CN101299581A - Neural network generalized inverse coordination control frequency transformer for two induction machines and construction method thereof - Google Patents
Neural network generalized inverse coordination control frequency transformer for two induction machines and construction method thereof Download PDFInfo
- Publication number
- CN101299581A CN101299581A CNA2008100196312A CN200810019631A CN101299581A CN 101299581 A CN101299581 A CN 101299581A CN A2008100196312 A CNA2008100196312 A CN A2008100196312A CN 200810019631 A CN200810019631 A CN 200810019631A CN 101299581 A CN101299581 A CN 101299581A
- Authority
- CN
- China
- Prior art keywords
- generalized inverse
- neural network
- neural net
- input
- induction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention discloses a neural net general inverse coordination control frequency converter of two induction machines and a manufacture method thereof consisting of a linear closed-loop controller, a neural net general inverse and a compound controlled object composing of a magnetic linkage observer, two controlled induction machines and a flow control inverter including the coordinate transformation and the common load connected together. The neural net general inverse is arranged at the front of the compound controlled object to compose the pseudolinear system, also the linear closed-loop controller are provided on the basis of the pseudolinear system. The invention realizes the dynamic decoupling of the rotor magnetic linkage and the speed of each induction machine and the decoupling control between the speed and the strain of the two induction machines, obtains the excellent adjustment performance of the speed and the strain, advances the robustness of the motor parametric variation, the load disturbance and the network delay variation.
Description
Technical field
The present invention relates to a kind of The synchronized Coordinative Control frequency converter and building method thereof of two AC induction motor systems, be applicable to that two induction alternating current (AC) motors drive the control of common load (band shape load), and can realize networking, belong to the technical field of Electric Drive control appliance.
Background technology
At present, in industrial production, exist in a large number by two induction alternating current (AC) motors (abbreviation induction machine) even multiple electric motors and drive common load (as the band shape load etc.), the most of direct current machine that adopts of system with identical speed synchronous coordination operation.Because the DC motor structure complexity is difficult in maintenance, and has the commutation problem, this brings inconvenience to use.Adopt frequency converter to drive many fields that induction machine has been widely used in the transmission of former employing direct current machine at present, a frequency converter can be controlled an induction machine preferably and drive load running.But concerning system's (being called for short the two induction system) of two common load runnings of driven by motor, only adopt frequency converter can't satisfy the constant tension force of band shape load maintenance and the actual requirement of system synchronization coordinated operation.Because in two induction, not only have the speed of every induction machine and intercoupling of rotor flux, and because the tension force of two induction is relevant with the difference of the speed of two induction machines, so there be intercoupling between the speed of two induction and the tension force.The two induction system is the coupled system of the multivariable nonlinearity of a complexity, adopt constant voltage and frequency ratio control of conversion device or adopt vector-control frequency converter all to be difficult to make two induction machines that drive common load to realize high performance synchronous coordination operation, particularly networked system no matter be.
The employing method is to increase the The synchronized Coordinative Control device outside the two induction system at present, clearly increases the The synchronized Coordinative Control device cost of system is uprised, and realizes simultaneously difficulty being difficult to the high performance synchronous coordinated operation that reaches real.Therefore concerning two induction, adopting frequency converter to add the working method that the induction motor adds the The synchronized Coordinative Control device usually is not the most effective control mode.
For from improve in essence that two induction changes parameter and the network delay of adaptability, robustness and the networked system of disturbance to the influence of system, realize the speed of two induction and the decoupling zero control of tension force, and then the runnability of raising two induction The synchronized Coordinative Control, realize real high performance synchronous coordinated operation, need to adopt some new control technologys and new control method.
Neural net is a subject that develops rapidly in recent years, it is by the network of a large amount of processing unit PE by the complexity that is connected to form widely, nerual network technique has been incorporated in parameter Estimation and the System Discrimination, the neural net estimating techniques have been applied to subjects such as biology, medical treatment, electronics, mathematics, physics and engineering, realize by setting up model.Generalized regression nerve networks commonly used is made up of input layer, one deck or multilayer hidden layer and output layer, adopt totally interconnected connecing between each layer, but with not interconnecting between one deck unit, as long as abundant neuron is arranged in hidden layer, multitiered network just can be used for approaching almost any one nonlinear function, the output that just can obtain requiring according to one group of specific input.
Summary of the invention
The purpose of this invention is to provide and a kind ofly both can make every motor have good dynamic and static control performance, anti-parameter of electric machine variation and anti-load disturbance ability are strong, can improve every control performance index of two induction again effectively, as the neural net generalized inverse The synchronized Coordinative Control frequency converter of the two induction of dynamic responding speed, steady-state tracking precision, parameter robustness and network delay influence; Another object of the present invention provides the building method of the neural net generalized inverse The synchronized Coordinative Control frequency converter of this two induction.
The technical scheme that the neural net generalized inverse coordination control frequency transformer of two induction of the present invention adopts is: comprise flux observer, front end at two induction machines connects frequency converter, the rear end connects common load by conveyer belt, described frequency converter is by the linear closed-loop controller, the neural net generalized inverse The synchronized Coordinative Control frequency converter that neural net generalized inverse and composite controlled object connect and compose, described composite controlled object is by flux observer, controlled two induction machines with comprise that the Flow Control inverter and the common load of the expansion of coordinate transform connect to form, place composite controlled object to form pseudo-linear system before described neural net generalized inverse, this pseudo-linear system is by two linear subsystems that the magnetic linkage single order is stable, a stable linear subsystem and the stable linear subsystem of tension force second order of speed single order is formed; On the pseudo-linear system basis, constitute two magnetic linkage control devices, a speed control and a tension controller of linear closed-loop controller.
The technical scheme that the building method of the neural net generalized inverse Synchronization Control frequency converter of two induction of the present invention adopts, adopt electric current commonly used, speed flux observation model and Clarke Clark conversion to form two flux observers earlier, also in turn include the following steps
(1) forms the two Flow Control inverters of expanding jointly by current-controlled voltage source inverter, contrary Parker Park conversion and contrary Clarke Clark conversion;
(2) with the Flow Control inverters of two expansions and two induction machines and load thereof as a composite controlled object;
(3) composite controlled object is adopted the static neural network of 9 input nodes, 4 output nodes add 4 and pass letters and an integration comes the constructing neural network generalized inverse, make the neural net generalized inverse realize the generalized inverse systemic-function of composite controlled object by each weight coefficient of adjusting static neural network;
(4) the neural net generalized inverse is serially connected in before the composite controlled object, it is two rotor flux subsystems and a speed subsystem and the pseudo-linear system that the second order subsystem constitutes that neural net generalized inverse and composite controlled object synthesize by three single order subsystems;
(5) on the pseudo-linear system basis, make two magnetic linkage control devices, a speed control and a tension controller respectively and form the linear closed-loop controller, finally form neural net generalized inverse The synchronized Coordinative Control frequency converter.
The present invention is by the constructing neural network generalized inverse, will be to this multivariable of two induction, close coupling, the time become non linear system control be converted into two rotor fluxs, the control of the second order regulated linear subsystem of the first-order linear stabistor system of a speed and a tension force, correspondingly just can design linear closed loop controller easily, owing to really realized speed and the control of the decoupling zero between the tension force to rotor flux and the dynamic decoupling between the speed and the two induction of each induction machine, thereby not only can distinguish the effective control that realizes independently two induction machine speed and rotor flux, and can distinguish the effective control that independently realizes two induction speed and tension force, obtain good speed and tension adjustment performance.Owing to adopted the neural net that does not rely on the controlled device Mathematical Modeling to realize the generalized inverse systemic-function, thereby improved the robustness that parameter of electric machine variation, load disturbance and network delay are changed greatly.
The invention has the advantages that:
1. with this controlled volume of two induction (speed of each induction machine and rotor flux, the speed of two induction and tension force) four inputs (given and two velocity setting of two rotor fluxs of two induction machines) four that intercouple export (speed of conveyer belt, the rotor flux of the tension force of conveyer belt and two induction machines) control problem of complex nonlinear coupled system is converted into simple four stabilized pseudo linear subsystems (two rotor flux linear subsystems, a speed linear subsystem and a tension force linear subsystem) control problem, further appropriate design linear closed-loop controller can obtain the runnability of high performance The synchronized Coordinative Control and anti-load disturbance.
2. add the generalized inverse system that letter and integration are realized composite controlled object that passes with static neural network, constructing neural network generalized inverse synchronous coordination frequency converter is realized the control to two induction, be completely free of the dependence of traditional Control of Induction Motors method for Mathematical Modeling, reduced the influence of parameter of electric machine variation, load disturbance and network delay effectively, improved the performance index of two induction control significantly two induction.
3. can be used for constructing the novel synchronous coordination control frequency transformer two induction is carried out high performance control, not only in the The synchronized Coordinative Control system that with the induction machine is power set, very high using value is arranged, and be in the The synchronized Coordinative Control system of power set at the alternating current machine with the networking of other type, have a extensive future.
Description of drawings
Below in conjunction with the drawings and specific embodiments the present invention is described in further detail.
The rotor flux observer 11 that Fig. 1 is made up of electric current, speed magnetic linkage estimation model 11 and Clarke Clark conversion;
Fig. 2 forms the structure charts of the Flow Control inverter 3 of expansion jointly by coordinate transform 31, current-controlled voltage source inverter 32, wherein by current-controlled voltage source inverter 32 with by contrary Parker Park conversion and the coordinate transform 31 formed against Clarke Clark conversion;
Fig. 3 is the principle assumption diagram that two induction 2 that the Flow Control inverter 3 with expansion drives drives common load 4, and the Flow Control inverter 3 and the common load 4 of two induction electromotor rotor flux observers 1, two induction machines 2, two expansions are wherein arranged;
Fig. 4 is each the induction electromotor rotor flux observer 1 of two induction 2 correspondences and the concrete principle assumption diagram of the induction machine 2 of Flow Control inverter 3 drivings of expanding;
Fig. 5 is each rotor flux observer 1 of two induction 2 correspondences and the Mathematical Modeling schematic diagram and the isoboles thereof of the induction machine 2 of Flow Control inverter 3 drivings of expanding;
Fig. 6 is the equivalent control block diagram of two induction four inputs and four outputs;
Fig. 7 is the schematic diagram and the isoboles thereof of the pseudo-linear system 7 of neural net generalized inverse 6 and composite controlled object 5 compound formations; The letter of biography, integrator, static neural network 61 are wherein arranged, pseudo-linear system 7;
Fig. 8 is two current component signals that the Flow Control inverter input that is added in the composite controlled object shown in Figure 35 No. 1 expansion is used to obtain the neural metwork training data;
The structure chart of the closed-loop control system that Fig. 9 is made up of linear closed-loop controller 8 and pseudo-linear system 7; Wherein pseudo-linear system 7 comprises two rotor flux subsystems 71 and 73, speed subsystems 72 and a tension force subsystem 74; The linear closed-loop controller comprises two rotor flux controllers 81 and 83, speed controls 82 and a tension controller 84;
Figure 10 is the control principle block diagram of integral body of the present invention.
Figure 11 is the complete principle block diagram that adopts 9 pairs of two inductions 5 of neural net generalized inverse The synchronized Coordinative Control frequency converter to control;
Figure 12 adopts DSP to form schematic diagram as apparatus of the present invention of neural net generalized inverse The synchronized Coordinative Control frequency converter.DSP10, photoelectric encoder 12 are wherein arranged;
Figure 13 is to be the realization systems soft ware block diagram of the present invention of controller with DSP.
Embodiment
As shown in figure 11, coordination control frequency transformer of the present invention comprises flux observer 1, connects frequency converter at the front end of two induction machines 2, and the rear end connects common load 4 by conveyer belt.Frequency converter is by linear closed-loop controller 8, the neural net generalized inverse The synchronized Coordinative Control frequency converter 9 that neural net generalized inverse 6 and composite controlled object 5 connect and compose, described composite controlled object 5 is by flux observer 1, controlled two induction machines 2 with comprise that the Flow Control inverter 3 and the common load 4 of the expansion of coordinate transform connect to form, place composite controlled object 5 to form pseudo-linear system 7 before described neural net generalized inverse 6, this pseudo-linear system 7 is by two linear subsystems 71 that the magnetic linkage single order is stable, 73, a stable linear subsystem 72 and the stable linear subsystem 74 of tension force second order of speed single order is formed; On pseudo-linear system 7 bases, constitute 81,83, speed controls 82 of two magnetic linkage control devices and a tension controller 84 of linear closed-loop controller 8.The Flow Control inverter 3 of described expansion is made up of current-controlled voltage source inverter 32 and coordinate transform 31, and coordinate transform 31 is to be concatenated into by two contrary Parker Park conversion and two contrary Clarke Clark conversion.
Shown in Fig. 1-10, the building method that control is coordinated in the neural net generalized inverse of two induction is, at first adopt two flux observers 1 of electric current commonly used, speed flux observation model and Clarke Clark conversion composition, obtain the rotor flux information of two required induction machines of magnetic linkage closed-loop control; Form the Flow Control inverter 3 that coordinate transform 31 forms expansion jointly by current-controlled voltage source inverter 32, contrary Parker Park conversion and contrary Clarke Clark conversion again.The Flow Control inverter 3 of this expansion will be as a part of whole neural net generalized inverse 6 The synchronized Coordinative Control frequency converters.Secondly with the Flow Control inverters 3 of two expansions and two induction machines 2 and load thereof as a composite controlled object 5, these composite controlled object 5 equivalences be five rank Differential Equation Models under the rotor flux coordinate system, the vectorial relative rank of system be 1,1,1,2}.Adopt the static neural network 61 (static neural network 61 is multitiered network MLN) of 9 inputs nodes, 4 output nodes to add 4 and pass the neural net generalized inverse 6 that letters and integration are constructed composite controlled object 5.And make neural net generalized inverse 6 realize the generalized inverse systemic-function of composite controlled object 5 by each weight coefficient of adjusting static neural network 61.Neural net generalized inverse 5 is serially connected in before the composite controlled object 5 again, neural net generalized inverse 6 and composite controlled object 5 synthesize by i.e. two 71,73, linear subsystems that the speed single order is stable 72 of linear subsystem and the stabilized pseudo linear systems 7 that the stable linear subsystem 74 of tension force second order is formed that the magnetic linkage single order is stable of three single order subsystems, thereby the control of the nonlinear multivariable systems of a complexity is converted into the control that three simple single order stabistor systems add a simple second-order stabistor system.Three single order subsystems and a second order subsystem for decoupling zero, adopt a kind of simple linear system synthesis method, as PID or POLE PLACEMENT USING etc., make two magnetic linkage control devices 81 respectively, 83, a speed control 82 and a tension controller 84, two magnetic linkage control devices 81,83, a speed control 82 and the linear closed loop controller 8 of tension controller 84 mutual group, finally form by neural net generalized inverse 6, linear closed-loop controller 8, the Flow Control inverter 3 of expansion and flux observer 1 be totally 4 neural net generalized inverse The synchronized Coordinative Control frequency converters 9 that part is formed, and comes two induction 2 is controlled.According to different control requirements, can select different hardware and softwares to realize.
The specific embodiment of building method of the present invention is divided following 9 steps:
1. construct rotor flux observer 1 as shown in Figure 1.Two induction machines are constructed rotor flux observer 1 respectively, and flux observer 1 is made up of electric current, speed magnetic linkage estimation model 11 and Clarke Clark conversion commonly used.Flux observer 1 be input as induction machine 2 stator phase current i
a, i
bAnd speed omega
r, be output as rotor flux angle θ and rotor flux ψ
rWherein rotor flux angle θ will be used to realize Parker Park transform operation and contrary Parker Park transform operation, rotor flux ψ
rWill be as the feedback quantity of magnetic linkage closed-loop control.Flux observer 1 will be as a part of whole neural net generalized inverse The synchronized Coordinative Control frequency converter 9.
2. construct the Flow Control inverter 3 of expansion as shown in Figure 2.At first form coordinate transform 31 by contrary Parker Park conversion and contrary Clarke Clark conversion, with this coordinate transform 31 and the current-controlled voltage source inverter 32 common Flow Control inverters 3 of forming expansion commonly used, the Flow Control inverter 3 of this expansion is its input with two stator current components afterwards.The Flow Control inverter 3 of expansion will be as a part of whole neural net generalized inverse The synchronized Coordinative Control frequency converter 9.
3. form composite controlled object 5 as shown in Figure 3.The Flow Control inverter 3 of two expansions that structure is good, two induction machines 2, flux observers 1 are formed composite controlled object 5 with controlled two induction machines 2 and common load 4 (band shape load), this composite controlled object 5 is its input with four stator current components, and the speed and the tension force of two induction machine 2 rotor flux measured values, two induction 2 are output.
As shown in Figure 4 by analyze, equivalence and derivation, for the structure of neural net generalized inverse 6 and learning training provide basis on the method.At first set up the Mathematical Modeling of each compound controlled subobject 5, promptly set up the Mathematical Modeling of Flow Control inverter 3 of each induction machine 2, each expansion and the Mathematical Modeling of each induction machine 2 rotor flux observers 1, and through equivalence shown in Figure 5.Consider the common load 4 of two induction machines 2 and two induction machines 2 then, get the equivalent mathematical model of composite controlled object 5, schematic diagram as shown in Figure 6.Be the five rank differential equations under the rotor flux coordinate system, its vector relatively rank be 1,1,1,2}.Generalized inverse system through provable this five rank Differential Equation Model of deriving exists, and four inputs can determining its generalized inverse system are the first derivative of the first derivative of No. 1 induction machine 2 rotor fluxs, the first derivative of two induction 2 speed, No. 2 induction electromotor rotor magnetic linkages and the second dervative of two induction 2 tension force, and four outputs are respectively four inputs of composite controlled object 5.Need to prove, this step only provides basis on the method for the structure of following neural net generalized inverse 6 and learning training, in concrete enforcement of the present invention, this step, comprise theoretical proof and some corresponding equivalent transformations and derivation etc. that composite controlled object 5 generalized inverse systems are existed, can skip.
5. adopt static neural network to add 4 and pass letter and integration constructing neural network generalized inverse 6, shown in the frame of broken lines of the left figure of Fig. 7.Wherein static neural network 61 adopts 3 layers MLN network, and the input layer number is 9, and the hidden layer node number is 17, and output layer node number is 4, and the hidden neuron activation primitive uses S type hyperbolic tangent function
The neuron of output layer adopts pure linear function f (x)=x, and x is neuronic input, and the weight coefficient of static neural network 61 will be determined in next step off-line learning.Add 4 with static neural network 61 then and pass letters and an integration constitutes neural net generalized inverse 6, shown in the frame of broken lines of the left figure of Fig. 7 with 9 inputs nodes, 4 output nodes.Wherein first of static neural network 61 is input as first input of neural net generalized inverse 6, and it passes second input that letter is output as static neural network 61 through first; The 3rd second input that is input as neural net generalized inverse 6 of static neural network 61, it passes the 4th input that letter is output as static neural network 61 through second.The 5th the 3rd input that is input as neural net generalized inverse 6 of static neural network 61, it passes the 6th input that letter is output as static neural network 61 through the 3rd; The 7th the 4th input that is input as neural net generalized inverse 6 of static neural network 61, it passes the 8th input that letter is output as static neural network 61 through the 4th, and the 8th input of described static neural network 61 is the 9th input of static neural network 61 again through first integration.Static neural network 61 passes letter with four and an integration is formed neural net generalized inverse 6, and the output of static neural network 61 is exactly the output of neural net generalized inverse 6.
6. adjust as shown in Figure 8, the weight coefficient of static neural network 61.(a) two current components are added to the Flow Control inverter of expanding for No. 1 and No. 23 (being the input of composite controlled object 5) respectively with the form of importing, with 6 milliseconds sampling periods collection induction machine speed omega
R1, ω
R2With current i
A1, i
B1And i
A2, i
B2And tension force F, according to ω
R1, ω
R2With i
A1, i
B1And i
A2, i
B2, obtain two rotor flux ψ by two rotor flux observers 1
R1And ψ
R2, and preserve data { ψ
R1, ω
R1, ψ
R2, F}.(b) two rotor fluxs and rate signal off-line are asked its first derivative respectively, the tension signal off-line is asked its single order, second dervative respectively, and signal is done standardization processing, the training sample set of composition neural net
ψ
R1,
ω
R1,
ψ
R2,
F, i
Sm1 *, i
St1 *, i
Sm2 *, i
St2 *.(c) adopt the error anti-pass BP algorithm that drives quantifier and learning rate changing that static neural network 61 is trained, through 600 training, neural net output mean square error meets the demands less than 0.001, thereby has determined each weight coefficient of static neural network 61.
7. form two rotor flux subsystems, a speed subsystem and a tension force subsystem.Pass letters and 1 integration constitutes the neural net generalized inverse by the static neural network 61 of having determined each weight coefficient and 4, shown in the frame of broken lines among the left figure of Fig. 7, neural net generalized inverse 6 composes in series pseudo-linear system 7 with composite controlled object 5, shown in the right figure of Fig. 7, this pseudo-linear system 7 is by two linear subsystems 71 that the magnetic linkage single order is stable, 73, a stable linear subsystem 72 and the stable linear subsystem 74 of tension force second order of speed single order is formed jointly, thereby reached between speed and the rotor flux, the control that is converted into simple four single argument linear systems is controlled complicated nonlinear multivariable systems in decoupling zero between speed and the tension force.
8. make the linear closed-loop controller.Shown in the right figure of Fig. 7, two rotor flux subsystems, a speed subsystem and a tension force subsystem are made linear closed-loop controller 8 respectively.Linear closed-loop controller 8 adopts proportional plus integral plus derivative controller PID, POLE PLACEMENT USING or the most excellent method of quadratic performance in the lineary system theory to design as shown in Figure 9, in the embodiment that the present invention provides, 81,83, speed controls 82 of two magnetic linkage control devices and a tension controller 84 have all been selected proportional integral PI controller for use, its parameter tuning is that two magnetic linkage control devices 81,83 are PI=600+12/s, speed control 82 is PI=85+12/s, tension controller 84 is PI=40+6/s, and whole system as shown in figure 10.
9. form neural net generalized inverse The synchronized Coordinative Control frequency converter 9.As shown in big frame of broken lines among Figure 11, with the Flow Control inverter 3 and two flux observers, 1 common composition neural net generalized inverse The synchronized Coordinative Control frequency converters of neural net generalized inverse 6, linear closed-loop controller 8, two expansions.Can require to adopt different hardware or software to realize according to different control.
As shown in figure 12, wherein neural net generalized inverse 6, closed loop controller 8, coordinate transform 31 and flux observer 1 are that dsp controller is realized by software by digital signal processor; The Flow Control inverter 3 of expansion adopts Intelligent Power Module to realize.The system program block diagram as shown in figure 13.Controlled induction machine model is Y90S-4, and the parameter of electric machine is P
e=1.1kW; U
e=220/380V; I
e=2.7A; f
e=50Hz; n
p=2; ω
e=1400rpm.
Claims (5)
1, a kind of neural net generalized inverse coordination control frequency transformer of two induction, comprise flux observer (1), front end at two induction machines (2) connects frequency converter, the rear end connects common load (4) by conveyer belt, it is characterized in that: described frequency converter is by linear closed-loop controller (8), the neural net generalized inverse The synchronized Coordinative Control frequency converter (9) that neural net generalized inverse (6) and composite controlled object (5) connect and compose, described composite controlled object (5) is by flux observer (1), controlled two induction machines (2) with comprise that the Flow Control inverter (3) and the common load (4) of the expansion of coordinate transform connect to form, place composite controlled object (5) to form pseudo-linear system (7) before described neural net generalized inverse (6), this pseudo-linear system (7) is by two linear subsystems (71 that the magnetic linkage single order is stable, 73), a stable linear subsystem (72) and the stable linear subsystem (74) of tension force second order of speed single order is formed; On pseudo-linear system (7) basis, constitute two magnetic linkage control devices (81,83), a speed control (82) and a tension controller (84) of linear closed-loop controller (8).
2, the neural net generalized inverse coordination control frequency transformer of two induction according to claim 1, it is characterized in that: the Flow Control inverter (3) of described expansion is made up of current-controlled voltage source inverter (32) and coordinate transform (31), and coordinate transform (31) is to be concatenated into by two contrary Parker Park conversion and two contrary Clarke Clark conversion.
3, a kind of building method of neural net generalized inverse coordination control frequency transformer of two induction, adopt electric current commonly used, speed flux observation model and Clarke Clark conversion to form two flux observers (1) earlier, it is characterized in that also in turn including the following steps
(1) forms the two Flow Control inverters of expanding (3) jointly by current-controlled voltage source inverter (32), contrary Parker Park conversion and contrary Clarke Clark conversion;
(2) with Flow Control inverters (3) and two induction machines (2) of two expansions and load thereof as a composite controlled object (5);
(3) composite controlled object (5) is adopted the static neural network (61) of 9 input nodes, 4 output nodes add 4 and pass letters and an integration comes constructing neural network generalized inverse (6), make neural net generalized inverse (6) realize the generalized inverse systemic-function of composite controlled object (5) by each weight coefficient of adjusting static neural network (61);
(4) neural net generalized inverse (6) is serially connected in composite controlled object (5) before, it is two rotor flux subsystems and a speed subsystem and the pseudo-linear system (7) that the second order subsystem constitutes that neural net generalized inverse (6) synthesizes by three single order subsystems with composite controlled object (5);
(5) on pseudo-linear system (7) basis, make two magnetic linkage control devices (81,83), a speed control (82) and a tension controller (84) respectively and form linear closed-loop controller (8), finally form neural net generalized inverse The synchronized Coordinative Control frequency converter (9).
4, the building method of the neural net generalized inverse coordination control frequency transformer of two induction according to claim 3, it is characterized in that: in the step (3), neural net generalized inverse (6) has 4 input nodes, 4 output nodes, wherein: first of static neural network (61) is input as first input of neural net generalized inverse (6), and it passes second input that letter is output as static neural network (61) through first; The 3rd second input that is input as neural net generalized inverse (6) of static neural network (61), it passes the 4th input that letter is output as static neural network (61) through second; The 5th the 3rd input that is input as neural net generalized inverse (6) of static neural network (61), it passes the 6th input that letter is output as static neural network (61) through the 3rd; The 7th the 4th input that is input as neural net generalized inverse (6) of static neural network (61), it passes the 8th input that letter is output as static neural network (61) through the 4th, and the 8th input of described static neural network (61) is the 9th input of static neural network (61) again through first integration; Static neural network (61) passes letter with four and an integration is formed neural net generalized inverse (6), and the output of static neural network (61) is exactly the output of neural net generalized inverse (6).
5, the building method of the neural net generalized inverse coordination control frequency transformer of two induction according to claim 3 is characterized in that: in the step (3), the method for adjustment of each weight coefficient of static neural network (61) is with step excitation signal i
Sm1 *, i
St1 *And i
Sm2 *, i
St2 *Be added to the input of composite controlled object (5); Gather the speed omega of No. 1 induction machine
R1With two stator phase current i
A1, i
B1Speed omega with No. 2 induction machines
R2With two stator phase current i
A2, i
B2And tension force F; According to ω
R1, ω
R2With i
A1, i
B1And i
A2, i
B2, obtain two rotor flux ψ by two rotor flux observers (1)
R1And ψ
R2With two rotor flux ψ
R1, ψ
R2And speed omega
R1, ω
R2Off-line is asked its first derivative respectively, and the tension signal off-line is asked its single order, second dervative, and signal is done standardization processing, the training sample set of composition neural net
ψ
R1,
ω
R1,
ψ
R2,
F, i
Sm1 *, i
St1 *, i
Sm2 *, i
St2 *; Static neural network (61) is trained, thus each weight coefficient of definite static neural network (61).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008100196312A CN101299581B (en) | 2008-03-10 | 2008-03-10 | Neural network generalized inverse coordination control frequency transformer for two induction machines and construction method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008100196312A CN101299581B (en) | 2008-03-10 | 2008-03-10 | Neural network generalized inverse coordination control frequency transformer for two induction machines and construction method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101299581A true CN101299581A (en) | 2008-11-05 |
| CN101299581B CN101299581B (en) | 2010-09-15 |
Family
ID=40079297
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2008100196312A Expired - Fee Related CN101299581B (en) | 2008-03-10 | 2008-03-10 | Neural network generalized inverse coordination control frequency transformer for two induction machines and construction method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101299581B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917150A (en) * | 2010-06-24 | 2010-12-15 | 江苏大学 | Robust controller of permanent magnet synchronous motor based on fuzzy-neural network generalized inverse and construction method thereof |
| CN102361429A (en) * | 2011-09-13 | 2012-02-22 | 江苏大学 | Bearing-free asynchronous motor control method based on neural network inverse system theory |
| CN102437816A (en) * | 2011-10-25 | 2012-05-02 | 武汉鑫通科创科技发展有限公司 | Adaptive motor motion control apparatus based on neural network |
| CN102790583A (en) * | 2012-08-06 | 2012-11-21 | 江苏大学 | Constructing method for internal model controller of radial generalized inverter of bearingless permanent magnet synchronous motor |
| CN105186938A (en) * | 2015-08-28 | 2015-12-23 | 江苏大学 | Sensorless tension identification method for two-motor speed regulating system |
| CN105262379A (en) * | 2015-11-16 | 2016-01-20 | 厦门理工学院 | Controller design method for dual motor drive system |
| CN105634356A (en) * | 2016-01-07 | 2016-06-01 | 江苏大学 | PLC-based generalized neural network inverse internal model implementation method for multi-motor speed regulating system |
-
2008
- 2008-03-10 CN CN2008100196312A patent/CN101299581B/en not_active Expired - Fee Related
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917150A (en) * | 2010-06-24 | 2010-12-15 | 江苏大学 | Robust controller of permanent magnet synchronous motor based on fuzzy-neural network generalized inverse and construction method thereof |
| CN102361429A (en) * | 2011-09-13 | 2012-02-22 | 江苏大学 | Bearing-free asynchronous motor control method based on neural network inverse system theory |
| CN102437816A (en) * | 2011-10-25 | 2012-05-02 | 武汉鑫通科创科技发展有限公司 | Adaptive motor motion control apparatus based on neural network |
| CN102437816B (en) * | 2011-10-25 | 2014-05-07 | 武汉鑫通科创科技发展有限公司 | Adaptive motor motion control apparatus based on neural network |
| CN102790583A (en) * | 2012-08-06 | 2012-11-21 | 江苏大学 | Constructing method for internal model controller of radial generalized inverter of bearingless permanent magnet synchronous motor |
| CN105186938A (en) * | 2015-08-28 | 2015-12-23 | 江苏大学 | Sensorless tension identification method for two-motor speed regulating system |
| CN105186938B (en) * | 2015-08-28 | 2018-02-27 | 江苏大学 | A kind of two motor speed regulation systems are without sensor tension force discrimination method |
| CN105262379A (en) * | 2015-11-16 | 2016-01-20 | 厦门理工学院 | Controller design method for dual motor drive system |
| CN105262379B (en) * | 2015-11-16 | 2018-12-28 | 厦门理工学院 | The controller design method of twin drive system |
| CN105634356A (en) * | 2016-01-07 | 2016-06-01 | 江苏大学 | PLC-based generalized neural network inverse internal model implementation method for multi-motor speed regulating system |
| CN105634356B (en) * | 2016-01-07 | 2018-04-17 | 江苏大学 | More motor speed regulation system neural network generalized inverse internal model implementation methods based on PLC |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101299581B (en) | 2010-09-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101299581B (en) | Neural network generalized inverse coordination control frequency transformer for two induction machines and construction method thereof | |
| CN101630936B (en) | Neural network inverse controller of brushless DC motor and construction method thereof | |
| CN100433537C (en) | Method for controlling bearing-less AC asynchronous motor neural network inverse decoupling controller | |
| CN103259479B (en) | A kind of permanent magnet synchronous electric machine neural network left inverse state observation method | |
| Ramesh et al. | Type-2 fuzzy logic control based MRAS speed estimator for speed sensorless direct torque and flux control of an induction motor drive | |
| CN1212702C (en) | Neural network reversal control frequency converter of induction motor and structure method | |
| CN101741297B (en) | Method and device for inversely controlling fuzzy compensation of radial position of bearing-free synchronous reluctance motor | |
| CN1655438A (en) | Magnetic levitation switch reluctance motor radial neural network reversed decoupling controller and method for constructing same | |
| CN109120014A (en) | A kind of micro-capacitance sensor frequency voltage power collaboration compensation method | |
| CN105048910A (en) | Sensorless high-voltage cascaded frequency converter vector control system and control method thereof | |
| CN101299580A (en) | Cooperative controller for two-motor vector control frequency control system and construction method thereof | |
| CN105406784A (en) | Single-winding bearingless motor torque and suspension force direct controller and construction method | |
| CN107370429B (en) | Fuzzy neural network inverse decoupling controller for bearingless permanent magnet synchronous motor | |
| CN104779873A (en) | Prediction function control method for PMSM servo system | |
| CN1216454C (en) | Construction method for contradictorily synchronous control frequency-converter of two-phase induction motor neural net | |
| CN101640513A (en) | Support vector machine (SVM) inverse controller of two-motor variable-frequency speed-regulating system and construction method thereof. | |
| CN102790577A (en) | Constructing method for suspended subsystem controller of bearingless permanent magnet synchronous motor | |
| CN1226824C (en) | Method for structuring neural network reversal composite controller of variable frequency speed control system with constant ratio of voltage and frequency | |
| CN1216455C (en) | Method for structuring mervo network inverse synchronous controller of tow-frequenct changing speed regulating motor system | |
| CN104022701B (en) | Mould method for control speed in a kind of permanent magnetic linear synchronous motor Newton method | |
| CN110061676A (en) | A kind of bearing-free permanent magnet synchronous motor controller based on flux observer | |
| CN113098033A (en) | Adaptive virtual inertia control system and method based on flexible direct current power transmission system | |
| CN105915094A (en) | LCL type grid-connected inverter control method based on power grid current single sensor | |
| CN104767452A (en) | Self-adaptative inverse decoupling control method based on non-linear filters for bearing-free asynchronous motor | |
| CN102497149B (en) | Direct decoupling control method of permanent magnet linear synchronous motor-driven suspension platform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Assignee: JIANGSU MONTMERY ELEVATOR CO., LTD. Assignor: Jiangsu University Contract record no.: 2012320000838 Denomination of invention: Neural network generalized inverse coordination control frequency transformer for two induction machines and construction method thereof Granted publication date: 20100915 License type: Exclusive License Open date: 20081105 Record date: 20120627 |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100915 Termination date: 20200310 |