CN1299082A - PID nerve network controller - Google Patents

PID nerve network controller Download PDF

Info

Publication number
CN1299082A
CN1299082A CN 99117277 CN99117277A CN1299082A CN 1299082 A CN1299082 A CN 1299082A CN 99117277 CN99117277 CN 99117277 CN 99117277 A CN99117277 A CN 99117277A CN 1299082 A CN1299082 A CN 1299082A
Authority
CN
China
Prior art keywords
circuit
pid
input
network 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.)
Granted
Application number
CN 99117277
Other languages
Chinese (zh)
Other versions
CN1137419C (en
Inventor
舒怀林
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangzhou University
Original Assignee
Guangzhou University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangzhou University filed Critical Guangzhou University
Priority to CNB991172779A priority Critical patent/CN1137419C/en
Publication of CN1299082A publication Critical patent/CN1299082A/en
Application granted granted Critical
Publication of CN1137419C publication Critical patent/CN1137419C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Landscapes

  • Feedback Control In General (AREA)

Abstract

A PID nerve network controller is forward multilayer nerve net work constituted by interconnectdd proportional neurone, integral neurone and defferential neurone. It is used in several kinds of control system, and it is superior to conventional controller in that its use needs measurement on identification of the internal structure and parameters of the controlled object.

Description

The PID nerve network controller
The present invention relates to nerve network controller, particularly a kind of PID nerve network controller belongs to automation field.
In prior art, there is more deficiency in the nerve network controller that is applied in automatic control system, and for example: studying convergence speed is very slow; Be absorbed in local minimum point easily; Only have the static mappings characteristic, the incompatibility control system is to the requirement of dynamic perfromance in itself; Structure is uncertain, the actual difficulty of using; The control performance index does not match, be difficult to fast with the desired response of control system, overshoot is little, floating etc. dynamically and static performance index interrelate, system's overall process stability is difficult to guarantee.On the other hand, controller (PID controller) in the ratio of deviation, integration, differential is with the longest history, most widely used general, but because the PID control law is a kind of control law of linearity, it only has in simple linear single-variable system controls effect preferably, and in the control of complication system poor effect.
The objective of the invention is to propose a kind of PID nerve network controller, it combines the advantage of traditional PID control theory and neural network.
Elaborate below in conjunction with 9 pairs of ultimate principles of the present invention of accompanying drawing 1~accompanying drawing: the present invention adopts computer software to combine with hardware circuit and realizes.Computer software is pressed the establishment of Fig. 7 program flow diagram and is carried out according to following network structure.The PID neural network is a kind of multilayer feedforward neural network, is connected to each other formation by ratio (P) neuron, integration (I) neuron and differential (D) neuron, and these neuronic input one output functions are respectively ratio, integration and differentiation function.The PID neural network is divided three layers of composition, and wherein input layer comprises two neurons, and middle layer (being called hidden layer again) comprises three neurons, and output layer has a neuron.Network structure such as Fig. 1.The PID neural network realizes that by following computing formula two neurons of input layer are accepted outside input information, and one of them accepts desired value or set-point u 1, another accepts controlled variable value u 2The neuron input-output characteristic of input layer is up and down the unit proportion function of amplitude limit, formula as follows:
Figure 9911727700031
The output weighted sum that is input as previous stage of hidden layer neuron.Computing formula is u j ″ ( t ) = Σ i = 1 2 w ij x i ( t ) - - - j = 1 , 2 , 3
Three neurons of hidden layer are completed percentage, integration and differentiate respectively, and their input-output characteristic is respectively ratio, integration and differentiation function, and is following various:
Figure 9911727700042
Figure 9911727700043
Figure 9911727700044
The output layer neuron is input as the output summation of previous stage, and computing formula is: u ′ ′ ( t ) = Σ i = 1 3 w i x ′ i ( t )
The neuronic input-output characteristic of output layer also is up and down the unit proportion function of amplitude limit, formula as follows
Figure 9911727700046
The PID neural network is revised algorithm by the recursion that connects weights and is realized the network control optimization in Properties, and this weights recursive algorithm adopts the back-propagation algorithm of neural network.According to above-mentioned network theory and formula, the present invention adopts the combination of computer hardware and software to be achieved.It is the core devices of controller that single chip microcomputer U1 in the hardware circuit of the present invention adopts the 89C51 type, and it finishes all calculating of controller and input, output Control work; D/A U2 adopts the ADC0809 type under the control of single-chip microcomputer U1, with the actual value V of controlled device IWith set-point V PBe converted to digital quantity, send into single-chip microcomputer (Fig. 3); Photoelectric coupled device U6, U7 in the photoelectric coupling switch amount input circuit (Fig. 4), U8, resistance R 7, R8, R9, R12, R13, R14, R15, R16, R17, capacitor C 5, C6, C7, light emitting diode D10, D11, D12 realize the isolation transfer function between external switch signal and the single-chip microcomputer.The external switch signal is sent into by K1, K2, K3, and as K1, K2, when the K3 end is low level, P13, P14, P15 end are respectively low level, and as K1, K2, when the K3 end is high level, P13, P14, P15 end are respectively high level; Not gate U4, photoelectric coupled device U5, amplifier U3, resistance R 1, R2, R3, R4, R5, R6, R10, R11, electrolytic condenser C11 constitute controlled quentity controlled variable output circuit (Fig. 5).When the P10 end was input as the pulsating wave of different in width by single-chip microcomputer U1, the different magnitude of voltage of VO end output was delivered to topworks; It is direct current that commutation diode D1, D2 in the voltage-stabilized power supply circuit (Fig. 6), D3, D4 will exchange 9 volts of voltage commutations, capacitor C 3, electrochemical capacitor C12 carry out filtering, voltage stabilizer V1 carries out voltage stabilizing, again by capacitor C 4, electrochemical capacitor C13 filtering, obtain 5 volts of voltages of stable direct current, use for single-chip microcomputer, D/A, photoelectricity coupled circuit, controlled quentity controlled variable output circuit.It is direct current that commutation diode D5, D6, D7, D8 will exchange 15 volts of voltage commutations, capacitor C 9, electrochemical capacitor C15 carry out filtering, voltage stabilizer V2 carries out voltage stabilizing, again by capacitor C 8, electrochemical capacitor C14 filtering, obtain 12 volts of voltages of stable direct current, use for photoelectricity coupled circuit, controlled quentity controlled variable output circuit.
Advantage of the present invention: the PID neural network belongs to the category of multilayer feedforward neural network, and it possesses any non-linear approximation capability and other good performance of multilayer feedforward neural network.Comprised the processing unit with differential and integration dynamic perfromance in the PID neural network, it is a kind of internal dynamic network, and the requirement of Adaption Control System easily needn't add other parts again when constituting control system.The existence of ratio unit, integration unit and differential unit in the PID neural network makes the response of PID nerve network control system fast, little, the floating of overshoot, meets control system to dynamically and the static characteristics requirement.
Fig. 1 is PID neural network structure figure of the present invention.
Fig. 2 is a PID nerve network controller system hardware theory of constitution block diagram of the present invention.
Fig. 3 is single-chip microcomputer and simulated measurement input circuit figure.U1 is that single-chip microcomputer adopts the 89C51 type among Fig. 3, and U2 is that analog to digital converter adopts the ADC0809 type.
Fig. 4 is photoelectric coupling switch amount input circuit figure.U6 among Fig. 4, U7, U8 are photoelectrical couplers.
Fig. 5 is controlled quentity controlled variable output circuit figure.
Fig. 6 is voltage-stabilized power supply circuit figure.
Fig. 7 is a PID nerve network controller software program flow chart.
Fig. 8 is that the PID Shen is through the network controller board schematic layout pattern.
Fig. 9 is a system chart of controlling tube type resistance furnace among the embodiment with the PID nerve network controller.
Embodiment: PID nerve network controller of the present invention is realized by the circuit board type structure with reference to Fig. 1-Fig. 9 design.The layout of various piece on circuit board is shown in 8 figure.Controlled device is a tube type resistance furnace in the example, 1000 watts of power, and 220 volts of voltages require its fire box temperature of control constant.A PID nerve network controller that adopts the present invention to realize by the temperature sensor detected temperatures, has been realized the control requirement by topworks's (bidirectional triode thyristor) regulation voltage (Fig. 9).

Claims (4)

1, a kind of PID nerve network controller, comprise computer control software and hardware circuit, it is characterized in that computer software is a multilayer feedforward neural network, structure is divided into three layers, wherein input layer comprises two neurons, middle layer (being called hidden layer again) comprises three neurons, and output layer has a neuron.These neuronic input-output functions are respectively ratio, integration and differentiation function.
2, controller according to claim 1 is characterized in that hardware circuit comprises single-chip microcomputer (contain and reset and crystal oscillator) circuit, simulated measurement input circuit, photoelectric coupling switch amount input circuit, controlled quentity controlled variable output circuit and mu balanced circuit.
3,, it is characterized in that described single chip circuit is to adopt 89C51 type single-chip microcomputer according to claim 1,2 described controllers.
4,, it is characterized in that described simulated measurement input circuit adopts the ADC0809 analog to digital converter according to claim 1,2 described controllers.
CNB991172779A 1999-12-07 1999-12-07 PID nerve network controller Expired - Fee Related CN1137419C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CNB991172779A CN1137419C (en) 1999-12-07 1999-12-07 PID nerve network controller

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CNB991172779A CN1137419C (en) 1999-12-07 1999-12-07 PID nerve network controller

Publications (2)

Publication Number Publication Date
CN1299082A true CN1299082A (en) 2001-06-13
CN1137419C CN1137419C (en) 2004-02-04

Family

ID=5279913

Family Applications (1)

Application Number Title Priority Date Filing Date
CNB991172779A Expired - Fee Related CN1137419C (en) 1999-12-07 1999-12-07 PID nerve network controller

Country Status (1)

Country Link
CN (1) CN1137419C (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009097728A1 (en) * 2008-02-04 2009-08-13 Nanjing University Of Science And Technology Robust control method for asphalt cement mixing device dispensing error
CN103235503A (en) * 2013-01-05 2013-08-07 太原科技大学 Novel multi-neuron PID (proportion, integration and differentiation) controller
CN108555914A (en) * 2018-07-09 2018-09-21 南京邮电大学 A kind of DNN Neural Network Adaptive Control methods driving Dextrous Hand based on tendon
CN110647078A (en) * 2019-09-26 2020-01-03 西安科技大学 Underground unattended drainage system for coal mine and control method

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009097728A1 (en) * 2008-02-04 2009-08-13 Nanjing University Of Science And Technology Robust control method for asphalt cement mixing device dispensing error
CN103235503A (en) * 2013-01-05 2013-08-07 太原科技大学 Novel multi-neuron PID (proportion, integration and differentiation) controller
CN108555914A (en) * 2018-07-09 2018-09-21 南京邮电大学 A kind of DNN Neural Network Adaptive Control methods driving Dextrous Hand based on tendon
CN108555914B (en) * 2018-07-09 2021-07-09 南京邮电大学 DNN neural network self-adaptive control method based on tendon-driven dexterous hand
CN110647078A (en) * 2019-09-26 2020-01-03 西安科技大学 Underground unattended drainage system for coal mine and control method
CN110647078B (en) * 2019-09-26 2020-10-16 西安科技大学 Underground unattended drainage system for coal mine and control method

Also Published As

Publication number Publication date
CN1137419C (en) 2004-02-04

Similar Documents

Publication Publication Date Title
Su et al. A fast-computation Hopfield method to economic dispatch of power systems
Kim et al. Fuzzy precompensation of PID controllers
Nelson A theory of the low-level equilibrium trap in underdeveloped economies
Dillon et al. Short term load forecasting using an adaptive neural network
Kumar et al. Application of fractional order PID controller for AGC under deregulated environment
CN105676649A (en) Control method for sewage treatment process based on self-organizing neural network
CN105759607A (en) Design method for PAC controller based on intelligent control algorithms
Padhee et al. A novel evolutionary tuning method for fractional order PID controller
CN107272754B (en) Water and fertilizer accurate proportioning control system based on fuzzy PID
Chen et al. Fuzzy controller design by using neural network techniques
CN1137419C (en) PID nerve network controller
CN109217664B (en) A kind of Fuzzy PI of boost breadboardin load unit
Fliess et al. State feedbacks without asymptotic observers and generalized PID regulators
HORIKAWA et al. A learning fuzzy controller using a neural network
CN110429835A (en) A kind of RBFNN segmentation on-line optimization Passive Shape Control system and method based on LCL filtering
Becker et al. A note on approximating correlations from odds ratios
CN115987086A (en) Single-switch DC-DC converter on-line control method based on neural network
Shaoyuan et al. Adaptive fuzzy sliding mode control for a class of uncertain dynamic systems
EP0509796A2 (en) Defuzzifier circuit
TW200930284A (en) Photoelectric detection based control method and system adopting fuzzy theory
CN112350316A (en) Variable-discourse-domain cloud PI load frequency control method based on cloud scaling factor
Barsoum Artificial neuron controller for DC drive
Hasegawa et al. An application of fuzzy neural networks to design of adaptive fuzzy controllers
Alazzam et al. A speed control system using fuzzy logic
CN2074472U (en) Deviation-controlled anti-integral-saturation pid regulator

Legal Events

Date Code Title Description
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C06 Publication
PB01 Publication
C14 Grant of patent or utility model
GR01 Patent grant
C19 Lapse of patent right due to non-payment of the annual fee
CF01 Termination of patent right due to non-payment of annual fee