CN103886131B - A kind of switched reluctance motor flux linkage line modeling method based on extreme learning machine - Google Patents

Abstract

The present invention discloses a kind of extreme learning machine magnetic linkage line modeling method of switched reluctance machines, belongs to the technical field of switched reluctance machines Based Intelligent Control.Present invention static data based on switched reluctance machines collection, use extreme learning machine, set up the magnetic linkage off-line model of switched reluctance machines, on this basis, regulating in real time according to error, set up the magnetic linkage on-time model of switched reluctance machines, magnetic linkage on-time model has real-time online and improves the function of flux linkage model accuracy, switched reluctance machines dynamic property can be described more accurately, adapt to the different operating environment of switched reluctance machines.

Description

A kind of switched reluctance motor flux linkage line modeling method based on extreme learning machine

Technical field

The present invention relates to a kind of switched reluctance motor flux linkage line modeling method based on extreme learning machine, belong to switch magnetic Resistance intelligent motor control field.

Background technology

Flux linkage characteristic is the key property of switched reluctance machines.Due to switched reluctance machines double-salient-pole structure and run time The features such as magnetic circuit is the most saturated, accurate perception flux linkage characteristic is for optimizing design of electrical motor, improving runnability, realize position sensorless Deng the most significant.The flux linkage characteristic of switched reluctance machines can be obtained by the method that FEM calculation or experiment are measured. Finite Element Method Consideration is complicated, computationally intensive, and therefore measurement method is still that the main method obtaining flux linkage characteristic at present.

The stator magnetic linkage of switched reluctance machines is the nonlinear function of rotor-position and winding current, sets up accurately practical Flux linkage model interesting.Although traditional look-up table has higher precision, but the cycle of calculating is long, it is impossible to meet control in real time System and the requirement of design of electrical motor rapid modeling.Function analytic method can optimize systematic function to a certain extent, but to load and The adaptability of environmental change is the strongest.Along with the development of artificial intelligence technology, learning algorithm answering in Modeling of Switched Reluctance Motors With more and more extensive, the switched reluctance motor flux linkage modeling method being suggested has: BP neutral net, RBF neural, Hold vector machine etc., and BP neutral net, RBF neural need mass data ability real in switched reluctance motor flux linkage modeling Existing, and pace of learning is slow；Without intelligent algorithm (such as particle cluster algorithm, genetic algorithm etc.), support vector machine parameter is being optimized In the case of, the flux linkage model precision that support vector machine study obtains is general；Extreme learning machine is a kind of single hidden layer feed forward neural Network, without mass data in the magnetic linkage modeling process realize switched reluctance machines, study " extremely " rapidly, and opening of obtaining Closing reluctance motor flux linkage model accuracy the highest, above reason makes switched reluctance motor flux linkage modeling side based on extreme learning machine Method, has the highest researching value.Meanwhile, the modeling of existing switched reluctance motor flux linkage is substantially off-line modeling, the magnetic obtained Chain off-line model is difficult in adapt to the change of switched reluctance machines working environment, and some scholars is online at the magnetic linkage of switched reluctance machines Modeling aspect is made that research, wherein has a kind of method to be: with the magnetic linkage off-line model that pre-builds to switched reluctance machines Magnetic linkage is predicted, it was predicted that result magnetic linkage data real-time with switched reluctance machines are compared, and obtain the magnetic pre-build Chain off-line model real-time estimate error, re-establishes flux linkage model online according to real-time Flux estimation error, having of the method Effect property will necessarily require: online switched reluctance machines modeling process again needs quickly to realize, and this point requires at switching magnetic-resistance When motor runs up especially prominent, and this magnetic linkage line modeling method uses, and pace of learning is relatively slow, need mass data Carry out the RBF neural learnt, largely limit this magnetic linkage line modeling method at switched reluctance machines in higher Speed, run up under application, extreme learning machine the most well solves this problem.

Summary of the invention

It is an object of the invention to provide a kind of switched reluctance motor flux linkage line modeling method based on extreme learning machine.

The discrete integration formula of switched reluctance machines winding magnetic linkage is as follows:

ψ (k)=ψ (k-1)+0.5T [u (k)-ri (k)+u (k-1)-ri (k-1)] (1)

In formula (1),

ψ phase winding magnetic linkage value,

U phase winding terminal voltage,

The T sampling period,

I phase current,

R phase winding internal resistance,

K sampled point sequence number.

Magnetic linkage data acquisition needed for the foundation of switched reluctance motor flux linkage model, is based on having delivered disclosed switch magnetic Resistance motor flux linkage characteristic detecting system, its voltage using step voltage measurement phase winding and electric current, core is DSP.Drive line In the module of road, 220V alternating voltage through transformer pressure-reducing, four diodes easy rectification device rectification after obtain required DC voltage；For maintaining voltage constant and preventing power supply and winding to constitute lc circuit vibration, a relatively great Rong in parallel after power supply The electrochemical capacitor C, disconnecting link S of value is electric capacity C charging when closing；Diode VD is used for being that phase winding continues when MOS switch pipe disconnects Stream；R is current-limiting resistance, prevents that winding current is excessive burns device.In detection line module, R₁、R₂Survey for phase winding terminal voltage Amount, resistance is bigger so that R₁、R₂Line current is less, so that being negligible All other routes impact；R_CFor current sample Resistance.Rotor end fixes a rotary encoder, in real time rotor Angle Position is converted into electric impulse signal.If By R₂、R_CBoth end voltage is designated as u respectively₂、u_c, then formula (1) is rewritable is:

$\mathrm{\ψ}\left(k\right)=\mathrm{\ψ}(k-1)+0.5T[\frac{R_1+R_2}{R_2}{u}_2\left(k\right)-r\frac{u_c\left(k\right)}{R_C}+\frac{R_1+R_2}{R_2}{u}_2(k-1)-r\frac{u_c(k-1)}{R_C}]---\left(2\right)$

DSP, as the core of whole flux measurement system, carries out periodic sampling to all kinds of signals of telecommunication in real time, is calculated Real-time magnetic linkage data, are then transferred to host computer and carry out magnetic linkage modeling.

The simplest and the clearest, clear for making following summary of the invention introduce, spy makees following parameter and defines:

Ω static data collection,

ψ_k1The real-time estimate result of the flux linkage model having built up,

ψ_k0The real-time sampling result of magnetic linkage data,

δ_kThe absolute relative error of flux linkage model real-time estimate result,

The ε flux linkage model Relative Error absolute value upper limit.

On the basis of switched reluctance machines static data collection Ω, limit of utilization learning machine successfully carries out switched reluctance machines Magnetic linkage off-line modeling.Magnetic linkage data are carried out real when still switch word reluctance motor being run up with above-mentioned flux linkage characteristic detecting system Time sampling, carry out magnetic linkage comparing with the flux linkage model having built up, draw the Relative Error that flux linkage model is real-time Absolute value δ_k:

${\mathrm{\δ}}_k=\left|\frac{{\mathrm{\ψ}}_{k1}-{\mathrm{\ψ}}_{k0}}{{\mathrm{\ψ}}_{k0}}\right|\×100\%---\left(3\right)$

Experiment proves: the absolute relative error that the flux linkage model that extreme learning machine is set up predicts the outcome can reach 0.001 order of magnitude, simultaneously take account of setting bigger flux linkage model predict the outcome the absolute relative error upper limit can reduce right Host computer hardware resource takies and the requirement to flux linkage model precision of prediction of the motor workplace of reality, can set magnetic linkage Model prediction absolute relative error higher limit ε=1%～5%.Work as δ_kDuring ＞ ε, just the magnetic linkage sampled data in this moment is added In Ω, re-start magnetic linkage modeling, repeat constantly this operation, until δ_k≤ ε, stops the online of magnetic linkage and again models, and with this Time the switched reluctance motor flux linkage model that re-establishes as the flux linkage model under this working environment, to improve flux linkage model pair The adaptation ability of different operating environment.

It is an advantage of the current invention that:

1. extreme learning machine learning process is without mass data, and the model prediction accuracy of foundation is high, and learning process is fast Speed, therefore, operating limit learning machine carries out the magnetic linkage modeling of switched reluctance machines, it is to avoid use traditional neural network is (such as BP Neutral net, RBF neural), support vector machine carry out the problem existing for magnetic linkage modeling.

2. establish online switched reluctance motor flux linkage modeling method, fully taken into account in motor practical engineering application Situation about being likely encountered, can describe the switched reluctance motor flux linkage characteristic in real work exactly, have the strongest transplantability.

3. the switched reluctance machines online magnetic linkage modeling method illustrated by this patent realizes, to liking entity motor, being different from The existing magnetic linkage modeling method based on the Realization of Simulation of part, has higher application.

Accompanying drawing explanation

Fig. 1 is the structural representation of the switched reluctance motor flux linkage characteristic detecting system with DSP as core；

Fig. 2 is detection line module；

In figure: 1, driver circuit module；2, detection line module；3, switched reluctance machines；4、DSP；5, host computer.

Detailed description of the invention

The discrete integration formula of switched reluctance machines winding magnetic linkage is as follows:

ψ (k)=ψ (k-1)+0.5T [u (k)-ri (k)+u (k-1)-ri (k-1)] (1)

In formula (1),

ψ is phase winding magnetic linkage value,

U is phase winding terminal voltage,

T is the sampling period,

I is phase current,

R is phase winding internal resistance,

K is sampled point sequence number.

Magnetic linkage data acquisition needed for the foundation of switched reluctance motor flux linkage model, is based on having delivered disclosed switch magnetic Resistance motor flux linkage characteristic detecting system, its voltage using step voltage measurement phase winding and electric current, core is DSP, such as Fig. 1 institute Show.In driver circuit module 1,220V alternating voltage through transformer pressure-reducing, four diodes easy rectification device rectification after To required DC voltage；For maintaining voltage constant and preventing power supply and winding to constitute lc circuit vibration, in parallel after power supply It it is electric capacity C charging during the electrochemical capacitor C of one bigger capacitance, disconnecting link S Guan Bi；Diode VD is for when MOS switch pipe disconnects For phase winding afterflow；R is current-limiting resistance, prevents that winding current is excessive burns device.In detection line module 2, R₁、R₂For phase Winding terminal voltage measurement, resistance is bigger so that R₁、R₂Line current is less, so that being negligible All other routes impact； R_CFor current sampling resistor.Rotor end fixes a rotary encoder, is converted into by rotor Angle Position in real time Electric impulse signal.If by R₂、R_CBoth end voltage is designated as u respectively₂、u_c, as in figure 2 it is shown, then formula (1) is rewritable it is:

$\mathrm{\ψ}\left(k\right)=\mathrm{\ψ}(k-1)+0.5T[\frac{R_1+R_2}{R_2}{u}_2\left(k\right)-r\frac{u_c\left(k\right)}{R_C}+\frac{R_1+R_2}{R_2}{u}_2(k-1)-r\frac{u_c(k-1)}{R_C}]---\left(2\right)$

DSP, as the core of whole flux measurement system, carries out periodic sampling to all kinds of signals of telecommunication in real time, is calculated Real-time magnetic linkage data, are then transferred to host computer and carry out magnetic linkage modeling.

The simplest and the clearest, clear for making following summary of the invention introduce, spy makees following parameter and defines:

Ω is static data collection,

ψ_k1The real-time estimate result of the flux linkage model for having built up,

ψ_k0For the real-time sampling result of magnetic linkage data,

δ_kFor the absolute relative error of flux linkage model real-time estimate result,

ε is the flux linkage model Relative Error absolute value upper limit.

On the basis of switched reluctance machines static data collection Ω, limit of utilization learning machine successfully carries out switched reluctance machines Magnetic linkage off-line modeling.Magnetic linkage data are carried out real when still switch word reluctance motor being run up with above-mentioned flux linkage characteristic detecting system Time sampling, carry out magnetic linkage comparing with the flux linkage model having built up, draw the Relative Error that flux linkage model is real-time Absolute value δ_k:

${\mathrm{\δ}}_k=\left|\frac{{\mathrm{\ψ}}_{k1}-{\mathrm{\ψ}}_{k0}}{{\mathrm{\ψ}}_{k0}}\right|\×100\%---\left(3\right)$

Experiment proves: the absolute relative error that the flux linkage model that extreme learning machine is set up predicts the outcome can reach 0.001 order of magnitude, simultaneously take account of setting bigger flux linkage model predict the outcome the absolute relative error upper limit can reduce right Host computer hardware resource takies and the requirement to flux linkage model precision of prediction of the motor workplace of reality, can set magnetic linkage Model prediction absolute relative error higher limit ε=1%～5%.Work as δ_kDuring ＞ ε, just the magnetic linkage sampled data in this moment is added In Ω, re-start magnetic linkage modeling, repeat constantly this operation, until δ_k≤ ε, stops the online of magnetic linkage and again models, and with this Time the switched reluctance motor flux linkage model that re-establishes as the flux linkage model under this working environment, to improve flux linkage model pair The adaptation ability of different operating environment.

Embodiment of the present invention are specifically divided into following 4 steps:

1. use the switched reluctance motor flux linkage characteristic detecting system with DSP as core to gather the quiet of switched reluctance machines 3 State data, obtain static data collection, and the learning training method off-line of limit of utilization learning machine sets up the magnetic linkage of switched reluctance machines 3 Model.

2., when switched reluctance machines 3 runs up, gather real-time magnetic linkage data ψ of switched reluctance machines 3_k0, and built Magnetic linkage data ψ that the flux linkage model of the switched reluctance machines 3 stood is predicted_k1Contrast, according to formula It is calculated the Relative Error absolute value δ that flux linkage model is real-time_k。

3. set the Relative Error absolute value δ that flux linkage model is real-time_kHigher limit ε, ε takes in the range of 1%～5% One fixed value.

4. by Relative Error absolute value δ real-time for flux linkage model_kWith the higher limit ε comparison set, work as δ_kDuring ＞ ε, will The magnetic linkage data in this moment add static data and concentrate, and re-establish flux linkage model；Repeat constantly this operation, until δ_k≤ ε, stops Only the online of magnetic linkage models again, and using the switched reluctance motor flux linkage model that now re-establishes as this working environment under Flux linkage model.

Claims (1)

1. a switched reluctance motor flux linkage line modeling method based on extreme learning machine, it is characterised in that include walking as follows Rapid:

1) the switched reluctance motor flux linkage characteristic detecting system with DSP as core is used to gather the static state of switched reluctance machines (3) Data, obtain static data collection, and the learning training method off-line of limit of utilization learning machine sets up the magnetic linkage of switched reluctance machines (3) Model；

2), when switched reluctance machines (3) runs up, real-time magnetic linkage data ψ of switched reluctance machines (3) are gathered_k0, and built Magnetic linkage data ψ that the flux linkage model of the switched reluctance machines (3) stood is predicted_k1Contrast, according to formulaIt is calculated the Relative Error absolute value δ that flux linkage model is real-time_k；

3) the Relative Error absolute value δ that flux linkage model is real-time is set_kHigher limit ε, ε takes one in the range of 1%～5% Individual fixed value；

4) by Relative Error absolute value δ real-time for flux linkage model_kWith the higher limit ε comparison set, work as δ_kDuring ＞ ε, during by this The magnetic linkage data carved add static data and concentrate, and re-establish flux linkage model；Constantly repeat this operation, until δ_k≤ ε, stops magnetic The online of chain models again, and using the switched reluctance motor flux linkage model that now re-establishes as the magnetic under this working environment Chain model.