Fault-tolerant direct thrust-force control method for five-phase permanent-magnet linear motor with one open-circuit phase fault

Abstract

In the fault-tolerant direct thrust-force control (FT-DTC) method, the generalized Clarke transform matrix and its inverse matrix are derived according to the fault-tolerant phase currents. The stator fluxes in α-β plane are deduced based on these. Based on the requirement of circular stator flux trajectory, virtual stator fluxes are defined, and then compensatory voltages in the α-β plane are obtained. Actual stator voltages in the α-β plane are calculated by modulation function of voltage source inverter. Combining with the compensatory voltages, the actual stator voltages and the stator currents, the virtual stator fluxes and the thrust-force are estimated by the flux and thrust-force observers. The thrust-force reference, the stator flux amplitude reference, the observed thrust-force and virtual stator flux are applied to predict virtual stator voltage references. The actual stator voltage references is calculated according to virtual voltage references and compensatory voltages, and are fed to voltage source inverter.

Description

CROSS REFERENCE TO THE RELATED APPLICATION

This application is the national stage entry of International Application No. PCT/CN2018/074639, filed on Jan. 31, 2018, which is based upon and claims priority to Chinese Patent Application No. 201810025607.3, filed on Jan. 11, 2018, the entire contents of which are incorporated herein by reference.

TECHNICAL FIELD

The invention relates to a fault-tolerant direct thrust-force control (FT-DTC) method for a linear motor under one-phase open-circuit fault condition, especially to a FT-DTC method for a five-phase permanent-magnet linear motor (PMLM) with an open-circuit fault. It is suitable for aerospace, electric vehicles and so on, which have strict requirements on reliability and dynamic performance.

BACKGROUND

With the improvement of science, technology and living standards, people are requiring more from transportation, especially safety, reliability and comfort of driving. The performance of a suspension system is of great significance for vehicle driving performance, so researches on steady and dynamic performance of active suspension systems have received more and more attention. The tubular linear motor is a key component for active suspension system. The steady and dynamic performance of a tubular linear motor with a fault is directly related to the reliability of electromagnetic suspension and the capability of disturbance-free operation.

When a motor has a one phase open-circuit fault, although it can still output certain thrust/torque and speed, it results in deteriorated phase currents, and fluctuations of thrust/torque and speed when compared to those in healthy operation. Meanwhile, the noise and loss become larger, which will deteriorate operational performance and decrease service life, and even cause irreparable damage to main components of the motor drive. When an open-circuit fault occurs, it is possible to ensure the motor reaches the same steady and dynamic performance of healthy operation if the suitable fault-tolerant algorithm is injected in time. For example, the fluctuations of thrust/torque and speed in fault operation can be significantly reduced. The application number of a Chinese patent is 201410477291.3—“Fault-tolerant torque control method for six-phase permanent-magnet synchronous motor with biased-60-degree under one-phase loss”. It presented a fault-tolerant direct torque control (FT-DTC) method for a six-phase permanent-magnet synchronous motor with a one-phase open-circuit fault, which is the improved method of traditional direct torque control (DTC). Virtual voltage sector and zero-sequence subspace are divided into 36 and 46 sectors respectively, where the appropriate switching vector is selected to realize FT-DTC operation. However, the complexity of control algorithm was increased. Particularly, sector division is related to trigonometric functions and irrational numbers, which increases control error and affects FT-DTC performance. Meanwhile, hysteresis control is used to control the flux amplitude and torque, which also caused problems such as error and difficulty of regulation. The paper “Open-phase fault-tolerant direct torque control technique for five-phase induction motor drives, IEEE Transactions on Industrial Electronics 64(2): 902-911, 2017” proposed a FT-DTC method for a five-phase induction motor with a one-phase open-circuit fault. When a one-phase open-circuit fault occurs, the motor drive has large control error and noise, disordered switching frequency and deteriorative dynamic performance due to hysteresis control. Meanwhile, according to the stator flux and the output of two hysteresis comparators, the suitable sector and switching vector are selected, which causes complex fault-tolerant control algorithm and low control precision. In addition, current in the zero-sequence space is not controlled to zero, which results in increasing loss, large ripples of torque and phase currents. At present, the main idea of the FT-DTC method for a multi-phase motor with a one-phase open-circuit fault is that components of stator currents in the two-phase stationary frame are obtained by a traditional transformation matrix, and voltage components in the two-phase stationary frame are obtained according to the switching state, and then electromagnetic torque, stator flux and its angle are observed. Finally, FT-DTC operation is achieved by selecting a suitable switching vector based on hysteresis comparators and a switching table. However, due to adoption of the hysteresis comparators and switching table, it results in unequal phase current amplitudes, large switching frequencies, existing zero-sequence space currents, large noise and loss, and deteriorated steady and dynamic performance. Therefore, this method is not suitable for applications with strict requirements on steady and dynamic performance. Moreover, most FT-DTCs are employed for multi-phase induction/permanent-magnet rotating machines, and there is no FT-DTC method for a five-phase PMLM.

SUMMARY

It is aimed at shortcomings of the existing fault-tolerant control technology, and characteristics of a five-phase PMLM and its open-circuit fault. The goal of invention is to overcome disadvantages of existing fault-tolerant control strategy when a five-phase linear motor has a one-phase open-circuit fault, such as an unequal phase current amplitude, large thrust-force/torque ripples, variable inverter switching frequency, low control accuracy, severe noise, poor dynamic performance and so on, caused by a traditional Clarke transform matrix, hysteresis comparators and switching tables. The invention can not only achieve high fault-tolerant performance, high dynamic performance, excellent tracking performance of thrust-force and speed when the motor drive has a one-phase open-circuit fault, but can also reduce the control algorithm's complexity and CPU's cost. Meanwhile, it can achieve a constant inverter switching frequency, equal phase current amplitude, low noise, and simplified EMC design. Finally, high reliability, high steady-state and dynamic performance of a five-phase PMLM with an open-circuit fault can be achieved.

The control object of the invention is the five-phase PMLM, whose five phase are defined as phase-A, phase-B, phase-C, phase-D and phase-E, respectively. When the motor has an open-circuit fault, the FT-DTC method includes the following steps:

(12) A five-phase PMLM model is established.

(13) When an open-circuit fault occurs in phase-A, according to the principle that Magneto-motive Force (MMF) of stator windings keeps constant both in healthy and fault operation, the constraint that the sum of remaining healthy phase currents is zero, and the constraint that the amplitudes of remaining healthy phase currents should be equal, the fault-tolerant currents i_B, i_C, i_D, i_E in the remaining healthy phases are deduced. Based on the obtained results, the generalized Clarke transform matrix T_4s/2s and its inverse transform matrix T_2s/4s are derived under open-circuit fault conditions, where T_4s/2s is adopted to transform variations from the natural frame into the two-phase stationary frame named as α-β plane.

(14) According to the relationship between stator flux and permanent-magnet flux: ψ_s=L_si_s+ψ_f, T_4s/2s is applied to transform stator fluxes ψ_B, ψ_C, ψ_D, ψ_E to the stator fluxes in the α-β plane and ψ_z in the zero-sequence subspace.

(15) Based on the principle that trajectory of the stator flux vector and MMF vector are circular in the α-β plane under fault-tolerant operation, virtual stator flux ψ_xs in the α-β plane is defined according to the Step 3.

(16) The relationship between actual and virtual stator voltages in the α-β plane is derived by T_4s/2s and ψ_xs:

$\left[\begin{array}{c}{u}_{x\alpha }\\ u_{x\beta }\end{array}\right]=\left[\begin{array}{c}{u}_{\alpha }\\ u_{\beta }\end{array}\right]+{\lambda }_m\omega \left[\begin{array}{c}-0.5\sin \theta \\ 0.053\cos \theta \end{array}\right]$

(17) In order to keep a circular trajectory of stator flux vector and MMF vector under open-circuit fault conditions, the stator compensatory voltages u_compα, u_compβ in the α-β plane are defined based on Step 5.

(18) T_4s/2s is employed to transform the remaining healthy phase currents i_B, i_C, i_D, i_E to the currents i_α, i_β in the α-β plane and the current i_z in the zero-sequence subspace. Based on these, the observers of flux and thrust-force are designed to estimate the amplitude of virtual stator flux ψ_xs, the virtual stator flux angle θ_s and the thrust-force F_e. Meanwhile, a Proportional Integral (PI) controller is adopted to control i_z to zero, whose output is the zero-sequence voltage reference u*_z.

(19) The difference between the thrust-force reference F*_e and the estimated thrust-force F_e is used to obtain the thrust-force incremental angle Δδ by a PI regulator. Then, the differences Δψ_xα, Δψ_xβ between the flux references and the estimated virtual stator fluxes in the α-β plane are achieved according to the observed virtual stator flux amplitude ψ_xs, the flux reference ψ*_s and the virtual stator flux angle θ_s.

(20) The virtual stator voltages are defined as:

$\left[\begin{array}{c}{u}_{x\alpha }\\ u_{x\beta }\end{array}\right]=R\left[\begin{array}{c}{i}_{\sigma }\\ i_{\beta }\end{array}\right]+\frac{d}{dt}\left[\begin{array}{c}{\psi }_{x\alpha }\\ {\psi }_{x\beta }\end{array}\right],$ then the virtual stator voltage references u*_xα, u*_xβ are deduced in the α-β plane.

(21) According to step 6, the actual stator voltage references u*_α, u*_β in fault-tolerant operation are obtained.

(22) T_2s/4s is adopted to transform the actual voltage references u*_α, u*_β, u*_z obtained from step (7) and (10) to the phase voltage references u*_B, u*_C, u*_D, u*_E in the natural fame. Since phase-A is open-circuit, the voltage reference of phase-A is defined as u*_A=0.

(12) Combining with voltage source inverter and the CPWM method, the phase voltage references u*_A, u*_B, u*_C, u*_p, u*_E obtained from step (11) are employed to realize disturbance-free operation of the five-phase PMLM with an open-circuit fault.

The invention has the following beneficial effects:

8. The proposed FT-DTC method in the present invention is quite different from traditional FT-DTC, which adopts a hysteresis comparator to select a voltage vector in a switching table. The traditional FT-DTC method brings voltage error. Additionally, the switching table query and sector selection and sector division, calculation on trigonometric function and irrational function greatly increase complexity and results in large thrust-force ripples. The FT-DTC method in the present invention adopts the CPWM method based on zero-sequence voltage signal injection, which has the same effect as the space vector pulse width modulation (SVPWM), then the proposed method can be achieved without sector selection and complex calculations. Meanwhile, CPU cost is saved, in which calculation time is effectively reduced, and pulsating thrust-force is greatly restrained.9. The combination of the generalized Clarke transformation matrix and FT-DTC strategy proposed in the present invention realizes that the amplitude of currents in α-β plane is equal and the angle between them is 90 degrees in open-circuit fault operation, i.e., the trajectory of the current vector in the α-β plane is circular. At the same time, currents in zero-sequence space are extracted, then they can be restrained to improve the quality of the sinusoidal waveform of phase currents and reduce copper and core loss using a simple PI controller, and it is no longer needed to set a special voltage vector to suppress zero-sequence current as in the traditional FT-DTC method.10. The generalized Clarke transform matrix and stator compensatory voltages in the present invention not only realize circular trajectories of the virtual stator flux vector and MMF vector in the α-β plane, but also achieve fault-tolerant direct thrust-force control.11. The combination of virtual stator flux and direct thrust-force control (DTC) based on a space vector in the present invention effectively avoids a too large and complex voltage vector switching table of the traditional FT-DTC. The voltage vector can be obtained accurately, and then a CPWM strategy can be employed to achieve precise modulation. Thus, sinusoidal waveforms of phase currents are improved, thrust-force pulsations are effectively suppressed, steady-state performance is improved, and vibration and noise are restrained.12. The amplitude of the virtual stator flux obtained by the flux observer is equal to the stator flux amplitude in healthy operation, which ensures consistency of stator fluxes and thrust-force both in healthy and fault operation.13. Virtual voltages in the present invention are reconstructed to compensate stator voltages when a phase-A open-circuit fault occurs, then it is capable of keeping sinusoidal stator currents, equal amplitude of stator currents, 90 degree phase difference between the two current phasors in the α-β plane, and circular trajectories of the stator flux vector and the MMF vector are also achieved without reconstructing virtual stator currents, virtual flux and so on in the α-β plane; when an open-circuit fault occurs, this method can effectively switch from healthy operation to fault-tolerant operation without disturbance. Compared to fault operation, it effectively reduces thrust-force ripples and distortion of the phase currents. The proposed control method in the present invention adopts voltage vectors produced by the remaining healthy phases to realize instantaneous control of thrust-force in the α-β plane, hence possessing excellent dynamic response and accurate following performance.14. The generalized Clarke transform matrix in the present invention is used to transform the variations in remaining healthy phases into the two-phase stationary frame (α-β), compensatory voltages are employed to realize circular stator flux trajectory. These create a precondition for FT-DTC disturbance-free operation under open-circuit fault condition. On the other hand, zero-sequence current is controlled to zero to reduce copper and core loss, to increase efficiency and to restrain fluctuations of thrust-force. Combining with the generalized Clarke transform matrix, its inverse matrix, the CPWM method based on zero-sequence voltage injection, compensatory voltages in the two-phase stationary frame, reconstruction of virtual stator voltages, the observers of virtual stator flux and thrust-force, the proposed FT-DTC strategy can not only realize DTC disturbance-free operation under an open-circuit fault condition, but also increase utilization of inverter bus voltage. Meanwhile, it avoids the complexity of the traditional SVPWM algorithm. The invention also adopts the CPWM method, which is a clear and concise way to highlight simple and effective characteristics of DTC. Additionally, the proposed FT-DTC method in the present invention can also improve the tracking performance of thrust-force, dynamic and steady-state performance under an open-circuit fault condition, so that dynamic and steady-state performance are almost the same as that in healthy operation.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 shows the schematic diagram of a five-phase fault-tolerant PMLM.

FIG. 2 shows the schematic diagram of the DTC strategy for the five-phase fault-tolerant PMLM of the present invention.

FIG. 3 shows the schematic diagram of a FT-DTC method for the five-phase fault-tolerant PMLM with the phase-A open-circuit fault of the present invention.

FIG. 4 shows the waveforms of phase currents under the phase-A open-circuit fault condition without and with the FT-DTC of the present invention.

FIG. 5 shows the waveforms of thrust-force under the phase-A open-circuit fault condition without and with the FT-DTC of the present invention.

FIG. 6 shows the waveforms of stator currents in the α-β plane under the phase-A open-circuit fault condition without and with the FT-DTC of the present invention.

FIG. 7 shows the waveforms of stator fluxes in the α-β plane under the phase-A open-circuit fault condition without and with the FT-DTC of the present invention.

FIG. 8 shows the trajectory of stator flux vector in the α-β plane under the phase-A open-circuit fault without and with the FT-DTC of the present invention.

FIG. 9 shows the waveforms of phase currents when the thrust-force reference steps in healthy operation of the present invention.

FIG. 10 shows the waveform of thrust-force when the thrust-force reference steps in healthy operation of the present invention.

FIG. 11 shows the waveforms of stator currents in the α-β plane when the thrust-force reference steps in healthy operation of the present invention.

FIG. 12 shows the waveforms of phase currents when the thrust-force reference steps under a phase-A open-circuit fault condition with the FT-DTC of the present invention.

FIG. 13 shows the waveform of the output thrust when the thrust reference steps under a phase-A open-circuit fault condition with the FT-DTC of the present invention.

FIG. 14 shows the waveforms of the stator currents in the α-β plane when the thrust-force reference steps under a phase-A open-circuit fault condition with the FT-DTC of the present invention.

In the figures: 1 represents a stator, 2 represents a mover, 3 represents an armature tooth, 4 represents fault-tolerant teeth, 5 represents a coil winding, 6 represents a permanent magnet, and 7 represents magnetic material.

DETAILED DESCRIPTION OF THE EMBODIMENTS

In order to more simply and clearly illustrate the structural characteristics and beneficial effects of the FT-DTC method for a five-phase PMLM in the present invention, a five-phase fault-tolerant PMLM is described in the following.

FIG. 1 shows the topology of the five-phase permanent-magnet fault-tolerant tubular linear motor in the present invention. The windings are placed in the stator. The permanent magnets are located on the mover, and each permanent magnet mounted on the mover in the form of a spoke is isolated by a magnetically permeable material. Then, it can generate a large thrust-force and a small detent force. The fractional-slot single-layer concentrated windings are used and coil windings are located on both sides of the armature teeth which decrease the harmonics of back EMF and improve winding utilization rate. The fault-tolerant teeth are introduced to achieve electrical, magnetic and thermal isolation between phases. When the motor has an open-circuit fault, the remaining healthy phases are not affected by the faulty phase, thus possessing strong fault-tolerance.

Part One: Derivation of Constant MMF Under an Open-Circuit Fault Condition

The five-phase fault-tolerant PMLM is powered by a voltage source inverter. The five phases are divided into phase-A, phase-B, phase-C, phase-D, and phase-E respectively. When the motor is in healthy operation, it is assumed that the zero-sequence space currents i_z1 and i_z2 have been controlled to zero. Then phase-A, phase-B, phase-C, phase-D and phase-E currents can be expressed as:

$\begin{array}{cc}\{\begin{array}{c}{i}_A=i_{\alpha }\\ i_B=i_{\alpha }\cos a+i_{\beta }\sin a\\ i_C=i_{\alpha }\cos 2a+i_{\beta }\sin 2a\\ i_D=i_{\alpha }\cos 3a+i_{\beta }\sin 3a\\ i_E=i_{\alpha }\cos 4a+i_{\beta }\sin 4a\end{array}& \left(1\right)\end{array}$

where i_A, i_B, i_C, i_D, i_E are the currents in phase-A, phase-B, phase-C, phase-D and phase-E, respectively. i_α and i_β are the currents in the α-β plane, respectively, a=2π/5, θ is the electrical angle,

$\theta =\int \frac{\pi \upsilon }{\tau }\mathrm{dt},$ ν is me electrical velocity of mover, and τ is the pole pitch of the five-phase fault-tolerant PMLM.

The traveling wave MMF of the five-phase PMLM is

$\begin{array}{cc}\begin{array}{c}\mathrm{MMF}=\sum _{i=A}^E{\mathrm{MMF}}_i\\ ={\mathrm{Ni}}_A+\varepsilon {\mathrm{Ni}}_B+{\varepsilon }^2{\mathrm{Ni}}_C+{\varepsilon }^3{\mathrm{Ni}}_D+{\varepsilon }^4{\mathrm{Ni}}_E\end{array}& \left(2\right)\end{array}$ where ε=e^j2π/5, and N is the effective turns of each phase winding.

When a phase-A open-circuit fault occurs, its phase current is zero, so the traveling wave MMF of the remaining healthy phases can be expressed as

$\begin{array}{cc}\begin{array}{c}\mathrm{MMF}=\sum _{i=B,C,D,E}{\mathrm{MMF}}_i\\ =\varepsilon {\mathrm{Ni}}_B+{\varepsilon }^2{\mathrm{Ni}}_c+{\varepsilon }^3{\mathrm{Ni}}_D+{\varepsilon }^4{\mathrm{Ni}}_E\end{array}& \left(3\right)\end{array}$

When phase-A is in an open-circuit fault, to reach the purpose of disturbance-free operation, it is necessary to keep the same amplitude and speed of the MMF both in healthy and fault operation. Hence, (2) should be equal to (3). Because the windings are star-connected and its center point is not connected to the neutral point of the DC bus, the sum of healthy phase currents should be zero. Then, according to the constraint that amplitudes of the remaining healthy phase currents are equal, the remaining healthy phase currents can be expressed as:

$\begin{array}{cc}\{\begin{array}{c}{I}_B=I_C=I_D=I_E\\ i_B+i_C+i_D+i_E=0\end{array}& \left(4\right)\end{array}$ where I_B, I_C, I_D, I_E are the amplitudes of currents in phase-B, phase-C, phase-D, phase-E, respectively.

Based on the aforementioned constraints, the remaining healthy phase currents (i_B, i_C, i_D, i_E) under fault tolerance can be deduced:

$\begin{array}{cc}\left[\begin{array}{c}{i}_B\\ i_C\\ i_D\\ i_E\end{array}\right]=1.382\left[\begin{array}{cc}\cos 05a& \sin 0.5a\\ \cos 2a& \sin 2a\\ \cos 3a& \sin 3a\\ \cos 4.5a& \sin 4.5a\end{array}\right]\left[\begin{array}{c}{i}_{\alpha }\\ i_{\beta }\end{array}\right]& \left(5\right)\end{array}$

When an open-circuit fault occurs to phase-A, there remains only three degrees of freedom in the PMLM drive system, two of which are in the fundamental subspace and another is in the zero-sequence subspace. Since back-EMFs mainly contain fundamental components, the electromechanical energy conversion occurs in the fundamental subspace, and does not occur in the zero-sequence subspace. Therefore, according to (5), the basis of fundamental subspace can be defined as:

$\begin{array}{cc}\{\begin{array}{c}{T}_1=\left[\cos 0.5a\cos 2a\cos 3a\cos 4.5a\right]\\ T_2=\left[\sin 0.5a\sin 2a\sin 3a\sin 4.5a\right]\end{array}& \left(6\right)\end{array}$

The basis of zero-sequence subspace is defined as z. Then, based on the orthogonal principle of the two subspaces and (5), the generalized Clarke transformation matrix T_4s/2s, which is applied to transform variables in the remaining healthy phases into the two-phase stationary frame, and its inverse transformation matrix T_2s/4s are deduced as:

$\begin{array}{cc}{T}_{4s/2s}=\left[\begin{array}{cccc}\cos 0.5a/3.618& \cos 2a/3.618& \cos 3a/3.618& \cos 4.5a/3.618\\ \sin 0.5a/1.91& \sin 2a/1.91& \sin 3a/1.91& \sin 4.5a/1.91\\ \sin a/5& \sin 4a/5& \sin 6a/5& \sin 9a/5\end{array}\right]& \left(7\right)\\ T_{2s/4s}=1.382\left[\begin{array}{ccc}\cos 0.5a& \sin 0.5a& \sin a\\ \cos 2a& \sin 2a& \sin 4a\\ \cos 3a& \sin 3a& \sin 6a\\ \cos 4.5a& \sin 4.5a& \sin 9a\end{array}\right]& \left(8\right)\end{array}$

Part Two: Derivation of a Circular Trajectory of the Stator Flux Vector in Fault Operation

When the motor has an open-circuit fault, the compensatory voltages are designed to ensure a circular flux trajectory under the precondition of keeping a circular trajectory of the MMF vector.

When a phase-A open-circuit fault occurs, the relationship between the stator flux, permanent-magnet flux and remaining healthy phase currents can be still expressed as ψ_s=L_si_s+ψ_f, which can be further expressed as:

$\begin{array}{cc}\left[\begin{array}{c}{\psi }_B\\ {\psi }_C\\ {\psi }_D\\ {\psi }_E\end{array}\right]=L_s\left[\begin{array}{c}{i}_B\\ i_C\\ i_D\\ i_E\end{array}\right]+{\lambda }_m\left[\begin{array}{c}\cos \left(\theta -a\right)\\ \cos \left(\theta -2a\right)\\ \cos \left(\theta -3a\right)\\ \cos \left(\theta -4a\right)\end{array}\right]& \left(9\right)\end{array}$ where ψ_s are the stator fluxes, ψ_s=[ψ_B ψ_C ψ_D ψ_E]^T, ψ_B, ψ_C, ψ_D, ψ_E are the stator fluxes of phase-B, phase-C, phase-D, phase-E, respectively. i_s are the stator currents, i_s=[i_B i_C i_D i_E]^T, λ_m is the amplitude of permanent-magnet flux, and L_s is the phase inductance which is assumed as to be constant.

T_4s/2s is adopted to transform the stator fluxes in the remaining healthy phases (ψ_B, ψ_C, ψ_D, ψ_E) to the actual stator flux components (ψ_α, ψ_β) in the α-β plane and the flux ψ_z in the zero-sequence subspace, which can be expressed as:

$\begin{array}{cc}\{\begin{array}{c}{\psi }_s=\sqrt{{\psi }_{\alpha }^2+{\psi }_{\beta }^2}\\ {\psi }_{\alpha }=L_s{i}_{\alpha }+0.5{\lambda }_m\cos \theta \\ {\psi }_{\beta }=L_s{i}_{\beta }+0.947{\lambda }_m\sin \theta \\ {\psi }_z=L_s{i}_z+0.138{\lambda }_m\sin \theta \end{array}& \left(10\right)\end{array}$ where ψ_s is the actual stator flux amplitude, ψ_α, ψ_β are the actual stator fluxes in the α-β plane, and i_z is the current in the zero-sequence subspace.

Equation (10) shows that the trajectory of the stator flux vector in the α-β plane is not circular when a phase-A open-circuit fault occurs. Then, if a suitable fault-tolerant algorithm is not employed to ensure the circular trajectory of the stator flux vector, the currents (i_α, i_β) are no longer sinusoidal and their amplitudes are no longer equal. It will inevitably cause a large pulsation on the MMF. In order to keep a circular trajectory of the stator flux vector and the MMF vector simultaneously, the virtual stator flux ψ_xs in the α-β plane is defined as:

$\begin{array}{cc}\{\begin{array}{c}{\psi }_{\mathrm{xs}}=\sqrt{{\psi }_{x\alpha }^2+{\psi }_{x\beta }^2}\\ {\psi }_{x\alpha }={\lambda }_m\cos \theta +L_s{i}_{\alpha }\\ {\psi }_{x\beta }={\lambda }_m\sin \theta +L_s{i}_{\beta }\end{array}& \left(11\right)\end{array}$ where ψ_xα and ψ_xβ are the components of virtual stator fluxes ψ_xs in the α-β plane, and ψ_xs is the amplitude of the virtual stator flux.

According to (10)-(11), the relationship between the actual and virtual stator fluxes is derived as:

$\begin{array}{cc}\{\begin{array}{c}{\psi }_{\alpha }={\psi }_{x\alpha }-0.5{\lambda }_m\cos \theta \\ {\psi }_{\beta }={\psi }_{x\beta }-0.053{\lambda }_m\sin \theta \end{array}& \left(12\right)\end{array}$

When a phase-A open-circuit fault occurs, the voltage model of the remaining healthy phases in the natural frame can be obtained as:

$\begin{array}{cc}\{\begin{array}{c}{u}_B={\mathrm{Ri}}_B+\frac{d{\psi }_B}{dt}\\ u_C={\mathrm{Ri}}_c+\frac{d{\psi }_C}{dt}\\ u_D={\mathrm{Ri}}_D+\frac{d{\psi }_D}{dt}\\ u_E={\mathrm{Ri}}_E+\frac{d{\psi }_E}{dt}\end{array}& \left(13\right)\end{array}$ where R is stator resistance which is recognized as constant value, and u_B, u_C, u_D, u_E are the voltages in the remaining healthy phases.

The voltage model under fault condition is transformed into the α-β plane by T_4s/2s, and then it can be expressed as:

$\begin{array}{cc}\left[\begin{array}{c}{u}_{\alpha }\\ u_{\beta }\end{array}\right]=R\left[\begin{array}{c}{i}_{\alpha }\\ i_{\beta }\end{array}\right]+\frac{d}{dt}\left[\begin{array}{c}{\psi }_{\alpha }\\ {\psi }_{\beta }\end{array}\right]& \left(14\right)\end{array}$ where u_α, u_β are the actual stator voltages in the α-β plane, and ψ_α, ψ_β are the actual stator fluxes in the α-β plane. When (12) is introduced into (14), the voltage model represented by the virtual stator fluxes can be expressed as:

$\begin{array}{cc}\left[\begin{array}{c}{u}_{\alpha }\\ u_{\beta }\end{array}\right]=R\left[\begin{array}{c}{i}_{\alpha }\\ i_{\beta }\end{array}\right]+\frac{d}{dt}\left[\begin{array}{c}{\psi }_{x\alpha }\\ {\psi }_{x\beta }\end{array}\right]-{\lambda }_m\omega \left[\begin{array}{c}-0.5\sin \theta \\ 0.053\cos \theta \end{array}\right]& \left(15\right)\end{array}$ where ω is the electric angular velocity,

$\omega =\frac{\pi v}{\tau }.$

The virtual stator voltages are defined as:

$\begin{array}{cc}\left[\begin{array}{c}{u}_{x\alpha }\\ u_{x\beta }\end{array}\right]=R\left[\begin{array}{c}{i}_{\alpha }\\ i_{\beta }\end{array}\right]+\frac{d}{dt}\left[\begin{array}{c}{\psi }_{x\alpha }\\ {\psi }_{x\beta }\end{array}\right]& \left(16\right)\end{array}$ where u_xα, u_xβ are the components of virtual stator voltages in the α-β plane.

According to (15)-(16), the relationship between actual and virtual stator voltages can be derived as:

$\begin{array}{cc}\left[\begin{array}{c}{u}_{x\alpha }\\ u_{x\beta }\end{array}\right]=\left[\begin{array}{c}{u}_{\alpha }\\ u_{\beta }\end{array}\right]+{\lambda }_m\omega \left[\begin{array}{c}-0.5\sin \theta \\ 0.053\cos \theta \end{array}\right]& \left(17\right)\end{array}$

When the motor has an open-circuit fault, in order to keep the circular trajectories of the stator flux vector and the MMF vector, according to (17), the stator compensatory voltages u_compα, u_compβ in the α-β plane are defined as:

$\begin{array}{cc}\{\begin{array}{c}{u}_{\mathrm{comp}\alpha }=-0.5{\lambda }_m\omega \sin \theta \\ u_{\mathrm{comp}\beta }=0.053{\lambda }_m\omega \cos \theta \end{array}& \left(18\right)\end{array}$

Part Three: The FT-DTC Strategy Under an Open-Circuit Fault Condition

T_4s/2s is adopted to transform the currents in the remaining healthy phases to the currents i_α, i_β in the α-β plane and i_z in the zero-sequence frame:

$\begin{array}{cc}\left[\begin{array}{c}{i}_{\alpha }\\ i_{\beta }\\ i_z\end{array}\right]=T_{4s/2s}\left[\begin{array}{c}{i}_B\\ i_C\\ i_D\\ i_E\end{array}\right]& \left(19\right)\end{array}$

The PI controller is designed to control i_z to zero, whose output is defined as the zero-sequence voltage reference u*_z.

Since the CPWM method is applied in the present invention, components of voltages u_α, u_β in the α-β plane in fault tolerant operation are derived by T_4s/2s:

$\begin{array}{cc}\{\begin{array}{c}{u}_{\alpha }=0.2764U_{\mathrm{dc}}\left(u_b\cos 0.5a+u_c\cos 2a+u_d\cos 3a+u_e\cos 4.5a\right)\\ u_{\beta }=0.5236U_{\mathrm{dc}}\left(u_b\sin 0.5a+u_c\sin 2a+u_d\sin 3a+u_e\sin 4.5a\right)\end{array}& \left(20\right)\end{array}$ where U_dc is the DC-bus voltage, and u_b, u_c, u_d, u_e are the phase functions derived from the modulation of the phase-voltage references u*_B, u*_C, u*_D, u*_E by the CPWM method.

According to (17), the virtual voltages u_xα and u_xβ are obtained as:

$\begin{array}{cc}\{\begin{array}{c}{u}_{x\alpha }=u_{\alpha }+u_{\mathrm{comp}\alpha }\\ u_{x\beta }=u_{\beta }+u_{\mathrm{comp}\beta }\end{array}& \left(21\right)\end{array}$

The back EMF method is applied to construct a flux observer, and then the virtual stator flux ψ_xα and ψ_xβ in the α-β plane are observed as:

$\begin{array}{cc}\{\begin{array}{c}{\psi }_{x\alpha }=\frac{1}{s+{\omega }_c}\left(u_{x\alpha }-i_{\alpha }R\right)+\frac{{\omega }_c}{s+{\omega }_c}{z}_{x\alpha }\\ {\psi }_{x\beta }=\frac{1}{s+{\omega }_c}\left(u_{x\beta }-i_{\beta }R\right)+\frac{{\omega }_c}{s+{\omega }_c}{z}_{x\beta }\\ {\psi }_{\mathrm{xs}}=\sqrt{{\psi }_{x\alpha }^2+{\psi }_{x\beta }^2}\\ {\theta }_s=\mathrm{arc}\mathrm{tg}\frac{{\psi }_{x\beta }}{{\psi }_{x\alpha }}\end{array}& \left(22\right)\\ z_{xa}=\{\begin{array}{cc}{\psi }_{x\alpha }& \left|{\psi }_{xa}\right|\le {\psi }_{\max }\\ {\psi }_{\max }& \left|{\psi }_{xa}\right|>{\psi }_{\max }\end{array}& \left(23\right)\\ z_{x\beta }=\{\begin{array}{cc}{\psi }_{x\beta }& \left|{\psi }_{x\beta }\right|\le {\psi }_{\max }\\ {\psi }_{\max }& \left|{\psi }_{x\beta }\right|>{\psi }_{\max }\end{array}& \left(24\right)\end{array}$ where ω_c is the cutoff frequency of the low pass filter, ψ_max is the limited value of stator flux, θ_s is the virtual stator flux angle, and s is the integral operator.

By employing the magnetic co-energy method, a thrust-force observer in fault-tolerant operation can be built as:

$\begin{array}{cc}{F}_e=\frac{2.5\pi }{\tau }\left({\psi }_{x\alpha }{i}_{\beta }-{\psi }_{x\beta }{i}_{\alpha }\right)+\frac{0.955\pi }{\tau }{\lambda }_m{i}_z\cos \theta & \left(25\right)\end{array}$

The difference between thrust-force reference F*_e and estimated value F_e is adjusted to generate an incremental angle of thrust-force Δδ by a PI regulator. Then, according to the observed virtual stator flux amplitude ψ_xs, the flux reference ψ*_s and the virtual stator flux angle θ_s, the difference Δψ_xα, Δψ_xβ between the flux reference and observed virtual stator flux in the αβ plane can be derived as:

$\begin{array}{cc}\{\begin{array}{c}\Delta {\psi }_{x\alpha }={\psi }_s^{*}\cos \left({\theta }_s+\Delta \delta \right)-{\psi }_{x\alpha }\\ {\Delta }_{{\psi }_{x\beta }}={\psi }_s^{*}\sin \left({\theta }_s+\Delta \delta \right)-{\psi }_{x\beta }\end{array}& \left(26\right)\end{array}$

Based on (16)-(26), the components of the virtual stator voltage references (u*_xα, u*_xβ) in the α-β plane are deduced as:

$\begin{array}{cc}\{\begin{array}{c}{u}_{x\alpha }^{*}=R{i}_{\alpha }+\frac{\Delta {\psi }_{x\alpha }}{T}\\ u_{x\beta }^{*}=R{i}_{\beta }+\frac{\Delta {\psi }_{x\beta }}{T}\end{array}& \left(27\right)\end{array}$ where Tis the sampling period.

On the basis of the relationship between actual and virtual stator voltages in (17), the actual stator voltage references u*_α, u*_β in the α-β plane in fault-tolerant operation are derived as:

$\begin{array}{cc}\{\begin{array}{c}{u}_{\alpha }^{*}=u_{x\alpha }^{*}-u_{\mathrm{comp}\alpha }\\ u_{\beta }^{*}=u_{x\beta }^{*}-u_{\mathrm{comp}\beta }\end{array}\mathrm{and}{U}_s^{*}=\left[\begin{array}{c}{u}_{x\alpha }^{*}\\ u_{x\beta }^{*}\end{array}\right].& \left(28\right)\end{array}$

T_2s/4s is adopted to transform the actual voltage references u*_α, u*_β and the zero-sequence voltage reference u*_z to the references u*_B, u*_C, u*_D, u*_E in the natural frame, which can be expressed as:

$\begin{array}{cc}\left[\begin{array}{c}{u}_B^{*}\\ u_C^{*}\\ u_D^{*}\\ u_E^{*}\end{array}\right]=T_{2s/4s}\left[\begin{array}{c}{u}_{\alpha }^{*}\\ u_{\beta }^{*}\\ u_z^{*}\end{array}\right]& \left(29\right)\end{array}$ where the voltage reference of phase-A is set as u*_A=0, since an open-circuit fault occurs in phase-A.

On the basis of the traditional CPWM method with a sinusoidal modulation wave, the same flux control effect as the five-phase SVPWM method can be achieved when zero-sequence voltage harmonics c₀=max(u_i)+min(u_i))/2 is injected into five-phase sinusoidal modulation waves (u_i is five-phase sinusoidal modulating wave function for each phase). Hence, this invention performs the CPWM method with the injection of zero-sequence voltage harmonics.

By using the CPWM method with the injection of zero-sequence voltage harmonics, the phase voltage references u*_A, u*_B, u*_C, u*_D, u*_E (29) fed to voltage source inverter are used to realize FT-DTC disturbance-free operation when a five-phase fault-tolerant PMLM is under an open-circuit fault condition. The proposed FT-DTC strategy for a fault-tolerant PMLM with a one-phase open-circuit fault is shown in FIG. 3.

According to FIGS. 2 and 3, the simulation model of a five-phase fault-tolerant PMLM is established in Matlab/Simulink, and then simulated results are obtained when the five-phase fault-tolerant PMLM has a one-phase open-circuit fault.

FIG. 4 shows phase current waveforms without and with the proposed FT-DTC method in phase-A open-circuit fault operation. When a phase-A open-circuit fault occurs at 0.12 s, the waveforms of the phase currents are distorted. The proposed FT-DTC strategy in the present invention is injected at 0.18 s, and then the sinusoidal waveforms of currents are obtained. FIG. 5 shows thrust-force waveforms without and with the proposed FT-DTC method under phase-A open-circuit fault conditions. When a phase-A open-circuit fault occurs at 0.12 s, the fluctuations of the thrust-force are significant. The proposed FT-DTC strategy in the present invention is injected at 0.18 s, then the thrust-force pulsation is effectively suppressed, and there are almost no fluctuations. FIG. 6 shows waveforms of stator currents in the α-β plane without and with the proposed FT-DTC under phase-A open-circuit fault conditions. When a phase-A open-circuit occurs at 0.12 s, the current distortion is significant, and the proposed FT-DTC strategy in the present invention is injected at 0.18 s. Then the sinusoidal waveform of currents is obtained, and their distortion is significantly decreased. FIG. 7 shows waveforms of stator fluxes in the α-β plane without and with the proposed FT-DTC under phase-A open-circuit fault conditions. When a phase-A open-circuit fault occurs at 0.12 s, although the difference between the flux amplitudes in healthy and fault operation is small, the waveforms of stator fluxes in the α-β plane are distorted seriously. The proposed FT-DTC strategy in the present invention is applied at 0.18 s, and then the sinusoidal waveforms of stator fluxes in the α-β plane are almost the same as those in healthy operation. FIG. 8 shows flux trajectory in the α-β plane without and with the proposed FT-DTC strategy under phase-A open-circuit fault conditions. The stator flux trajectory is no longer circular until the proposed FT-DTC in the present invention is injected. FIGS. 9-11 show the responses of phase currents, thrust-force and stator current components in the α-β plane, respectively, when the thrust-force reference steps in healthy operation, and their response time is 1 ms. FIGS. 12-14 show responses of phase currents, thrust-force and stator current components in the α-β plane, respectively, when thrust-force reference steps in FT-DTC operation under a phase-A open-circuit fault, and their response time is still 1 ms. It can be concluded that the motor with the proposed FT-DTC in fault operation has the same dynamic performance as that in healthy operation when a phase-A open-circuit fault occurs. The thrust-force has almost no fluctuation both in healthy and fault operation, the phase currents also have almost no distortion. Therefore, fault-tolerant disturbance-free operation is realized. Meanwhile, the dynamic performance is not affected, which is the same as that in healthy operation.

If an open-circuit fault occurs in any other phase, the electrical angular difference between the phase and phase-A is ka. k=0, 1, 2, 3, 4. k=0 corresponds to the phase-A open-circuit fault. k=1 corresponds to the phase-B open-circuit fault. k=2 corresponds to the phase-C open-circuit fault. k=3 corresponds to the phase-D open-circuit fault. k=4 corresponds to the phase-E open-circuit fault. Then the natural frame needs to be counterclockwise rotated by 0.4 kπ electrical angle to ensure the axis of the faulty phase overlaps with the axis of phase-A in healthy operation. Hence, θ in the FT-DTC strategy should be replaced by θ−0.4 kπ, and then the stator compensation voltages are expressed as:

$\begin{array}{cc}\{\begin{array}{c}{u}_{\mathrm{comp}\alpha }=-0.5{\lambda }_m\mathrm{\omega sin}\left(\theta -0.4k\pi \right)\\ u_{\mathrm{comp}\beta }=0.053{\lambda }_m\omega \cos \left(\theta -0.4k\pi \right)\end{array}& \left(30\right)\end{array}$

It can be seen from the aforementioned statement that the five-phase PMLM with the proposed FT-DTC strategy in open-circuit fault operation can not only output the same thrust-force as that with DTC in healthy operation, but also restrain the thrust-force fluctuation significantly when the currents are less than the limited current. Last but not least, the PMLM with the proposed FT-DTC in fault operation has the same dynamic performance, steady-state performance and tracking accuracy of thrust-force as those in healthy operation. It is suitable for any other phase open-circuit fault, and there is no complicated calculation, and a low CPU overhead. Compared with field-oriented control strategy, the invention has a simple structure, small CPU overhead and faster dynamic response, and it has good application prospects in electromagnetic active suspension systems with high requirements on operational reliability and high dynamic performance.

Although the present invention has been disclosed as the above implemented example, the example is not used to limit the invention. Any equivalent change or modification within the spirit and field of the present invention belongs to the scope of the invention.

Claims

1. A fault-tolerant direct thrust-force control (FT-DTC) method for a five-phase permanent-magnet linear motor (PMLM) with a one-phase open-circuit fault, comprising the following steps: (1) establishing a five-phase PMLM model;(2) dividing windings of the five-phase PMLM into phase-A, phase-B, phase-C, phase-D, and phase-E, and when a phase-A open-circuit fault occurs, according to a principle wherein a traveling wave magneto-motive force (MMF) is constant both in a healthy operation and a fault operation, and according to a constraint wherein a sum of remaining healthy phase currents is zero, and amplitudes of the remaining healthy phase currents are equal, obtaining currents iB, iC, iD, iE in remaining healthy phases, and then, based on the currents, deriving a generalized Clarke transformation T4S/2S, wherein the generalized Clarke transformation is employed to transform variations in a natural frame into a two-phase stationary frame named as an α-β plane, and inverse transformation matrix T2s/4s of the generalized Clarke transformation;(3) according to a relationship between a stator flux and a permanent-magnet flux: ψs=Lsis+ψf, using T4s/2s to transform the stator fluxes ψB, ψC, ψD, ψE in the natural frame into the α-β plane and a zero-sequence subspace by the following equations:

2. The FT-DTC method of claim 1, wherein the step (2) further comprises: (2.1) a design object of the FT-DTC method is the five-phase PMLM, wherein the windings of the five-phase PMLM are star-connected and a center point of the five-phase PMLM is not connected to a DC bus, according to the principle wherein the traveling wave MMF remains constant both in the healthy operation and the fault operation, and the sum of the remaining healthy phase currents is zero, and the amplitudes of the remaining healthy phase currents are equal, the currents iB, iC, iD, iE in the remaining healthy phases under the phase-A open-circuit fault conditions are deduced as:

3. The FT-DTC method of claim 2, wherein the generalized Clarke transformation matrix T4s/2s and the inverse transformation matrix T2s/4s of the generalized Clarke transform matrix are deduced as respectively:

4. The FT-DTC method of claim 1, wherein the step (5) further comprises: (5.1) when the phase-A open-circuit fault occurs, a voltage model of the five-phase PMLM in the natural frame is:

5. The FT-DTC method of claim 4, wherein the relationship between the actual stator voltages and the virtual stator voltages is:

6. The FT-DTC method of claim 1, wherein the step (7) further comprises: (7.1) T4s/2s is adopted to transform the healthy phase currents from the natural frame into the α-β plane and a zero-sequence frame, by the following equation:

7. The FT-DTC method of claim 1, wherein if an open-circuit fault occurs in any other phase, an electrical angular difference between a faulty phase and the phase-A is ka, k=0,1,2,3,4, k=0 corresponds to the phase-A open-circuit fault; k=1 corresponds to a phase-B open-circuit fault; k=2 corresponds to a phase-C open-circuit fault; k=3 corresponds to a phase-D open-circuit fault; k=4 corresponds to a phase-E open-circuit fault, then the natural frame is rotated counterclockwise by 0.4 kπ electrical angle to ensure an axis of the faulty phase overlaps with an axis of the phase-A in the healthy operation, hence, θ in a FT-DTC strategy is replaced by θ−0.4 kπ, and then the stator compensatory voltages are changed as:

8. The FT-DTC method of claim 1, wherein the FT-DTC method is also suitable for a five-phase permanent-magnet rotating motor drive.