Hybrid model predictive current control method for dual three-phase open winding permanent magnet synchronous generator

Abstract

The invention belongs to a motor control technology, and relates to a hybrid closed-loop model predictive current control method for a dual three-phase open-winding permanent magnet synchronous generator. Comprising the following steps: constructing an alternative voltage vector set and a value function under three subspaces; the voltage vector of the fundamental wave surface is composed of three alternative voltages, one alternative voltage is generated by a rectifier 1, and the other two alternative vectors are generated by a rectifier 2; selecting alternative vectors acting in the two harmonic subspaces and action time of the alternative vectors in combination with simplified optimization and classification compensation strategies; and according to the optimal alternative voltage of the three subspaces and the action time of the optimal alternative voltage, the control of the three subspaces is combined into a control period and acts on the rectifier. According to the method, the fundamental wave subspace and the two harmonic wave subspaces are controlled at the same time in one control period, harmonic current is effectively restrained, current pulsation is reduced, and the steady-state performance of the dual three-phase PMSG system is improved.

Description

Technical Field

[0001] The invention relates to a control technology of a dual three-phase open-winding permanent magnet synchronous generator, in particular to a hybrid model prediction current control method of a dual three-phase open-winding permanent magnet synchronous generator. Background Art

[0002] Due to the full electrification trend of defense-specific equipment such as military ships, electric aircraft and multi-purpose combat vehicles, the DC power supply system with generator as the core has broad application prospects. Compared with traditional three-phase motors, multi-phase motors have the advantages of low voltage, high power output and strong fault tolerance. Among them, the dual three-phase PMSG has attracted much attention due to its small current pulsation and torque. However, due to the increase in the number of motor phases, the corresponding harmonic subspace will also increase, which will generate larger harmonic currents, resulting in increased motor losses and poor steady-state performance.

[0003] The open winding structure with dual rectifiers provides more options for voltage vectors and more flexibility. The open winding structure does not change the structure of the motor and rectifier and has good adaptability. According to the power supply method at both ends of the stator winding, the open winding motor system is divided into three topological structures: single power supply DC bus, dual power supply isolated DC bus and hybrid power supply. Among them, the common DC bus structure only requires one DC power supply, which can reduce the system size and cost. However, under the common DC bus, the system has a zero-sequence path, which in turn generates a large zero-sequence current, reducing the efficiency and steady-state performance of the system. Therefore, in the open winding motor of the common DC bus dual rectifier system, measures must be taken to suppress the zero-sequence current.

[0004] In order to effectively suppress the zero-sequence current of the open-winding motor system with a common DC bus, scholars from various countries have proposed a series of control methods. Reference 1 (An Quntao, Sun Li, Sun Lizhi. Research on a new open-winding permanent magnet synchronous motor vector control system [J]. Proceedings of the CSEE, 2015, 35(22): 5891-5898.) proposed that the zero-sequence voltage is not generated by combining the voltage vectors, that is, the zero-sequence current is suppressed by applying the same zero-sequence voltage to the rectifiers on both sides. Reference 2 (C. Xue, W. Song, and X. Feng, "Finite control-set model predictive current control of five-phase permanent-magnet synchronous machine based on virtual voltage vectors," IETElectr. Power Appl., vol. 11, no. 5, pp. 836–846, May. 2017.) proposed that a single virtual voltage vector is synthesized to have a large voltage amplitude on the fundamental wave plane and a zero voltage amplitude on the harmonic wave plane, thereby suppressing the harmonic current. Reference 3 (C.Xue, W.Song, X.Wu, and X.Feng, “A constant switching frequency finite-control-set predictive current control scheme of a five-phase inverter with duty-ratio optimization,” IEEE Trans.Power Electron., vol.33, no.4, pp.3583–3594, Apr.2018.) proposes a duty cycle modulation strategy based on a single virtual voltage, and adds a zero vector by calculating the action time of the effective virtual voltage, thereby improving the current control accuracy. Although the above method effectively suppresses the harmonic current and improves the steady-state performance of the system, it does not apply a voltage vector separately to the harmonic subspace to further suppress the harmonic current generated by the nonlinear factors of dead zone, back electromotive force, and tube voltage drop. The harmonic subspace is essentially an open-loop control. In actual systems, the nonlinear factors of dead zone, back electromotive force, and tube voltage drop will cause the amplitude of the virtual voltage in each harmonic subspace to be non-zero, which will lead to larger harmonic currents, limit the efficiency of the system, and affect the stability of the system. Summary of the invention

[0005] In view of the shortcomings of the existing dual three-phase open-winding PMSG system, the present invention proposes a hybrid model predictive current control method for a dual three-phase open-winding permanent magnet synchronous generator, which simultaneously considers the fundamental subspace and two harmonic subspaces, suppresses harmonics through a virtual voltage vector of the fundamental subspace, and classifies the virtual voltage vector according to the common mode voltage. At the same time, a closed loop of a dual subplane and a closed loop of classified compensation can be added to the harmonic surface respectively, thereby effectively suppressing harmonic currents and improving the steady-state performance of the system.

[0006] To achieve the purpose of the invention, the hybrid closed-loop model predictive current control method of the dual three-phase open-winding permanent magnet synchronous generator of the present invention comprises the following steps:

[0007] Step 1: Derive the discretized prediction current model of the dual three-phase open-winding PMSG in the rotating coordinate system;

[0008] Step 2: construct the candidate voltage vector set and cost function of the dq fundamental subspace, the candidate voltage vector set and cost function of the xy harmonic subspace, and the candidate voltage vector set of the o1-o2 harmonic subspace;

[0009] Step 3: By simplifying the optimization, the optimal voltage vector and its action time of the xy harmonic subspace are selected based on the principle of minimum value function, and the dual-subplane closed-loop control of the xy harmonic subspace is constructed;

[0010] Step 4: Calculate the reference voltage of the o1-o2 harmonic subspace according to the current deadbeat, compensate the actual voltage value to the corresponding reference value, calculate the required compensation voltage vector and its action time, and realize the classified compensation closed-loop control of the o1-o2 harmonic subspace;

[0011] Step 5: According to the zero-beat current prediction and sector judgment method, the voltage vector of the dq fundamental plane is composed of three alternative voltages, rectifier 1 generates one alternative voltage vector, and the other two alternative vectors are generated by rectifier 2; according to the voltage vectors of the dq fundamental subspace, xy harmonic subspace and o1-o2 harmonic subspace and their action time, the control of the dq fundamental subspace, xy harmonic subspace and o1-o2 harmonic subspace is merged into one control cycle and acts on the two rectifiers.

[0012] Furthermore, in step 1, the derivation process of the discrete prediction model of the dual three-phase open-winding PMSG in the dq, xy, and o1-o2 coordinate systems is as follows:

[0013] The voltage equation of the dual three-phase open-winding PMSG in the dq, xy, o1-o2 coordinate system is:

[0014]

[0015] Where: ud 、u q 、u x 、u y 、u o1 、u o2 are d, q, x, y, o1, o2 axis voltages respectively; i d 、i q 、i x 、i y 、i o1 、i o2 are d, q, x, y, o1, o2 axis currents respectively; L d , L q are the d-axis and q-axis inductances respectively, and L d =L q ; L0 is the stator self-leakage inductance; ω e is the electrical angular velocity; f is the fundamental flux amplitude of the permanent magnet; R s is the stator resistance; f3 is the third harmonic flux amplitude of the permanent magnet;

[0016] According to the forward Euler approximation method, the discrete state equation of the motor can be obtained, and the current prediction value at the next sampling moment can be obtained by calculation:

[0017]

[0018] Where: i d (k+1), i q (k+1), i x (k+1), i y (k+1), i o1 (k+1), i o2 (k+1) are the predicted current values ​​of the axes d, q, x, y, o1, and o2 at the next sampling moment respectively; i d (k), i q (k), i x (k), i y (k), i o1 (k), i o2 (k) are the current sampling values ​​of the axes d, q, x, y, o1, and o2 at the current moment; T s is the sampling period; u d (k) and u q (k) and u x (k) and u y (k) and u o1 (k) and u o2 (k) is the current d, q, x, y, o1, and o2 axis voltages.

[0019] Furthermore, step 2 includes:

[0020] ① Voltage vector construction and value function design of fundamental wave plane:

[0021] The voltage vectors of the rectifiers on both sides are selected respectively, and a single rectifier is selected to generate the outermost and second outermost voltage vectors in the dq fundamental subspace, and 12 xy virtual voltage vectors are synthesized through the principle of amplitude-second balance. When the ratio of the outermost voltage vector to the second outermost voltage vector satisfies the action time of 0.732:0.268, the voltage amplitude of the single rectifier corresponding to the xy harmonic subspace is zero;

[0022] The common mode voltages are grouped as follows:

[0023] According to the vector space decoupling coordinate transformation, the voltage generated by a single rectifier is mapped to the mutually orthogonal α-β fundamental subspace, xy harmonic subspace and o1-o2 harmonic subspace, and the mapping relationship is as follows:

[0024]

[0025] According to the above formula, the voltage vector distribution diagram of a single rectifier in the o1-o2 harmonic subspace is derived to obtain the corresponding common-mode voltage. Through virtual synthesis, the 12 xy virtual voltage vectors are classified according to the size of the generated common-mode voltage.

[0026] d.q-axis value function g dq design:

[0027] After classification according to the virtual voltage vector, the voltage amplitude mapped in the two harmonic subspaces is zero, and only the tracking of the d and q axis currents is considered in the cost function:

[0028] g dq =|i d * -i d (k+1)| 2 +|i q * -i q (k+1)| 2

[0029] Where: i d * 、i q * are the reference values ​​of d-axis and q-axis currents respectively;

[0030] ② Voltage vector construction and value function design of x-y harmonic surface:

[0031] According to the outermost and second outermost voltage vectors generated by a single rectifier in the xy harmonic subspace, 12 virtual voltage vectors are synthesized through the principle of amplitude-second balance; when the ratio of the outermost voltage vector to the second outermost voltage vector satisfies the action time of 0.732:0.268, the voltage amplitude of the voltage vector generated by a single rectifier corresponding to the dq fundamental subspace is zero;

[0032] According to the voltage vector distribution diagram of a single rectifier, among the above 12 virtual voltage vectors, three virtual voltages every 120 degrees have the same common mode voltage. 12 Find 12 virtual voltage vectors among the voltage vectors so that they have amplitude in the xy harmonic subspace, while the corresponding amplitudes in the dq fundamental subspace and the o1-o2 harmonic subspace are zero;

[0033] The value function g of the xy harmonic surface xy design:

[0034] g xy =|i x * -i x (k+1)| 2 +|i y * -i y (k+1)| 2

[0035] Where: i x * 、i y * are the reference values ​​of x-axis and y-axis currents respectively;

[0036] ③Voltage selection of o1-o2 harmonic subspace

[0037] According to the voltage vector distribution diagram of a single rectifier in the o1-o2 harmonic subspace, the basic voltage vector u 07 and u 70 As an alternative voltage, o1-o2 harmonic subspace adopts closed-loop control of classification compensation, without enumeration and optimization process, and does not need to design value function.

[0038] Furthermore, the process of step 3 is:

[0039] The alternative voltage vector set of the xy harmonic subspace is brought into the current prediction equation, and the optimal alternative voltage vector in the xy harmonic subspace is selected through the value function. By optimizing the duty cycle, the action time of the optimal alternative voltage vector and the zero vector is allocated so that the current at time k+1 reaches a given value, that is,

[0040]

[0041] Where: t opt1 is the action time of the optimal candidate voltage vector corresponding to the xy harmonic subspace; s x 、s y They are respectively the optimal alternative voltage vector i x 、i y The slope of x0 、s y0 i is zero vector x 、i y The slope of

[0042] Substitute the above formula into the value function of the xy harmonic subspace design, and let the value function act on the optimal candidate vector for time t opt1 The partial derivative is 0, and the optimal action time can be obtained:

[0043]

[0044] Furthermore, the specific process of step 4 is:

[0045] Based on the deadbeat method, the o1 axis voltage u corresponding to the o1-o2 harmonic subspace can be obtained: o1 and o2 axis voltage u o2 Reference value of

[0046]

[0047] Compensation is categorized as follows:

[0048] ①When u o1ref >0 and u o2ref >0:

[0049] Rectifier 1 select u 70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 1 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 ;

[0050] According to the compensation time t1, t2, set t3 = (t1 + t2); during the action time t3, the rectifier 2 adds a zero vector u 00 , zero vector u 00The action time of is t4=t3. This makes the action time of the classification compensation of the two groups of rectifiers the same in the o1-o2 subspace;

[0051] ②When u o1ref <0andu o2ref <0 hours:

[0052] Rectifier 2 Select u 70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 2 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s ; That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 ;

[0053] According to the compensation time t1, t2, set t3 = (t1 + t2); during the action time t3, the rectifier 1 adds a zero vector u 00 , zero vector u 00 The action time of is t4=t3. This makes the action time of the classification compensation of the two groups of rectifiers the same in the o1-o2 subspace;

[0054] ③When u o1ref >0 and u o2ref <0 hours:

[0055] Rectifier 1 select u 70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 2 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 ;

[0056] Secondly, according to the compensation time t1, t2, set t3 = max(t1, t2); within the action time t3, the rectifier with less compensation time adds a zero vector u on the basis of the original compensation. 00 , zero vector u 00The action time is t4 = t 3— min(t1, t2); so that the classification compensation time of the two groups of rectifiers is the same in the o1-o2 subspace;

[0057] ④When u o1ref <0andu o2ref >0:

[0058] Rectifier 1 select u 07 To compensate u o2ref , compensation time t1=|u o2ref / U dc |*T s , rectifier 2 select u 70 To compensate u o1ref , compensation time t2=|u o1ref / U dc |*T s ; That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 ;

[0059] According to the compensation time t1, t2, set t3 = max(t1, t2); within the action time t3, the rectifier with less compensation time adds a zero vector u on the basis of the original compensation. 00 , zero vector u 00 The action time is t4=t3-min(t1, t2); so that the action time of the classification compensation of the two groups of rectifiers in the o1-o2 subspace is the same.

[0060] Furthermore, the specific process of step 5 is as follows:

[0061] Rectifier 1 is in the remaining time T s -t opt1 The optimal voltage vector acting within -t3 is obtained by the d-axis and q-axis current deadbeat formula:

[0062]

[0063] Where u dref 、u qref They are the d-axis and q-axis reference voltage components in the two-phase rotating coordinate system, and the Clarke transformation is performed on them:

[0064]

[0065] u αref 、u βref The position angle θ in the α-β plane is:

[0066]

[0067] According to the position of the reference voltage in the sector, the rectifier 1 is selected for the remaining time T s -t opt1 -t3; secondly, according to the rectifier 1 in the remaining time T s -t opt1 -t3, rectifier 2 selects the optimal voltage vector within the same group of two xy virtual voltage vectors, enumerates three times, and obtains rectifier 2 in the remaining time T s -t opt1 - Two virtual voltage vectors acting within t3 and their action time; the action time of the two virtual voltage vectors is calculated by the q-axis current deadbeat;

[0068] The calculation method is as follows:

[0069]

[0070] Where: s0, s i 、s j The changes in the slope of the q-axis current caused by the 0 voltage vector, the rectifier 2 selects the i voltage vector, and the rectifier 2 selects the j voltage vector; d opt is the duty cycle of the i voltage vector selected by rectifier 2; then the j voltage vector time selected by rectifier 2 is (T s -t opt1 -t3)*(1-d opt );

[0071] In a control cycle T s Within, there is an effective voltage vector action time greater than T s In order to ensure the control of the active component of the system, the compensation time t of the two harmonic subspace closed loops is opt1 , t3 imposes restrictions:

[0072]

[0073] That is, when t opt1 +t 3≥ T s When t , the harmonics are suppressed by using the virtual voltage vector of the fundamental subspace and classifying the virtual voltage vector according to the common mode voltage. opt1 +t 3< T s When the closed-loop control of two harmonic subspaces is added, the influence of nonlinear factors is reduced and the harmonic current is suppressed;

[0074] Finally, the voltage vectors acting on the three subspaces and their action time are determined, and the switch state and the corresponding opening time corresponding to the voltage vector are applied to the rectifier to complete the output of the switching sequence.

[0075] The beneficial effects of the present invention are as follows:

[0076] 1. The present invention constructs alternative voltage vector sets for the fundamental subspace and two harmonic subspaces respectively, and the three subspaces are independently controlled. The dual-subplane closed-loop control of the xy harmonic surface and the classified closed-loop control of the o1-o2 harmonic surface are added to solve the problem of excessive harmonic current caused by nonlinear factors such as dead zone, back electromotive force, and tube voltage drop, thereby improving the steady-state performance of the system.

[0077] 2. The fundamental wave plane of the present invention is classified by virtual voltage, and the required virtual voltage and its action time can be selected by enumerating three times. The xy harmonic plane only needs to be enumerated 12 times, and the o1-o2 harmonic plane classification compensation closed loop is obtained by zero-beat calculation, which reduces the amount of calculation caused by expanding the harmonic subspace vector set and alleviates the calculation burden of the controller. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 Predictive current control block diagram for the hybrid closed-loop model of a dual three-phase open-winding PMSG.

[0079] Figure 2 This is the topological structure diagram of the dual three-phase open-winding PMSG system.

[0080] Figure 3 This is the spatial voltage vector distribution diagram of a single rectifier in an open-winding dual three-phase PMSG system.

[0081] Figure 4 The vector distribution of alternative voltages for a single rectifier in the fundamental subspace of an open-winding dual three-phase converter.

[0082] Figure 5 It is the alternative voltage vector distribution diagram in the xy harmonic subspace.

[0083] Figure 6 It is the alternative voltage vector distribution diagram in the o1-o2 harmonic subspace.

[0084] Figure 7 This is the bus voltage waveform under the open-winding dual three-phase PMSG hybrid MPCC.

[0085] Figure 8 Shown are the d and q axis current waveforms of the open-winding dual three-phase PMSG hybrid MPCC.

[0086] Fig. 9 This is the a-phase current waveform under the open-winding dual three-phase PMSG hybrid MPCC.

[0087] Fig.10 This is the harmonic content diagram of phase a current under open-winding dual three-phase PMSG hybrid MPCC. DETAILED DESCRIPTION

[0088] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0089] like Figure 1 As shown, the present invention is a hybrid closed-loop MPCC control strategy for an open-winding dual three-phase PMSG, including the selection of alternative voltage vectors in the fundamental subspace and two harmonic subspaces, and the process of adding two harmonic surfaces to the closed loop respectively. The specific measures are as follows:

[0090] First, the discrete prediction model of the dual three-phase open-winding PMSG in the dq, xy, and o1-o2 coordinate systems is derived:

[0091]

[0092] Where: u d 、u q 、u x 、u y 、u o1 、u o2 are d, q, x, y, o1, o2 axis voltages respectively; i d 、i q 、i x 、i y 、i o1 、i o2 are d, q, x, y, o1, o2 axis currents respectively; L d , L d are the d-axis and q-axis inductances respectively, and L d =L d ; L0 is the stator self-leakage inductance; ω e is the electrical angular velocity; f is the fundamental flux amplitude of the permanent magnet; R s is the stator resistance; f3 is the third harmonic flux amplitude of the permanent magnet;

[0093] According to the forward Euler approximation method, the discrete state equation of the motor can be obtained, and the current prediction value at the next sampling moment can be obtained by calculation:

[0094]

[0095] Where: i d (k+1), i q (k+1), i x (k+1), i y (k+1), i o1 (k+1), i o2 (k+1) are the predicted current values ​​of the axes d, q, x, y, o1, and o2 at the next sampling moment respectively; i d(k), i q (k), i x (k), i y (k), i o1 (k), i o2 (k) are the current sampling values ​​of the axes d, q, x, y, o1, and o2 at the current moment; T s is the sampling period; u d (k) and u q (k) and u x (k) and u y (k) and u o1 (k) and u o2 (k) is the current d, q, x, y, o1, and o2 axis voltages.

[0096] Secondly, the candidate voltage vector set and value function of the fundamental subspace, xy harmonic subspace and the candidate voltage vector set of the o1-o2 harmonic subspace are constructed:

[0097] ① Voltage vector construction and value function design of fundamental wave plane:

[0098] The voltage vectors of the rectifiers on both sides are selected respectively, and a single rectifier is selected to generate the outermost and second outermost voltage vectors in the dq fundamental subspace, and 12 xy virtual voltage vectors are synthesized through the principle of amplitude-second balance. When the ratio of the outermost voltage vector to the second outermost voltage vector satisfies the action time of 0.732:0.268, the voltage amplitude of the single rectifier corresponding to the xy harmonic subspace is zero. Figure 3 and Figure 4 The spatial voltage vector distribution diagram of a single rectifier in an open-winding dual three-phase PMSG system and the vector distribution of the alternative voltage of a single rectifier in the fundamental subspace are given respectively.

[0099] The common mode voltages are grouped as follows:

[0100] According to the vector space decoupling coordinate transformation, the voltage generated by a single rectifier is mapped to the mutually orthogonal α-β fundamental subspace, xy harmonic subspace and o1-o2 harmonic subspace, and the mapping relationship is as follows:

[0101]

[0102]

[0103] According to the above formula, the voltage vector distribution diagram of a single rectifier in the o1-o2 harmonic subspace is derived, and the corresponding common mode voltage is obtained. Through virtual synthesis, the 12 xy virtual voltage vectors are classified according to the size of the common mode voltage generated. The classification is as follows:

[0104] Group ① <![CDATA[vv2(u 64 ,in 46 )]]> <![CDATA[vv6(u 32 ,in 23 )]]> <![CDATA[vv 10 (in 51 ,in 15 )<!-- 8 --> ]]> Group ② <![CDATA[vv3(u 66 ,in 24 )]]> <![CDATA[vv7(u 33 ,in 12 )]]> <![CDATA[vv 11 (in 55 ,in 41 )]]> Group ③ <![CDATA[vv4(u 26 ,in 62 )]]> <![CDATA[vv8(u 13 ,in 31 )]]> <![CDATA[vv 12 (in 45 ,in 54 )]]> Group ④ <![CDATA[vv5(u 22 ,in 36 )]]> <![CDATA[vv9(u 11 ,in 53 )]]> <![CDATA[vv1(u 44 ,in 65 )]]>

[0105] Value function design:

[0106] After classification according to the virtual voltage vector, the voltage amplitude mapped in the two harmonic subspaces is zero, and only the tracking of the d and q axis currents is considered in the cost function:

[0107] g dq =|i d * -i d (k+1)| 2 +|i q * -i q (k+1)| 2

[0108] Where: i d * 、i q * They are the reference values ​​of d-axis and q-axis currents respectively.

[0109] ② Voltage vector construction and value function design of x-y harmonic surface:

[0110] According to the outermost and second outermost voltage vectors generated by a single rectifier in the xy harmonic subspace, 12 virtual voltage vectors are synthesized through the principle of amplitude-second balance. When the ratio of the outermost voltage vector to the second outermost voltage vector satisfies the action time of 0.732:0.268, the voltage vector generated by a single rectifier corresponds to a voltage amplitude of zero in the dq fundamental subspace.

[0111] According to the voltage vector distribution diagram of a single rectifier, among the above 12 virtual voltage vectors, three virtual voltages every 120 degrees have the same common mode voltage. 12 Find 12 virtual voltage vectors among the voltage vectors so that they have amplitudes in the xy harmonic subspace, while the corresponding amplitudes in the dq fundamental subspace and the o1-o2 harmonic subspace are zero.

[0112] Figure 5 The alternative voltage vector distribution diagram in the xy harmonic subspace is given (where vv3'(u 52.61 ,u 46.23 ) means that rectifier 1 selects u 52 , rectifier 2 select u 61 Construct a complete voltage vector, rectifier 1 selects u 46 , rectifier 2 select u 23 Another complete set of voltage vectors is formed. When the two sets of voltage vectors satisfy the action time ratio of 0.732:0.268, a set of alternative voltage vectors of the xy harmonic subspace is formed).

[0113] Value function design:

[0114] g xy =|i x * -i x (k+1)| 2 +|i y * -i y (k+1)| 2

[0115] Where: i x * 、i y * They are the reference values ​​of x-axis and y-axis currents respectively.

[0116] ③Voltage selection of o1-o2 harmonic subspace

[0117] According to the voltage vector distribution diagram of a single rectifier in the o1-o2 harmonic subspace, the basic voltage vector u 07 and u 70 As an alternative voltage, o1-o2 harmonic subspace adopts closed-loop control of classification compensation, without enumeration and optimization process, and does not need to design value function. Figure 6 The alternative voltage vector distribution diagram in the o1-o2 harmonic subspace is given.

[0118] The purpose of constructing the voltage vectors of the three subspaces is to make the voltage vectors selected by the three subspaces independent of each other, that is, the selected voltage has an amplitude in its own space, and the voltage amplitude mapped to each other in the subspace is zero. Then, on the basis of independent control, the closed-loop control of the two harmonic surfaces is introduced.

[0119] According to the candidate voltage vector sets of the three subspaces, a dual-subplane closed-loop control is first applied to the xy subspace:

[0120] The set of candidate voltage vectors in the xy harmonic subspace is brought into the current prediction equation, and the optimal candidate voltage vector in the xy harmonic subspace is selected through the value function. By optimizing the duty cycle, the action time of the optimal candidate voltage vector and the zero vector is allocated so that the current at time k+1 reaches a given value.

[0121]

[0122] Where: t opt1 is the action time of the optimal candidate voltage vector corresponding to the xy harmonic subspace; s x 、s y They are respectively the optimal alternative voltage vector i x 、i yThe slope of x0 、s y0 i is zero vector x 、i y The slope of

[0123] Substitute the above formula into the value function of the xy harmonic subspace design, and let the value function act on the optimal candidate vector for time t opt1 The partial derivative is 0, and the optimal action time can be obtained:

[0124]

[0125] Secondly, a closed-loop control of classification compensation is applied to the o1-o2 subspace:

[0126] Through the idea of ​​no-beat, we can get the corresponding u o1 The reference value and u o2 Reference value:

[0127]

[0128] Compensation is categorized as follows:

[0129] ①When u o1ref >0 and u o2ref >0:

[0130] Rectifier 1 select u 70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 1 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 .

[0131] Secondly, according to the compensation time t1, t2, set t3 = (t1 + t2). During the action time t3, the rectifier 2 adds a zero vector u 00 , zero vector u 00 The action time of is t4=t3, so that the action time of the classification compensation of the two groups of rectifiers in the o1-o2 subspace is the same.

[0132] ②When u o1ref <0andu o2ref <0 hours:

[0133] Rectifier 2 Select u70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 2 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 .

[0134] Secondly, according to the compensation time t1, t2, set t3 = (t1 + t2). During the action time t3, the rectifier 1 adds a zero vector u 00 , zero vector u 00 The action time of is t4=t3, so that the action time of the classification compensation of the two groups of rectifiers in the o1-o2 subspace is the same.

[0135] ③When u o1ref >0 and u o2ref <0 hours:

[0136] Rectifier 1 select u 70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 2 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 .

[0137] Secondly, according to the compensation time t1, t2, set t3 = max(t1, t2). Within the action time t3, the rectifier with less compensation time adds a zero vector u on the basis of the original compensation. 00 , zero vector u 00 The action time of is t4=t3-min(t1, t2). This makes the action time of the classification compensation of the two groups of rectifiers the same in the o1-o2 subspace.

[0138] ④When u o1ref <0andu o2ref >0:

[0139] Rectifier 1 select u 07 To compensate u o2ref , compensation time t1=|u o2ref / U dc |*T s , rectifier 2 select u 70 To compensate u o1ref , compensation time t2=|u o1ref / U dc |*T s That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 .

[0140] Secondly, according to the compensation time t1, t2, set t3 = max(t1, t2). Within the action time t3, the rectifier with less compensation time adds a zero vector u on the basis of the original compensation. 00 , zero vector u 00 The action time of is t4=t3-min(t1, t2), so that the action time of the classification compensation of the two groups of rectifiers in the o1-o2 subspace is the same.

[0141] Secondly, after adding the closed loop of the two harmonic planes, the d and q axis currents are tracked on the fundamental wave plane:

[0142] Rectifier 1 is in the remaining time T s -t opt1 The optimal voltage vector acting within -t3 is obtained by the d-axis and q-axis current deadbeat formula:

[0143]

[0144] Where u dref 、u qref They are the d-axis and q-axis reference voltage components in the two-phase rotating coordinate system, and the Clarke transformation is performed on them:

[0145]

[0146] u αref 、u βref The position angle θ in the α-β plane is:

[0147]

[0148] According to the position of the reference voltage in the sector, the rectifier 1 is selected for the remaining time T s -t opt1 -t3. Secondly, according to the rectifier 1 in the remaining time T s -topt1 -t3, rectifier 2 selects the optimal voltage vector within the same group of two xy virtual voltage vectors, enumerates three times, and obtains rectifier 2 in the remaining time T s -t opt1 - Two virtual voltage vectors acting within t3 and their action time. The action time of the two virtual voltage vectors is calculated by the q-axis current deadbeat.

[0149] The calculation method is as follows:

[0150]

[0151] Where: s0, s i 、s j The changes in the slope of the q-axis current caused by the 0 voltage vector, the rectifier 2 selects the i voltage vector, and the rectifier 2 selects the j voltage vector; d opt is the duty cycle of the i voltage vector selected by rectifier 2. Then the j voltage vector time selected by rectifier 2 is (T s -t opt1 -t3)*(1-d opt );

[0152] In a control cycle T s Within, there is an effective voltage vector action time greater than T s In order to ensure the control of the active component of the system, the compensation time t of the two harmonic subspace closed loops is opt1 , t3 imposes restrictions:

[0153]

[0154] That is, when t opt1 +t 3≥ T s When t is used, the harmonics can be suppressed by using the virtual voltage vector of the fundamental subspace and classifying the virtual voltage vector according to the common mode voltage. opt1 +t 3< T s When the closed-loop control of two harmonic subspaces is added, the influence of nonlinear factors is reduced and the harmonic current is suppressed.

[0155] Finally, the voltage vectors acting on the three subspaces and their action time are determined, and the switch state and the corresponding opening time corresponding to the voltage vector are applied to the rectifier to complete the output of the switching sequence.

[0156] In order to illustrate the hybrid closed-loop model predictive current control method of a dual three-phase open-winding permanent magnet synchronous generator of the present invention, a simulation test is carried out on the implementation method, and the parameters of the permanent magnet synchronous generator used are as follows: stator resistance R s =0.67Ω; d, q inductance: Ld =2.46mL, L d =2.46mL; stator self-leakage inductance L0 = 0.492mL; permanent magnet fundamental wave flux amplitude Ψ f =0.0885Wb; Permanent magnet third harmonic flux amplitude Ψ f3 =0.004425Wb; Motor pole pair number p = 5; Rated speed n = 750r / min; Bus capacitance C = 2200μF; Bus control voltage U dc =80V;

[0157] Figure 7 The bus voltage waveform under the open-winding dual three-phase PMSG hybrid MPCC. It can be seen that the DC bus voltage can stably track the given value. Figure 8 The d and q axis current waveforms under the open-winding dual three-phase PMSG hybrid MPCC. It can be seen that the d and q axis currents can reach the reference values. Fig. 9 Fig.10 It can be concluded that the proposed method significantly improves the sinusoidality of phase current and significantly suppresses harmonics.

[0158] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A hybrid closed-loop model predictive current control method for a dual three-phase open-winding permanent magnet synchronous generator, comprising the following steps: Step 1: Derive the discretized prediction current model of the dual three-phase open-winding PMSG in the rotating coordinate system; Step 2: construct the candidate voltage vector set and cost function of the dq fundamental subspace, the candidate voltage vector set and cost function of the xy harmonic subspace, and the candidate voltage vector set of the o1-o2 harmonic subspace; Step 3: By simplifying the optimization, the optimal voltage vector and its action time of the xy harmonic subspace are selected based on the principle of minimum value function, and the dual-subplane closed-loop control of the xy harmonic subspace is constructed; Step 4: Calculate the reference voltage of the o1-o2 harmonic subspace according to the current deadbeat, compensate the actual voltage value to the corresponding reference value, calculate the required compensation voltage vector and its action time, and realize the classified compensation closed-loop control of the o1-o2 harmonic subspace; Step 5: According to the zero-beat current prediction and sector judgment method, the voltage vector of the dq fundamental plane is composed of three alternative voltages, rectifier 1 generates one alternative voltage vector, and the other two alternative vectors are generated by rectifier 2; according to the voltage vectors of the dq fundamental subspace, xy harmonic subspace and o1-o2 harmonic subspace and their action time, the control of the dq fundamental subspace, xy harmonic subspace and o1-o2 harmonic subspace is merged into one control cycle and acts on the two rectifiers.

2. A dual three-phase open-winding PMSG hybrid model predictive current control method according to claim 1, characterized in that: The derivation process of the mathematical model and discrete predictive control model of the dual three-phase open-winding PMSG in the rotating coordinate system is as follows: The voltage equation of the dual three-phase open-winding PMSG in the dq, xy, o1-o2 coordinate system is: Where: u d 、u q 、u x 、u y 、u o1 、u o2 are d, q, x, y, o1, o2 axis voltages respectively; i d 、i q 、i x 、i y 、i o1 、i o2 are d, q, x, y, o1, o2 axis currents respectively; L d , L q are the d-axis and q-axis inductances respectively, and L d =L q ; L0 is the stator self-leakage inductance; ω e is the electrical angular velocity; f is the fundamental flux amplitude of the permanent magnet; R s is the stator resistance; f3 is the third harmonic flux amplitude of the permanent magnet; According to the forward Euler approximation method, the discrete state equation of the motor can be obtained, and the current prediction value at the next sampling moment can be obtained by calculation: Where: i d (k+1), i q (k+1), i x (k+1), i y (k+1), i o1 (k+1), i o2 (k+1) are the predicted current values ​​of the axes d, q, x, y, o1, and o2 at the next sampling moment respectively; i d (k), i q (k), i x (k), i y (k), i o1 (k), i o2 (k) are the current sampling values ​​of the axes d, q, x, y, o1, and o2 at the current moment; T s is the sampling period; u d (k),u q (k),u x (k),u y (k),u o1 (k),u o2 (k) is the current d, q, x, y, o1, and o2 axis voltages.

3. A dual three-phase open-winding PMSG hybrid model predictive current control method according to claim 1, characterized in that: The specific process of step 2 is: ① Voltage vector construction and value function design of fundamental wave plane: The voltage vectors of the rectifiers on both sides are selected respectively, and a single rectifier is selected to generate the outermost and second outermost voltage vectors in the dq fundamental subspace, and 12 xy virtual voltage vectors are synthesized by the principle of amplitude-second balance; when the ratio of the outermost voltage vector to the second outermost voltage vector satisfies the action time of 0.732:0.268, the voltage amplitude of the single rectifier corresponding to the xy harmonic subspace is zero; The common mode voltages are grouped as follows: According to the vector space decoupling coordinate transformation, the voltage generated by a single rectifier is mapped to the mutually orthogonal α-β fundamental subspace, xy harmonic subspace and o1-o2 harmonic subspace, and the mapping relationship is as follows: According to the above formula, the voltage vector distribution diagram of a single rectifier in the o1-o2 harmonic subspace is derived to obtain the corresponding common-mode voltage. Through virtual synthesis, the 12 xy virtual voltage vectors are classified according to the size of the generated common-mode voltage. Value function design: After classification according to the virtual voltage vector, the voltage amplitude mapped in the two harmonic subspaces is zero, and the value function g dq In this case, we only need to consider tracking the d and q axis currents: g dq =|i d * -i d (k+1)| 2 +|i q * -i q (k+1)| 2 Where: i d * 、i q * are the reference values ​​of d-axis and q-axis currents respectively; ② Voltage vector construction and value function design of x-y harmonic surface: According to the outermost and second outermost voltage vectors generated by a single rectifier in the xy harmonic subspace, 12 virtual voltage vectors are synthesized through the principle of amplitude-second balance; when the ratio of the outermost voltage vector to the second outermost voltage vector satisfies the action time of 0.732:0.268, the voltage amplitude of the voltage vector generated by a single rectifier corresponding to the dq fundamental subspace is zero; According to the voltage vector distribution diagram of a single rectifier, among the above 12 virtual voltage vectors, three virtual voltages every 120 degrees have the same common mode voltage. 12 Find 12 virtual voltage vectors among the voltage vectors so that they have amplitude in the xy harmonic subspace, while the corresponding amplitudes in the dq fundamental subspace and the o1-o2 harmonic subspace are zero; Value function g xy design: g xy =|i x * -i x (k+1)| 2 +|i y * -i y (k+1)| 2 Where: i x * 、i y * are the reference values ​​of x-axis and y-axis currents respectively; ③Voltage selection of o1-o2 harmonic subspace According to the voltage vector distribution diagram of a single rectifier in the o1-o2 harmonic subspace, the basic voltage vector u 07 and u 70 As an alternative voltage.

4. A dual three-phase open-winding PMSG hybrid model predictive current control method according to claim 1, characterized in that: The specific process of step 3 is: The alternative voltage vector set of the xy harmonic subspace is brought into the current prediction equation, and the optimal alternative voltage vector in the xy harmonic subspace is selected through the value function. By optimizing the duty cycle, the action time of the optimal alternative voltage vector and the zero vector is allocated so that the current at time k+1 reaches a given value, that is, Where: t opt1 is the action time of the optimal candidate voltage vector corresponding to the xy harmonic subspace; s x 、s y They are respectively the optimal alternative voltage vector i x 、i y The slope of x0 、s y0 i is zero vector x 、i y The slope of Substitute the above formula into the value function of the xy harmonic subspace design, and let the value function act on the optimal candidate vector for time t opt1 The partial derivative is 0, and the optimal action time can be obtained:

5. The dual three-phase open-winding PMSG hybrid model predictive current control method according to claim 1, characterized in that: The specific process of step 4 is: Through the idea of ​​no-beat, we can get the corresponding u o1 The reference value u o1ref and u o2 The reference value u o2ref : Compensation is categorized as follows: ①When u o1ref >0 and u o2ref >0: Rectifier 1 select u 70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 1 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s , U dc is the DC bus voltage; that is, the effect of classification compensation is achieved, so that u o1ref =u o1 ,u o2ref =u o2 ; Secondly, according to the compensation time t1, t2, set t 3＝ (t1+t2); During the action time t3, rectifier 2 adds zero vector u 00 , zero vector u 00 The action time is t 4＝ t3; Make the classification compensation time of the two groups of rectifiers in the o1-o2 subspace the same; ②When u o1ref <0andu o2ref <0 hours: Rectifier 2 Select u 70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 2 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s ; That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 ; Secondly, according to the compensation time t1, t2, set t 3＝ (t1+t2); During the action time t3, rectifier 1 adds zero vector u 00 , zero vector u 00 The action time is t 4＝ t3; Make the classification compensation time of the two groups of rectifiers in the o1-o2 subspace the same; ③When u o1ref >0 and u o2ref <0 hours: Rectifier 1 select u 70 To compensate u o1ref , compensation time t1=|u o1ref / U dc |*T s , rectifier 2 select u 07 To compensate u o2ref , compensation time t2=|u o2ref / U dc |*T s ; That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 ; Secondly, according to the compensation time t1, t2, set t 3＝ max(t1, t2); within the action time t3, the rectifier with less compensation time adds a zero vector u on the basis of the original compensation 00 , zero vector u 00 The action time is t4 = t 3— min(t1, t2); so that the classification compensation time of the two groups of rectifiers is the same in the o1-o2 subspace; ④When u o1ref <0andu o2ref >0: Rectifier 1 select u 07 To compensate u o2ref , compensation time t1=|u o2ref / U dc |*T s , rectifier 2 select u 70 To compensate u o1ref , compensation time t2=|u o1ref / U dc |*T s ; That is, to achieve the effect of classification compensation, so that u o1ref =u o1 ,u o2ref =u o2 ; Secondly, according to the compensation time t1, t2, set t3 = max(t1, t2); within the action time t3, the rectifier with less compensation time adds a zero vector u on the basis of the original compensation. 00 , zero vector u 00 The action time is t4 = t 3— min(t1, t2); makes the classification compensation time of the two groups of rectifiers the same in the o1-o2 subspace.

6. A dual three-phase open-winding PMSG hybrid model predictive current control method according to claim 1, characterized in that: The specific process of step 5 is: Rectifier 1 is in the remaining time T s -t opt1 The optimal voltage vector acting within -t3 is obtained by the d-axis and q-axis current deadbeat formula: Where u dref 、u qref They are the d-axis and q-axis reference voltage components in the two-phase rotating coordinate system, and the Clarke transformation is performed on them: u αref 、u βref The position angle θ in the α-β plane is: According to the position of the reference voltage in the sector, the rectifier 1 is selected for the remaining time T s -t opt1 -t3; secondly, according to the rectifier 1 in the remaining time T s -t opt1 -t3, rectifier 2 selects the optimal voltage vector within the same group of two xy virtual voltage vectors, enumerates three times, and obtains rectifier 2 in the remaining time T s -t opt1 - Two virtual voltage vectors acting within t3 and their action time; the action time of the two virtual voltage vectors is calculated by the q-axis current deadbeat; The calculation method is as follows: Where: s0, s i 、s j The changes in the slope of the q-axis current caused by the 0 voltage vector, the rectifier 2 selects the i voltage vector, and the rectifier 2 selects the j voltage vector; d opt is the duty cycle of the i voltage vector selected by rectifier 2; then the j voltage vector time selected by rectifier 2 is (T s -t opt1 -t3)*(1-d opt ); In a control cycle T s Within, there is an effective voltage vector action time greater than T s In order to ensure the control of the active component of the system, the compensation time t of the two harmonic subspace closed loops is opt1 , t3 imposes restrictions: That is, when t opt1 +t 3≥ T s When t , the harmonics are suppressed by using the virtual voltage vector of the fundamental subspace and classifying the virtual voltage vector according to the common mode voltage. opt1 +t 3< T s When the closed-loop control of two harmonic subspaces is added, the influence of nonlinear factors is reduced and the harmonic current is suppressed; Finally, the voltage vectors acting on the three subspaces and their action time are determined, and the switch state and the corresponding opening time corresponding to the voltage vector are applied to the rectifier to complete the output of the switching sequence.

Images