CN104158456B - A kind of position sensorless control method for driving motor for electric automobile - Google Patents

Abstract

The invention discloses a kind of position sensorless control method for driving motor for electric automobile, use pulse voltage injection method to driving original position of electric motor's rotator to carry out detection formula before electric automobile starts；After electric automobile starts, in order to detect accurately, the rotary high frequency signal injection method adopted at low speed segment detects driving motor rotor position in real time, and in order to avoid the impact of the multiple saliency of motor, the structure of high frequency signal injection method adds dual salient pole decoupling observer, use mode that high-frequency signal injection and counter electromotive force method combine during middling speed and process detection through nerve network controller and drive motor rotor position, counter electromotive force is used to detect driving motor rotor position during high speed, simultaneously for solving to cause observer to converge to the state of rotor magnetic pole opposite location at high speeds due to external interference, add rotor-position Robust Observers.The present invention can be accurate and effective detection driving motor for electric automobile rotor position information.

Description

A kind of position sensorless control method for driving motor for electric automobile

Technical field

The present invention relates to a kind of position sensorless control method for driving motor for electric automobile, pulse voltage is injected Method, rotation high frequency signal injection method, counter electromotive force method, nerve network controller, dual salient pole decoupling observer and rotor-position The position sensor technology that Robust Observers combines.

Background technology

The development of electric automobile is oil crisis and people's inevitable outcome to environmental requirement, compared with internal-combustion engines vehicle, Electric automobile is with vehicle power as power, uses motor-driven wheels travel, and meets each to automobile of road safety regulation The vehicle that item requires.Owing to permagnetic synchronous motor has high power density and the highest control performance so that it is become The first-selection of electric automobile.Permagnetic synchronous motor (PMSM) relies on its high operational efficiency and high power density, is widely used in On electric automobile.The most ripe vector control technology that surely belongs in current high performance Permanent-magnet Synchronous-motor Speed Servo System, but logical Often will be at electric machine main shaft side installation site sensor, in order to detect real time position and the velocity information of rotor.But position sensor Existence not only add the cost of whole system, also reduce the reliability of system, so Sensorless Control Technique Become the most important research direction.

The core of sensor is that the real time position of rotor and speed can be estimated by control system accurately, conventional without passing The control method of sensor can be divided into 3 classes:

(1) open loop using motor ideal model calculates method, such as direct computing method, counter electromotive force integration method etc.；Based on opening The computational methods of ring are simply direct, dynamic property is preferable；But rely on the parameter of electric machine when calculating, and motor runtime parameter is always in Among change, so will certainly affect the accuracy of rotor position estimate；And when motor speed is the lowest, counter electromotive force is very Little, together with easy and various interference signal is entrained in, signal to noise ratio step-down so that back-emf is difficult to detect.Institute is the most also Be not suitable for that motor is static or estimate without sensing station during low speed.

(2) the rotor-position identification scheme injected based on outside high-frequency signals, as rotated high frequency signal injection method, rotating height Frequently voltage injection method and rotation high frequency current injection.High Frequency Injection is by injecting high frequency letter to motor three-phase windings Number (voltage or current signal), relies on the saliency of rotor self or due to the saturated saliency caused, makes high frequency believe Number magnetic field produced, therefore will be with rotor position information in high-frequency signal by the modulating action of rotor with salient pole, then by high frequency Signal demodulates out the positional information that just can extract rotor from stator current or voltage.This method relies on additional sharp Encouraging signal, be not rely on rotating speed, but the time required for estimation rotor-position is longer, position quantity renewal frequency is the highest, so High Frequency Injection is motor is static and has during low speed and preferably estimates effect.

(3) closed loop algorithm based on state observer, such as sliding mode observer method (SMO), model reference adaptive systems approach (MRAS), extended Kalman filter method (EKF) etc..The essence of observer is exactly system mode reconstruct, i.e. reconfigures one System, utilizes the output vector that directly can measure in original system and input vector as its input signal, and makes reconstruct system The output signal of system is equivalent to the state of original system under certain conditions, and this system reconfigured is known as observer.

Above method all has the respective scope of application, does not also have a kind of method can make various permagnetic synchronous motor under full speed Can perfect stable operation.

Summary of the invention

Goal of the invention: in order to overcome the deficiencies in the prior art, the present invention provides a kind of and drives for electric automobile The position sensorless control method of motor, is divided into initial position detection, the detection of low speed segment position, medium velocity position full speed period Put detection and high regime position is detected, thus realize driving motor for electric automobile sensorless strategy in full speed range, can To select detection method and by pulse voltage injection method, rotation high frequency signal injection method, anti-electricity for different velocity shootings Kinetic potential method, nerve network controller, dual salient pole decoupling observer and rotor-position Robust Observers combine, Ke Yizhun Really, the effective rotor position information detecting driving motor for electric automobile.

Technical scheme: for achieving the above object, the technical solution used in the present invention is:

A kind of position sensorless control method for driving motor for electric automobile, including initial position detection, low speed segment Position detection, the detection of middling speed fragment position and the detection of high regime position, specifically include following steps:

(1) initial position detection: use pulse voltage injection method detection rotor-position, apply sky to the armature winding of motor Between voltage vector, utilize the difference of equivalent circuit time constant, by comparing the die-away time of response current, determine that rotor is initial Position, the voltage vector angle corresponding to the shortest die-away time is rotor initial angle；Final guarantee electric automobile realizes Start without reversion with torque capacity；

(2) low speed segment position detection: use rotary high frequency signal injection method to detect rotor-position in real time, pass through software locks Phase ring realizes the tracking of the phase place to negative phase-sequence high frequency electric, thus obtains azimuth error, uses pi regulator regulation to vow simultaneously The error of angulation is allowed to go to zero, and makes the estimated value of rotor position angleConverge on actual value θ_r, rightMake time diffusion, it is thus achieved that Rotor velocityIn order to avoid the impact of the multiple saliency of motor, the structure of high frequency signal injection method add dual Salient pole decoupling observer；

(3) middling speed fragment position detection, uses mode that high-frequency signal injection and counter electromotive force method combine and passes through nerve net Network controller processes and detects rotor-position in real time, high-frequency signal injection testing result, counter electromotive force method testing result and both inspections Surveying resultant error signal after nerve network controller processes uses pi regulator to be adjusted, and its output is as rotor position Confidence ceases；

(4) high regime position detection, uses counter electromotive force to detect rotor-position in real time, takes synovial membrane observer to obtain and turns Sub-positional information；In order to slacken the chattering phenomenon of synovial membrane observer, use saturation function to replace traditional switch function z, obtain Equivalent electromotive force, thus obtain rotor-position detected value；In order to solve may cause at high speeds synovial membrane due to external interference Observer converges to the state of rotor magnetic pole opposite location, adds rotor-position Robust Observers.

Beneficial effect: the position sensorless control method for driving motor for electric automobile that the present invention provides, by pulse Voltage injection method, rotate high frequency signal injection method, counter electromotive force method, nerve network controller, dual salient pole decoupling observer and Rotor-position Robust Observers combines, and has the advantage that 1, initial position detection uses pulse voltage injection method Mode, applies space voltage vector to the armature winding of driving motor for electric automobile, it is possible to detect electric automobile the most accurately Drive the initial position of rotor of motor, it is achieved the smooth starting of electric automobile；2, inspection can be selected for different velocity shootings Survey method, improves stability and the degree of accuracy of drive system of electric automobile；3, at low speed segment, use high-frequency signal injection, and join Close dual salient pole decoupling observer, effectively solve at low speed segment, hydraulic performance decline, the problems such as control accuracy is the highest；4, in, High section takes the mode that high-frequency signal injection and counter electromotive force combine, and is simultaneously introduced rotor-position Robust Observers and solves magnetic pole receipts The problem held back, effectively solves the bitter difficulty of high speed lower rotor part position detection and the problem of magnetic pole convergence；5, hardware one-tenth has been saved Originally with maintenance adult, improve anti-interference and the robustness of system simultaneously.

Accompanying drawing explanation

Fig. 1 is two groups of d, q windings schematic diagrams when diverse location；

Fig. 2 is space voltage vector figure；

Fig. 3 is initial position detection flow chart and operation figure；

Fig. 4 is rotary high frequency signal injection method schematic diagram；

Fig. 5 is the dual salient pole decoupling observer model figure for high frequency signal injection method；

Fig. 6 is the counter electromotive force detection method schematic diagram with sliding mode observer；

Fig. 7 is the rotor-position Robust Observers Observation principle figure for counter electromotive force method

Fig. 8 is that middling speed section drives motor rotor position Cleaning Principle figure.

Detailed description of the invention

Below in conjunction with the accompanying drawings the present invention is further described.

A kind of position sensorless control method for driving motor for electric automobile, including initial position detection, low speed segment Position detection, the detection of middling speed fragment position and the detection of high regime position, specifically include following steps:

(1) initial position detection: use pulse voltage injection method detection rotor-position, apply sky to the armature winding of motor Between voltage vector, utilize the difference of equivalent circuit time constant, by comparing the die-away time of response current, determine that rotor is initial Position, the voltage vector angle corresponding to the shortest die-away time is rotor initial angle；Final guarantee electric automobile realizes Start without reversion with torque capacity；

(2) low speed segment position detection: use rotary high frequency signal injection method to detect rotor-position in real time, pass through software locks Phase ring realizes the tracking of the phase place to negative phase-sequence high frequency electric, thus obtains azimuth error, uses pi regulator regulation to vow simultaneously The error of angulation is allowed to go to zero, and makes the estimated value of rotor position angleConverge on actual value θ_r, rightMake time diffusion, it is thus achieved that Rotor velocityIn order to avoid the impact of the multiple saliency of motor, the structure of high frequency signal injection method add dual Salient pole decoupling observer；

(3) middling speed fragment position detection, uses mode that high-frequency signal injection and counter electromotive force method combine and passes through nerve net Network controller processes and detects rotor-position in real time, high-frequency signal injection testing result, counter electromotive force method testing result and both inspections Surveying resultant error signal after nerve network controller processes uses pi regulator to be adjusted, and its output is as rotor position Confidence ceases；

(4) high regime position detection, uses counter electromotive force to detect rotor-position in real time, takes synovial membrane observer to obtain and turns Sub-positional information；In order to slacken the chattering phenomenon of synovial membrane observer, use saturation function to replace traditional switch function z, obtain Equivalent electromotive force, thus obtain rotor-position detected value；In order to solve may cause at high speeds synovial membrane due to external interference Observer converges to the state of rotor magnetic pole opposite location, adds rotor-position Robust Observers.

Below with regard to the present invention implement thought and process is illustrated.

Initial position detection

The Cleaning Principle of rotor-position is nonlinear magnetization characteristic based on stator core.As it is shown in figure 1, set rotor permanent magnet The magnetic linkage that body produces is ψ_f, direction and d¹Axle overlaps, then at d²Axial component is ψ_fcos△θ；Motor d axle magnetic under d, q axle Chain equation is:

ψ_d=L_di_d+ψ_f (1)

Wherein, θ is rotor position angle, L_dFor d axle inductance, i_dFor d shaft current.

Obtain according to formula (1):

ψ_d1=L_d1i_d1+ψ_f

(2)

ψ_d2=L_d2i_d2+ψ_fcosΔθ

Comparison expression (1) and formula (2) have:

ψ_d1>ψ_d2 (3)

As shown in Figure 1, because L_d1Winding flow direction is consistent with permanent magnetism magnetic pole direction, therefore works as L_d1、L_d2Electricity in winding When stream increases simultaneously, it is in d¹Axial magnetic circuit more tends to saturated, according to inductor saturation effect, it can be deduced that L_d1<L_d2.Root According to zero state response current formula:

I (t)=U [1-e^-(R/L)t]/R (4)

Wherein, U is the voltage magnitude applied, and L, R are self-induction and the resistance of stator winding.

Then have:

i_d1>i_d2 (5)

By above-mentioned analysis it can be concluded that when the synthesis magnetic linkage that 2 groups of identical windings produce is equal, magnetic flux side The highest to winding equivalent inductance saturation immediate with rotor magnetic pole, its inductance value is minimum, and electric current is maximum, therefore can pass through The amplitude size of current-responsive produced by detection potential pulse, determines initial position of rotor, but the method needs detection electricity Stream peak value, requires higher to sample circuit, and sample frequency also influences whether its accuracy judged, patent of the present invention is this it On, utilize the characteristic that under different voltage vector, d axle equivalent circuit time constant is different, decay to 0 by detection response current Time is different, it is judged that the initial position of rotor, it is not necessary to detect current peak, decreases the dependency to sample circuit, Concrete principle is as follows:

When driving motor for electric automobile PMSM is static, as the potential pulse direction being passed through and d¹When direction is equal, L=L_d1, Then the time constant of circuit is τ_d1；As the potential pulse direction passed through and d²Time identical, L=L_d2, then the time constant of circuit is τ_d2.τ is understood according to inductor saturation effect_d1<τ_d2, then t_d1<t_d2.Needed for response current produced by two potential pulses decays to 0 The time wanted is respectively t_d1、t_d2, by comparing the size of both times, it is known that the initial position of rotor is closer to d¹.According to this Principle, shown in Fig. 2, is passed through 12 different voltage vectors in direction, voltage vector equidirectional with rotor magnetic pole N in order (when the voltage vector angle being i.e. passed through is rotor actual angle), corresponding circuit time constant τ is minimum, then t is minimum.

Therefore, it can decay to the time of 0 by comparing d shaft current under constant voltage vector effect, it is judged that rotor Initial position, minimal time value t of measurement_{d_min}The angle of corresponding voltage vector is the initial angle of rotor.

According to above-mentioned former topic, its concrete operating process as it is shown on figure 3, particularly as follows:

Whole detection process can be divided into two steps, and Fig. 3 is the flow chart of detection process, wherein wherein in n representative graph 2 Individual space voltage vector, θ_nIt it is the angle of space voltage vector n.

The first step, applies 12 different voltage vectors according to the order (1 → 2 → 3 →...→ 12) of Fig. 2 (a) to motor, And detect that d shaft current decays to the time t of 0 from steady-state value.Along with voltage vector moves closer to rotor N pole, then because of magnetic Saturated phenomenon, die-away time, t can be gradually reduced.Finally, t_{d_min}Corresponding voltage vector angular range is rotor initial angle Degree scope.

Second step, first with θ_M2As the intermediate value of the voltage vector angular range tentatively judged, on the basis of the first step On, applying angle to motor respectively is θ_M2-△ θ, θ_M2And θ_M2Tri-kinds of voltage vectors of+△ θ, the initial value of △ θ is 7.5 °, therefore Second step is applied to shown in voltage vector such as Fig. 2 (b) of motor.Finally, t'_{d_min}Corresponding voltage vector angular range is i.e. For precision higher rotor initial angle scope.

Subsequent to obtain rotor-position the most accurately, with θ '_M2As precision higher voltage vector angular range Intermediate value, repeat second step, be applied to shown in voltage vector such as Fig. 2 (c) of motor, according to same angle detecting make new advances turn Sub-position angle θ "_M2, finally give precision and meet the rotor initial angle of requirement.

Low speed segment position is detected

After electric automobile starts, as shown in Figure 4, if the angular frequency rotating high-frequency voltage signal is ω_i, amplitude be v_si, Then rotate high-frequency voltage signalIt is expressed as:

$v_{\mathrm{qdsi}}^s=\left[\begin{array}{c}{v}_{\mathrm{qsi}}^s\\ v_{\mathrm{dsi}}^s\end{array}\right]=v_{\mathrm{si}}\left[\begin{array}{c}\cos \left({\mathrm{\&omega;}}_{i}t\right)\\ -\sin \left({\mathrm{\&omega;}}_{i}t\right)\end{array}\right]=v_{\mathrm{si}}{e}^{j{\mathrm{\&omega;}}_{i}t}---\left(6\right)$

Wherein:Q axle component for high frequency voltage；D axle component for high frequency voltage.

The DC response rotating the three-phase inverter outfan motor under high-frequency voltage signal excitation isWillPass through After band filter BPF filtering, obtain dq axle high frequency electricFor:

$i_{\mathrm{qdi}}=i_{\mathrm{dqs}\_\mathrm{ip}}^s+i_{\mathrm{dqs}\_\mathrm{in}}^s=i_{\mathrm{ip}}{e}^{j({\mathrm{\&theta;}}_t\left(t\right)-\mathrm{\&pi;}/2)}+i_{\mathrm{in}}{e}^{j({2\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/2)}---\left(7\right)$

Wherein: $i_{\mathrm{dqs}\_\mathrm{ip}}^s=i_{\mathrm{ip}}{e}^{j({\mathrm{\&theta;}}_t\left(t\right)-\mathrm{\&pi;}/2)}$ For positive-phase-sequence current component； $i_{\mathrm{dqs}\_\mathrm{in}}^s=i_{\mathrm{in}}{e}^{j({2\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/2)}$ For negative-phase sequence electricity Flow component；i_ipFor positive-phase-sequence current DC component；i_inFor negative phase sequence current DC component；θ_tT () is mover speed；θ_r1For note The rotor position angle reflected after entering high frequency voltage, i.e. the position angle of mover during low speed segment position；Owing to only negative phase sequence current divides AmountComprise position angle information θ of mover_r1, therefore first pass through high pass filter SFF by positive-phase-sequence current componentFilter Remove, then allow negative phase sequence current componentFirst it is multiplied byObtainAfter, then be multiplied byAfter, azimuth error ε can be obtained For:

Wherein:For negative phase sequence current q axle component；For negative phase sequence current d axle component；During for low speed segment position The estimated value of rotor position, θ_r1For the actual value of mover during low speed segment position；Use pi regulator regulation azimuth simultaneously Error is allowed to go to zero so thatConverge on actual value θ_r1, rightMake time diffusion to obtain mover angular velocity

In formula (7), only to only account for depending on the space of rotor structure convex for the mathematical model of driving motor for electric automobile Pole.And in practice, motor has multiple salient pole, including the non-linear produced multiple salient pole of rotor, stator and inverter And the saturated salient pole caused.

The driving motor for electric automobile multiple salient pole under rotating the injection that high-frequency voltage signal is good can be by electric current Complex vector and expression, i.e. formula (7) can be write as:

$i_{\mathrm{qdi}}=i_{\mathrm{dqs}\_\mathrm{ip}}^s+i_{\mathrm{dqs}\_\mathrm{in}}^s=i_{\mathrm{ip}\_1}{e}^{j({\mathrm{\&theta;}}_t\left(t\right)-\mathrm{\&pi;}/2)}+\underset{k}{\mathrm{\&Sigma;}}{i}_{\mathrm{in}\_2k}{e}^{j({2\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/2)}---\left(9\right)$

Wherein: i_{ip_1}≈i_ip；i_{in_2}≈i_in。

Salient pole during k=0 is that the negative phase-sequence carrier wave caused by the unsymmetry by motor dissymmetrical structure and current measurement is sat DC offset in mark system；As k=1, space salient pole componentIt is by the inductance difference of d axle and q axle Caused.Magnetic saturation impact under other salient poles (k=± 1, ± 2, ± 3, ± 4) hands loading condition, the most saturated causes Salient pole, verifies through test of many times, obtains electric current complex vector that driving motor for electric automobile causes by multiple salient pole and is similar to:

$\begin{array}{c}{i}_{\mathrm{qdi}}=i_{\mathrm{ip}\_1}{e}^{j({\mathrm{\&theta;}}_t\left(t\right)-\mathrm{\&pi;}/2)}+i_{\mathrm{in}\_0}{e}^{j({2\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/2)+}\\ i_{\mathrm{in}\_2}{e}^{j({2\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/4)}+i_{\mathrm{in}\_-2}{e}^{j({-2\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/4)}+\\ i_{\mathrm{in}\_4}{e}^{j({4\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/8)}+i_{\mathrm{in}\_-4}{e}^{j({-4\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/8)}+\\ i_{\mathrm{in}\_6}{e}^{j({2\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/12)}+i_{\mathrm{in}\_8}{e}^{j({2\mathrm{\&theta;}}_{r1}-{\mathrm{\&theta;}}_t\left(t\right)+\mathrm{\&pi;}/24)}\end{array}---\left(10\right)$

The 1st of formula (10) is the positive-sequence component of high frequency electric, and the 2nd is as k=0, the static salient pole of motor, the 3rd Item is as k=1, by the electric current negative sequence component caused by the inductance difference of d axle and q axle.It is k respectively from the 4th to the 8th =-1, affected by the magnetic saturation under loading condition when k=± 2, k=3 and k=4, by the salient pole of saturated introduction, remaining salient pole Because amplitude is the least, its impact can be left in the basket, to driving the method that can be taked decoupling by saturated other salient poles caused in motor, Fig. 5 is to the 6th the tracking observer schematic diagram decoupled in formula (10) under rest frame.In like manner, it is also possible to use Other are decoupled by same method.

High regime position is detected

When being in electric automobile high regime, sliding mode observer is used to obtain driving motor rotor position information, at d-q The voltage equation driving motor in rotating coordinate system is:

$\left[\begin{array}{c}{u}_d\\ u_q\end{array}\right]=\left[\begin{array}{cc}R+{\mathrm{DL}}_d& {-w}_r{L}_d\\ w_r{L}_d& R+{\mathrm{DL}}_q\end{array}\right]\left[\begin{array}{c}{i}_d\\ i_q\end{array}\right]+\left[\begin{array}{c}0\\ w_r{K}_E\end{array}\right]---\left(10\right)$

Wherein, [u_d u_q]^TFor voltage under rotating coordinate system；[i_d i_q]^TFor electric current under rotating coordinate system；R is stator resistance； D is differential operator；w_rFor rotor velocity (electrical angle)；K_EFor back electromotive-force constant；L_dFor d axle inductance；L_qFor q axle inductance.

Formula (11) is transformed under alpha-beta rest frame, obtains:

$\left[\begin{array}{c}{u}_{\mathrm{\&alpha;}}\\ u_{\mathrm{\&beta;}}\end{array}\right]=\left[\begin{array}{cc}R+{\mathrm{DL}}_{\mathrm{\&alpha;}}& {-w}_r{L}_{\mathrm{\&alpha;\&beta;}}\\ w_r{L}_{\mathrm{\&alpha;\&beta;}}& R+{\mathrm{DL}}_{\mathrm{\&beta;}}\end{array}\right]\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}\end{array}\right]+w_r{K}_E\left[\begin{array}{c}-\sin {\mathrm{\&theta;}}_{r2}\\ \cos {\mathrm{\&theta;}}_{r2}\end{array}\right]---\left(12\right)$

Wherein, [u_α u_β]^TFor voltage under rotating coordinate system；[i_α i_β]^TFor electric current under rotating coordinate system；L_α=L_o+L₁cos2 θ_r2；L_αβ=L₁sin2θ_r2；L_β=L_o-L₁cos2θ；L_o=(L_d+L_q)/2；L₁=(L_d-L_q)/2；θ_r2For electric automobile in fortune at a high speed PMSM position angle during row.

Including θ, 2 θ items in formula (11), wherein 2 θ will bring the biggest difficulty to the calculating in later stage, therefore, it can pass through Suitable conversion makes it eliminate, it can be seen that the main cause of the asymmetric appearance being 2 θ of inductance matrix from formula (12), because of And, the voltage equation (11) driving motor under d-q axle is rewritten as:

$\left[\begin{array}{c}{u}_d\\ u_q\end{array}\right]=\left[\begin{array}{cc}R+{\mathrm{DL}}_d& {-w}_r{L}_q\\ w_r{L}_q& R+{\mathrm{DL}}_d\end{array}\right]\left[\begin{array}{c}{i}_d\\ i_q\end{array}\right]\left[\begin{array}{c}0\\ w_r{K}_E+(L_d-L_q)(w_r{i}_d-{\mathrm{di}}_q/\mathrm{dt})\end{array}\right]---\left(13\right)$

Formula (13) transforms under alpha-beta rest frame:

$\left[\begin{array}{c}{u}_{\mathrm{\&alpha;}}\\ u_{\mathrm{\&beta;}}\end{array}\right]=\left[\begin{array}{cc}R+{\mathrm{DL}}_d& w_r\left(L_d{-L}_q\right)\\ {-w}_r(L_d-L_q)& R+{\mathrm{DL}}_d\end{array}\right]\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}\end{array}\right]+\left[(w_r{K}_E+(L_d-L_q)(w_r{i}_d-\frac{{\mathrm{di}}_q}{\mathrm{dt}})\right]\left[\begin{array}{c}-\sin {\mathrm{\&theta;}}_{r2}\\ \cos {\mathrm{\&theta;}}_2\end{array}\right]---\left(14\right)$

For the ease of using synovial membrane observer that counter electromotive force is observed, voltage equation (11) is rewritten into the shape of electric current State equation form:

$\frac{d}{\mathrm{dt}}\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}\end{array}\right]=A\&CenterDot;\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}\end{array}\right]+\frac{1}{L_d}\left[\begin{array}{c}{u}_{\mathrm{\&alpha;}}\\ u_{\mathrm{\&beta;}}\end{array}\right]+\frac{E}{L_d}\left[\begin{array}{c}\sin {\mathrm{\&theta;}}_{r2}\\ -{\cos \mathrm{\&theta;}}_{r2}\end{array}\right]---\left(15\right)$

Wherein,

$A=\frac{1}{L_d}\left[\begin{array}{cc}-R& {-\mathrm{\&omega;}}_r(L_d-L_q)\\ {\mathrm{\&omega;}}_r(L_d-L_q)& -R\end{array}\right]---\left(16\right)$

Wherein, ${\left[\begin{array}{cc}{\hat{i}}_{\mathrm{\&alpha;}}& {\hat{i}}_{\mathrm{\&beta;}}\end{array}\right]}^T$ For stator α and β shaft current observation；Z_αβ=[k/L_dsinθ_r2,-k/L_dcosθ_r2], k is for sliding Mould gain.

Formula (16) subtracts formula (15), and the state equation obtaining electric current observation error is:

$\frac{d}{t}\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}-{\hat{i}}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}-{\hat{i}}_{\mathrm{\&beta;}}\end{array}\right]=A\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}-{\hat{i}}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}-{\hat{i}}_{\mathrm{\&beta;}}\end{array}\right]-\frac{1}{L_d}(Z-E)---\left(17\right)$

When completely following condition, sliding mode observer entrance sliding formwork state:

$[i_{\mathrm{\&alpha;}}-{\hat{i}}_{\mathrm{\&alpha;}},i_{\mathrm{\&beta;}}-{\hat{i}}_{\mathrm{\&beta;}}]\frac{d}{t}\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}-{\hat{i}}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}-{\hat{i}}_{\mathrm{\&beta;}}\end{array}\right]<0---\left(18\right)$

If gain k is sufficiently large for sliding formwork, inequality (18) is set up, and system enters synovial membrane state, has:

$\frac{d}{t}\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}-{\hat{i}}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}-{\hat{i}}_{\mathrm{\&beta;}}\end{array}\right]=\left[\begin{array}{c}{i}_{\mathrm{\&alpha;}}-{\hat{i}}_{\mathrm{\&alpha;}}\\ i_{\mathrm{\&beta;}}-{\hat{i}}_{\mathrm{\&beta;}}\end{array}\right]=0---\left(19\right)$

Bring formula (19) into formula (17), obtain:

Z=E (20)

Z wherein includes discontinuous high-frequency signal, therefore for removing discontinuous high-frequency signal, is passed into low-pass filtering Equivalence control amount is obtained after device, it may be assumed that

$\left[\begin{array}{c}{Z}_{\mathrm{\&alpha;}}\\ Z_{\mathrm{\&beta;}}\end{array}\right]=\left[\begin{array}{c}{E}_{\mathrm{\&alpha;}}\\ E_{\mathrm{\&beta;}}\end{array}\right]=\left[(w_r{K}_E+(L_d-L_q)(w_r{i}_d-\frac{{\mathrm{di}}_q}{\mathrm{dt}})\right]\left[\begin{array}{c}-\sin {\mathrm{\&theta;}}_{r2}\\ \cos {\mathrm{\&theta;}}_{r2}\end{array}\right]---\left(21\right)$

By formula (21), can obtain driving the motor rotor position angle θ when high-speed cruising_r:

${\widehat{\mathrm{\&theta;}}}_{r2}=\arctan (-\frac{E_{\mathrm{\&alpha;}}}{E_{\mathrm{\&beta;}}})---\left(22\right)$

In order to reduce observation error, the error producing low accepter delayed phase compensates, and offset is:

${\widehat{\mathrm{\&theta;}}}_{\mathrm{re}}=\arctan \left(\frac{w_r}{w_{\mathrm{cutoff}}}\right)---\left(23\right)$

Wherein, w_cutoff=1/ τ₀It is the cut-off frequency of low accepter, τ₀It it is the time constant of low pass filter.

In order to prevent electric automobile when high-speed cruising, due to wind speed resistance, rotor-position observation is restrained Situation, in order to solve this problem, patent of the present invention adds rotor-position Robust Observers, and its concrete operation principle is as follows:

Detection θ_r2For T₁The position angle that moment is detected,For T₂Position angle detected by moment, and T₂-T₁=20ms, If position error signalAnd using ξ as input, and add matrix feedforward input, according to the machinery driving motor Motion model, can the equivalent structure figure of the rotor-position Robust Observers shown in structural map 7.Position error signal is by linear anti- Feedback structural regime observation, thus realize the observation to rotor-position.The second-order differential item of electromagnetic torqueRegard observer as Equivalent inpnt, such that it is able to take into account the wind speed disturbance situation that different lower the produced rate of change of wind speed is different, makes observer have enough Hole disturb ability.

The machine performance equation of Fig. 7 top half is represented by:

$\stackrel{\&CenterDot;}{X}=\mathrm{AX}+\mathrm{Bu}---\left(24\right)$

Y=CX (25)

Wherein: $\stackrel{\&CenterDot;}{X}={\left[\begin{array}{cccc}{\stackrel{\&CenterDot;}{T}}_e& T_e& w_r& {\mathrm{\&theta;}}_r\end{array}\right]}^T;u={\stackrel{\&CenterDot;\&CenterDot;}{T}}_e;$ Y=θ_r； $A=\left[\begin{array}{cccc}0& 0& 0& 0\\ 1& 0& 0& 0\\ 0& \frac{1}{J}& 0& 0\\ 0& 0& P_n& 0\end{array}\begin{array}{}\end{array}\right];B=\left[\begin{array}{c}1\\ 0\\ 0\\ 0\end{array}\right];C=\left[\begin{array}{c}0\\ 0\\ 0\\ 1\end{array}\right];$ P_nNumber of pole-pairs；J is rotary inertia.

The state equation of rotor-position observer can be expressed as:

$\stackrel{\stackrel{\&CenterDot;}{^}}{X}=\left[\begin{array}{cccc}0& 0& 0& 0\\ 1& 0& 0& 0\\ 0& \frac{1}{\hat{J}}& 0& 0\\ 0& 0& P_n& 0\end{array}\begin{array}{}\end{array}\right]\hat{X}=\left[\begin{array}{c}1\\ 0\\ 0\\ 0\end{array}\right]u+\left[\begin{array}{c}{l}_1\\ l_2\\ \frac{l_3}{\hat{J}}\\ \frac{l_4}{\hat{J}}\end{array}\right](y-\hat{y})---\left(26\right)$

In definition (26) $\left[\begin{array}{cccc}{l}_1& l_2& \frac{l_3}{\hat{J}}& \frac{l_4}{\hat{J}}\end{array}\right]$ For feedback matrix.

According to the structure of the position detection device shown in Fig. 7, the transmission letter of position detection error and perturbing torque can be set up Number relational expression:

${\mathrm{\&Delta;\&theta;}}_{r2}\left(s\right)={\widehat{\mathrm{\&theta;}}}_{r2}\left(s\right)-{\mathrm{\&theta;}}_{r2}\left(s\right)=\frac{-\frac{\hat{J}}{J}{s}^2}{\hat{J}{s}^4+l_4{s}^3+l_3{s}^2+l_{2}s+l_1}{T}_d\left(s\right)---\left(27\right)$

In formula, T_dS () is the perturbing torque caused because of wind speed.

From formula (27), when causing wind speed torque generation continuous print to change because of wind speed, the disturbance of observer Steady-state error is 0, observer hole load disturbance ability is available effective to be improved, it is to avoid electric automobile when high-speed cruising, The torque disturbance produced because wind speed is excessive, so that observer error increases the problem causing observation to converge to S pole.

Middling speed fragment position detects

When being in electric automobile middling speed section, mode process that employing high-frequency signal injection and counter electromotive force method combine are refreshing Processing detection through network controller and drive motor rotor position, Speed Controller of Networks requires reaction high-frequency signal injection detection ResultCounter electromotive force method testing resultAnd both testing result error effects to controller；Above signal is as god Input ε through network controller_r, after nerve network controller processes, its outfan is driving motor for electric automobile Positional informationAs shown in Figure 8, the work of its nerve network controller is former as follows for whole system block diagram.

Design of Neural Network Controller uses 3 layer networks: input layer, hidden layer and output layer.Input layer has 3 input quantitiesε_r；Hidden layer has 6 neurons；Output layer is

Input layer: be made up of 3 neurons

$\begin{array}{cc}{\mathrm{\&sigma;}}_1={\widehat{\mathrm{\&theta;}}}_{r1}& {\mathrm{\&sigma;}}_2={\widehat{\mathrm{\&theta;}}}_{r2}\end{array}$ σ₃=ε_r

o_i(t)=σ_iI=1,2,3

$n_{2j}\left(t\right)=\underset{i=1}{\overset{6}{\mathrm{\&Sigma;}}}{w}_{\mathrm{ji}}{o}_i\left(t\right)+{\mathrm{\&theta;}}_{2j},i=\mathrm{1,2,3}$

Hidden layer: be made up of 6 neurons

o_2j(t)=f₁[n_2j(t)] j=1,2 ..., 6

$n_3\left(t\right)=\underset{j=1}{\overset{6}{\mathrm{\&Sigma;}}}{w}_{3j}{o}_{2j}\left(t\right)+{\mathrm{\&theta;}}_3,j=\mathrm{1,2},...,6$

Output layer: be made up of 1 neuron

o₃(t)=f₂[n₃(t)]

Select different output functions can strengthen the mapping function of network, and improve network convergence speed.Hidden layer Output function is logsigmoid function, and the output function of output layer is trnsig moid function.

$f_1\left(x\right)=\log \mathrm{sig}\left(x\right)=\frac{1}{1+e^{-x}}$

$f_2\left(x\right)=\tan \mathrm{sig}\left(x\right)=\frac{1-e^{-2x}}{1+e^{-2x}}$

Through above process processing, final inspection to driving motor for electric automobile rotor is at the positional information of middling speed section

The above is only the preferred embodiment of the present invention, it should be pointed out that: for the ordinary skill people of the art For Yuan, under the premise without departing from the principles of the invention, it is also possible to make some improvements and modifications, these improvements and modifications also should It is considered as protection scope of the present invention.

Claims (2)

1. the position sensorless control method for driving motor for electric automobile, it is characterised in that: include that initial position is examined Survey, the detection of low speed segment position, the detection of middling speed fragment position and the detection of high regime position, specifically include following steps:

(1) initial position detection: use pulse voltage injection method detection rotor-position, apply space electricity to the armature winding of motor Pressure vector, utilizes the difference of equivalent circuit time constant, by comparing the die-away time of response current, determines rotor initial bit Putting, the voltage vector angle corresponding to the shortest die-away time is rotor initial angle；Finally ensure that electric automobile realizes nothing Invert and start with torque capacity；

(2) low speed segment position detection: use rotary high frequency signal injection method to detect rotor-position in real time, pass through software phase-lock loop Realize the tracking of the phase place to negative phase-sequence high frequency electric, thus obtain azimuth error, use pi regulator regulation azimuth simultaneously Error be allowed to go to zero, make the estimated value of rotor position angleConverge on actual value θ_r, rightMake time diffusion, it is thus achieved that rotor Angular velocityIn order to avoid the impact of the multiple saliency of motor, the structure of high frequency signal injection method adds dual salient pole Decoupling observer；

(3) middling speed fragment position detection, uses mode that high-frequency signal injection and counter electromotive force method combine and passes through neutral net control Device processed processes and detects rotor-position in real time, high-frequency signal injection testing result, counter electromotive force method testing result and both detect knot Really error signal after nerve network controller processes uses pi regulator to be adjusted, and its output is believed as rotor-position Breath；

(4) high regime position detection, uses counter electromotive force to detect rotor-position in real time, takes synovial membrane observer to obtain rotor position Confidence ceases；In order to slacken the chattering phenomenon of synovial membrane observer, use saturation function to replace traditional switch function z, obtain equivalence Electromotive force, thus obtain rotor-position detected value；In order to solve to cause synovial membrane to be observed at high speeds due to external interference Device converges to the state of rotor magnetic pole opposite location, adds rotor-position Robust Observers.

Position sensorless control method for driving motor for electric automobile the most according to claim 1, it is characterised in that: In described step (1), initial position detection implements process and is: (11) first apply to the armature winding of motor every 30 ° One space voltage vector, applies 12 different voltage vectors altogether；(12) then for each voltage vector, d is detected Shaft current decays to the time t of 0 from steady-state value, and the time t of note minimum is t_{d_min}, t_{d_min}Corresponding voltage vector angle model Enclose and be rotor initial angle scope；(13) the rotor initial angle scope obtained is carried out n to divide equally, every an equal score value to The armature winding of motor applies a space voltage vector, altogether applies n+1 different voltage vector；Repeat step (12), Until the rotor initial angle in acquisition accuracy rating.