A zero-vector predictive control method for a four-leg inverter of a new energy vehicle

Abstract

The present invention relates to a zero-vector predictive control method for a four-leg inverter for a new energy vehicle, comprising the following steps: collecting three-phase current at time t; decoupling the ABC stationary coordinate system to obtain the dq0 axis current and dq0 axis voltage through park variation; inputting the initial sampling period, the phase voltage corresponding to the initial switching state, and the decoupled current into a current prediction model to obtain a predicted value of the current at time t1; optimizing vector selection to obtain predicted output current values ​​and sampling periods corresponding to each switching state at time t2; substituting the current predicted values ​​at time t2 under different switching states and a given reference value into a cost function for tracking the reference current, and selecting the switching state that minimizes the cost function as the optimal switching state and sampling period at time t1. The present invention simplifies the inverter control process, adds a cost function for tracking the reference current, and reduces the output current distortion rate from 3.9% and 2.54% to 1.43% while suppressing the common-mode voltage.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy vehicle motor control, and in particular to a zero-vector-free predictive control method for a four-bridge-arm inverter of a new energy vehicle. Background Art

[0002] The development of intelligent and electrified vehicles has led to the integration of more electronic devices and cables into vehicles, resulting in more stray distributed parameters. The development of power electronics has enabled efficient energy conversion and higher switching frequencies. The resulting high rates of change in voltage and current act on stray parameters, causing extremely strong electromagnetic interference. Furthermore, with the increasing number of electromagnetically sensitive devices in vehicles, electromagnetic interference, through conduction and radiation, affects various performance characteristics of new energy vehicles, such as power, communication quality, and safety, potentially causing serious safety accidents.

[0003] Electromagnetic interference (EMI) generated by power electronic equipment generally couples to various electromagnetically sensitive devices through conducted interference and radiated interference. Conducted interference is the primary source of radiated interference, making conducted interference suppression a major research area in electromagnetic compatibility. Conducted interference can be divided into common-mode interference and differential-mode interference based on its propagation path. Common-mode interference is generated by the rapidly changing voltages at the midpoints of the inverter's bridge arms and the load acting on parasitic capacitances, propagating through the live, neutral, and ground wires. Differential-mode interference originates from the currents generated by power switch switching and the oscillating currents in the parallel loop, propagating between the neutral and signal wires. Generally, EMI suppression considers three aspects: reducing the interference source, cutting off or weakening the interference within the propagation path, and improving the anti-interference capabilities of electronic equipment.

[0004] Currently, the largest source of electromagnetic interference in electric vehicles comes from the inverter in the electric drive system. In traditional three-phase, three-leg inverters, due to circuit topology limitations, the number of switches conducting in the upper and lower legs cannot be equal at any given time. Therefore, even adding an EMI filter to the circuit cannot eliminate the common-mode interference caused by circuit asymmetry at the source. Current research on suppressing common-mode interference focuses on circuit topology and inverter modulation methods. The four-leg topology proposed in 1999 is simple and can theoretically eliminate common-mode voltage, but it also brings complex modulation and degraded output current waveform quality, which urgently need to be addressed. Summary of the Invention

[0005] The object of the present invention is to provide a zero-vector predictive control method for a four-bridge-arm inverter of a new energy vehicle, which can improve the quality of the output current and reduce the distortion rate of the current while reducing the common-mode voltage at the load end.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a zero-vector-free predictive control method for a four-leg inverter of a new energy vehicle, the method comprising the following steps in sequence:

[0007] (1) Set time t as the initial time, apply the initial switch state and initial sampling period at time t, and collect the three-phase current i at time t a (t), i b (t), i c (t);

[0008] (2) The collected three-phase current i a (t), i b (t), i c (t) and the three-phase voltage u corresponding to the initial switching state a (t),u b (t),u c (t), decoupling from the abc stationary coordinate system by park change, and obtaining the dq0 axis current i d (t), i q (t), i0(t) and dq0 axis voltage u d (t),u q (t), u0(t);

[0009] (3) The initial sampling period T s (t), the decoupled dq0 axis voltage u corresponding to the initial switching state d (t),u q (t), u0(t), and the decoupled dq0 axis current i d (t), i q (t) and i0(t) are input into the current prediction model to obtain the predicted value of the current at time t+1

[0010] (4) The predicted value of the current at time t+1 The phase voltage u corresponding to the optimally selected switching state d (t+1),u q (t+1) and u0(t+1) are input into the optimization formula of the sampling period, and the optimal sampling period T corresponding to each switch state is obtained respectively. s (t+1), the optimal sampling period T corresponding to each switch state s (t+1), the phase voltage u corresponding to each switch state at time t+1 d (t+1),u q (t+1), u0(t+1), predicted value of current at time t+1 In the input current prediction model, the output current prediction value at time t+2 corresponding to each switch state is obtained

[0011] (5) The current prediction value at time t+2 under different switching states With the given reference value Substitute it into the cost function of tracking the current reference value and select the switching state that minimizes the cost function, that is, the optimal switching state at time t+1 and the corresponding optimal sampling period.

[0012] In step (2), the formula of the park transformation is as follows:

[0013]

[0014]

[0015] Among them, θ is the angle between the a-axis and the d-axis, i a 、i b 、i c They are three-phase current, u a 、u b 、u c are the three-phase voltages, i d 、i q , i0 is the dq0 axis current obtained by decoupling, u d 、u d , u0 is the dq0 axis voltage obtained by decoupling.

[0016] In step (3), the current prediction model is constructed based on the state equation of the four-leg inverter topology, and the state equation based on the four-leg inverter topology is:

[0017]

[0018] Among them, R is the load resistance, L is the filter inductor, i d 、i q , i0 is the dq0 axis current obtained by decoupling, u d 、u d , u0 is the dq0 axis voltage obtained by decoupling;

[0019] By solving the state equation to accurately discretize the output current, the current prediction model is obtained:

[0020]

[0021] Among them, T s (t) is the sampling period at time t, and i(t) is the load current at time t.

[0022] In step (4), the switching state after the optimization selection refers to: defining the upper bridge arm of each bridge arm of the four-bridge-arm inverter topology as 1, the lower bridge arm is turned on, and the output phase voltage is On the contrary, the upper bridge arm is disconnected and the lower bridge arm is turned on, which is 0. The output phase voltage is Therefore, there are 16 switching states in the four-leg inverter topology. The common-mode voltage at the load end is defined as the voltage difference between the midpoint of the motor's three-phase load and the ground:

[0023]

[0024] Among them U cm is the common mode voltage, U a , U b , U c , U f are the phase voltages of each bridge arm, U d is the DC source voltage;

[0025] When the upper bridge arm of the two bridge arms is always turned on and the lower bridge arm is turned off, the common-mode voltage at the load end is zero. Six vectors 0011, 0101, 0110, 1001, 1010, and 1100 are selected as a finite set of switching states to reduce the amount of calculation while suppressing the common-mode voltage to 0.

[0026] Optimize the sampling period through the optimization formula of the sampling period:

[0027]

[0028] Substituting the cost function and current prediction model into the above formula, we can get the optimal sampling period T s (t+1) value, and then T s Substitute (t+1) into the current prediction model to obtain the current prediction value at time t+2, that is:

[0029]

[0030] Among them, R is the load resistance and L is the filter inductor.

[0031] The cost function uses the least squares method to approximate the reference value, that is, for the current prediction value at time t+2, the cost function is:

[0032]

[0033] The current prediction value at time t+2 under different switching states obtained in step (4) With the given reference value Substituting into the above cost function, we can obtain g1(t+2), g2(t+2), g3(t+2), g4(t+2), g5(t+2), and g6(t+2) respectively. The switching state and sampling period corresponding to the smallest value are selected as the optimal choice and applied to time t+1.

[0034] It can be seen from the above technical solution that the beneficial effects of the present invention are: First, the inverter structure in the present invention adopts a three-phase four-bridge arm topology, optimizes the finite state set, removes the zero vector in the switching state, keeps the inverter always has two upper bridge arms turned on, and the other two lower bridge arms are disconnected, thereby suppressing the common-mode voltage at the load end to be basically zero; Second, the present invention uses zero-vector-free predictive control to replace the modulation process, which simplifies the control process of the inverter and makes the prediction model more accurate, increases the cost function of tracking the reference current, and reduces the distortion rate of the output current while suppressing the common-mode voltage. The distortion rate of the output current is reduced from 3.9% and 2.54% to 1.43%. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is the structural diagram of the four-leg inverter topology;

[0036] Figure 2 is a flow chart of the method of the present invention;

[0037] Figure 3 is the common-mode voltage waveform of the four-leg inverter using the present invention;

[0038] Figure 4 is the output current waveform of the three-phase four-bridge inverter using SVPWM modulation;

[0039] Figure 5 FFT analysis of the output current of a three-phase four-bridge inverter using SVPWM modulation;

[0040] Figure 6 Output current waveform and FFT analysis of current waveform of three-phase three-bridge inverter using SVPWM modulation method;

[0041] Figure 7 FFT analysis of the output current of a three-phase three-bridge inverter using the SVPWM modulation method;

[0042] Figure 8 The output current waveform and current waveform of the three-phase four-bridge inverter using the present invention;

[0043] Figure 9 FIG. 1 is an FFT analysis of the output current of the three-phase four-bridge inverter using the present invention. DETAILED DESCRIPTION

[0044] like Figure 2As shown, a zero-vector-free predictive control method for a four-leg inverter of a new energy vehicle comprises the following steps in sequence:

[0045] (1) Set time t as the initial time, apply the initial switch state and initial sampling period at time t, and collect the three-phase current i at time t a (t), i b (t), i c (t);

[0046] (2) The collected three-phase current i a (t), i b (t), i c (t) and the three-phase voltage u corresponding to the initial switching state a (t),u b (t),u c (t), decoupling from the abc stationary coordinate system by park change, and obtaining the dq0 axis current i d (t), i q (t), i0(t) and dq0 axis voltage u d (t),u q (t), u0(t);

[0047] (3) The initial sampling period T s (t), the decoupled dq0 axis voltage u corresponding to the initial switching state d (t),u q (t), u0 ( t), and the decoupled dq0 axis current i d (t), i q (t) and i0(t) are input into the current prediction model to obtain the predicted value of the current at time t+1

[0048] (4) The predicted value of the current at time t+1 The phase voltage u corresponding to the optimally selected switching state d (t+1),u q (t+1) and u0(t+1) are input into the optimization formula of the sampling period, and the optimal sampling period T corresponding to each switch state is obtained respectively. s (t+1), the optimal sampling period T corresponding to each switch state s (t+1), the phase voltage u corresponding to each switch state at time t+1 d (t+1),u q (t+1), u0(t+1), predicted value of current at time t+1 In the input current prediction model, the output current prediction value at time t+2 corresponding to each switch state is obtained

[0049] (5) The current prediction value at time t+2 under different switching states With the given reference value Substitute it into the cost function of tracking the current reference value and select the switching state that minimizes the cost function, that is, the optimal switching state at time t+1 and the corresponding optimal sampling period.

[0050] In step (2), the formula of the park transformation is as follows:

[0051]

[0052]

[0053] Among them, θ is the angle between the a-axis and the d-axis, i a 、i b 、i c They are three-phase current, u a 、u b 、u c are the three-phase voltages, i d 、i q , i0 is the dq0 axis current obtained by decoupling, u d 、u d , u0 is the dq0 axis voltage obtained by decoupling.

[0054] In step (3), the current prediction model is constructed based on the state equation of the four-leg inverter topology, and the state equation based on the four-leg inverter topology is:

[0055]

[0056] Among them, R is the load resistance, L is the filter inductor, i d 、i q , i0 is the dq0 axis current obtained by decoupling, u d 、u d , u0 is the dq0 axis voltage obtained by decoupling;

[0057] By solving the state equation to accurately discretize the output current, the current prediction model is obtained:

[0058]

[0059] Among them, T s (t) is the sampling period at time t, and i(t) is the load current at time t.

[0060] In step (4), the switching state after the optimization selection refers to: defining the upper bridge arm of each bridge arm of the four-bridge-arm inverter topology as 1, the lower bridge arm is turned on, and the output phase voltage is On the contrary, the upper bridge arm is disconnected and the lower bridge arm is turned on, which is 0. The output phase voltage is Therefore, there are 16 switching states in the four-leg inverter topology. The common-mode voltage at the load end is defined as the voltage difference between the midpoint of the motor's three-phase load and the ground:

[0061]

[0062] Among them U cm is the common mode voltage, U a , U b , U c , U f are the phase voltages of each bridge arm, U d is the DC source voltage;

[0063] When the upper bridge arm of the two bridge arms is always turned on and the lower bridge arm is turned off, the common-mode voltage at the load end is zero. Six vectors 0011, 0101, 0110, 1001, 1010, and 1100 are selected as a finite set of switching states to reduce the amount of calculation while suppressing the common-mode voltage to 0.

[0064] Since the selected switching states are reduced and there is no zero vector transition, the cost function deviates from the reference value in the same sampling period and cannot accurately track the current. Therefore, the sampling period is optimized by the optimization formula of the sampling period:

[0065]

[0066] Substituting the cost function and current prediction model into the above formula, we can get the optimal sampling period T s (t+1) value, and then T s Substitute (t+1) into the current prediction model to obtain the current prediction value at time t+2, that is:

[0067]

[0068] Among them, R is the load resistance and L is the filter inductor.

[0069] The cost function uses the least squares method to approximate the reference value, that is, for the current prediction value at time t+2, the cost function is:

[0070]

[0071] The current prediction value at time t+2 under different switching states obtained in step (4) With the given reference value Substituting into the above cost function, we can obtain g1(t+2), g2(t+2), g3(t+2), g4(t+2), g5(t+2), and g6(t+2) respectively. The switching state and sampling period corresponding to the smallest value are selected as the optimal choice and applied to time t+1.

[0072] Figure 1 It is a four-leg inverter topology, and the three-phase current i is output from the midpoint ABC of the first three legs. a ,i b ,i c , where N is the load neutral point, and the output midpoint of the fourth bridge arm is connected to the three-phase bridge arm through an LC filter.

[0073] When using the traditional SVPWM modulation method to control the three-phase four-bridge inverter, although it is possible to always keep the two upper bridge arms turned on and the two lower bridge arms turned off, so that the common mode voltage is basically zero, but because there is no zero vector transition, there are three or four switches switching at the same time, so the output current distortion rate increases, and the output effect is as follows: Figure 4 、 Figure 5 shown.

[0074] The traditional three-phase three-leg inverter also uses the SVPWM modulation method without zero vector. Because there are only three legs, the common mode voltage can only be reduced from 1 / 2Ud to 1 / 6Ud in theory. At the same time, because there is no zero vector transition, there are two or three switches switching at the same time, which increases the distortion rate of the output current. The output current effect is as follows: Figure 6 、 Figure 7 shown.

[0075] The present invention builds an accurate current prediction model, optimizes vector selection, and applies it to a three-phase four-bridge-arm inverter structure. While suppressing the common-mode voltage at the load end to be essentially zero, it improves the quality of the output current, that is, reduces the current distortion rate. The effect is as follows: Figure 3 、 Figure 8 and Figure 9 As shown in the figure, the common mode voltage is basically 0, and the current distortion rate is reduced from 2.54% to 1.43%.

[0076] In summary, the inverter structure in the present invention adopts a three-phase four-bridge arm topology, optimizes the finite state set, removes the zero vector in the switching state, keeps the inverter always with two upper bridge arms turned on, and the other two lower bridge arms disconnected, thereby suppressing the common-mode voltage at the load end to be basically zero; the present invention uses zero-vector-free predictive control to replace the modulation process, simplifies the control process of the inverter, and at the same time increases the cost function of tracking the reference current, thereby suppressing the common-mode voltage while reducing the distortion rate of the output current, and the distortion rate of the output current is reduced from 3.9% and 2.54% to 1.43%.

Claims

1. A zero-vector-free predictive control method for a four-leg inverter of a new energy vehicle, characterized by: The method comprises the following steps in sequence: (1) Set time t as the initial time, apply the initial switch state and initial sampling period at time t, and collect the three-phase current i at time t a (t), i b (t), i c (t); (2) The collected three-phase current i a (t), i b (t), i c (t) and the three-phase voltage u corresponding to the initial switching state a (t),u b (t),u c (t), decoupling from the abc stationary coordinate system by park transformation, the dq0 axis current i is obtained respectively d (t), i q (t), i0(t) and dq0 axis voltage u d (t),u q (t), u0(t); (3) The initial sampling period T s (t), the decoupled dq0 axis voltage u corresponding to the initial switching state d (t),u q (t), u0(t), and the decoupled dq0 axis current i d (t), i q (t) and i0(t) are input into the current prediction model to obtain the predicted value of the current at time t+1 (4) The predicted value of the current at time t+1 The phase voltage u corresponding to the optimally selected switching state d (t)+1),u q (t+1) and u0(t+1) are input into the optimization formula of the sampling period, and the optimal sampling period T corresponding to each switch state is obtained respectively. s (t+1), the optimal sampling period T corresponding to each switch state s (t+1), the phase voltage u corresponding to each switch state at time t+1 d (t+1),u q (t+1), u0(t+1), predicted value of current at time t+1 In the input current prediction model, the output current prediction value at time t+2 corresponding to each switch state is obtained The switching state after the optimization selection refers to: defining the upper bridge arm of each bridge arm of the four-bridge-arm inverter topology as 1, the lower bridge arm is turned on, and the output phase voltage is On the contrary, the upper bridge arm is disconnected and the lower bridge arm is turned on, which is 0. The output phase voltage is Therefore, there are 16 switching states in the four-leg inverter topology; Optimize the sampling period through the optimization formula of the sampling period: (5) The current prediction value at time t+2 under different switching states With the given reference value Substitute this into the cost function for tracking the current reference value and select the switch state that minimizes the cost function, which is the optimal switch state at time t+1 and the corresponding optimal sampling period. In step (3), the current prediction model is constructed based on the state equation of the four-leg inverter topology, and the state equation based on the four-leg inverter topology is: Among them, R is the load resistance, L is the filter inductor, i d 、i q , i0 is the dq0 axis current obtained by decoupling, u d 、u d , u0 is the dq0 axis voltage obtained by decoupling; By solving the state equation to accurately discretize the output current, the current prediction model is obtained: Among them, T s (t) is the sampling period at time t, and i(t) is the load current at time t.

2. The zero-vector-free predictive control method for a four-arm inverter of a new energy vehicle according to claim 1, characterized in that: In step (2), the formula of the park transformation is as follows: Among them, θ is the angle between the a-axis and the d-axis, i a 、i b 、i c They are three-phase current, u a 、u b 、u c are the three-phase voltages, i d 、i q , i0 is the dq0 axis current obtained by decoupling, u d 、u d , u0 is the dq0 axis voltage obtained by decoupling.

3. The zero-vector-free predictive control method for a four-arm inverter of a new energy vehicle according to claim 1, characterized in that: In step (4), the common-mode voltage at the load end is defined as the voltage difference between the midpoint of the motor's three-phase load and the ground: Among them U cm is the common mode voltage, U a , U b , U c , U f are the phase voltages of each bridge arm, U d is the DC source voltage; When the upper bridge arm of the two bridge arms is always turned on and the lower bridge arm is turned off, the common-mode voltage at the load end is zero. Six vectors 0011, 0101, 0110, 1001, 1010, and 1100 are selected as a finite set of switching states to reduce the amount of calculation while suppressing the common-mode voltage to 0. Substituting the cost function and current prediction model into the above formula, we can get the optimal sampling period T s (t+1) value, and then T s Substitute (t+1) into the current prediction model to obtain the current prediction value at time t+2, that is: Among them, R is the load resistance and L is the filter inductor.

4. The zero-vector-free predictive control method for a four-arm inverter of a new energy vehicle according to claim 1 or 3, characterized in that: The cost function uses the least squares method to approximate the reference value, that is, for the current prediction value at time t+2, the cost function is: The current prediction value at time t+2 under different switching states obtained in step (4) With the given reference value Substituting into the above cost function, we can obtain g1(t+2), g2(t+2), g3(t+2), g4(t+2), g5(t+2), and g6(t+2) respectively. The switching state and sampling period corresponding to the smallest value are selected as the optimal choice and applied to time t+1.

Images