CN101488031A - High-precision magnetic bearing axial control method based on interference observer - Google Patents

Abstract

The invention provides a method for controlling axial direction of a high-precision magnetic bearing based on an interference observer. The method consists of a controller and the interference observer, wherein the interference observer comprises a Q filter and a rationalization generalized object inverse QGn<-1> part; the controller in a control system calculates according to displacement deviation to obtain basic control quantity so as to form a position closed loop control system; an interference value obtained by observation of the interference observer is negatively fed back into the basic control quantity to compensate exterior interference; and the formed current control quantity drives power amplification to realize high precision suspension of the magnetic bearing. The method leads differences caused by both exterior interference and objective parameter variation to be equivalent to a control input end, and introduces equivalent compensation into the control quantity to realize interference inhabitation. The method can carry out online observation and effective inhabitation against exterior interference which is not modeled or known, thereby improving control precision of suspension, and contributing to stability of the system.

Description

A kind of high-precision magnetic bearing axial control method based on interference observer

Technical field

The present invention is a kind of magnetic bearing axial control method, relates to disturbance observation and inhibition technology in the High Accuracy Control, for external disturbance online observation and inhibition automatically, can be used for the High Accuracy Control and the Disturbance Rejection of magnetic bearing in the suspension control moment gyro system.

Background technology

(Control Moment Gyroscope CMG) is the crucial topworks that spacecraft carries out attitude control to control-moment gyro.The supporting of CMG high speed magnetic bearing is a critical component, and mechanical ball bearing and magnetic bearing dual mode are arranged usually.The magnetic suspension bearing mode has solved wearing and tearing and vibration problem that mechanical support brings, has long-life advantage, and allows to increase substantially the magnetic bearing rotating speed, can significantly dwindle the volume of CMG under same angular momentum prerequisite.Simultaneously, control the support accuracy that can improve magnetic bearing, support magnetic bearing in the relatively-stationary position of rotor case by active vibration.But magnetic levitation is a kind of gapped elastic bearing mode, and the shortcoming of magnetic bearing is will inevitably produce under the perturbation action displacement of transient state or stable state, has influenced the precision that suspends.Therefore, must keep under the stable prerequisite, realize the minimization of magnetic bearing displacement disturbance response reaching the purpose that improves magnetic bearing suspension precision by disturbance compensation.The disturbance that influences precision mainly contains can survey and can not survey two kinds of forms.For surveying disturbance, can directly carry out feedforward compensation by detection signal.For can not survey, probabilistic disturbance, then need online observation is carried out in the disturbance of system, extract disturbance information, and then realize compensation, to improve precision.

At present, mainly contain two classes at the method for magnetic bearing High Accuracy Control, the one, optimal controller allows disturbance response function minimization, comprises sliding mode variable structure control method etc.; The 2nd, equivalent compensation is carried out in disturbance, comprise feed forward control method.Sliding mode variable structure control method solved the interference that the amount of unbalance of system brings, but calculated amount is very big, has limited the application in real system.Feed forward control method carries out feedforward compensation by direct detection signal, has solved deterministic, measurable disturbance, but is the compensation to the modeling disturbance.The greatest problem of these methods be not at can not survey, uncertain disturbance carries out online observation and compensation, for example carrier disturbance, special Disturbance Rejection is not carried out in outside noise interference etc., so can not these disturbances that influence precision be suppressed.The disturbance situation that can not real-time analysis be subjected to of these methods simultaneously also lacks the means of corresponding disturbance record and quantitative test.

Summary of the invention

Technology of the present invention is dealt with problems: propose method uncertain at the outside in a kind of magnetic suspension bearing High Accuracy Control, that unknown disturbance carries out online observation, disturbance suppression, realize the effective inhibition to disturbing, improved the suspension precision of magnetic bearing.

Technical solution of the present invention: a kind of high-precision magnetic bearing axial control method based on interference observer, realize that there are frequency sweep circuit and digital control hardware in the system of this method, digital control hardware has the A/D module, DSP module and FPGA module; Frequency sweep circuit is passed to the A/D module after with the stack of sensor signal and pumping signal, the FPGA module receives the digital quantity of A/D module converts, send to the DSP module again, the DSP module software comprises that interference observer algorithm and controller K algorithm computation obtain the Current Control amount, be sent to the FPGA module subsequently, the FPGA module converts the Current Control amount output of to PWM form then, drives power amplifier generation electromagnetic force and acts on the magnetic bearing, and the inventive method specifically may further comprise the steps:

(a) at first add frequency sweep circuit, do the frequency sweep experiment respectively for power amplifier and the broad sense controlled device that comprises power amplifier, obtain the broad sense object parameters, set up the contrary Gn-1 transport function of generalized object, remove the frequency sweep circuit part then, sensor output is directly inserted the A/D module;

(b) interference observer in the initialization DSP module and controller K parameter are provided with the sampled data storage space, and the sampling pattern of setting the FPGA module is that clock interrupts;

(c) FPGA module controls A/D module samples obtains sensor output, and receives the digital quantity result that the A/D module converts obtains;

(d) the FPGA module is sent to the DSP module with digital quantity, according to given suspension center, calculates the offset deviation amount of input displacement value correspondence in the DSP module;

(e) the controller K in the DSP module is input as the offset deviation amount, and controller K adopts controller algorithm to calculate the basic controlling amount;

(f) interference observer in the DSP module is made of Q wave filter and the contrary QGn-1 of rationalization generalized object, interference observer comprises Current Control amount and two inputs of displacement, wherein, Current Control amount input Q wave filter calculates, the contrary QGn-1 of displacement input rationalization generalized object calculates, and above-mentioned both result calculated are kept to the Interference Estimation amount mutually;

(g) utilize Interference Estimation amount and basic controlling amount to obtain the Current Control amount in the DSP module, then the Current Control amount is passed to the FPGA module, form the PWM waveform in the FPGA module, the control power amplifier produces electromagnetic force, realizes magnetic bearing high precision stable suspersion.

The contrary G of the generalized object in the described step (a) _n ^-1Transport function is:

$G_n^{-1}\left(s\right)=\frac{{\mathrm{ms}}^2-k_h}{k_i{k}_w{k}_s}$

Wherein: k _wBe the power amplifier coefficient, k _sBe transducer sensitivity, m is the quality of magnetic bearing, k _hBe the displacement rigidity of magnetic bearing, k _iCurrent stiffness for magnetic bearing.

Q wave filter in the described step (f) is inferred-zero, the limit number low-pass filter than broad sense controlled device exponent number higher order, and its transport function is:

$Q\left(s\right)=\frac{1}{{\left(\mathrm{\&tau;s}\right)}^3+3{\left(\mathrm{\&tau;s}\right)}^2+3\mathrm{\&tau;s}+1}$

Wherein, the Laplace transform form of Q (s) expression Q wave filter, s is a Laplace operator, τ is the Q filter parameter, can adjust the Q filter cutoff frequency.

The contrary QG of rationalization generalized object in the described step (f) in the interference observer _n ^-1Contrary G for generalized object _n ^-1Product with the Q wave filter.

Frequency sweep circuit in the described step (a) is made up of reverse totalizer and one-level reverser, and oppositely two of totalizer are input as sensor output and pumping signal, and pumping signal is a sinusoidal signal.

Principle of the present invention is: external disturbance and object parameters are changed the difference that causes, and all equivalence promptly observes equivalence and disturbs to the control input end, introduces the compensation of equivalent in control, realizes suppressing disturbing fully.Simultaneously, the broad sense object parameters changes the model difference cause and also is observed and is fed back to the control input end, helps to eliminate non-linear in the certain limit like this.

As shown in Figure 7, the transport function of input end u that can obtain in the broad sense controlled device and broad sense controlled device output terminal y is as follows:

$G_{\mathrm{uy}}\left(s\right)=\frac{Y\left(s\right)}{U\left(s\right)}=\frac{G\left(s\right)G_n\left(s\right)}{G_n\left(s\right)+Q\left(s\right)(G\left(s\right)-G_n\left(s\right))}$

Interference d that obtains from Fig. 7 in input and the transport function of broad sense controlled device output terminal y are as follows:

$G_{\mathrm{dy}}\left(s\right)=\frac{Y\left(s\right)}{D\left(s\right)}=\frac{G\left(s\right)G_n\left(s\right)(1-Q\left(s\right))}{G_n\left(s\right)+Q\left(s\right)(G\left(s\right)-G_n\left(s\right))}$

Wherein, the Laplace transform form of the output y of Y (s) expression broad sense controlled device, s is a Laplace operator.The Laplace transformation form of d is disturbed in D (s) expression, the Laplace transformation form of the input u of U (s) expression broad sense controlled device, G (s) expression broad sense controlled device, G _n(s) the nominal model of expression broad sense controlled device, i.e. the generalized object model that adopts in the interference observer algorithm.If the cutoff frequency of low-pass filter Q (s) is f _q, as f≤f _qWhen promptly being in low-frequency range, can think | Q (jw) | ≈ 1, then G _Uy=G _n, G _Dy=0.Wherein, G _Uy≈ G _nEven it is uncertain to illustrate that model exists, i.e. G ≠ G _n, the response of the practical object that interference observer is recognized is consistent with the response of nominal model, and promptly variation has certain robustness to controller to the broad sense object parameters; G _Dy=0 explanation interference observer has the complete inhibition ability for the low-frequency disturbance in Q (s) frequency band.

As f 〉=f _qPromptly, can think at high band | Q (jw) | ≈ 0, then G _Uy=G, G _Dy=G.G _Uy=G explanation interference observer for the perturbation of image parameter without any effect.G _Dy=G explanation is not because the low-pass filter effect of interference observer suppresses effect for the interference that is in Q (s) high band, still is the transport function of original system based on the ssystem transfer function of interference observer.

As the above analysis, design is the Q wave filter based on the committed step of the high-precision magnetic bearing axial control method of interference observer, if the Q wave filter has the ideal low-pass filter form, then reach above-mentioned analytical performance fully, but ideal low-pass filter physics can not be realized.Therefore, design the form of Q wave filter and the dynamic property that cutoff frequency has determined whole interference observer.Simultaneously, also need consider inhibition ability,, following constraint condition be arranged with the robustness of interference observer to disturbing:

(1) exponent number of Q (s) wave filter satisfies Q G _n ^-1(s) canonical, physics can be realized.Promptly require the relative rank of Q (s) wave filter should be more than or equal to the relative rank of broad sense plant model G (s), Q (s) wave filter should be too not high simultaneously, the exponent number that increases wave filter can cause uncertain edge effect, and in fact this reduced near the robust stability of the control system of Q (s) wave filter peak value.The operand of controller is increased, unfavorable to real-time control.So select inferred-zero, limit number low-pass filter than broad sense controlled device exponent number higher order.

(2) interference performance is strong more outside the wide more inhibition of the frequency band system of Q (s) wave filter, but the high band gain is bigger, and then the robust stability variation increases the susceptibility of measuring noise; Otherwise the narrow more robust stability of frequency band is good more, and it is insensitive more to measure noise, and externally the inhibition ability of disturbing is also weak more.Can be with reference to Q (s) general formula of following form:

$Q_{\mathrm{NM}}\left(s\right)=\frac{\underset{k=0}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_k{\left(\mathrm{\&tau;s}\right)}^k}{{(\mathrm{\&tau;s}+1)}^N}$ (M＝0，1，…，N-1)

Wherein, ${\mathrm{\&alpha;}}_k=\frac{N!}{(N-k)!k!}$ Be coefficient, N is the exponent number of denominator, and M is the exponent number of molecule, and N-M is relative rank.

Can know by the transfer function analysis that disturbs, ideally, if the transport function of broad sense controlled device is consistent with nominal model, then above-mentioned analysis has ideal results, but exist under the situation of modeling error perturbation and external disturbance, between real system and the nominal model error must be arranged.Can set up relatively accurate nominal model in the hope of the actual parameter of generalized object by frequency sweep experiment to system.Remaining model error is in the realization of interference observer, and to the control signal end, equivalence is an external disturbance, compensates with this error converting.

Because use prerequisite of the present invention is to set up nominal accurately model by the frequency sweep experiment, obtain model parameter, and frequency sweep is obtained is the frequency response curve of single-input single-output system, then on the axial single channel of magnetic bearing, can realize, so the hyperchannel for radially coupling then can't use the method for frequency sweep, the scope that then defines use is single pass axial control system.

In sum, the high-precision magnetic bearing axial control method based on interference observer of the present invention can be uncertain at the outside, unknown disturbance carries out online observation, and effective disturbance suppression, realizes the purpose of High Accuracy Control.

The solution of the present invention is compared with existing scheme, and major advantage is:

(1) owing to adopted the algorithm of interference observer, thus realized to can not modeling, the observation and the inhibition of uncertain disturbance, improved the precision of magnetic bearing axial control;

(2) by online observation and the inhibition of interference observer realization for disturbance, algorithm operation quantity is less, and operation time is shorter, is easy to on-the-spot realization, and debug process is more flexible;

(3) in interference observer by changing the cutoff frequency of Q wave filter, can effectively suppress the external disturbance of various different frequency bands;

(4) interference observer of the present invention partly has independence, and the interference value that observes is realized record and quantitative test, has expanded range of application.

Description of drawings

The system principle diagram that Fig. 1 realizes for the inventive method;

Fig. 2 is an A/D modular circuit schematic diagram among Fig. 1;

Fig. 3 is a DSP module hardware partial circuit schematic diagram among Fig. 1;

Fig. 4 is a FPGA modular circuit schematic diagram among Fig. 1;

Fig. 5 is the realization flow figure of the inventive method;

Fig. 6 is the frequency sweep circuit schematic diagram among Fig. 1;

Fig. 7 is a theory diagram of the present invention;

Fig. 8 is an interference observer algorithm flow chart of the present invention.

Embodiment

As shown in Figure 1, realize that the system of the inventive method comprises frequency sweep circuit 5 and digital control hardware 6, wherein digital control hardware 6 comprises A/D module 8, DSP module 9 and FPGA module 7; Frequency sweep circuit 5 is passed to A/D module 8 after with the output signal of sensor 4 and pumping signal stack, FPGA module 7 receives the digital quantity of A/D module 8 conversions, send to DSP module 9 again, DSP module 9 adopts the interference observer algorithm computation to obtain the Current Control amount, be sent to FPGA module 7 subsequently, FPGA module 7 converts the Current Control amount output of to PWM form then, drives power amplifier 2 generation electromagnetic forces and acts on the magnetic bearing, realizes the suspension function.

Be illustrated in figure 2 as the schematic diagram of A/D module 8 of the present invention, this A/D module 8 is used to gather shift value and is converted to digital quantity.A/D module 8 is selected the AD7938 chip of TI company for use, and this chip precision is 12, and line output, can sample simultaneously by 8 passages.AD7938 single channel sampling rate is: 25M/ (17*8)=183.8235KHz, can satisfy the demand that bearing is controlled required sampling rate (10KHz).This AD chip Adjustable Output Voltage, it is 3.3V that the present invention designs the output high level, can save level shifting circuit required when being connected with FPGA module 7.

Shown in Fig. 3,4, for DSP module 9 of the present invention and FPGA module 7 are the main control chip of digital control hardware components.FPGA module 7 is handled external signals as master control system control A/D module 8, by bus external signal is sent into to enter DSP module 9 and carry out computing, and the result of computing sends back FPGA module 7 by bus again and forms pwm signals and drive power amplifiers.

The hardware components of DSP module 8 adopts the TMS320VC33 chip of selecting TI company for use, dominant frequency can reach 150MHz, word length is 32, extended precision is 40, program bus, data bus and dma bus separately makes to be got finger, reads and writes data and can carry out simultaneously, integrated synchronous serial interface can be realized high speed communication between the DSP.

The hardware components of FPGA module 7 is selected the Spartan-3 series of X C3S400 chip of Xilinx for use, and integrated 400,000 gate circuits of this chip satisfy the required resource of management control peripheral hardware, and this chip I/O port supply voltage is 3.3V, and the kernel supply voltage is 1.2V, and is low in energy consumption.Support is led and is gone here and there, leads also, the JTAG downloading mode, debugs flexible.

As shown in Figure 5, this high-precision magnetic bearing axial control method specific implementation step based on interference observer of the present invention is as follows:

Step 1: utilize frequency sweep circuit 5, do the frequency sweep experiment respectively, obtain the broad sense object parameters, set up the contrary G of generalized object for power amplifier 2 and the broad sense controlled device 1 that comprises power amplifier _n ^-1Transport function removes frequency sweep circuit 5 parts then, and sensor 4 is directly inserted A/D module 8;

By the principle of inventing as can be known, whether design of the present invention depends on the broad sense object parameters accurate, so before using, must do the frequency sweep experiment to the broad sense controlled device, to obtain the actual parameter of object model.

As shown in Figure 6, be the frequency sweep circuit 5 that the present invention adds.Electric connecting point when as shown in Figure 1, the A among the figure, B, C, D are for the different link of test.A, B, C, D represent four test points in the closed-loop control system respectively: the A point is the value after the conditioning of displacement transducer process, and the B point is the input signal values of controller, and the C point is digital control amount output, and the D point is a current magnitude.

Adopt the Agilent35670A kinetic analyzer that the power amplifier link is done the frequency sweep experiment, as shown in Figure 1, the pumping signal of frequency sweep and sensor signal are imported the reverse totalizer of being made up of operational amplifier simultaneously, through becoming output signal after the one-level reverser computing of forming by operational amplifier, be sent to the A/D module port again.Operational amplifier is selected the TL084 of TI company for use.The input end of kinetic analyzer is received the C point among Fig. 7, i.e. the digital quantity of controller output, and the output terminal of kinetic analyzer is received the D point among Fig. 7, i.e. the current magnitude of power amplifier output.The pumping signal that the excitation end adds is V _{Peak value}=40mV, V _BiasingThe sinusoidal signal of=0V, the frequency of sinusoidal signal is tried to achieve the frequency sweep curve of power amplifier thus during frequency sweep from 0.1Hz to 2kHz.The theoretical formula of power amplifier is as follows:

G _w＝k _wg _wLPF

Wherein, k _wBe power amplifier direct current enlargement factor, g _WLPFBe the power amplifier low-pass filter function.The frequency sweep curve obtains the analytical expression of realistic model after over-fitting, compare with theoretical formula, obtains actual k _wParameter value.

Adopt identical method that the broad sense controlled device 1 that comprises power amplifier 2 is carried out the frequency sweep experiment, input end is at digital control amount output C point, and output terminal obtains the analytical expression of broad sense controlled device at sensor output A point.According to formula:

$G_n^{-1}\left(s\right)=\frac{{\mathrm{ms}}^2-k_h}{k_i{k}_w{k}_s}$

Try to achieve the contrary G of generalized object _n ^-1(s), r:k wherein _sBe transducer sensitivity, m is the quality of magnetic bearing, k _hBe the displacement rigidity of magnetic bearing, k _iBe the current stiffness of magnetic bearing, k _wBe the power amplifier coefficient, adopt above-mentioned parameter after measured.

Setting up the contrary G of generalized object _n ^-1After the transport function,, must remove the frequency sweep circuit part, sensor is directly inserted the A/D module, form independently closed-loop system operation for the assurance system is not subjected to the interference of other signals.

Step 2: interference observer 11 in this control method of initialization in the DSP module and controller K parameter, the storage space of setting sampled data, the sampling pattern that FPGA module 7 is set is that clock interrupts.

Initialized parameter is included in parameter and the rationalization generalized object QG that controller K adopts the parameters in the PID control and tested the Q wave filter 12 that uses that obtains by above-mentioned frequency sweep in interference observer _n ^-113 parameter etc., specifically comprise: (a) parameter of initialized controller K10 comprises: given magnetic bearing suspension center location parameter, its value gets half for the magnetic bearing end positions, as the center; The ratio of controller, integration, differential parameter, the specified rate of its value for regulating in advance; The time constant of controller, its value is consistent with the sampling time; The control variable storage unit of controller all is initialized as zero.(b) parameter of initialized Q wave filter 12 comprises: it is 20Hz that the Q filter cutoff frequency of setting, its value are set at minimum cutoff frequency; According to the filter parameter that cutoff frequency calculates, its value is according to the transport function of Q wave filter and cutoff frequency and obtains the corresponding multiplication coefficient of being done for input, output; The storage unit of input quantity and output quantity is initialized as zero in the Q wave filter.(c) the contrary QG of initialized rationalization generalized object _n ^-113 parameter comprises: input and output multiplication weighting parameter, and its value calculates according to the cutoff frequency of above-mentioned Q wave filter; The storage unit of input quantity and output quantity is initialized as zero.The sampling time of A/D module 8 is 7kHz, and the storage space of opening up in controller K10 is mapped among the distributed ROM of FPGA module 7 inside.

Step 3: 8 samplings of FPGA module 7 control A/D modules obtain the displacement of sensor 4 outputs, are transformed into the DSP module 9 of digital quantity input hardware control; A/D module 8 and FPGA module 7 adopt the design of above-mentioned digital control hardware.

Step 4: FPGA module 7 is sent to DSP module 9 with digital quantity, and according to given suspension center, the displacement value of calculating input obtains corresponding offset deviation amount in DSP module 9; Given levitation position is the center of magnetic bearing in the DSP module 9, before system debug, adopt and float, the end positions that A/D module 8 collects is got half, as the center, determine direction be if the center on then the offset deviation amount on the occasion of, otherwise be negative value.

Step 5: the offset deviation amount of input adopts in the controller K algorithm and calculates the basic controlling amount in DSP module 9; The algorithm of controller K uses the Decentralized PID control algolithm, and its transport function is:

$G\left(s\right)=K_P+\frac{1}{T_{I}s}+\frac{T_{D}s}{1+T_{f}s}$

Wherein, K _pBe scale-up factor, T _iBe integral coefficient, T _DBe differential coefficient, T _fThe time constant of the inertial element that adds for differentiation element, T is the sampling period.

Step 6: the interference observer algorithm in the DSP module 9 has two inputs, and Current Control amount input Q wave filter calculates, the contrary QG of displacement input rationalization generalized object _n ^-1Calculate, the two result of calculation is subtracted each other the Interference Estimation amount of trying to achieve;

Be illustrated in figure 7 as the theory diagram based on interference observer of the present invention, wherein, y is the output of broad sense controlled device, and d is for disturbing, and u is the input of broad sense controlled device, G (s) expression broad sense controlled device, G _n(s) the nominal model of expression broad sense controlled device, i.e. the generalized object model that adopts in the interference observer algorithm, interference observer carries out disturbance observation and compensation to system, is practical object G (s) output that external disturbance causes is added to G _n ^-1(s), recurrent disturbance and error signal sum deduct error signal again, observe disturbance quantity, and introduce the compensation of equivalent in the control input, realize the elimination to disturbing.By principle analysis as can be known, key of the present invention is to try to achieve the nominal model G based on magnetic bearing system _n, and realize G _n ^-1(s) Q wave filter.Frequency sweep experiment has before obtained nominal model G _nParameter, next select suitable Q wave filter, and realize the interference observer algorithm.

Described in inventive principle, the exponent number of broad sense controlled device is 2 rank, the relative rank of the Q wave filter of selecting must be more than or equal to 2 rank, so elect the form of general formula as, N=3 wherein, M=0, then rank are N-M=3 relatively, satisfy the requirement on relative rank, and guaranteed good robust performance index, shown in the following form:

$Q\left(s\right)=\frac{1}{{\left(\mathrm{\&tau;s}\right)}^3+3{\left(\mathrm{\&tau;s}\right)}^2+3\mathrm{\&tau;s}+1}$

Wherein, τ is the Q filter parameter, can adjust the Q filter cutoff frequency.In the process that realizes, can adjust the cutoff frequency of low-pass filter according to interference characteristic, its scope is 20Hz ~ 1630Hz, corresponding τ=0.004～0.00005.

The contrary QG of rationalization generalized object _n ^-1The form of the Q wave filter of choosing in 13 12 and above-mentioned Q wave filter 12 are in full accord, and parameter is identical.The contrary G of generalized object _n ^-1Be the transport function that step 1 is tried to achieve, the contrary QG of rationalization generalized object _n ^-1(s) transport function is the contrary G of generalized object _n ^-1(s) with Q (s) product of Q wave filter, be shown below:

${\mathrm{QG}}^{-1}\left(s\right)=Q\left(s\right)*G^{-1}\left(s\right)$

$=\frac{1}{{\left(\mathrm{\&tau;s}\right)}^3+3{\left(\mathrm{\&tau;s}\right)}^2+3\mathrm{\&tau;s}+1}*\frac{{\mathrm{ms}}^2-k_h}{k_i{k}_w{k}_s}$

$=\frac{{\mathrm{ms}}^2-k_h}{({\left(\mathrm{\&tau;s}\right)}^3+3{\left(\mathrm{\&tau;s}\right)}^2+3\mathrm{\&tau;s}+1)k_i{k}_w{k}_s}$

As Fig. 8 is interference observer algorithm flow chart of the present invention.Transport function to design is carried out discretize, and its method can be a backward difference method, also can be forward difference method, Tustin converter technique or revises the Tustin converter technique in advance.Adopted the method for backward difference to write corresponding algorithm among the present invention.In the interference observer algorithm is realized, at first according to the cutoff frequency of setting, the parameter that initialization algorithm uses, the shift value that is input as axial sensor that the rationalization generalized object is contrary calculates the result; Then the Q wave filter be input as axial controlled quentity controlled variable, calculate corresponding result; Two results subtract each other, and obtain the Interference Estimation value that interference observer observes, and the negative feedback of Interference Estimation value adds in the controlled quentity controlled variable, and equivalent compensation is done in disturbance to external world, realize the High Accuracy Control of magnetic bearing.

Step 7: in DSP module 9, utilize Interference Estimation amount and basic controlling amount to obtain the Current Control amount, send the Current Control amount to FPGA module 7 then, form the PWM waveform in FPGA module 7, control power amplifier drive magnetic produces electromagnetic force, realizes magnetic bearing high precision stable suspersion.

In a word, the present invention can effectively control the stability that magnetic bearing suspends, and the extraneous disturbance unknown, not modeling of compensation is disturbed axial magnetic bearing.By to the relatively calculating between controlled quentity controlled variable and the displacement, the disturbance that the system of obtaining is subjected to adds system by negative feedback, reaches the effect of disturbance suppression.Simultaneously, can do performance analysis with oscillograph or spectrum analyzer, the disturbance situation that the quantitative solving magnetic bearing is subjected to for the disturbance that observes.In using the process of interference observer, can effectively suppress the disturbance of different frequency bands by the cutoff frequency of regulating the Q wave filter, strengthened the dirigibility of system debug.Has the advantage that significantly improves magnetic-suspension axial magnetic bearing suspension precision.

The content that is not described in detail in the instructions of the present invention belongs to this area professional and technical personnel's known prior art.

Claims (5)

1, a kind of high-precision magnetic bearing axial control method based on interference observer is characterized in that step is as follows:

(a) at first add frequency sweep circuit, do the frequency sweep experiment respectively for power amplifier and the broad sense controlled device that comprises power amplifier, obtain the broad sense object parameters, set up the contrary Gn-1 transport function of generalized object, remove the frequency sweep circuit part then, sensor output is directly inserted the A/D module;

(b) interference observer in the initialization DSP module and controller K parameter are provided with the sampled data storage space, and the sampling pattern of setting the FPGA module is that clock interrupts;

(c) FPGA module controls A/D module samples obtains sensor output, and receives the digital quantity result that the A/D module converts obtains;

(d) the FPGA module is sent to the DSP module with digital quantity, according to given suspension center, calculates the offset deviation amount of input displacement value correspondence in the DSP module;

(e) the controller K in the DSP module is input as the offset deviation amount, and controller K adopts controller algorithm to calculate the basic controlling amount;

(f) interference observer in the DSP module is made of Q wave filter and the contrary QGn-1 of rationalization generalized object, interference observer comprises Current Control amount and two inputs of displacement, wherein, Current Control amount input Q wave filter calculates, the contrary QGn-1 of displacement input rationalization generalized object calculates, and above-mentioned both result calculated are kept to the Interference Estimation amount mutually;

(g) utilize Interference Estimation amount and basic controlling amount to obtain the Current Control amount in the DSP module, then the Current Control amount is passed to the FPGA module, form the PWM waveform in the FPGA module, the control power amplifier produces electromagnetic force, realizes magnetic bearing high precision stable suspersion.

2, the high-precision magnetic bearing axial control method based on interference observer according to claim 1 is characterized in that: the contrary G of the generalized object in the described step (a) _n ^-1Transport function is:

$G_n^{-1}\left(s\right)=\frac{{\mathrm{ms}}^2-k_h}{k_i{k}_w{k}_s}$

Wherein: k _wBe the power amplifier coefficient, k _sBe transducer sensitivity, m is the quality of magnetic bearing, k _hBe the displacement rigidity of magnetic bearing, k _iCurrent stiffness for magnetic bearing.

3, the high-precision magnetic bearing axial control method based on interference observer according to claim 1, it is characterized in that: the Q wave filter in the described step (f) is inferred-zero, the limit number low-pass filter than broad sense controlled device exponent number higher order, and its transport function is:

$Q\left(s\right)=\frac{1}{{\left(\mathrm{\&tau;s}\right)}^3+3{\left(\mathrm{\&tau;s}\right)}^2+3\mathrm{\&tau;s}+1}$

Wherein, the Laplace transform form of Q (s) expression Q wave filter, s is a Laplace operator, τ is the Q filter parameter, can adjust the Q filter cutoff frequency.

4, the high-precision magnetic bearing axial control method based on interference observer according to claim 1 is characterized in that: the contrary QG of the rationalization generalized object in the described step (f) in the interference observer _n ^-1Contrary G for generalized object _n ^-1Product with the Q wave filter.

5, the high-precision magnetic bearing axial control method based on interference observer according to claim 1, it is characterized in that: the frequency sweep circuit in the described step (a) is made up of reverse totalizer and one-level reverser, oppositely two of totalizer are input as sensor output and pumping signal, and pumping signal is a sinusoidal signal.

Images