CN103795066A - Control method of electricity distribution network active filtering static reactive compensation - Google Patents

Abstract

The invention discloses a control method of an electricity distribution network active filtering static reactive compensation, and belongs to the technical field of the active filter control. A DSP chip is used to control a PWM; during each sampling period, the DSP chip uses a deadbeat algorithm to generate the PWM according to collected data; the PWM is compensated through a dead zone compensator and is sent to an electricity distribution network control circuit; and the DSP chip stops sending the PWM only when a load is underloaded. According to the invention, the deadbeat PWM algorithm is combined with the dead zone control; the sleep mode is carried out when the load is underloaded, so as to effectively improve the waveform of the output current; and the APFC is prevented from unnecessary loss caused by continuous opening and closing of the switching tube to maintain the bus voltage.

Description

The control method of distribution network active filtering quiescent reactive compensation

Technical field

The invention discloses the control method of distribution network active filtering quiescent reactive compensation, belong to the technical field of active filter control.

Background technology

Along with the develop rapidly of power electronic technology, in electrical network, increase a large amount of nonlinear loads, being widely used of especially large capacity convertor equipment, causes a large amount of harmonic waves to inject electrical network, and line voltage and current waveform are distorted, and the quality of power supply declines day by day.Harmonic wave has become large public hazards of electrical network.

In order to suppress mains by harmonics, conventional method adopts passive filter.That this method has is simple in structure, equipment investment is low, operational reliability is high and the advantage such as easy to maintenance, but also there are many defects, as can only eliminate specific subharmonic, existence and electrical network generation resonance may, filtering characteristic is subject to the impact of system parameters larger, it is large etc. to consume.

At present, the key technology of obstruction active filter development has 2 points: the generation of electric current detecting method and offset current, the method for current detecting mainly contains: (1) method based on frequency domain computing.This is class methods for instruction current detecting the earliest.Its basic thought is first to isolate fundametal compoment and the harmonic component in load current with band pass filter, and then uses the computational methods in Circuit theory that fundamental current is decomposed into first-harmonic real component and fundamental reactive component.The method can be extracted fundametal compoment adaptively from the fluctuation of motion tracking electrical network frequency domain, but still there is larger delay, the problem such as real-time is poor, compensation effect is bad.(2) instantaneous space vector method.Instantaneous space vector method based on instantaneous reactive power theory is a kind of instruction current detection method most widely used in current APF.The method is only applicable to symmetrical three phase circuit at first, has now comprised p-q method, ip-iq method and d-p method etc. by updating.Wherein p-q method is applied the earliest, is only applicable to symmetrical and distortionless electrical network; And ip-iq method is both effective to line voltage distortion, and be applicable to asymmetric three phase network; D-p method based on synchronous rotary park conversion has not only been simplified symmetrical instruction current computing under undistorted, and is applicable to asymmetric, to have distortion electrical network.(3) detection method based on modern control theory.Application has a method based on P-I controller the earliest.Because the characteristic of P-I controller can not adapt to the variation of load and electrical network, therefore the modern control method such as sliding formwork control and fuzzy control were proposed again afterwards.They are all directly according to the voltage of inverter direct-flow side or electric current, obtain the first-harmonic real component amplitude of required power network current, then calculate the value of required offset current.(4) adaptive detective method and based on the new detection method such as ANN Control method

The generation of offset current adopts the voltage source inverter based on PWM conventionally, and its current control method mainly contains following several: (1) triangular carrier Linear Control.It,, using the difference between instruction current and actual offset current as modulation signal, is compared with high frequency triangular carrier, thereby obtains the needed control signal of inverter switch device.Its advantage is that dynamic response is good, and switching frequency is fixed, and circuit is simple, and shortcoming is that switching loss is larger, and in output waveform, contains the high frequency distortion component of carrier frequency and harmonic frequency thereof.(2) stagnant chain rate is controlled.It is that the difference of instruction current value and inverter actual current output valve is input to the comparator with hysteresis characteristic, carrys out the folding of control switch by the output of comparator, thereby reaches inverter output valve real-time tracking instruction current value.Compared with triangular carrier Linear Control, to have switching loss little compared with controlling for stagnant chain rate, the feature such as dynamic response is fast.Shortcoming is that the tracking accuracy of switching frequency, response speed and the electric current of system can be subject to that stagnant endless belt is wide to be affected.When bandwidth is fixed, switching frequency can change with offset current, thereby causes larger pulsating current and switching noise.(3) dead beat control.This is a kind of control technology of total digitalization, and it utilizes instruction current and the actual compensation current of previous moment, calculates according to space vector theory the switching mode that next moment of inverter should be satisfied.Advantage is that dynamic response is very fast, is easy to computer and carries out, and shortcoming is that amount of calculation is large, and larger to system parameters dependence.

In above control method, triangular carrier Linear Control method based on analog control technique and control method of stagnant chain rate are the methods that current Active Power Filter-APF generally adopts, can overcome intrinsic defect by corrective measures such as multiple technology, the stagnant rings of self adaptation, improve its service efficiency.Dead beat control method based on full-digital control technology, along with digital signal processor (DSP) arithmetic speed improves constantly, also will be widely used in Active Power Filter-APF.In recent years, some scholars have proposed to be again applied to based on the nonlinear control method of internal model control, sliding formwork control and ANN Control etc. the scheme that offset current produces.These nonlinear Control have a good application prospect, the further research but compensation characteristic under various loading conditions is still needed.

Due to power switch pipe non-ideal switches, and the turn-off time is longer than service time.In order to make power switch pipe work safety, before it is opened, insert dead band, allow the individual switching tube of this brachium pontis within the time all in off state.Due to the existence of Dead Time, the ON time of power switch pipe is shortened, cause the loss of pwm pulse width and current potential, and then have influence on the average voltage of inverter output.

For the operating mode of load variations, for example: have compared with heavy load some on-the-spot daytime, and almost do not have loaded situation night, when load is under underloading condition of work, if APF is also in normal mode of operation, owing to will maintaining busbar voltage, IGBT need to ceaselessly cut-off, this loss is unnecessary, must be improved.

Summary of the invention

Technical problem to be solved by this invention is the deficiency for above-mentioned background technology, and the control method of distribution network active filtering quiescent reactive compensation is provided.

The present invention, for solving the problems of the technologies described above, adopts following technical scheme:

The control method of distribution network active filtering quiescent reactive compensation, utilize dsp chip to realize the control of PWM, within each sampling period, dsp chip utilizes dead beat algorithm to generate PWM according to image data, PWM is sent to power distribution network control circuit after the compensation of dead-zone compensator, and dsp chip only stops sending PWM in the time of load underloading.

As the further prioritization scheme of the control method of described distribution network active filtering quiescent reactive compensation, the three-phase current error of calculating in dead beat algorithm is transient current value, predicted current variable poor of current reference value and current sampling instant.

Generate as dead-zone compensator described in the further prioritization scheme of the control method of described distribution network active filtering quiescent reactive compensation the reverse compensation signal equating with PWM error signal amplitude.

Further prioritization scheme as the control method of described distribution network active filtering quiescent reactive compensation: load underloading is determined by the following method: gather power distribution network electric current, relatively power distribution network electric current and rated current, the in the situation that in set time section, power distribution network electric current being less than rated current, load is in light condition.

The present invention adopts technique scheme, there is following beneficial effect: in sum, the present invention is by controlling dead beat PWM algorithm to combine with dead band, in the time of system underloading, enter park mode, effectively improve the waveform of output current, avoided APFC and do not stopped disconnected switching tube and cause unnecessary loss because maintain busbar voltage.

Accompanying drawing explanation

Fig. 1 judges that DSP enters the schematic diagram of resting state.

Embodiment

Below in conjunction with accompanying drawing, technical scheme of the present invention is elaborated:

The control method of distribution network active filtering quiescent reactive compensation, utilize dsp chip to realize the control of PWM, within each sampling period, dsp chip utilizes dead beat algorithm to generate PWM according to image data, PWM is sent to power distribution network control electric current after the compensation of dead-zone compensator, and dsp chip only enters resting state and stops sending PWM in the time of load underloading.

Dead-zone compensator generates the reverse compensation signal equating with PWM error signal amplitude.

Load underloading is by the following method as shown in Figure 1: gather power distribution network current i, relatively power distribution network current i and rated current i _o, the in the situation that in set time section, power distribution network electric current being less than rated current, load is in light condition.

Dead beat PWM algorithm, based on three-phase current reference value

the current instantaneous value i of current sampling instant _a, i _b, i _ccalculate each phase PWM duty ratio of next control cycle T, can derive following formula based on dead beat theory

${\mathrm{\Δd}}_u=\frac{{1.5U}_{\mathrm{dc}}+2(U_{\mathrm{AN}}-U_{\mathrm{BN}}+L\frac{i_a^{*}-i_a}{T}-L\frac{i_b^{*}-i_b}{T})+(U_{\mathrm{BN}}-U_{\mathrm{CN}}+L\frac{i_b^{*}-i_b}{T}-L\frac{i_c^{*}-i_c}{T})}{{3U}_{\mathrm{dc}}};$

${\mathrm{\Δd}}_v=\frac{{1.5U}_{\mathrm{dc}}-(U_{\mathrm{AN}}-U_{\mathrm{BN}}+L\frac{i_a^{*}-i_a}{T}-L\frac{i_b^{*}-i_b}{T})+(U_{\mathrm{BN}}-U_{\mathrm{CN}}+L\frac{i_b^{*}-i_b}{T}-L\frac{i_c^{*}-i_c}{T})}{{3U}_{\mathrm{dc}}};$

${\mathrm{\Δd}}_w=\frac{{1.5U}_{\mathrm{dc}}-(U_{\mathrm{AN}}-U_{\mathrm{BN}}+L\frac{i_a^{*}-i_a}{T}-L\frac{i_b^{*}-i_b}{T})-2(U_{\mathrm{BN}}-U_{\mathrm{CN}}+L\frac{i_b^{*}-i_b}{T}-L\frac{i_c^{*}-i_c}{T})}{{3U}_{\mathrm{dc}}};$

${\mathrm{\Δi}}_a=i_a^{*}-(i_a+{\mathrm{\Δi}}_{a0});$

${\mathrm{\Δi}}_b=i_b^{*}-(i_b+{\mathrm{\Δi}}_{b0});$

${\mathrm{\Δi}}_c=i_c^{*}-(i_c+{\mathrm{\Δi}}_{c0});$

${\mathrm{\Δi}}_n=i_n^{*}-(i_n+{\mathrm{\Δi}}_{n0});$

Δi _a0=0.3*Δi _a；

Δi _b0=0.3*Δi _b；

Δi _c0=0.3*Δi _c；

Δi _n0=0.3*Δi _n

Wherein: U _aN, U _bN, U _cNfor three-phase phase voltage, U _dcfor DC voltage, L is grid-connected filter inductance, Δ i _a0, Δ i _b0, Δ i _c0the predictive variable of three-phase current.

Implementation in DSP: this chip includes a PWM module, and the burst length of adjusting waveform generator before Dead Time counter compensates dead time effect.The method and carrier frequency are irrelevant, only relevant to load current polarity.Drive signal as example take certain phase IGBT, with polarity and a computing mode variable CNT who represents down/up of load current be according to programming, down represents that switching tube opens, up represents that switching tube closes.Decide thus timing whether to need to add deduct time of pulse.Utilize software to produce the ton and toff switching time of desirable operation.Dead band td is pre-stored within PWM module controls register, ceaselessly upgrades current polarity by data/address bus, by the more state of new variables CNT of the interrupt signal from waveform generator by current detector.

In the time of i>0, when CNT is down state, software need add a td pulse on ton, and is stored in ton, then delivers in waveform generator, after Dead Time counter is processed, is applied in load and goes.Dead band counter provides two complementary pwm control signals to remove to control upper and lower two power switch pipes.In the time that i>0 and CNT are up, toff does not need to proofread and correct, and toff directly delivers in waveform generator, after dead hour counter processing, is applied in load.

In the time that i<0 and CNT are up, ton does not need to proofread and correct, and directly delivers in waveform generator, after dead hour counter processing, is applied in load.In the time that i0 and CNT are up, toff need deduct a pulsewidth td, is stored in toff, then delivers in waveform generator, after Dead Time counter is processed, is applied in load.

After adding dead band to control, can effectively reduce the distortion of the burr and the zero crossing that send electric current, send electric current effect and be improved.

In sum, the present invention, by dead beat PWM algorithm and dead band are controlled and combined, enters park mode in the time of system underloading, has effectively improved the waveform of output current, has avoided APFC and has not stopped disconnected switching tube and cause unnecessary loss because maintain busbar voltage.

Claims (4)

1. the control method of distribution network active filtering quiescent reactive compensation, utilize dsp chip to realize the control of PWM, it is characterized in that: within each sampling period, dsp chip utilizes dead beat algorithm to generate PWM according to image data, PWM is sent to power distribution network control circuit after the compensation of dead-zone compensator, and dsp chip only stops sending PWM in the time of load underloading.

2. the control method of distribution network active filtering quiescent reactive compensation according to claim 1, is characterized in that: the three-phase current error of calculating in described dead beat algorithm is transient current value, predicted current variable poor of current reference value and current sampling instant.

3. the control method of distribution network active filtering quiescent reactive compensation according to claim 1 and 2, is characterized in that: described dead-zone compensator generates the reverse compensation signal equating with PWM error signal amplitude.

4. the control method of distribution network active filtering quiescent reactive compensation according to claim 3, it is characterized in that: load underloading is determined by the following method: gather power distribution network electric current, relatively power distribution network electric current and rated current, the in the situation that in set time section, power distribution network electric current being less than rated current, load is in light condition.

Images