CN106253279A - A kind of anti-resonance control algolithm being applied to active filter - Google Patents

Abstract

The present invention provides a kind of rapid extraction current on line side harmonic components and carries out the algorithm suppressed, and judges by carrying out detection contrast, effectively suppresses the resonance of active filter and electrical network.A kind of anti-resonance control algolithm being applied to active filter, is detected in real time load current, is extracted the harmonic component in load current by sliding window DFT algorithm, provides current reference for control module, it is achieved the harmonic restraining function of active filter.The present invention is capable of, in the case of electrical network damping is more weak, protecting equipment, preventing resonance from equipment causing damage, simultaneously can the harmonic components of rapid extraction resonance current, internal module quickly responds, thus reaches the purpose of resonance suppression.

Description

A kind of anti-resonance control algolithm being applied to active filter

Technical field

The present invention relates to active filter based on FPGA there is control method during resonance with electrical network, be related specifically to The most quickly carry out dynamic tracking control resonance.

Background technology

Active Power Filter-APF (APF) is a kind of dynamic compensating device, it is possible to the harmonic wave in electrical network, idle carry out quickly Response, and increasingly receive publicity.And electrical network typically exhibits underdamping characteristic, the big inductance of its equivalence makes active filter LCL resonant frequency changes, and easily causes low-frequency resonant, when resonance occurs in electrical network, usually introduces series of malpractice:

(1) resonance is easily caused overvoltage.Electrical network is worked the mischief greatly by resonance overvoltage, such as causes voltage transformer Fuse failure, voltage transformer burn, grid equipment insulation damage, even cause phase fault, protection device misoperation etc.；

(2) active filter equipment itself is adversely affected by resonance, and resonance current is excessive, may cause DC side electricity Pressure fluctuation acutely, even occurs that overvoltage or DC bus capacitor damage；

The method solving resonance has a lot, can on filter capacitor series damping resistor, but this way can cause setting Standby loss increases, and the efficiency of complete machine reduces, and resistance heating adds the trouble point of complete machine, and system reliability reduces, it addition, increase Add cost；It addition, also there are the more state variables such as research introducing curtage that are devoted to carry out feedback algorithm, in practice Middle acquirement certain effect, but when discrete system, almost without clear and definite parametric stability scope and concrete setting method.

Summary of the invention

1, technical problem to be solved:

The present invention proposes a kind of dynamic control algolithm based on FPGA, dynamic realtime detection current on line side, carries with load The harmonic components taken is analyzed, it may be judged whether there is resonance, and quickly takes corresponding measure to suppress.

2, technical scheme:

The present invention provides a kind of rapid extraction current on line side harmonic components and carries out the algorithm suppressed, by detecting Contrast judges, effectively suppression active filter and the resonance of electrical network.

A kind of anti-resonance control algolithm being applied to active filter, detects in real time load current, passes through sliding window Harmonic component in load current is extracted by DFT algorithm, provides current reference to realize active filter for control module Harmonic restraining function, comprises the following steps:

Step 1: detect load current in real time, by sliding window DFT algorithm, extracts each harmonic component in load current；

Step 2: being overlapped by each harmonic component in the load current of extraction, its result is as controller unit Instruction current；

Step 3: monitor actual current on line side in real time, is carried out sliding window DFT algorithm and is extracted actual current on line side therein Each harmonic component, the harmonic component corresponding with the instruction current of current period compares respectively, through comparing unit Monitoring identification, it is judged that actual net surveys whether current system exists resonance；

Step 4: if resonance having been detected, then remove this corresponding subharmonic in instruction, carry out resonance suppression；Realize Resonance can be carried out quick dynamic tracking suppression.

The algorithm extracting each harmonic component in networking load current in described step 1 is:

According to Fourier decomposition, by load current i_LIt is expressed as form:

$i_L\left(t\right)=a_0+\underset{n=1}{\overset{\mathrm{\∞}}{\Σ}}(a_n\cos nwt+b_n\sin nwt)$

$a_n=\frac{1}{T}\underset{0}{\overset{T}{\∫}}u\left(t\right)cos\left(nwt\right)dt,b_n=\frac{1}{T}\underset{0}{\overset{T}{\∫}}u\left(t\right)sin\left(nwt\right)dt$

Wherein: when n take 3 respectively, 5,7... etc. time, networking load current each harmonic component can be drawn,

In described step 2, after each harmonic superposition that will extract, can be used as the current-order of master control algorithm, i.e. system During resonance free, after active filter equipment runs, inverter output current should be consistent with the instruction current described in step 2, it may be assumed that

ILg=∑ a_nsin(nwt)+b_ncos(nwt)

N=2k+1, k=1,2,3...

Ask for algorithm according to Fourier Transform Coefficients, obtain the coefficient of instruction current each harmonic component, the calculating of coefficient Method is:

$a_n=\frac{2}{N}\underset{n=0}{\overset{N-1}{\Σ}}{i}_L\·cos\left(\frac{kn\mathrm{\π}}{N}\right)$

$b_n=\frac{2}{N}\underset{n=0}{\overset{N-1}{\Σ}}{i}_L\·sin\left(\frac{kn\mathrm{\π}}{N}\right)$

Gained coefficient can obtain corresponding instruction current individual harmonic current AC compounent, respectively it is carried out record, By data inputting relief area, calculate active filter instruction current simultaneously:

I_ref=∑ (a_h+b_h) h be compensate overtone order

In described step 3, also include: when by actual current on line side each harmonic component, with the instruction current of current period In corresponding harmonic component compare respectively, comparative result is sent into recognition unit, when detect both difference amplitudes surpass Crossing preset range, control unit then thinks that system exists resonance.

In described step 4, described this subharmonic is removed in instruction, carries out being calculated as of resonance suppression:

I_{ref_adj}=I_ref-(a_m+b_m) m is overtone order at the resonant frequency detected

Above-mentioned algorithm resonance anti-for active filter has played defencive function, and its reason is: on the one hand be based on The parallel running characteristic of FPGA, greatly speeds up rapidity and the real-time of algorithm, and on the other hand, resonance suppression function strengthens and sets Standby reliability and the stability of electrical network, be substantially improved the grid-connected performance of active filter.

3, beneficial effect:

The present invention first detects load current in real time, is carried the harmonic components in load current by sliding window DFT algorithm Take, provide current reference i.e. instruction current for control module, and the harmonic components of the actual current detected in real time is carried Take, follow the tracks of the most rapidly and suppress, it is possible in the case of electrical network damping is more weak, protects equipment, prevent resonance Equipment is caused damage, simultaneously can the harmonic components of rapid extraction resonance current, internal module quickly responds, thus reaches humorous Shake suppression purpose.

Accompanying drawing explanation

The equivalent circuit of the grid-connected monophase system of Fig. 1

Detailed description of the invention

The present invention uses following control program: first pass through sliding window DFT algorithm, extracts each harmonic in load current, Specific as follows:

(1) according to Fourier decomposition, by load current i_LIt is expressed as form:

$i_L\left(t\right)=a_0+\underset{n=1}{\overset{\mathrm{\∞}}{\Σ}}(a_n\cos nwt+b_n\sin nwt)$

$a_n=\frac{1}{T}\underset{0}{\overset{T}{\∫}}u\left(t\right)cos\left(nwt\right)dt,b_n=\frac{1}{T}\underset{0}{\overset{T}{\∫}}u\left(t\right)sin\left(nwt\right)dt$

N takes 3 respectively, 5,7...

When n take 3 respectively, 5,7... etc. time, each harmonic composition being in networking load current.

After each harmonic composition superposition in networking load current, can be used as the current-order of master control algorithm.It is i.e. During system resonance free, after active filter equipment runs, inverter output current should be consistent with described instruction current, it may be assumed that

ILg=∑ a_n sin(nwt)+b_n cos(nwt)

N=2k+1, k=1,2,3...

Ask for algorithm according to Fourier Transform Coefficients, obtain the coefficient of instruction current each harmonic, its calculating side of coefficient Method is:

$a_n=\frac{2}{N}\underset{n=0}{\overset{N-1}{\Σ}}{i}_L\·cos\left(\frac{kn\mathrm{\π}}{N}\right)$

$b_n=\frac{2}{N}\underset{n=0}{\overset{N-1}{\Σ}}{i}_L\·sin\left(\frac{kn\mathrm{\π}}{N}\right)$

The coefficient of instruction current each harmonic is corresponding instruction current individual harmonic current alternating component, will refer to respectively Make electric current individual harmonic current AC compounent carry out record, by data inputting relief area, now calculate active filter instruction Electric current is:

I_ref=∑ (a_h+b_h) h be compensate overtone order

(3) monitor grid side actual current in real time, and will monitor grid side actual current through Fourier transformation It is carried out sliding window and extracts wherein each subharmonic composition, the one-tenth corresponding with harmonic components in the instruction current of current period Divide and compare respectively；

(4) comparative result is sent into recognition unit, after detecting that both difference amplitudes exceed preset range, control unit Then think that system exists resonance, and by following calculating, this subharmonic removed in instruction, carry out resonance suppression:

I_{ref_adj}=I_ref-(a_m+b_m) m is overtone order at the resonant frequency detected

Through above process, utilize the parallel running characteristic of FPGA, active filter can system for rapidly identifying whether There is vibration, it is possible to dynamic monitoring system output state, analyze harmonic components therein simultaneously, humorous by resonant frequency Ripple processes the purpose reaching resonance suppression, and rapidity is high, and resonance inhibition is obvious.

Although the present invention is open as above with preferred embodiment, but they are not for limiting the present invention, any ripe Practise this those skilled in the art, without departing from the spirit and scope of the invention, can make various changes or retouch from working as, the therefore guarantor of the present invention The scope of protecting should be with being as the criterion that claims hereof protection domain is defined.

Claims (5)

1. the anti-resonance control algolithm being applied to active filter, it is characterised in that:

Load current is detected in real time, by sliding window DFT algorithm, the harmonic component in load current is extracted, for control Molding block provides current reference to realize the harmonic restraining function of active filter, comprises the following steps:

Step 1: detect load current in real time, by sliding window DFT algorithm, extracts each harmonic component in load current；

Step 2: being overlapped by each harmonic component in the load current of extraction, its result is as the instruction of controller unit Electric current；

Step 3: monitor actual current on line side in real time, is carried out sliding window DFT algorithm and is extracted actual current on line side therein each time Harmonic component, the harmonic component corresponding with the instruction current of current period compares respectively, through the prison of comparing unit Survey and identify, it is judged that actual net surveys whether current system exists resonance；

Step 4: if resonance having been detected, then remove this corresponding subharmonic in instruction, carry out resonance suppression；It is right to achieve Resonance can carry out quick dynamic tracking suppression.

A kind of anti-resonance control algolithm being applied to active filter the most according to claim 1, it is characterised in that: described The algorithm extracting each harmonic component in networking load current in step 1 is:

According to Fourier decomposition, by load current i_LIt is expressed as form:

Wherein: when n take 3 respectively, 5,7... etc. time, networking load current each harmonic component can be drawn.

A kind of anti-resonance control algolithm being applied to active filter the most according to claim 1, it is characterised in that: described In step 2,

After each harmonic superposition that will extract, can be used as the current-order of master control algorithm, i.e. during system resonance free, active power filtering After device equipment runs, inverter output current should be consistent with the instruction current described in step 2, it may be assumed that

ILg=∑ a_nsin(nwt)+b_n cos(nwt)

N=2k+1, k=1,2,3...

Ask for algorithm according to Fourier Transform Coefficients, obtain the coefficient of instruction current each harmonic component, the computational methods of coefficient For:

Gained coefficient can obtain corresponding instruction current individual harmonic current AC compounent, respectively it is carried out record, by number According to typing relief area, calculating active filter instruction current simultaneously:

I_ref=Σ (a_h+b_h) h be compensate overtone order.

A kind of anti-resonance control algolithm being applied to active filter the most according to claim 1, it is characterised in that: described In step 3, also include: when by actual current on line side each harmonic component, corresponding with the instruction current of current period is humorous Wave component compares respectively, and comparative result is sent into recognition unit, when detecting that both exceed preset range at difference amplitude, and control Unit processed then thinks that system exists resonance.

A kind of anti-resonance control algolithm being applied to active filter the most according to claim 1, it is characterised in that: described In step 4,

Described this subharmonic is removed in instruction, carries out being calculated as of resonance suppression:

I_{ref_adj}=I_ref-(a_m+b_m) m is overtone order at the resonant frequency detected.