CN112436773A - Improved dead-beat prediction control method for permanent magnet synchronous motor single current sensor - Google Patents

Abstract

The invention provides an improved deadbeat predictive control method for a single current sensor of a permanent magnet synchronous motor. First, the traditional deadbeat predictive control is improved, and the improved deadbeat predictive control can effectively suppress the effects caused by motor parameter changes. influences. Then the reconstructed phase current based on a single current sensor is applied to improve dead-beat predictive control, which reduces the cost of the control system, reduces the volume of the control system, and avoids the measurement error caused by the difference between the current sensors. In addition, the control method has a certain inhibitory effect on the parameter changes of the motor.

Description

Improved dead-beat prediction control method for permanent magnet synchronous motor single current sensor

Technical Field

The invention relates to the technical field of permanent magnet synchronous motor control, in particular to single current sensor dead-beat prediction control of a permanent magnet synchronous motor based on a bus current sensor, which is suitable for inhibiting parameter disturbance under the condition of parameter mismatch of the permanent magnet synchronous motor.

Background

In the existing permanent magnet synchronous motor control technology, by means of traditional control strategies such as PI control, dead beat model predictive control and the like, the dependence degree on system models and motor parameters is very high, the inherent problem of poor parameter robustness exists, when the parameters change, the control effect is rapidly reduced, the control strategies depending on models and parameter accuracy cannot meet the requirement of high-performance control, and the application range of the control strategies is limited.

At present, the main methods for controlling the permanent magnet synchronous motor comprise vector control and direct torque control, and most of the methods are based on closed-loop control and require the acquisition of three-phase current of the motor for feedback to form a closed loop. Accurate acquisition of three-phase current is particularly important for the control strategies, the simplest mode is to detect currents of two-phase windings respectively, but the increase of torque fluctuation of the whole system is brought by direct current bias difference or current gain difference between different sensors, and the improvement of system performance is limited. This problem does not exist if a single current sensor is used, and the hardware cost of the single current sensor is the lowest. However, due to the non-simultaneity of sampling of the single current sensor, the current reconstruction needs a complex algorithm and needs to consider the compensation of the voltage vector.

Disclosure of Invention

In view of this, the present invention provides an improved method for predicting dead beat of a single current sensor of a permanent magnet synchronous motor, which specifically includes the following steps:

step one, collecting three-phase current i of permanent magnet synchronous motor at current k moment in real time on line_a、i_b、i_cA rotor speed w and a rotor position angle theta;

establishing mathematical models of the permanent magnet synchronous motor in an alpha-beta coordinate system and a d-q coordinate system, and establishing a traditional dead-beat current prediction control model; improving the traditional deadbeat current prediction control model based on the difference between the current measured value and the predicted value at the current k moment caused by parameter disturbance to obtain an improved deadbeat current prediction control model corresponding to the k +1 moment; calculating reference voltage within the SVPWM output voltage range at the k +1 moment in real time by combining the data acquired in the step one;

step three, the improved permanent magnet synchronous motor dead-beat prediction control and the single current sensor bus current sampling are combined, namely, the three-phase current i is reconstructed by utilizing the real-time switching state of the inverter and the acquired inverter bus current i_a、i_b、i_cReplacing the three-phase current i of the permanent magnet synchronous motor acquired in real time in the step one_a、i_b、i_cAnd calculating actual predicted values of d-axis current and q-axis current at the k moment based on a volt-second balance principle by using the actual values of the d-axis current and the q-axis current at the first rising edge moment in the switching period at the k-1 moment, and replacing corresponding current values in the improved deadbeat current prediction control model.

Further, in the second step, a mathematical model of the permanent magnet synchronous motor in the α - β coordinate system is as follows:

u_α＝R_si_α+L_spi_α-w_eψ_rsinθ

u_β＝R_si_β+L_spi_β+w_eψ_rcosθ

ψ_α＝L_si_α+ψ_rcosθ

ψ_β＝L_si_β+ψ_rsinθ

T_e＝1.5p_mψ_r(i_βcosθ-i_αsinθ)

in the formula u_α、u_βIs the stator voltage under an alpha-beta coordinate system; i.e. i_α、i_βIs in an alpha-beta coordinate systemA stator current; Ψ_rIs a rotor flux linkage; r_sIs a stator resistor; l is_sIs a stator inductance; w is a_e、w_mThe electrical angular velocity of the rotor and the mechanical angular velocity of the rotor, respectively; theta is a rotor position angle; p is a differential operator; t is_eIs an electromagnetic torque; t is_LIs the load torque; b is a viscosity coefficient; p is a radical of_mThe number of pole pairs of the motor is shown; Ψ_α、Ψ_βIs a stator flux linkage under an alpha-beta coordinate system; t is a time variable; j is load moment of inertia;

the mathematical model of the permanent magnet synchronous motor under a d-q coordinate system is as follows:

u_d＝R_si_d+pψ_d-w_eψ_q

u_q＝R_si_q+pψ_q+w_eψ_d

ψ_d＝L_di_d+ψ_r

ψ_q＝L_qi_q

T_e＝1.5p_m(ψ_ri_q+(L_d-L_q)i_di_q)

in the formula u_d、u_qIs the stator voltage under a d-q coordinate system; i.e. i_d、i_qIs the stator current under a d-q coordinate system; Ψ_d、Ψ_qIs a stator flux linkage under a d-q coordinate system; l is_d、L_qArmature inductances of d and q axes, respectively;

the traditional deadbeat current prediction control model is established as follows:

i_d-p(k+1)＝i_d(k)×(1-T_s×R_s/L_s)+i_q(k)×T_s×w_e+T_s/L_s×u_d(k)

i_q-p(k+1)＝i_q(k)×(1-T_s×R_s/L_s)-i_d(k)×T_s×w_e+T_s/L_s×u_q(k)-T_s×w_e×ψ_r/L_s

u_d-p(k+1)＝L_s/T_s×(i_d-ref-(1-T_s×R_s/L_s)×i_d-p(k+1)-T_s×i_q-p(k+1)×w_e)

u_q-p(k+1)＝L_s/T_s×(i_q-ref-(1-T_s×R_s/L_s)×i_q-p(k+1)+T_s×i_d-p(k+1)×w_e+T_s×w_e/L_s×ψ_r)

in the formula i_d-p(k +1) d-axis predicted current, i, at time k +1_q-p(k +1) is the predicted current of the q-axis at the time k +1, i_d-refFor d-axis reference current, i_q-refFor q-axis reference current, u_d-p(k +1) d-axis predicted voltage at time k +1, u_q-p(k +1) is the predicted voltage of the q-axis at time k +1, T_sIs a switching cycle;

the improved deadbeat current prediction control model is as follows:

i_d-error(k)＝i_d-p(k)-i_d(k)

i_q-error(k)＝i_q-p(k)-i_q(k)

K_d2＝(u_d(k-1)-u_d(k-2))

K_q2＝(u_q(k-1)-u_q(k-2))

X_d＝(i_d-err(k)-i_d-error(k-1))/K_d2/T_sX_q

＝(i_q-error(k)-i_q-err(k-1))/K_q2/T_s

K_d＝1/L_s-X_d

K_q＝1/L_s-X_q

i_d-p(k+1)

＝u_d(k)×K_d×T_s+i_d(k)×(1-T_s×R_s/L_s)+i_q(k)×T_s×w_e-(i_d-error(k)-X_d×T_s×u_d(k))

i_q-p(k+1)

＝u_q(k)×K_q×T_s+i_q(k)×(1-T_s×R_s/L_s)-i_d(k)×T_s×w_e-T_s×w_e×Ψ_f/L_s-(i_q-error(k)-X_q×T_s×u_q(k))

u_d-p(k+1)

＝(i_d-ref-i_d-p(k+1)×(1-T_s×R_s/L_s)-i_q-p(k+1)×T_s×w_e+i_d-err(k))/(K_d+X_d)/T_s

u_q-p(k+1)

＝(i_q-ref-i_q-p(k+1)×(1-T_s×R_s/L_s)+i_d-p(k+1)×T_s×w_e+T_s×w_e×Ψ_f/L_s+i_q-error(k))/(K_q+X_q)/T_s

in the formula i_d-err(k)、i_q-error(k) D and q axis current prediction errors i at time k_d-p(k)、i_q-p(k) D and q axis current predicted values i at the k time_d(k)、i_q(k) Actual measured values of d-and q-axis currents at time k, u_d(k-1) and u_d(k-2) d-axis voltages respectively acting at the k-1 moment and the k-2 moment, and u can be obtained by the same principle_q(k-1) and u_q(k-2)，i_d-p(k +1) is the improved d-axis predicted current at time k +1, i_q-pAnd (k +1) is the improved q-axis predicted current at the k +1 moment.

Further, the third step is specifically:

calculating the current at the k moment according to the reconstructed current at the k-1 switching cycle current updating momentPredicted value i of actual value_d′(k)、i_q' (k) replacing the actual value of the current at the time k, the current update time pair u of the k-1 cycle is calculated_d(k-1) and u_q(k-1) performing compensation;

i_d′(k)＝i_d(k-1，t₁)×(1-(T_s-t₁)×R_s/L_s)+i_q(k-1，t₁)×(T_s-t₁)×w_e+(T_s-t₁)/L_s×u_d(k-1)×T_s/(T_s-t₁)

i_q′(k)＝i_q(k-1，t₁)×(1-(T_s-t₁)×R_s/L_s)-i_d(k-1，t₁)×(T_s-t₁)×w_e+(T_s-t₁)/L_s×u_q(k-1)×T_s/(T_s-t₁)-(T_s-t₁)×w_e×Ψ_f/L_s

wherein i_d' (k) is a predicted value of the actual value of the d-axis current at time k, i_q' (k) is a predicted value of the actual value of the q-axis current at time k, t₁For the first rising edge instant i in a switching cycle_d(k-1，t₁) Is the d-axis current actual value, i, at the time of the k-1 period t1_q(k-1，t₁) Is k-1 period t₁An actual value of q-axis current at a time;

then using i_d' (k) and i_q' (k) replacement of i in the improved deadbeat current predictive control model in step two_d(k) And i_q(k)。

The improved single-current-sensor dead-beat prediction control method for the permanent magnet synchronous motor has the characteristic of good parameter disturbance suppression, and can completely replace the three-phase current of the motor to carry out prediction control by combining the improved dead-beat prediction control with the phase current reconstruction based on the bus current sensor, and simultaneously still maintain certain parameter disturbance suppression capability. It can be seen that the method of the present invention has at least the following beneficial effects compared to the prior art:

(1) the method effectively inhibits the influence caused by the change of the motor parameters by utilizing the improved dead-beat predictive control, so that the system has good control characteristics.

(2) The method combines improved dead-beat predictive control with phase current reconstruction based on the bus current sensor, reduces the cost of a control system, reduces the volume of the control system, and avoids the limitation of the difference between different sensors on the performance improvement of the whole system.

Drawings

FIG. 1 is a block diagram of a system model corresponding to the method of the present invention.

Fig. 2 is a comparison of d and q-axis currents and reference currents obtained based on the present invention in the case of a resistance mismatch (Rs ═ 10 Rs');

FIG. 3 shows flux linkage mismatch (psi)_f＝2ψ_f') the d and q-axis currents obtained based on the invention are compared with a reference current.

Detailed Description

The technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The improved single-current sensor dead-beat prediction control method for the permanent magnet synchronous motor, as shown in fig. 1, specifically comprises the following steps:

step one, collecting three-phase current i of permanent magnet synchronous motor at current k moment in real time through online data_a、i_b、i_cA rotor speed w and a rotor position angle theta;

establishing mathematical models of the permanent magnet synchronous motor in an alpha-beta coordinate system and a d-q coordinate system, establishing a traditional deadbeat current prediction control model, and improving the traditional deadbeat current prediction control model based on the difference between a current measured value and a predicted value at the current moment caused by parameter disturbance to obtain an improved deadbeat current prediction control model corresponding to the k +1 moment; calculating reference voltage within the SVPWM output voltage range at the next moment in real time by combining the data collected in the first step;

thirdly, reconstructing a three-phase current i by utilizing the real-time switching state of the inverter and the acquired bus current i of the inverter_a、i_b、i_cReplacing the three-phase current i of the permanent magnet synchronous motor acquired in real time in the step one_a、i_b、i_cAnd calculating actual predicted values of d-axis current and q-axis current at the k moment by using the actual values of the d-axis current and the q-axis current at the first rising edge moment in the switching period at the k-1 moment and a volt-second balance principle, and replacing corresponding current values in the improved deadbeat current prediction control model. .

Conventional deadbeat current predictive control is based on a current reference i_refAnd the voltage vector u (k) applied to the motor at the current moment and the motor parameters, and outputting the motor reference voltage u (k +1) at the next moment. At the k-th instant u (k +1) is calculated and applied at the next instant, so that the motor current reaches the current reference value at the instant k + 2. Thus, u (k +1) can be calculated at the kth time according to the following relation:

u_d(k+1)＝(2T_sR_s-2L_s)wi_q(k)-(L_s/T_s+T_sR_sR_s/L_s-2R_s)i_d(k)+T_sL_swwi_d(k)-T_swu_q(k)-(1-T_sR_s/L_s)u_d(k)+wwT_sψ_f

u_q(k+1)＝L_s/T_s×i_q-ref-(L_s/T_s+T_sR_sR_s/L_s-2R_s)i_q(k)-(2T_sR_s-2L_s)i_d(k)i_q(k)+T_sL_swwi_q(k)+T_swu_d(k)-(1-T_sR_s/L_s)u_q(k)+w(2ψ_f-T_sR_sψ_f/L_s)

in the formula T_sIs a control period; i.e. i_q-refIs a q-axis reference current.

When the calculated reference voltage exceeds the maximum output voltage limit of the SVPWM, the output reference voltage needs to be adjusted to obtain the reference voltage within the SVPWM output range:

in the formula u_d-j、u_q-jThe calculated stator reference voltage under the d-q coordinate system; u. of_d-x、u_q-xThe reference voltage within the corrected SVPWM output voltage range under the d-q coordinate system is obtained; u. of_dcIs the dc bus voltage.

It can be seen that the traditional dead-beat predictive control has high dependence on the motor model and parameter accuracy, and once the motor has parameter mismatch in the running process, the control effect is rapidly reduced. The invention changes the traditional dead-beat prediction control, and can be seen from the traditional dead-beat model that the current predicted value at the k +1 moment is different from the measured value due to parameter disturbance, and the error magnitude of the current predicted value is related to the variable quantity of the parameter:

for the d-axis, there are:

i_d-p(k)＝i_d(k-1)×(1-T_s×R_s′/L_s′)+i_q(k-1)×T_s×w_e+T_s×u_d(k-1)/L_s′

i_d(k)＝i_d(k-1)×(1-T_s×R_s/L_s)+i_q(k-1)×T_s×w_e+T_s×u_d(k-1)/L_s

in the formula, R_s′、L_s' is the theoretical value of the resistance and inductance of the stator of the motor, R_s、L_sIs its actual value.

Δi_d(k)＝i_d-p(k)-i_d(k)

＝u_d(k-1)×(T_s/L_s′-T_s/L_s)+i_d(k-1)×(T_s×R_s/L_s-T_s′×R_s′/L_s′)

In the same way, the method for preparing the composite material,

Δi_d(k-1)＝u_d(k-2)×(T_s/L_s′-T_s/L_s)+i_d(k-2)×(T_s×R_s/L_s-T_s′×R_s′/L_s′)

in the above two equations, the second term is very small compared with the first term and can be ignored, so that:

Δσ_d＝Δi_d(k)-Δi_d(k-1)＝(T_s/L_s′-T_s/L_s)×(u_d(k-1)-u_d(k-2))

then order:

K_d1＝Δσ_d/(u_d(k-1)-u_d(k-2))＝T_s/L_s′-T_s/L_s

the predicted value of the current at the k +1 moment is optimized for the first time:

i_d-p′(k+1)＝(T_s/L_s′-K_d1)×u_d(k)+i_d(k)×(1-T_s×R_s/L_s)+i_q(k)×T_s×ω_e

and then ordering:

K_d2＝Δi_d(k)-K_d1×u_d(k-1)

and (3) carrying out second optimization on the predicted value of the current at the k +1 moment:

i_d-p(k+1)＝i_d-p′(k+1)-K_d2

thus, the predicted current value at the time k +1 is optimized by using the error between the predicted current value and the actual current value at the time k, and the obtained optimized value can contain the information of the parameter disturbance, namely, the predicted current value at the time k +1 of the parameter disturbance is considered. The q-axis current can also be obtained based on the same principle.

Since the dead beat prediction control is a two beat prediction, i after optimization is used_d-pWhen calculating the reference voltage at the time k +1 from the reference currents at the times (k +1) and k +2, the parameter disturbance is also considered:

i_d-p′(k+2)＝(T_s/L_s′-K_d1)×u_d(k+1)+i_d-p(k+1)×(1-T_s×R_s/L_s)+i_q-p(k+1)×T_s×w_e

i_d-ref＝i_d-p(k+2)＝i_d-p′(k+2)-K_d2

＝(T_s/L_s′-K_d1)×u_d(k+1)+i_d-p(k+1)×(1-T_s×R_s/L_s)+i_q-p(k+1)×T_s×w_e-K_d2

solving the value of ud (k +1) according to the two formulas;

u_d(k+1)＝(i_d-ref+K_d2-i_d-p(k+1)×(1-T_s×R_s/L_s)-i_q-p(k+1)×T_s×w_e)/(T_s/L_s′-K_d1)

the q-axis voltage can also be obtained based on the same principle as described above.

The above derivation process is simplified, so that an improved deadbeat current prediction control model can be obtained:

i_d-error(k)＝i_d-p(k)-i_d(k)

i_q-e(k)＝i_q-p(k)-i_q(k)

K_d2＝(u_d(k-1)-u_d(k-2))

K_q2＝(u_q(k-1)-u_q(k-2))

X_d＝(i_d-error(k)-i_d-err(k-1))/K_d2/T_s

X_q＝(i_q-err(k)-i_q-err(k-1))/K_q2/T_s

K_d＝1/L_s-X_d

K_q＝1/L_s-X_q

i_d-p(k+1)

＝u_d(k)×K_d×T_s+i_d(k)×(1-T_s×R_s/L_s)+i_q(k)×T_s×w_e-(i_d-error(k)-X_d×T_s×u_d(k))

i_q-p(k+1)

＝u_q(k)×K_q×T_s+i_q(k)×(1-T_s×R_s/L_s)-i_d(k)×T_s×w_e-T_s×w_e×Ψ_f/L_s-(i_q-error(k)-X_q×T_s×u_q(k))

u_d-p(k+1)

＝(i_d-ref-i_d-p(k+1)×(1-T_s×R_s/L_s)-i_q-p(k+1)×T_s×w_e+i_d-error(k))/(K_d+X_d)/T_s

u_q-p(k+1)

＝(i_q-r-i_q-p(k+1)×(1-T_s×R_s/L_s)+i_d-p(k+1)×T_s×w_e+T_s×w_e×Ψ_f/L_s+i_q-err(k))/(K_q+X_q)/T_s

the improved dead-beat current prediction control model provided by the invention is applied to the bus current sensor, and the bus current sensor collects bus current twice at different voltage vector action moments in the first half period of a switching period according to the SVPWM seven-segment modulation mode. The specific acquisition method comprises the following steps: the current value at the moment when the switch state changes from 0 to 1 (i.e., the first two rising edges) is collected. Is provided with the firstA rising edge at time t₁The second rising edge at time t₂Since this time is not the initial time of each switching cycle, this means that only the current value reconstructed at the rising edge time of the previous cycle is available for the dead-beat prediction control of each cycle, and therefore, it is necessary to calculate the predicted value of the current actual value at the time k in place of the actual value of the current at the time k, based on the reconstructed current at the current update time of the previous switching cycle. And because the voltage vector in the dead-beat prediction model is an average value in a whole switching period, and when the predicted value of the current actual value at the k moment is calculated according to the reconstructed current at the current updating moment of the previous switching period, the time span is not the whole period, but T_s-t₁Therefore, the current updating time of k-1 period is needed to be calculated according to the volt-second balance principle_d(k-1) and u_q(k-1) compensate, i.e. multiply both of the equations by a factor: t is_s/(T_s-t₁). The method specifically comprises the following steps:

i_d′(k)＝i_d(k-1，t₁)×(1-(T_s-t₁)×R_s/L_s)+i_q(k-1，t₁)×(T_s-t₁)×w_e+(T_s-t₁)/L_s×u_d(k-1)×T_s/(T_s-t₁)

i_q′(k)＝i_q(k-1，t₁)×(1-(T_s-t₁)×R_s/L_s)-i_d(k-1，t₁)×(T_s-t₁)×w_e+(T_s-t₁)/L_s×u_q(k-1)×T_s/(T_s-t₁)-(T_s-t₁)×w_e×Ψ_f/L_s

fig. 2 and 3 show that in a preferred embodiment of the invention, the resistance mismatch (Rs ═ 10 Rs') and flux linkage mismatch (ψ) are present_f＝2ψ_f') in different cases, the d and q axis currents obtained on the basis of the invention are compared with a reference current.

It should be understood that, the sequence numbers of the steps in the embodiments of the present invention do not mean the execution sequence, and the execution sequence of each process should be determined by the function and the inherent logic of the process, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

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

Claims (3)

1. The improved dead-beat prediction control method of the permanent magnet synchronous motor single current sensor is characterized by comprising the following steps: the method specifically comprises the following steps:

step one, collecting three-phase current i of permanent magnet synchronous motor at current k moment in real time on line_a、i_b、i_cA rotor speed w and a rotor position angle theta;

establishing mathematical models of the permanent magnet synchronous motor in an alpha-beta coordinate system and a d-q coordinate system, and establishing a traditional dead-beat current prediction control model; improving the traditional deadbeat current prediction control model based on the difference between the current measured value and the predicted value at the current k moment caused by parameter disturbance to obtain an improved deadbeat current prediction control model corresponding to the k +1 moment; calculating reference voltage within the SVPWM output voltage range at the k +1 moment in real time by combining the data acquired in the step one;

thirdly, reconstructing a three-phase current i by utilizing the real-time switching state of the inverter and the acquired bus current i of the inverter_a、i_b、i_cReplacing the three-phase current i of the permanent magnet synchronous motor acquired in real time in the step one_a、i_b、i_c(ii) a And calculating actual predicted values of d-axis current and q-axis current at the k moment based on a volt-second balance principle by using the actual values of the d-axis current and the q-axis current at the first rising edge moment in the switching period at the k-1 moment, and replacing corresponding current values in the improved deadbeat current prediction control model.

2. The method of claim 1, wherein: the mathematical model of the permanent magnet synchronous motor in the second step under the alpha-beta coordinate system is as follows:

u_α＝R_si_α+L_spi_α-w_eψ_rsinθ

u_β＝R_si_β+L_spi_β+w_eψ_rcosθ

ψ_α＝L_si_α+ψ_rcosθ

ψ_β＝L_si_β+ψ_rsinθ

T_e＝1.5p_mψ_r(i_βcosθ-i_αsinθ)

in the formula u_α、u_βIs the stator voltage under an alpha-beta coordinate system; i.e. i_α、i_βIs stator current under an alpha-beta coordinate system; Ψ_rIs a rotor flux linkage; r_sIs a stator resistor; l is_sIs a stator inductance; w is a_e、w_mThe electrical angular velocity of the rotor and the mechanical angular velocity of the rotor, respectively; theta is a rotor position angle; p is a differential operator; t is_eIs an electromagnetic torque; t is_LIs the load torque; b is a viscosity coefficient; p is a radical of_mThe number of pole pairs of the motor is shown; Ψ_a、Ψ_βIs a stator flux linkage under an alpha-beta coordinate system; t is a time variable; j is load moment of inertia;

the mathematical model of the permanent magnet synchronous motor under the d-q coordinate system is as follows:

u_d＝R_si_d+pψ_d-w_eψ_q

u_q＝R_si_q+pψ_q+w_eψ_d

ψ_d＝L_di_d+ψ_r

ψ_q＝L_qi_q

T_e＝1.5p_m(ψ_ri_q+(L_d-L_q)i_di_q)

in the formula u_d、u_qIs the stator voltage under a d-q coordinate system; i.e. i_d、i_qIs the stator current under a d-q coordinate system; Ψ_d、Ψ_qIs a stator flux linkage under a d-q coordinate system; l is_d、L_qArmature inductances of d and q axes, respectively;

the traditional deadbeat current prediction control model is as follows:

i_d-p(k+1)＝i_d(k)×(1-T_s×R_s/L_s)+i_q(k)×T_s×w_e+T_s/L_s×u_d(k)

i_q-p(k+1)＝i_q(k)×(1-T_s×R_s/L_s)-i_d(k)×T_s×w_e+T_s/L_s×u_q(k)-T_s×w_e×ψ_r/L_s

u_d-p(k+1)＝L_s/T_s×(i_d-ref-(1-T_s×R_s/L_s)×i_d-p(k+1)-T_s×i_q-p(k+1)×w_e)

u_q-p(k+1)＝L_s/T_s×(i_q-ref-(1-T_s×R_s/L_s)×i_q-p(k+1)+T_s×i_d-p(k+1)×w_e+T_s×w_e/L_s×ψ_r)

in the formula i_d-p(k +1) d-axis predicted current, i, at time k +1_q-p(k +1) is the predicted current of the q-axis at the time k +1, i_d-refFor d-axis reference current, i_q-refFor q-axis reference current, u_d-p(k +1) d-axis predicted voltage at time k +1, u_q-p(k +1) is k +1 and time qAxial predicted voltage, T_sIs a switching cycle;

the improved deadbeat current prediction control model is as follows:

i_d-err(k)＝i_d-p(k)-i_d(k)

i_q-err(k)＝i_q-p(k)-i_q(k)

K_d2＝(u_d(k-1)-u_d(k-2))

K_q2＝(u_q(k-1)-u_q(k-2))

X_d＝(i_d-error(k)-i_d-error(k-1))/K_d2/T_s X_q＝(i_q-error(k)-i_q-error(k-1))/K_q2/T_s

K_d＝1/L_s-X_d

K_q＝1/L_s-X_q

i_d-p(k+1)

＝u_d(k)×K_d×T_s+i_d(k)×(1-T_s×R_s/L_s)+i_q(k)×T_s×w_e-(i_d-error(k)-X_d×T_s×u_d(k))

i_q-p(k+1)

＝u_q(k)×K_q×T_s+i_q(k)×(1-T_s×R_s/L_s)-i_d(k)×T_s×w_e-T_s×w_e×Ψ_f/L_s-(i_q-error(k)-X_q×T_s×u_q(k))

u_d-p(k+1)

＝(i_d-ref-i_d-p(k+1)×(1-T_s×R_s/L_s)-i_q-p(k+1)×T_s×w_e+i_d-error(k))/(K_d+X_d)/T_s

u_q-p(k+1)

＝(i_q-ref-i_q-p(k+1)×(1-T_s×R_s/L_s)+i_d-p(k+1)×T_s×w_e+T_s×w_e×Ψ_f/L_s+i_q-err(k))/(K_q+X_q)/T_s

in the formula i_d-error(k)、i_q-err(k) D and q axis current prediction errors i at time k_d-p(k)、i_q-p(k) D and q axis current predicted values i at the k time_d(k)、i_q(k) Actual measured values of d-and q-axis currents at time k, u_d(k-1) and u_d(k-2) d-axis voltages respectively acting at the k-1 moment and the k-2 moment, and u can be obtained by the same principle_q(k-1) and u_q(k-2)，i_d-p(k +1) is the improved d-axis predicted current at time k +1, i_q-pAnd (k +1) is the improved q-axis predicted current at the k +1 moment.

3. The method of claim 2, wherein: the third step is specifically as follows:

calculating a predicted value i of the current actual value at the k moment according to the reconstructed current at the k-1 switching period current updating moment_d′(k)、i_q' (k) replacing the actual value of the current at the time k, the current update time pair u of the k-1 cycle is calculated_d(k-1) and u_q(k-1) performing compensation;

i_d′(k)＝i_d(k-1，t₁)×(1-(T_s-t₁)×R_s/L_s)+i_q(k-1，t₁)×(T_s-t₁)×w_e+(T_s-t₁)/L_s×u_d(k-1)×T_s/(T_s-t₁)

i_q′(k)＝i_q(k-1，t₁)×(1-(T_s-t₁)×R_s/L_s)-i_d(k-1，t₁)×(T_s-t₁)×w_e+(T_s-t₁)/L_s×u_q(k-1)×T_s/(T_s-t₁)-(T_s-t₁)×w_e×Ψ_f/L_s

wherein i_d' (k) is a predicted value of the actual value of the d-axis current at time k, i_q' (k) is a predicted value of the actual value of the q-axis current at time k, t₁For the first rising edge instant i in a switching cycle_d(k-1，t₁) Is the d-axis current actual value, i, at the time of the k-1 period t1_q(k-1，t₁) Is k-1 period t₁An actual value of q-axis current at a time; then using i_d' (k) and i_q' (k) replacement of i in the deadbeat current prediction control model in step two_d(k) And i_q(k)。

Images