CN101719732A - five-level svpwm controller - Google Patents

Abstract

The invention relates to a five-level SVPWM controller, mainly comprising a sector judging module, a sector unity module, a new coordinate calculating module, an area judging module, an acting time calculating module, a pulse period selecting module, an on-off state generating module and a trigger pulse generating module. Each module of the five-level SVPWM controller adopts different neural network structures to complete the function thereof, and is realized by integrating on an FPGA chip EP1C6T144C. The five-level SVPWM controller has the advantages of applying a new coordinate system to judge the area of the reference voltage vector, simplifying the area judging method for the convenience of expanding to the SVPWM modulation of more multilevel inverters, integrating the five-level SVPWM controller on one FPGA chip to realize for offering a high-powered special SVPWM controller to high-voltage and high-capacity electrical energy conversion and speed regulation of an AC motor.

Description

Five-level SVPWM controller

Technical field

The present invention relates to a kind of SVPWM controller, especially a kind of five-level SVPWM controller.

Background technology

In recent years, diode clamp formula multi-electrical level inverter is used widely in high-voltage large-capacity transformation of electrical energy and speed regualtion of AC motor.Space voltage vector modulation method (SVPWM) is one of control technology of diode clamp formula multi-electrical level inverter, because it has less output harmonic wave content, higher voltage utilance and is easy to Digital Realization, become maximum, the most widely used control technology of research.Two level, 3 level space vector modulation controller are implementing more conveniently, but along with the increase of level number, space vector becomes geometric progression to increase, and this has brought very big difficulty for realization of space vector modulation.At present, space vector implementation algorithm at multi-electrical level inverter, mostly be on the basis of two level, 3 level space vector modulation technique, generally provide the general-purpose algorithm of multi-electrical level inverter, lack the specific algorithm of the space vector of multi-electrical level inverter and how to utilize the realization of problems such as concrete each switch vector, on off state to carry out careful research.

Summary of the invention

The objective of the invention is to overcome the deficiencies in the prior art, with the five-electrical level inverter is example, a kind of five-level SVPWM controller is provided, adopt new coordinate system to judge the residing zone of reference voltage vector, simplify the determination methods in zone, can conveniently expand in the SVPWM modulation of multi-electrical level inverter more, and five-level SVPWM controller is integrated on a slice fpga chip realizes, for high-voltage large-capacity transformation of electrical energy and speed regualtion of AC motor provide a high performance special-purpose SVPWM controller.

According to technical scheme provided by the invention, described five-level SVPWM controller comprises: sector discrimination module, sector normalizing module, new coordinate Calculation module, area judging module, action time, computing module, burst length section were selected module, on off state generation module, trigger impulse generation module;

Described sector discrimination module is according to the reference voltage vector u under the two-phase right angle rest frame _α ^*, u _β ^*Judge which of 6 sectors the synthesized reference space vector of voltage be arranged in; According to the symmetry of 6 sectors, adopt described sector normalizing module that reference voltage vector is carried out the place normalization rotation and obtain u _{α 1} ^*, u _{β 1} ^*, make u _{α 1} ^*, u _{β 1} ^*Be positioned at first sector; Described new coordinate Calculation module adopts the X1-X2-X3 coordinate system, to u _{α 1} ^*, u _{β 1} ^*Carry out coordinate Calculation, calculate u _{α 1} ^*, u _{β 1} ^*Synthesized reference space vector of voltage u ^*Distance with respect to three reference axis of X1-X2-X3; Described area judging module is according to synthesized reference space vector of voltage u ^*With respect to the distance of three reference axis of X1-X2-X3, judge synthesized reference space vector of voltage u ^*Be arranged in which of 16 little delta-shaped regions; Synthesized reference space vector of voltage u ^*Synthetic by three summit pairing three fundamental voltage space vectors effects of its residing delta-shaped region, described action time, computing module was then according to regional value and reference voltage vector u _{α 1} ^*, u _{β 1} ^*Calculate the action time of three fundamental voltage space vectors; Impulse waveform of five-level SVPWM was made of 25 time periods, and the on off state of three brachium pontis of the pairing inverter of different time sections in the zones of different is different, and described burst length section selection module is then calculated current time and is in which time period; Described on off state generation module is according to the on off state of time period, sector number, three brachium pontis of regional number decision inverter; Described trigger impulse generation module is exported 12 road trigger impulses according to the on off state of three brachium pontis of inverter, 12 road trigger impulses that generate can be given the pulse that peripheral drive circuit generates 24 tunnel complementary band dead band time-delays, receive the trigger end of 24 switching tubes of diode clamp formula five-electrical level inverter respectively;

Described X1-X2-X3 coordinate system is meant that the X1 axle overlaps with the α axle of two-phase rest frame, and the X2 axle is 60 ° of α axle inverse clock rotations, and the X3 axle is 60 ° of X2 axle inverse clock rotations.

Described new coordinate Calculation module adopts the neural net of 2-2-3 structure to realize that the weights of neural net hidden layer and input layer are [1,0; 1/2,

], threshold value is [0; 0]; The weights of output layer and hidden layer be [

], threshold value is [0; 0; 0], the activation primitive of all nodes all is a linear function; The output of described neural net divided by

And round up, obtain 3 integers of implicit area information, as the input signal of described area judging module.

Described area judging module adopts the BP neural net of 3-5-1 structure to realize, described BP neural net be input as synthesized reference space vector of voltage u ^*With respect to the distance of three reference axis of described X1-X2-X3 divided by

And 3 integers that obtain after rounding up; The weights and the threshold value of described neural net are obtained by off-line training, and training sample is 16 pairs of data, corresponding 16 little delta-shaped regions.

Described sector discrimination module, sector normalizing module, new coordinate Calculation module, area judging module, action time, computing module, burst length section selected module, on off state generation module and trigger impulse generation module to realize on fpga chip EP1C6T144C that all the input variable on the EP1C6T144C is the reference voltage u under the two-phase rest frame _α ^*, u _β ^*Pairing two 16 digital quantity is output as 12 road pwm pulse waveforms, and input, output signal take 44 pins of EP1C6T144C altogether.

Advantage of the present invention is: adopt new coordinate system to judge the residing zone of reference voltage vector, simplify the determination methods in zone, can conveniently expand in the SVPWM modulation of multi-electrical level inverter more; Adopt different neural network structures to finish the function of described each module, made full use of the parallel organization of neural net, the amount of calculation when having simplified work has greatly improved rapidity; Described five-level SVPWM controller is integrated into a slice fpga chip EP1C6T144C and goes up realization, utilize the parallel processing mechanism of FPGA that the parallel processing capability of described each neural net is demonstrated fully, for high-voltage large-capacity transformation of electrical energy and speed regualtion of AC motor provide a high performance special-purpose SVPWM controller, the computation burden of CPU in the minimizing system, convenient with the interface flexible of CPU in the system.

Description of drawings

Fig. 1 is the five-level SVPWM controller structured flowchart;

Fig. 2 is the sector distribution schematic diagram;

Fig. 3 is sector discrimination module neural net implementation structure figure;

Fig. 4 is sector normalizing modular neural network implementation structure figure;

Fig. 5 is new coordinate Calculation schematic diagram;

Fig. 6 is the reference voltage vector perspective view;

Fig. 7 is new coordinate Calculation modular neural network implementation structure figure;

Fig. 8 is area judging modular neural network implementation structure figure;

Fig. 9 is computing module neural net implementation structure figure action time;

Figure 10 is that the burst length section is selected modular neural network implementation structure figure;

Figure 11 is on off state generation module neural net implementation structure figure;

Figure 12 is trigger impulse generation module neural net implementation structure figure.

Embodiment

Below in conjunction with accompanying drawing specific embodiments of the present invention are described further.

As shown in Figure 1: the present invention mainly comprises: sector discrimination module, sector normalizing module, new coordinate Calculation module, area judging module, action time, computing module, burst length section were selected module, on off state generation module, trigger impulse generation module.

As shown in Figure 2: adopt neural net NN1, according to the reference voltage u under the two-phase rest frame _α ^*, u _β ^*Judge which of 6 sectors the synthesized reference space vector of voltage be arranged in.The structure of described NN1 is 2-3-4-1 (2-3-4-1 represents the layer structure of neural net, hereinafter all adopts this expression mode), wherein, the activation primitive of first hidden layer is the hard-limiting function, last amplitude limit is 1, the weights that following amplitude limit is 0, the first hidden layer and input layer be [ 1;

1; 1,0], threshold value is [0; 0; 0]; As input, the network that constitutes with second hidden layer and output layer is the BP network with the output of first hidden layer, adopts off-line training to obtain its weights and threshold value, and training sample is corresponding 6 sectors of 6 pairs of data.The output of network is through the integer of back output that rounds up, and its scope is 1～6, the corresponding I-VI of difference sector, as shown in Figure 3.Among Fig. 3: these 6 sectors of I-VI are divided into 6 equal portions to the plane, and the border of each sector is respectively 2 fundamental voltage space vectors.

Sector normalizing module realizes that mainly the vector among the II-VI of sector rotates among the I of sector, so that the action time of judgement zone, calculating fundamental voltage space vector etc.Sector normalizing module adopts the neural net NN2 of 1-3-4 structure as shown in Figure 4 and the neural net NN3 of 2-2 structure to realize.Among Fig. 4: NN2 is the BP network, and it is input as sector number, is output as the rotation transformation coefficient that different sectors rotate to sector I, and weights and the threshold value of NN2 are obtained by off-line training, and training sample is 6 pairs of data that sector number and conversion coefficient are formed, corresponding 6 sectors; The output of NN2 is as the weights of NN3, the input u of NN3 _α ^*, u _β ^*Be former reference voltage vector, output u _{α 1} ^*, u _{β 1} ^*Be the reference voltage vector behind the normalizing.

As shown in Figure 5: construct a new rest frame, the X1-X2-X3 coordinate system, wherein, the X1 axle overlaps with the α axle of two-phase right angle rest frame, and the X2 axle is 60 ° of α axle inverse clock rotations, and the X3 axle is 60 ° of X2 axle inverse clock rotations.Three fixed points of 2 space vector of voltage of I sector borders are coupled together, constitute a big triangle, the big triangle of I fan is divided into 16 five equilibriums, respectively a corresponding 1-16 delta-shaped region; V1～V14 is the pairing fundamental voltage space vector of the different conditions of inverter switching device pipe; Synthesized reference voltage vector u ^*Amplitude be U ^*, with the angle of α axle be θ; From synthesized reference voltage vector u ^*Terminal point make vertical line to X1, X2, X3 axle respectively, ask for that it is as follows apart from UX1, UX2, UX3:

For fear of ask apart from the time trigonometric function calculating, on the one hand the synthesized reference voltage vector is got n1, n2 to X1, X2 axial projection, as shown in Figure 6.Among Fig. 6:

Therefore, can solve according to formula (2):

$\left\{\begin{array}{c}\sin \mathrm{\&theta;}=1/\left(\sqrt{3}{U}^{*}\right)(2n_2-n_1)\\ \sin (\mathrm{\&theta;}+\mathrm{\&pi;}/3)=1/\left(\sqrt{3}{U}^{*}\right)({\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="H2009102323393E00012.png"\; state="\{\{state\}\}">n</figure-callout>}}_1+n_2)\\ \sin (\mathrm{\&theta;}-\mathrm{\&pi;}/3)=1/\left(\sqrt{3}{U}^{*}\right)(2n_1-n_2)\end{array}\right.---\left(3\right).$

On the other hand, the reference voltage u that will represent with the two-phase rest frame _{α 1} ^*, u _{β 1} ^*Project to X1, X2 axle respectively, its projection value is obviously identical with n1, n2, that is:

$\left\{\begin{array}{c}\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="H2009102323393E00012.png"\; state="\{\{state\}\}">n</figure-callout>}1=u_{\mathrm{\&alpha;}1}^{*}\\ n2=u_{\mathrm{\&alpha;}1}^{*}/2+\sqrt{3}{u}_{\mathrm{\&beta;}1}^{*}/2\end{array}\right.---\left(4\right).$

The trigonometric function of trying to achieve is updated to formula (1) can obtains distance value, with UX1, UX2, UX3 divided by

And round up, obtain 3 integers, represent the relative position information of reference voltage vector and three axles of X1-X2-X3 respectively, the integer span of output is 1～4, V _DcBe inverter direct-flow side voltage.New coordinate Calculation module realizes that with neural network structure as shown in Figure 7 among Fig. 7: NN4 is the neural net of 2-2-3 structure, and the activation primitive of hidden layer is a linear function, and the weights of hidden layer and input layer are [1,0; 1/2,

], threshold value is [0; 0]; The activation primitive of output layer is a linear function, the weights of output layer and hidden layer be [

], threshold value is [0; 0; 0].

It is that the BP neural net NN6 of 3-5-1 realizes that the area judging module adopts as shown in Figure 8 structure, among Fig. 8: UX_1, UX_2, UX_3, as the input signal of described neural net NN6, by UX1, UX2, UX3 divided by

And round up and obtain, the output of described neural net NN6 is through obtaining representing the integer of regional number after rounding up; Weights and the threshold value of described neural net NN6 are obtained by off-line training, and training sample is 16 pairs of data, corresponding 16 little delta-shaped regions.

The present invention is primarily aimed at the undermodulation pattern, synthesized reference space vector of voltage u ^*Three summit pairing three fundamental voltage space vectors effects by its residing triangular form zone are synthetic, according to the residing zone of reference voltage vector, in conjunction with the weber balance principle, calculate T action time of described three fundamental voltage space vectors _a, T _b, T _c, the trigonometric function value that relates in the computational process replaces with formula (2), does not therefore have the calculating of trigonometric function.Action time, generation module adopted neural network structure as shown in Figure 9 to realize that among Fig. 9: neural net NN7 is that structure is the BP network of 1-5-9, and weights and the threshold value of NN7 are obtained by off-line training; Training sample is in 16 zones, T _a, T _b, T _c16 pairs of data that the n1 that is comprised in the calculation expression, n2, Ts coefficient are constituted, Ts is the time of an impulse waveform, i.e. modulation period; Neural net NN8 is the BP network of 3-3 structure, and its weights are the output of NN7, and threshold value is 0 entirely.

Impulse waveform of five-level SVPWM was made of 25 time periods, the on off state of pairing three brachium pontis of different time sections in the zones of different is different, burst length section selection module is then calculated current time and is in which time period, adopts neural network structure shown in Figure 10 to realize.Among Figure 10: neural net NN9 structure is 4-24-1, and T is to be the output valve of the timer of Ts in the cycle, and the hidden layer activation primitive is the hard-limiting function, and last amplitude limit is 1, and following amplitude limit is 0; Weights between hidden layer and input layer can obtain according to coefficient calculations action time that impulse waveform is formed fragment, and threshold value is 0 entirely; The activation primitive of output layer is a linear function, and weights are 1 entirely, and threshold value is 25; The output area of NN9 is 1～25 integer, represents 25 time slices.

The on off state generation module then determines the on off state of three brachium pontis according to time period, sector number, regional number, the different different trigger impulses of on off state output adopts neural network structure shown in Figure 11 to realize the on off state generation module.Among Figure 11: neural net NN10 is the BP network of 3-20-20-3 structure, and input is respectively sector number, regional number and time period, is output as the on off state S of 3 brachium pontis _a, S _b, S _c, S _a, S _b, S _cValue all be 4,3,2,1,0 to represent five kinds of states respectively.Weights, the threshold value of NN10 are obtained by off-line training, and training sample is that 6 (sector number) * 16 (number of regions) * 25 (time hop count)=2400 pairs of data constitute.

The trigger impulse generation module is then exported 12 road trigger impulses according to the on off state of three brachium pontis, and 12 road trigger impulses can be given the pulse that peripheral drive circuit generates 24 tunnel complementary band dead bands, receives the trigger end of 24 switching tubes respectively.The trigger impulse generation module is realized by 3 identical neural nets, is example with A phase brachium pontis, as shown in figure 12.Among Figure 12: neural net NN12 is the BP network of 1-4-4 structure, and weights, the threshold value of NN12 are obtained by off-line training, and training sample is that different on off states and pairing triggering signal constitute 5 pairs of data, corresponding A phase brachium pontis on off state S _a5 states.

Described sector discrimination module, sector normalizing module, new coordinate Calculation module, area judging module, action time, computing module, burst length section were selected module, on off state generation module, trigger impulse generation module all to be integrated in a slice fpga chip EP1C6T144C to go up and realize that input variable is the reference voltage u under the two-phase rest frame _α ^*, u _β ^*Pairing two 16 digital quantity is output as 12 road pwm pulse waveforms, and input/output signal takies 44 pins of EP1C6T144C altogether.

The present invention has following characteristics: adopt new coordinate system to judge the residing zone of reference voltage vector, simplify the determination methods in zone, can conveniently expand in the SVPWM modulation of multi-electrical level inverter more; Adopt different neural network structures to finish the function of described each module, take full advantage of the parallel organization of neutral net, the amount of calculation when greatly having simplified the work of SVPWM controller has improved rapidity; Described five-level SVPWM controller is integrated into the upper realization of a slice fpga chip EP1C6T144C, utilize the parallel processing mechanism of FPGA that the parallel processing capability of described each neutral net is demonstrated fully, the controller of a high-performance special use is provided for high-voltage large-capacity transformation of electrical energy and Speed Adjustment of AC Motor, the computation burden of CPU in the minimizing system, convenient with the interface flexible of CPU in the system.

Claims (4)

1. five-level SVPWM controller is characterized in that comprising: sector discrimination module, sector normalizing module, new coordinate Calculation module, area judging module, action time, computing module, burst length section were selected module, on off state generation module, trigger impulse generation module;

Described sector discrimination module is according to the reference voltage vector u under the two-phase right angle rest frame _α ^*, u _β ^*Judge which of 6 sectors the synthesized reference space vector of voltage be arranged in; According to the symmetry of 6 sectors, adopt described sector normalizing module that reference voltage vector is carried out the place normalization rotation and obtain u _{α 1} ^*, u _{β 1} ^*, make u _{α 1} ^*, u _{β 1} ^*Be positioned at first sector; Described new coordinate Calculation module adopts the X1-X2-X3 coordinate system, to u _{α 1} ^*, u _{β 1} ^*Carry out coordinate Calculation, calculate u _{α 1} ^*, u _{β 1} ^*Synthesized reference space vector of voltage u ^*Distance with respect to three reference axis of X1-X2-X3; Described area judging module is according to synthesized reference space vector of voltage u ^*With respect to the distance of three reference axis of X1-X2-X3, judge synthesized reference space vector of voltage u ^*Be arranged in which of 16 little delta-shaped regions; Synthesized reference space vector of voltage u ^*Synthetic by three summit pairing three fundamental voltage space vectors effects of its residing delta-shaped region, described action time, computing module was then according to regional value and reference voltage vector u _{α 1} ^*, u _{β 1} ^*Calculate the action time of three fundamental voltage space vectors; Impulse waveform of five-level SVPWM was made of 25 time periods, and the on off state of three brachium pontis of the pairing inverter of different time sections in the zones of different is different, and described burst length section selection module is then calculated current time and is in which time period; Described on off state generation module is according to the on off state of time period, sector number, three brachium pontis of regional number decision inverter; Described trigger impulse generation module is exported 12 road trigger impulses according to the on off state of three brachium pontis of inverter, 12 road trigger impulses that generate can be given the pulse that peripheral drive circuit generates 24 tunnel complementary band dead band time-delays, receive the trigger end of 24 switching tubes of diode clamp formula five-electrical level inverter respectively;

Described X1-X2-X3 coordinate system is meant that the X1 axle overlaps with the α axle of two-phase rest frame, and the X2 axle is 60 ° of α axle inverse clock rotations, and the X3 axle is 60 ° of X2 axle inverse clock rotations.

2. five-level SVPWM controller according to claim 1 is characterized in that described new coordinate Calculation module adopts the neural net of 2-2-3 structure to realize that the weights of neural net hidden layer and input layer are [1,0; 1/2, ], threshold value is [0; 0]; The weights of output layer and hidden layer be [

], threshold value is [0; 0; 0], the activation primitive of all nodes all is a linear function; The output of described neural net divided by And round up, obtain 3 integers of inclusion region information, as the input signal of described area judging module.

3. five-level SVPWM controller according to claim 1 is characterized in that described area judging module adopts the BP neural net of 3-5-1 structure to realize, described BP neural net be input as synthesized reference space vector of voltage u ^*With respect to the distance of three reference axis of described X1-X2-X3 divided by And 3 integers that obtain after rounding up; The weights and the threshold value of described neural net are obtained by off-line training, and training sample is 16 pairs of data, corresponding 16 little delta-shaped regions.

4. five-level SVPWM controller according to claim 1, it is characterized in that described sector discrimination module, sector normalizing module, new coordinate Calculation module, area judging module, action time computing module, burst length section select module, on off state generation module and trigger impulse generation module to realize on fpga chip EP1C6T144C that all the input variable on the EP1C6T144C is the reference voltage u under the two-phase rest frame _α ^*, u _β ^*Pairing two 16 digital quantity is output as 12 road pwm pulse waveforms, and input, output signal take 44 pins of EP1C6T144C altogether.

Images