CN101630936B - Neural network inverse controller of brushless DC motor and construction method thereof - Google Patents
Neural network inverse controller of brushless DC motor and construction method thereof Download PDFInfo
- Publication number
- CN101630936B CN101630936B CN2009101843047A CN200910184304A CN101630936B CN 101630936 B CN101630936 B CN 101630936B CN 2009101843047 A CN2009101843047 A CN 2009101843047A CN 200910184304 A CN200910184304 A CN 200910184304A CN 101630936 B CN101630936 B CN 101630936B
- Authority
- CN
- China
- Prior art keywords
- motor
- brshless
- input
- controller
- brushless
- 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.)
- Active
Links
Images
Abstract
The invention discloses a speed regulation controller of a brushless DC motor and a construction method thereof. The speed regulation controller comprises a neural network inverse and a linear closed loop controller. The construction method comprises the following steps: using an inverse bridge and a brushless DC motor as a whole to form a brushless DC motor speed regulation system; connecting the neural network inverse in the front of the brushless DC motor speed regulation system in series to together compound a pseudo-linear system consisting of a speed subsystem; constructing the linear closed loop controller on the basis of the speed subsystem; and connecting the neural network inverse and the linear closed loop controller in series to form a neural network inverse controller. The brushless DC motor adopting the neural network inverse controller has good speed regulation performance, not only can be used for designing a new brushless DC speed regulation system, but also for reconstructing an original brushless DC motor system, and the invention has convenient system construction and low cost.
Description
Technical field
The present invention relates to a kind of brushless direct current motor controller and building method thereof, be applicable to that the high-performance speed governing of brshless DC motor is used, belong to the technical field of Electric Drive control appliance.
Background technology
At present, because the progressively reduction of power electronic device price, high performance permanent magnetic material constantly occurs simultaneously, a large amount of direct current machines that use permanent-magnet brushless DC electric machine to replace complex structure and have the commutation problem in the industrial production.Brshless DC motor has the mechanical property in direct current machine, and is simultaneously simple in structure stable, need not brush-commutatedly, is applicable to various operating modes.But because motor parameter perturbation in service and commutation process cause electromagnetic torque pulsation to limit its speed adjusting performance.Nonlinear multivariable systems as a complexity, brshless DC motor adopts the PID control strategy of traditional similar direct current machine to be difficult to satisfy the needs of high-performance speed governing, because the setting value of pid parameter is the optimal value in the certain limit, rather than optimal value of overall importance, the dynamic property and the stable state accuracy that can not in the scope of broad, obtain fundamentally, simultaneously as a kind of Linear Control strategy, PID control is difficult to better control to the nonlinear degree higher system, and this is a problem demanding prompt solution.
Summary of the invention
The purpose of this invention is to provide a kind of adaptability, the robustness that both can improve brshless DC motor to parameter variation and disturbance, simultaneously can improve every control performance index effectively again, as the nerve network reverse controller of dynamic responding speed, steady-state tracking precision.
Another object of the present invention provides the building method of the nerve network reverse controller of brshless DC motor.
The technical scheme that neural network inverse controller of brushless DC motor of the present invention adopts is: comprise that the inverter bridge that contains the buck chopper converter connects brshless DC motor, nerve network reverse is connected with the linear closed-loop controller, described nerve network reverse is connected in series the formation neural network inverse controller of brushless DC motor mutually with the linear closed-loop controller, wherein nerve network reverse adds 2 integral elements formations with the static person artificial neural networks, the linear closed-loop controller is a speed control, by the method for designing of linear system pseudo-linear system is made respectively; Pseudo-linear system is the speed subsystem, and the brshless DC motor governing system is done as a whole the composition by inverter bridge, brshless DC motor and load, the pseudo-linear system of nerve network reverse and the common compound composition equivalence of brshless DC motor governing system.
The technical scheme that the building method of the nerve network reverse controller of brshless DC motor of the present invention adopts is in turn include the following steps: 1. form brshless DC motor speed governing governing system; 2. do the equivalence of whole brshless DC motor speed governing governing system, input variable is the duty ratio of brshless DC motor governing system controller, output variable is the actual speed of motor, by the analysis and the Mathematical Modeling that can obtain whole brshless DC motor speed governing governing system of deriving is second order differential equation, speed phase match exponents is a second order, the input variable of fixed its inverse system is a rotating speed, the first derivative of rotating speed and second dervative, output variable are the duty ratio that is of brshless DC motor speed governing governing system; 3. the design (calculated) load torque observer is a reduced dimension observer; 4. constructing neural network is contrary; 5. rank step excitation signal is added to brshless DC motor speed governing governing system input; Gather pumping signal and actual speed; The actual speed signal that obtains is carried out off-line ask first derivative and second dervative, pass through the measured value of load torque observer simultaneously, train with the training sample set pair static person artificial neural networks that constitutes, determine each weight coefficient of static person artificial neural networks; 6. form the speed subsystem; 7. make behind the linear closed-loop controller being connected in series and constitute neural network inverse controller of brushless DC motor with neural net broad sense anti-phase.
The present invention turns to pseudo-linear system by the contrary link of constructing neural network with this non linear system linearity of former brshless DC motor governing system, by reasonably designing closed loop controller, can obtain good speed adjusting performance again.
The invention has the advantages that:
1, adopts the neural net inverse approach, solved the control problem of Complex Nonlinear System,, obtain high performance coordination control and anti-load disturbance runnability by further appropriate design linear closed-loop controller.
2, method of the present invention is than traditional brshless DC motor governing system, and hardware configuration changes little, and tectonic system is convenient, and cost is low, and application prospect is boundless.
Description of drawings
Fig. 1 is the principle assumption diagram of brshless DC motor speed governing.Brshless DC motor 1 is wherein arranged, the inverter bridge 2 of (buck chopper) circuit that contains BUCK.
Fig. 2 is the isoboles of the single output of single input (duty ratio ρ) (motor actual speed ω) of brshless DC motor governing system correspondence.
Fig. 3 is the schematic diagram and the isoboles thereof of the pseudo-linear system of nerve network reverse 4 and brshless DC motor governing system 3 compound formations.Wherein comprise integrator, static person artificial neural networks 41, pseudo-linear system 5.
The structure chart of the closed-loop control system that Fig. 4 is made up of linear closed-loop controller 6 and pseudo-linear system 5.Wherein pseudo-linear system is made of speed subsystem 51; The linear closed-loop controller is a speed control.
Fig. 5 is the complete principle block diagram that adopts 8 pairs of brshless DC motor governing systems 3 of neural network inverse controller of brushless DC motor to control.
Fig. 6 is the rotating speed response (given rotating speed be step signal) of brshless DC motor under the step signal excitation.
Fig. 7 adopts the DSP control board to form schematic diagram as apparatus of the present invention of nerve network reverse controller.Wherein comprise dsp controller 7, photoelectric encoder 9.
Fig. 8 is the program running block diagram when adopting dsp controller 7 as nerve network reverse controller.
Embodiment
As Fig. 1-2, brshless DC motor 1 and the inverter bridge circuit 2 that contains BUCK (buck chopper) circuit are made as a whole composition brshless DC motor governing system 3 jointly.According to analyzing as can be known: this brshless DC motor governing system 3 is non-linear single-input single-output system, is input as duty ratio ρ, is output as actual speed ω, and the relative rank of speed are second order, and the whole system segmentation is reversible.As Fig. 3, adopt the static person artificial neural networks 41 (as multitiered network BP or warp-wise primary function network RBF etc.) of 4 input nodes and 1 output node to add 2 integrators formation nerve network reverses 4.Again nerve network reverse 4 is serially connected in original system and is before the controlled brshless DC motor governing system 3, be combined into speed second order integral form pseudo-linear system 5, realize the linearisation of whole system.As Fig. 4, in order to eliminate the inverse system approximate error, adopt linear system design theory (as methods for designing such as proportion integration differentiation PID, POLE PLACEMENT USING) to carry out the design of linear closed-loop controller 6 at last to the linear subsystem 5 that obtains.The final nerve network reverse controller that forms comprises nerve network reverse 4 and linear closed-loop controller 5 two parts, can require to adopt different hardware or software to realize according to different control.Concrete enforcement following 7 steps of branch:
1. as shown in Figure 1, the hardware of forming the brshless DC motor governing system.The inverter bridge 2 that will contain BUCK (buck chopper) converter is combined into an integral body jointly with brshless DC motor 1, and this composite controlled object is input ρ with the duty ratio, and the actual speed of motor is output ω.
2. by analyze, equivalence and derivation, for the structure of nerve network reverse 4 and learning training provide foundation on the method.At first do to contain the inverter bridge 2 of BUCK (buck chopper) parallel operation and the equivalence of brshless DC motor 1, the inverter bridge circuit that contains BUCK (buck chopper) converter is shown in figure (2), both equivalences can be changed the non linear system of the single output of single input, and the relative rank of rotating speed are second order, the segmentation in the operation area of provable this system is reversible through deriving, and the input variable that can determine its inverse system is rotational speed omega, first derivative
And second dervative
Output variable is the input ρ of original system (brushless direct-current governing system 3).Need to prove that this step, the structure and the basis of learning to provide on the method for following nerve network reverse 4 only were provided, in concrete enforcement of the present invention, the proof of the analysis in this step, equivalence and inverse system etc. can be skipped.
3. the design (calculated) load torque observer 10.According to state feedback in the present control theory and observer theory, be input according to the linear mathematical model of brshless DC motor 1 with duty ratio ρ, dc bus current is for output design reduced dimension observer, in real time to load torque T
LObserve.
4. constructing neural network contrary 4.Shown in employing static person artificial neural networks 41 adds in the contrary frame of broken lines of seeing among the left figure of Fig. 3 of 2 integrator constructing neural networks, neural net adopts 3 layers BP network, the input layer number is 4, the hidden layer node number is 12,1 node of output layer, the hidden neuron function uses (0,1) S type function f (x)=(e
x-e
-x)/(e
x+ e
-x), the neuron of output layer adopts pure linear transformation function, and each weight coefficient of neural net will be determined in next step off-line learning; Add 2 integrations with this static neural network 41 then and constitute 4 input nodes with 4 input nodes, 1 output node, the nerve network reverse 4 of 1 output node, see shown in the frame of broken lines of Fig. 3, wherein: first is input as first input of nerve network reverse 4 static neural network 41, it is second input of static neural network 41 through first integrator, through second integrator is the 3rd input of static neural network 41, and the 4th of static neural network 41 is input as the measured value of 10 pairs of load torques of load observer
5. determine each weight coefficient of neural net.The steps include: that (A) selects the velocity setting signal as the study pumping signal, as shown in Figure 5, so that brshless DC motor governing system 3 can fully be encouraged in its working range; (B) selected pumping signal is added in brushless direct-current governing system 3 inputs of single der Geschwindigkeitkreis, the dynamic concept measured value of the duty ratio of sampling simultaneously ρ and actual speed ω and 10 pairs of loads of load observer
(C) off-line obtains actual speed ω differentiate
With
Thereby constitute the training sample of neural net
(D) adopt the error anti-pass BP algorithm of learning rate changings to train to static neural network 41, behind the learning training 2000 times, static neural network 41 output mean square errors meet the demands less than 0.01, thereby have determined each weight coefficient of static neural network 41.
6. form pseudo-linear system 5.The static neural network 41 that off-line training is good is mixed the nerve network reverse 4 that two integrators constitute, be connected in series compound with controlled brshless DC motor governing system shown in Figure 4, the linear subsystem of formation speed second order has been realized Complex Nonlinear System control is converted into simple linear subsystem control.
7. make linear closed-loop controller 6.The speed second order linear subsystem that obtains is carried out the closed loop controller design.
8. make the final nerve network reverse controller 8 that forms of linear closed-loop controller and comprise nerve network reverse 4 and linear closed-loop controller 6 two parts, as shown in Figure 7, can require to adopt different hardware or software to realize according to difference control.
Fig. 7 is a concrete schematic diagram of the present invention, and the master control chip adopts the control board based on the TMS320F2812DSP motor special integrated circuit of American TI Company among the figure.The control board part mainly comprises the detection of fault-signal, photoelectric isolating circuit, and the interface circuit that links to each other with the TMS320F2812DSP chip.
The flow chart of system comprises main program and interrupt service routine as shown in Figure 8.Main program is mainly realized the initialization of system, carries out real-time tracing trouble and warning according to speed control break in service and the given break in service of order simultaneously.The speed control break in service is as regularly interrupting the main defence program scene of realizing, real-time sampling signal simultaneously after sampled signal handled, carries out Neural network inverse control, recovers on-the-spot at last, withdraws from interruption.Order given interruption as human-computer interaction module, mainly realize given various control command.The parameter of controlled brshless DC motor is P=120W; U=36V; I=4A; ω=400rpm.
Claims (2)
1. neural network inverse controller of brushless DC motor, comprise that nerve network reverse (4) is connected with linear closed-loop controller (6), nerve network reverse (4) and linear closed-loop controller (6) are connected in series formation neural network inverse controller of brushless DC motor (8) mutually, wherein nerve network reverse (4) adds 2 integral elements formations with static person artificial neural networks (41), linear closed-loop controller (6) is a speed control, by the method for designing of linear system pseudo-linear system (5) is made respectively; Pseudo-linear system (5) is speed subsystem (51), the pseudo-linear system (5) of nerve network reverse (4) and the common compound composition equivalence of brshless DC motor governing system (3); It is characterized in that: brshless DC motor governing system (3) is by the inverter bridge that contains the buck chopper converter (2), as a whole composition is made in brshless DC motor (1) and load, the inverter bridge (2) that contains the buck chopper converter connects brshless DC motor (1), described static person artificial neural networks (41) has 4 input nodes and 1 output node, wherein first of static person artificial neural networks (41) input that is input as nerve network reverse (4), first input is second input of static person artificial neural networks (41) through first integral element, second input of static person artificial neural networks (41) is its 3rd input through an integral element, it the 4th is input as the measured value of load torque observer (10) to load torque, and static person artificial neural networks (41) is output as the output of nerve network reverse (4).
2. the building method of a neural network inverse controller of brushless DC motor is characterized in that in turn including the following steps:
1. form brshless DC motor governing system (3);
2. do the equivalence of whole brshless DC motor governing system (3), input variable is the duty ratio of brshless DC motor governing system controller, output variable is the actual speed of motor, by the analysis and the Mathematical Modeling that can obtain whole brshless DC motor governing system (3) of deriving is second order differential equation, speed phase match exponents is a second order, the input variable of fixed its inverse system is a rotating speed, the first derivative of rotating speed and second dervative, output variable are the duty ratio that is of brshless DC motor governing system (3);
3. design (calculated) load torque observer (10) is a reduced dimension observer;
4. constructing neural network is against (4);
5. the step excitation signal is added to brshless DC motor governing system (3) input; Gather pumping signal and actual speed; The actual speed signal that obtains is carried out off-line ask first derivative and second dervative, pass through the measured value of load torque observer (10) simultaneously, train with the training sample set pair static person artificial neural networks (41) that constitutes, determine each weight coefficient of static person artificial neural networks (41);
6. form speed subsystem (51);
7. make linear closed-loop controller (6) back and be connected in series formation neural network inverse controller of brushless DC motor (8) mutually with nerve network reverse (4).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009101843047A CN101630936B (en) | 2009-08-12 | 2009-08-12 | Neural network inverse controller of brushless DC motor and construction method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009101843047A CN101630936B (en) | 2009-08-12 | 2009-08-12 | Neural network inverse controller of brushless DC motor and construction method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101630936A CN101630936A (en) | 2010-01-20 |
| CN101630936B true CN101630936B (en) | 2011-09-14 |
Family
ID=41575915
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009101843047A Active CN101630936B (en) | 2009-08-12 | 2009-08-12 | Neural network inverse controller of brushless DC motor and construction method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101630936B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917150B (en) * | 2010-06-24 | 2012-06-20 | 江苏大学 | Robust controller of permanent magnet synchronous motor based on fuzzy-neural network generalized inverse and construction method thereof |
| CN101938246B (en) * | 2010-09-29 | 2012-07-04 | 重庆交通大学 | Fuzzy fusion identification method of rotating speed of sensorless motor |
| CN102055390B (en) * | 2011-01-10 | 2012-11-07 | 江苏大学 | Construction method for neural network Alpha-order inverse controller of bearing-free brushless DC motor |
| CN103151980B (en) * | 2012-12-29 | 2015-10-28 | 江苏大学 | Automobile EPS brushless direct current motor controller and its implementation |
| CN104300854A (en) * | 2014-11-07 | 2015-01-21 | 黑龙江省科学院科技孵化中心 | Brushless direct current motor drive circuit based on Buck convertor |
| CN104410341A (en) * | 2014-11-27 | 2015-03-11 | 江苏科技大学 | Low-speed torque ripple restraining device and restraining method based on direct current voltage adjustment |
| CN104868807B (en) * | 2015-05-06 | 2018-01-16 | 南京航空航天大学 | A kind of active damping method of Buck circuits brushless DC motor control system |
| CN105372075B (en) * | 2015-11-13 | 2018-06-08 | 武汉理工大学 | Brushless direct-current electronics water pump controller and diagnostic method with fault diagnosis functions |
| CN108712116B (en) * | 2017-04-10 | 2020-10-27 | 湖南工业大学 | Brushless direct current motor position sensorless control method based on extreme learning machine |
-
2009
- 2009-08-12 CN CN2009101843047A patent/CN101630936B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN101630936A (en) | 2010-01-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101630936B (en) | Neural network inverse controller of brushless DC motor and construction method thereof | |
| CN101917150B (en) | Robust controller of permanent magnet synchronous motor based on fuzzy-neural network generalized inverse and construction method thereof | |
| Xia et al. | Direct nonlinear primal–dual interior-point method for transient stability constrained optimal power flow | |
| CN103259479B (en) | A kind of permanent magnet synchronous electric machine neural network left inverse state observation method | |
| CN102361429A (en) | Bearing-free asynchronous motor control method based on neural network inverse system theory | |
| CN111525688A (en) | Electric power Internet of things power distribution network system based on 5G network and interconnection method | |
| CN101741297A (en) | Method and device for inversely controlling fuzzy compensation of radial position of bearing-free synchronous reluctance motor | |
| CN101299580A (en) | Cooperative controller for two-motor vector control frequency control system and construction method thereof | |
| CN106786529B (en) | A kind of distribution static security analysis method | |
| CN102664397B (en) | Electric power system transient stability simulation method based on implicit fine numerical integral | |
| CN105186958B (en) | The five mutually fault-tolerant magneto internal model control methods based on Neural Network Inverse System | |
| CN101299581A (en) | Neural network generalized inverse coordination control frequency transformer for two induction machines and construction method thereof | |
| CN106208879A (en) | A kind of method controlling asynchronous motor frequency-changing speed-regulating system based on automatic disturbance rejection controller | |
| Zhang et al. | The application of nonlinear PID controller in generator excitation system | |
| CN102521473B (en) | Grid-connected photovoltaic simulation generating system based on DIgSILENT (Digital Simulation and Electrical NeTwork) | |
| CN113569476B (en) | Method for realizing mixed dimension joint simulation by utilizing dynamic parameter mixing | |
| CN1212702C (en) | Neural network reversal control frequency converter of induction motor and structure method | |
| CN104600694A (en) | Micro-grid energy optimization method considering economic dispatch and loop current suppression | |
| CN104022701B (en) | Mould method for control speed in a kind of permanent magnetic linear synchronous motor Newton method | |
| CN1226824C (en) | Method for structuring neural network reversal composite controller of variable frequency speed control system with constant ratio of voltage and frequency | |
| CN101630940B (en) | Fuzzy neural network inverse robust controller of induction motor speed regulation system and construction method | |
| CN102655326B (en) | Forming method for neural network inverse controller of active power filter | |
| CN1216455C (en) | Method for structuring mervo network inverse synchronous controller of tow-frequenct changing speed regulating motor system | |
| CN105140957B (en) | Electromechanic oscillation mode evaluation method based on wind power plant and photovoltaic plant polymerization model | |
| CN102609576B (en) | Power system transient stability simulating method based on estimation-correction numerical integration |
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 |
Application publication date: 20100120 Assignee: Yancheng Yanhai Tractor Manufacturing Co., Ltd. Assignor: Jiangsu University Contract record no.: 2014320000075 Denomination of invention: Neural network inverse controller of brushless DC motor and construction method thereof Granted publication date: 20110914 License type: Exclusive License Record date: 20140225 |
|
| LICC | Enforcement, change and cancellation of record of contracts on the licence for exploitation of a patent or utility model |