;

Updating and storing the control input u and the system state variable y of the super local first order local model as a data register field in the form of u [0] … u [ n _T ] and y [0] … y [ n _T ] at each sampling period;

Then the estimated value of X _d,X_q Calculated by the following formula:

Where Γ is a variable time window of the data storage domain, defined as Γ=n _T·T_s. Wherein, T _s is the sampling period of the system, and n _T is a positive integer for representing the length of the time window; λ represents the λ sample point, and so on.

And respectively registering voltage and current data of the domain.

Further, in the second step, data acquired in real time are discretely sampled and registered according to fixed frequency, and data acquired at the moment k, (k-1), (k-2) … … (k- Γ -1) are jointly formed into a data register field corresponding to a sampling period at the moment k, and each sampling period is continuously updated and stored;

Considering the reasons that the instantaneous calculation load of the main control chip, the quick response requirement under the dynamic working condition, the historical data are favorable for improving the identification precision under the steady-state working condition and the like, the data dimension n _T used for calculation in the data register domain is set as follows:

Wherein, Is the reference current.

Further, the optimal incremental current vector of the next sampling period in the third stepOptimal predicted current vector under the action of each effective voltage vectorCalculated by:

where u _di(k)、u_qi (k) is the coordinates of the eight voltage vectors generated by the inverter in the dq coordinate system, i ε (0, 1,2,3,4,5,6, 7); Respectively referencing current vectors of dq axes at corresponding moments; predicting current vectors for dq axes at corresponding moments respectively; t _s is the sampling period.

Further, in step four, in order to eliminate the influence of the rotor position on the calculation accuracy, the optimal incremental current vector is first obtained by using the inverse Park transformationOptimal predicted current vectorThe rotating coordinate system is converted to an αβ stationary coordinate system and a cost function is constructed as follows:

The current increment corresponding to min (J _j) is set as the main vector, denoted by (Δi _aα,Δi_αβ). Two current increment vectors adjacent to the vector are set as candidate sub-vectors, and are denoted by (Δi _b1α,Δi_b1β)、(Δi_b2α,Δi_b2β). And setting a negative current increment formed by the zero voltage vector action as an optional auxiliary vector, and representing the negative current increment by (delta i _0α,Δi_0β); the symbol dc represents the duty cycle for the selected primary or secondary vector;

and establishing the following ternary once equation set by using the selected main vector and candidate auxiliary vector:

the optimal solution basis vector and the corresponding duty cycle are calculated based on the following conditions:

In the method, in the process of the invention, The optimal increment, the suboptimal increment and the negative increment in the optimal solution base vector are obtained; dc _ini (k+1) is the duty cycle corresponding to the optimal solution vector at time (k+1).

Based on the above results, in order to reduce the number of switching times per control period as much as possible, modulation can be performed using a 5-segment modulation mode.

The method for suppressing the current harmonic waves of the permanent magnet synchronous motor based on data driving can effectively identify the voltage interference item in real time by utilizing online data, and effectively overcomes the defect that the current control of the existing model prediction is seriously dependent on the accuracy of model parameters, so that the identification accuracy is not influenced by the initial value of system parameters. The identified voltage interference term is brought into a current increment prediction model, the optimal current increment and the corresponding duty ratio at the next moment are obtained through calculation, a novel modulation process and a five-segment modulation mode based on current circle tracking can be finally realized, effective control of motor stator current is completed, and current harmonic waves generated due to single effective vector modulation in traditional model prediction current control are effectively avoided.

Drawings

FIG. 1 is a general flow chart of the method provided by the present invention;

FIG. 2 is a diagram of a calculation process of an estimated value for a voltage disturbance term in the method provided by the present invention;

FIG. 3 is an example effect diagram of the present invention in dynamic and steady state conditions.

Detailed Description

The following description of the embodiments of the present invention will be made apparent and fully in view of the accompanying drawings, in which some, but not all embodiments of the invention are shown. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The invention provides a permanent magnet synchronous motor current harmonic suppression method based on data driving, which specifically comprises the following steps:

Step one, establishing a current differential equation model containing a voltage disturbance term under a dq rotating coordinate system aiming at a permanent magnet synchronous motor; expressing the current differential equation model into a super-local first-order local model, and solving an estimated value expression of the voltage disturbance term in a preset variable time window for storing data based on Laplace positive and negative transformation;

Step two, collecting stator current, rotor electric angular speed, rotor position angle data and inverter output voltage data of the permanent magnet synchronous motor in real time, and forming a data register domain for rolling update in each sampling period based on the preset variable time window; calculating an estimated value of the voltage disturbance item in real time by utilizing the data;

Step three, calculating an optimal incremental current vector of the next sampling period and an optimal predicted current vector under the action of each effective voltage vector by using the acquired stator current, the estimated value of the voltage disturbance item calculated in the step two and the reference current of the next sampling period;

Converting the optimal incremental current vector and the optimal predicted current vector obtained by the calculation in the step three into an alpha beta static coordinate system, and constructing a cost function according to the absolute value of the difference between the optimal incremental current vector and the optimal predicted current vector; and taking the optimal increment current vector which enables the cost function to reach the minimum as a main vector, taking an optimal increment current vector which is closest to the main vector and a negative current increment under the action of a zero voltage vector as auxiliary vectors respectively, solving the optimal solutions of the main vector and the auxiliary vectors and the corresponding duty ratios thereof, and outputting the optimal solutions and the corresponding duty ratios to an inverter for modulation in the next sampling period.

In a preferred embodiment of the present invention, in step one, a current differential equation model including the voltage disturbance term under the dq rotation coordinate system is built specifically for the surface mounted permanent magnet synchronous motor, and the following assumption is first based:

(1) The motor stator winding is a three-phase symmetrical winding;

(2) The magnetomotive force of electrons is distributed in a sine form along the air gap;

(3) Neglecting eddy current, hysteresis and core losses of the motor;

(4) The components in the motor stator inductance dq rotational coordinate system are equal, i.e., L _d＝L_q＝L_s.

On the basis of which the differential equation of the form:

Wherein u _d、u_q and i _d、i_q are respectively the d-axis and q-axis stator voltages and the d-axis and q-axis stator currents in the dq coordinate system; phi _r is the rotor flux linkage; r _s is the stator resistance; l _s is a stator inductance; omega _e is the electrical angular velocity of the rotor; u _d,par、U_q,par is a d-axis and q-axis voltage disturbance term generated by parameter mismatch; u _d,inv、U_q,inv is the d-axis and q-axis voltage disturbance term generated by the nonlinear characteristic of the inverter; t is a time variable;

The above description is rewritten as:

Wherein X _d、X_q is the sum of polynomials containing U _d,par、U_q,par、U_d,inv、U_q,inv in dq coordinate system; x _d、x_q is a scaling factor of the dq axis input stator voltage, typically set to (L _s)^-1.

In a preferred embodiment of the invention, the current differential equation model is expressed in the form of the following superlocal first order local model:

dy/dt＝F+αu

Wherein y and u represent system state variables and control inputs, respectively; f represents a time-varying unknown system state parameter. Due to the undetectable nature of F, it is assumed that a constant function is possible in each control period And (5) approximate estimation.

And carrying out Laplace transformation on the obtained product to obtain the following steps:

Wherein y ₀ is an initial value of a state variable corresponding to a time period [ t- Γ, t ], Γ representing the preset variable time window length; deriving s in the formula eliminates y ₀ to obtain:

to attenuate the influence of the noise domain on the s time domain calculation, the two ends of the above formula are multiplied by s ^-2 at the same time, so as to obtain:

Inverse Laplace transformation is carried out on the above method, and integral calculation is carried out, so that the time period of [0, Γ ] is obtained The calculated expression of (2) is as follows:

Wherein Γ=n _T·T_s,n_T is defined as an integer;

Updating and storing the control input u and the system state variable y of the super local first order local model as a data register field in the form of u [0] … u [ n _T ] and y [0] … y [ n _T ] at each sampling period;

Then the estimated value of X _d,X_q Calculated by the following formula:

Where Γ is a variable time window of the data storage domain, defined as Γ=n _T·T_s. Wherein, T _s is the sampling period of the system, and n _T is a positive integer for representing the length of the time window; λ represents the λ sample point, and so on.

And respectively registering voltage and current data of the domain.

In a preferred embodiment of the present invention, in the second step, data acquired in real time are discretely sampled and registered according to a fixed frequency, and data acquired at the time k, (k-1), (k-2) … … (k- Γ -1) are jointly formed into a data register field corresponding to a sampling period at the time k, and are continuously updated and stored for each sampling period;

Considering the reasons that the instantaneous calculation load of the main control chip, the quick response requirement under the dynamic working condition, the historical data are favorable for improving the identification precision under the steady-state working condition and the like, the data dimension n _T used for calculation in the data register domain is set as follows:

Wherein, Is the reference current.

In a preferred embodiment of the present invention, the optimal delta current vector for the next sampling period in step threeOptimal predicted current vector under the action of each effective voltage vectorCalculated by:

where u _di(k)、u_qi (k) is the coordinates of the eight voltage vectors generated by the inverter in the dq coordinate system, i ε (0, 1,2,3,4,5,6, 7); Respectively referencing current vectors of dq axes at corresponding moments; predicting current vectors for dq axes at corresponding moments respectively; t _s is the sampling period.

In a preferred embodiment of the present invention, in step four, to eliminate the influence of rotor position on the calculation accuracy, the optimal delta current vector is first transformed using an inverse Park transformationOptimal predicted current vectorThe rotating coordinate system is converted to an αβ stationary coordinate system and a cost function is constructed as follows:

The current increment corresponding to min (J _j) is set as the main vector, denoted by (Δi _aα,Δi_αβ). Two current increment vectors adjacent to the vector are set as candidate sub-vectors, and are denoted by (Δi _b1α,Δi_b1β)、(Δi_b2α,Δi_b2β). And setting a negative current increment formed by the zero voltage vector action as an optional auxiliary vector, and representing the negative current increment by (delta i _0α,Δi_0β); the symbol dc represents the duty cycle for the selected primary or secondary vector;

And establishing the following ternary once equation set by using the selected main vector and auxiliary vector:

the optimal solution basis vector and the corresponding duty cycle are calculated based on the following conditions:

In the method, in the process of the invention, The optimal increment, the suboptimal increment and the negative increment in the optimal solution base vector are obtained; dc _ini (k+1) is the duty cycle corresponding to the optimal solution vector at time (k+1).

Based on the above results, in order to reduce the number of switching times per control period as much as possible, modulation can be performed using a 5-segment modulation mode.

It should be understood that, the sequence number of each step in the embodiment of the present invention does not mean that the execution sequence of each process should be determined by the function and the internal logic of each process, and should not limit the implementation process of the embodiment of the present invention.

Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims (4)

1. A method for suppressing current harmonic waves of a permanent magnet synchronous motor based on data driving is characterized by comprising the following steps of: the method specifically comprises the following steps:

Step one, establishing a current differential equation model containing a voltage disturbance term under a dq rotating coordinate system aiming at a permanent magnet synchronous motor; expressing the current differential equation model into a super-local first-order local model, and solving an estimated value expression of the voltage disturbance term in a preset variable time window for storing data based on Laplace positive and negative transformation;

Step two, collecting stator current, rotor electric angular speed, rotor position angle data and inverter output voltage data of the permanent magnet synchronous motor in real time, and forming a data register domain for rolling update in each sampling period based on the preset variable time window; calculating an estimated value of the voltage disturbance item in real time by utilizing the data;

Step three, calculating an optimal incremental current vector of the next sampling period and an optimal predicted current vector under the action of each effective voltage vector by using the acquired stator current, the estimated value of the voltage disturbance item calculated in the step two and the reference current of the next sampling period;

Converting the optimal incremental current vector and the optimal predicted current vector obtained by the calculation in the step three into an alpha beta static coordinate system, and constructing a cost function according to the absolute value of the difference between the optimal incremental current vector and the optimal predicted current vector; taking the optimal increment current vector which enables the cost function to reach the minimum as a main vector, taking an optimal increment current vector which is closest to the main vector and negative current increment under the action of a zero voltage vector as auxiliary vectors respectively, solving the optimal solutions of the main vector and the auxiliary vectors and the corresponding duty ratios thereof, and outputting the optimal solutions to an inverter for modulation in the next sampling period;

in the first step, a current differential equation model containing a voltage disturbance term under a dq rotating coordinate system is specifically built for a permanent magnet synchronous motor, and the current differential equation model is firstly based on the following assumption:

(1) The motor stator winding is a three-phase symmetrical winding;

(2) The magnetomotive force of electrons is distributed in a sine form along the air gap;

(3) Neglecting eddy current, hysteresis and core losses of the motor;

(4) The components of the motor stator inductance under the dq rotation coordinate system are equal;

on the basis of which the differential equation of the form:

Wherein u _d、u_q and i _d、i_q are respectively the d-axis and q-axis stator voltages and the d-axis and q-axis stator currents in the dq coordinate system; phi _r is the rotor flux linkage; r _s is the stator resistance; l _s is a stator inductance; omega _e is the electrical angular velocity of the rotor; u _d,par、U_q,par is a d-axis and q-axis voltage disturbance term generated by parameter mismatch; u _d,inv、U_q,inv is the d-axis and q-axis voltage disturbance term generated by the nonlinear characteristic of the inverter; t is a time variable;

The above description is rewritten as:

Wherein X _d、X_q is the sum of polynomials containing U _d,par、U_q,par、U_d,inv、U_q,inv in dq coordinate system; xd, xq are the scale factors of dq axis input stator voltage, set to (L _s)^-1

And expressing the current differential equation model as a form of a super local first order local model:

dy/dt＝F+αu

Wherein y and u represent system state variables and control inputs, respectively; alpha is a non-physical factor selected according to actual conditions; f represents a time-varying unknown system state parameter, due to its undetectable nature, by a constant function Approximate estimation;

and carrying out Laplace transformation on the obtained product to obtain the following steps:

Wherein Y and U correspond to Y and U before transformation, respectively; wherein y ₀ is the initial value of the state variable corresponding to the time period [ t- Γ, t ], f representing the preset variable time window length; deriving s in the formula eliminates y ₀ to obtain:

to attenuate the influence of the noise domain on the s time domain calculation, the two ends of the above formula are multiplied by s ^-2 at the same time, so as to obtain:

Inverse Laplace transformation is carried out on the above method, and integral calculation is carried out, so that the time period of [0, Γ ] is obtained The calculated expression of (2) is as follows:

wherein sigma is an integral variable; defining a variable time window Γ=n _T·T_s,n_T of the data storage domain as a positive integer for representing the length of the time window, T _s being the sampling period of the system;

Updating and storing the control input u and the system state variable y of the super local first order local model as a data register field at each sampling period, expressed in the form of u [0]. U [ n _T ] and y [0]. Y [ n _T ];

then the estimated value of X _d,X_q Calculated by the following formula:

Where λ represents the λ sampling point, and so on;

and respectively registering voltage and current data of the domain.

2. The method of claim 1, wherein: in the second step, the data acquired in real time are discretely sampled and registered according to fixed frequency, the data acquired at the moment k, (k-1) and (k-2) … … (k-f-1) are jointly formed into a data register field corresponding to the sampling period at the moment k, and each sampling period is continuously updated and stored;

Considering the reasons that the instantaneous calculation load of the main control chip, the quick response requirement under the dynamic working condition and the historical data are favorable for improving the identification precision under the steady-state working condition, the data dimension n _T used for calculation in the data register domain is set as follows:

Wherein, Representing the reference current.

3. The method of claim 2, wherein: optimal incremental current vector for next sampling period in step threeOptimal predicted current vector under the action of each effective voltage vectorCalculated by:

where u _di(k)、u_qi (k) is the coordinates of the eight voltage vectors generated by the inverter in the dq coordinate system, i ε (0, 1,2,3,4,5,6, 7); Respectively referencing current vectors of dq axes at corresponding moments; predicting current vectors for dq axes at corresponding moments respectively; t _s is the sampling period.

4. A method as claimed in claim 3, wherein: in the fourth step, in order to eliminate the influence of the rotor position on the calculation accuracy, the optimal incremental current vector is first converted by using the inverse Park transformationOptimal predicted current vectorThe rotating coordinate system is converted to an αβ stationary coordinate system and the following cost function is constructed:

The current increment corresponding to min (J _j) is set as a main vector and is expressed by (delta i _aα,Δi_aβ); two current increment vectors adjacent to the vector are set as candidate auxiliary vectors, and are respectively indicated by (delta i _b1α,Δi_b1β)、(Δib_2α,Δib_2β); and setting a negative current increment formed by the zero voltage vector action as an optional auxiliary vector, and representing the negative current increment by (delta i _0α,Δi_0β); the symbol dc represents the duty cycle corresponding to the selected primary or secondary vector;

And establishing the following ternary once equation set by using the selected main vector and auxiliary vector:

The optimal solution basis vector and its corresponding duty cycle that minimizes the cost function J _j are calculated based on the following conditions:

In the method, in the process of the invention, The optimal increment, the suboptimal increment and the negative increment in the optimal solution base vector are obtained; dc _ini (k+1) is the duty cycle corresponding to the optimal solution vector at time (k+1).