CN103777601A - Shield door door-machine controller neural network PID method and controller controlled through the same - Google Patents
Shield door door-machine controller neural network PID method and controller controlled through the same Download PDFInfo
- Publication number
- CN103777601A CN103777601A CN201410020706.4A CN201410020706A CN103777601A CN 103777601 A CN103777601 A CN 103777601A CN 201410020706 A CN201410020706 A CN 201410020706A CN 103777601 A CN103777601 A CN 103777601A
- Authority
- CN
- China
- Prior art keywords
- neural network
- pid
- controller
- machine controller
- output
- 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.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Control Of Multiple Motors (AREA)
- Control Of Electric Motors In General (AREA)
- Control Of Ac Motors In General (AREA)
Abstract
The invention discloses a shield door door-machine controller controlled through a neural network PID control method. The neural network PID structure is a 2'3'1 structure which belongs to a BP neural network and includes an input layer, a hidden layer and an output layer. The neural network PID control method is characterized in that the method comprises the following steps: a reference rotating speed n* and a motor feedback rotating speed n are inputted to two neurons of the input layer respectively, outputted according to a particular function, outputted to three neurons including the proportion neuron, the integral neuron and the differential neuron of the hidden layer respectively and outputted to the output layer according to a particular function, and a reference current I* is outputted. The method and controller of the invention have the following advantages that the PID control based on a neural network has advantages of high tracking accuracy and good robustness, parameters can be effectively adjusted for system parameter drift, the adaptive control function can be realized, the advantage of significant superiority can be realized, and the shield door door-machine controller controlled through the N-PID control method also has advantages of high accuracy and good robustness.
Description
Technical field
The invention belongs to screen door in rail transit door machine opertaing device technical field, be specifically related to shield the Neural network PID method of every machine controller.
Background technology
Along with developing rapidly of urban track traffic, subway has become the most convenient, economic and one of the vehicles efficiently of people trip; Meanwhile, for the reliability service of shield door, more and more higher requirement has also been proposed.Shield door is mounted in the glass door at Subway Station Platform edge, and it separates platform and train operation region, has the features such as the investment that improves platform security, reduce station environmental control system.DC Brushless Motor has the advantages that AC motor structure is simple, reliable, easy to maintenance, the life-span is long, has again mechanical property and speed governing that conventional DC motor is good, and at present, the door machine that is widely used in subway shield door drives.Platform screen door machine control algolithm adopts conventional PID to control conventionally, due to the parametrical nonlinearity of brshless DC motor and time become, this just makes the response speed of shield door slower, again because traditional PID exists hyperharmonic oscillation problem in short-term, unstable while making shield door operation, antijamming capability is not strong simultaneously.
Summary of the invention
In order to address the above problem, the invention provides a kind of Neural network PID method and a kind of shielding every machine controller that utilizes the method that shields every machine controller, concrete technical scheme is as follows:
Shield a Neural network PID method for every machine controller, described Neural network PID structure is 2 ' 3 ' 1 structure, belongs to BP neural network, is respectively input layer, hidden layer and output layer, it is characterized in that, comprises the following steps:
Reference rotation velocity
be input to respectively two neurons of input layer with motor feedback rotation speed n, according to specific function output, output to respectively three neurons of ratio, integration, differential of hidden layer, output to output layer according to specific function, output reference current
.
A kind of shielding every machine controller that adopts this Neural network PID method control, comprise a central controller of machine controller DCU, cell controller PEDC, drive UPS, rectifier, inverter, current collector, with the brshless DC motor BLDCM of position transducer, wherein: central controller is connected with PEDC by asynchronous serial communication SCI module, rectifier is connected with driving UPS, inverter is connected with rectifier, BLDCM is connected with inverter, current collector is connected with inverter, central controller is connected with the position transducer of BLDCM by quadrature coding pulse circuit QEP module, be connected with current acquisition by D/A module, be connected with inverter by pulse-width modulation PWM module, the control of the inner employing of central controller speed and current double closed loop, current inner loop adopts PI to control, rotating speed outer shroud adopts Neural Network PID Control.
Further improve and be: described central controller connects master monitor MMS by CAN bus.
Further improve and be: current inner loop adopts PID to control.
Further improve and be: also comprise various kinds of protective circuit, described holding circuit comprises current protecting circuit, voltage protection circuit, virtual protection circuit.
Reference rotation velocity
with two inputs of rotating speed computing module output speed n as Neural network PID, the electric current that the output of Neural network PID and current collector collect is input to current comparator through the electric current of D/A module output, error current is through current PI (PID) controlled adjuster, it outputs to pulse-width modulation PWM controller, finally passes through the rotating speed of inverter control motor by pulse-width modulation PWM.
For the precision and the dirigibility that guarantee to control, central controller can be selected DSP, FPGA and special integrated circuit.
Beneficial effect of the present invention: compared with prior art, control that tracking accuracy is high, strong robustness, effectively adjust parameter for systematic parameter drift energy based on the PID of neural network, realized adaptive control function, there is obvious superiority.Therefore utilize shielding every machine controller of Neural Network PID Control to have advantages of that tracking accuracy is high, strong robustness.
Accompanying drawing explanation
Fig. 1 is circuit control principle drawing of the present invention.
Fig. 2 is the Neural Network PID Control structural drawing in central controller.
Fig. 3 is neural network-PID control flow.
Fig. 4 is neural network-PID and the comparison of PID tracking performance.
Embodiment
In order to deepen understanding of the present invention, below in conjunction with drawings and Examples, the present invention is described in further detail, this embodiment is only for explaining the present invention, and not right protection domain forms restriction.
Can see essential structure of the present invention clearly by Fig. 1: central controller 1 is connected with cell controller 2 rigid lines by asynchronous serial communication SCI module, be connected with master monitor 3 buses by CAN module, rectifier 8 is connected with driving UPS 7, inverter 4 is connected with rectifier 8, brshless DC motor BLDCM 6 is connected with inverter 4, and current collector 5 is connected with inverter 4; Central controller 1 is connected with the position transducer of BLDCM 6 by quadrature coding pulse circuit QEP module, is connected with current collector 5 by D/A module, is connected with inverter by pulse-width modulation PWM module.
Can see the inner speed and current double closed loop control structure 100 that adopts of central controller clearly by Fig. 1, its medium speed outer shroud adopts neural network-PID switching controls, and current inner loop adopts traditional PID control.Reference rotation velocity
with two inputs of rotating speed computing module 102 output speed n as Neural network PID 101, the electric current that the output of Neural network PID 101 and current collector device 5 are exported is input to current comparator 103, error current is through current PI D regulator 104, it outputs to pulse-width modulation PWM controller 105, is finally controlled the rotating speed of motor by inverter 4 by pulse-width modulation PWM module.
N-PID(Neural network PID as shown in Figure 2) structural drawing of controller, its network is 2 ' 3 ' 1 structure, belongs to BP neural network, comprises input layer, hidden layer and output layer.Input layer contains two neurons, and its input is respectively given rotating speed
with the sampling rotation speed n from BLDCM.Hidden layer comprises 3 neurons, corresponding ratio (P) respectively, and integration (I), differential (D) 3 parts, output layer completes the comprehensive of N-PID control law, P, I, D coefficient is embodied by the power of network.
The forward direction algorithm of N-PID is according to the value of control system set-point and controlled device output, by the output of network current weight and each layer of input-output function formation control device.Sampling instant K arbitrarily, the input and output of input layer are:
（2）
Hidden layer is the key stratum of neural network, wherein each neuronic being input as:
j=1,2,3 (4)
According to digital pid control algolithm, each output function of hidden layer neuron can be configured to:
Ratio
(5)
Integration
(6)
Differential
(7)
Wherein T is the sampling period.
Output layer only has a neuron, and its input and output are:
The back propagation algorithm of N-PID network completes the modification of network weight, completes the function of learning and memory, and this algorithm adopts method of steepest descent.If
in time, be total to p sampled point, N-PID network control target and training criterion are: adjust link weight coefficients, make cost function E minimum.
（10）
If
for the momentum term factor (0<
<1),
for learning rate (
>0),, in the K moment, hidden layer to the weights of output layer are:
(11)
Input layer to the weights of hidden layer are:
（12）
Hidden layer to the weights rate of change of output layer is:
=
（13）
Input layer to the weights rate of change of hidden layer is:
If Fig. 3 is N-PID control program process flow diagram.First initialization, can the initial weight of network is not only related to network reach overall smallest point, the length of e-learning time is had to larger impact simultaneously.Input layer to hidden layer power initial value meet (
, the n) mapping of e,
,
.Hidden layer to output layer power initial value can be by test time study obtain.Network initial value also should comprise the momentum term factor, learning efficiency, sampling number, assigned error size.N-PID is according to given rotating speed
calculate controlled quentity controlled variable with motor feedback rotation speed n, utilize method of steepest descent constantly to revise network weight, make error in setting range, then stop study.
Fig. 4 has illustrated neural network-PID and traditional PI D trace performance, and as can be seen from the figure, neural network-PID can respond setting speed rapidly, follows effect better, and traditional PID adjustment response is slower, follows weak effect.
Under normal mode of operation, PEDC according to train arrival/information leaving from station provide and open the door/close gate signal to the central controller of DCU, central controller is according to the speed curves starting BLDCM setting, by the switching of transmission mechanism control shield door.In the middle of the operation of shield door, constantly gather motor position and current information, the rotating speed of acquisition and setting speed, after speed and current double closed loop control, produce pulse-width modulation PWM control motor speed and make it follow setting speed.
Especially, in order to tackle the various emergency case in the middle of actual motion, door machine controller also should comprise various holding circuits, as current protecting circuit, voltage protection circuit and virtual protection circuit etc.
As mentioned above, the control algolithm of BLDCM of the present invention adopts speed and current double closed loop control, and its rotating speed outer shroud adopts neural network-PID switching controls, and current inner loop adopts conventional PID to regulate.Traditional PID regulates (to whole control system) to be difficult to overcome the hyperharmonic oscillation problem in short-term of brshless DC motor, Neural Network PID Control can utilize the self study of neural network, adaptation function to adjust online pid control parameter, thereby make shield door operation more stable, antijamming capability is strong.
Claims (5)
1. shield a Neural network PID method for every machine controller, described Neural network PID structure is 2 ' 3 ' 1 structure, belongs to BP neural network, is respectively input layer, hidden layer and output layer, it is characterized in that, comprises the following steps:
Reference rotation velocity
be input to respectively two neurons of input layer with motor feedback rotation speed n, according to specific function output, output to respectively three neurons of ratio, integration, differential of hidden layer, output to output layer according to specific function, output reference current
.
2. shielding every machine controller of the Neural network PID method control described in an application rights requirement 1, comprise a central controller of machine controller DCU, cell controller PEDC, drive UPS, rectifier, inverter, current collector, with the brshless DC motor BLDCM of position transducer, wherein: central controller is connected with PEDC by asynchronous serial communication SCI module, rectifier is connected with driving UPS, inverter is connected with rectifier, BLDCM is connected with inverter, current collector is connected with inverter, central controller is connected with the position transducer of BLDCM by quadrature coding pulse circuit QEP module, central controller is connected with current acquisition by D/A module, central controller is connected with inverter by pulse-width modulation PWM module, the control of the inner employing of central controller speed and current double closed loop, current inner loop adopts PI to control, rotating speed outer shroud adopts Neural Network PID Control.
3. shielding every machine controller according to claim 2, is characterized in that: described central controller connects master monitor MMS by CAN bus.
4. shielding every machine controller according to claim 2, is characterized in that: current inner loop adopts PID to control.
5. according to any one shielding every machine controller described in claim 2-4, it is characterized in that: also comprise various kinds of protective circuit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410020706.4A CN103777601A (en) | 2014-01-17 | 2014-01-17 | Shield door door-machine controller neural network PID method and controller controlled through the same |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410020706.4A CN103777601A (en) | 2014-01-17 | 2014-01-17 | Shield door door-machine controller neural network PID method and controller controlled through the same |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103777601A true CN103777601A (en) | 2014-05-07 |
Family
ID=50569976
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410020706.4A Pending CN103777601A (en) | 2014-01-17 | 2014-01-17 | Shield door door-machine controller neural network PID method and controller controlled through the same |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103777601A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105305895A (en) * | 2015-11-17 | 2016-02-03 | 吉林大学 | Torque feedback and commutation compensation-based brushless motor control method |
| CN106292631A (en) * | 2016-08-25 | 2017-01-04 | 哈尔滨理工大学 | A kind of PWM rectifier fault diagnosis system based on neutral net |
| CN108809167A (en) * | 2018-06-26 | 2018-11-13 | 长春工业大学 | A kind of BP neural network PID speed regulating control algorithms based on fuzzy control |
| CN109209527A (en) * | 2018-11-27 | 2019-01-15 | 济南大学 | A kind of steam turbine stress optimization control strategy |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004030200A1 (en) * | 2002-09-30 | 2004-04-08 | Sanyo Denki Co.,Ltd. | Brushless dc fan motor |
| JP2008005683A (en) * | 2006-06-26 | 2008-01-10 | Nidec Shibaura Corp | Drive unit for brushless dc motor |
| CN102063109A (en) * | 2010-11-29 | 2011-05-18 | 株洲南车时代电气股份有限公司 | Neural network-based subway train fault diagnosis device and method |
-
2014
- 2014-01-17 CN CN201410020706.4A patent/CN103777601A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004030200A1 (en) * | 2002-09-30 | 2004-04-08 | Sanyo Denki Co.,Ltd. | Brushless dc fan motor |
| JP2008005683A (en) * | 2006-06-26 | 2008-01-10 | Nidec Shibaura Corp | Drive unit for brushless dc motor |
| CN102063109A (en) * | 2010-11-29 | 2011-05-18 | 株洲南车时代电气股份有限公司 | Neural network-based subway train fault diagnosis device and method |
Non-Patent Citations (3)
| Title |
|---|
| 左旭坤: "基于DSP的直流无刷电动机神经网络控制***", 《深圳职业技术学院学报》 * |
| 左旭坤: "基于DSP的直流无刷电动机神经网络控制***", 《深圳职业技术学院学报》, no. 1, 31 January 2006 (2006-01-31), pages 11 - 14 * |
| 杨子河: "地铁屏蔽门监控***的研究与设计", 《万方数据库》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105305895A (en) * | 2015-11-17 | 2016-02-03 | 吉林大学 | Torque feedback and commutation compensation-based brushless motor control method |
| CN106292631A (en) * | 2016-08-25 | 2017-01-04 | 哈尔滨理工大学 | A kind of PWM rectifier fault diagnosis system based on neutral net |
| CN108809167A (en) * | 2018-06-26 | 2018-11-13 | 长春工业大学 | A kind of BP neural network PID speed regulating control algorithms based on fuzzy control |
| CN109209527A (en) * | 2018-11-27 | 2019-01-15 | 济南大学 | A kind of steam turbine stress optimization control strategy |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103777601A (en) | Shield door door-machine controller neural network PID method and controller controlled through the same | |
| Yaoqin et al. | A new maximum power point tracking control scheme for wind generation | |
| CN105141201B (en) | A kind of magnetic suspension control torque gyroscope high-speed electric expreess locomotive is without position commutation error correction control system and method | |
| CN102290814B (en) | Solar energy generation system | |
| Yan et al. | Moving horizon optimization of dynamic trajectory planning for high-speed train operation | |
| CN103414415B (en) | A kind of motor control method based on PI parameter self-tuning | |
| CN106685291A (en) | Position failure fault-tolerant driving control method for wind power generation pitch system | |
| CN102593862B (en) | Photovoltaic grid-connected inverter and control method thereof | |
| CN103532459A (en) | Linear servo motor control method for numerically-controlled machine tool driving | |
| Corradini et al. | Fully sensorless robust control of variable-speed wind turbines for efficiency maximization | |
| CN103440422A (en) | Bus behind-schedule recovering method based on arrival time predication with time window | |
| CN104378038A (en) | Permanent magnet synchronous motor parameter identification method based on artificial neural network | |
| CN107769642A (en) | A kind of driving speed governing integral type constrained forecast control method of direct current generator | |
| CN103309254A (en) | Motor control system of gate | |
| CN102662323A (en) | Adoptive sliding mode control method and adoptive sliding mode control system of wind power generation variable-pitch actuator | |
| CN103066603B (en) | Self-adaptive radial basis function (RBF) neural sliding mode control method for three-phase shunt active power filter | |
| CN114578726B (en) | Magnetic levitation train operation simulation system | |
| CN204761340U (en) | Three closed -loop control dc -inverter compressors based on DSP | |
| CN109639192A (en) | A kind of durface mounted permanent magnet synchronous motor time-sharing control system and Time-sharing control method | |
| CN104113241A (en) | Control device and method of swing angle motor operating mechanism of high-voltage GIS (Gas Insulated Switchgear) isolating switch | |
| Rhaili et al. | A new strategy based neural networks MPPT controller for five-phase PMSG based variable-speed wind turbine | |
| CN108964504A (en) | A kind of control system and control method of current transformer parallel system | |
| CN208907342U (en) | A kind of gate intelligent controlling device | |
| CN104993759B (en) | The quick field weakening control method of double-fed wind power generator | |
| Alnuman et al. | Electrical modelling of a metro system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140507 |
|
| RJ01 | Rejection of invention patent application after publication |