CN110492809B - Asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation - Google Patents
Asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation Download PDFInfo
- Publication number
- CN110492809B CN110492809B CN201910788111.6A CN201910788111A CN110492809B CN 110492809 B CN110492809 B CN 110492809B CN 201910788111 A CN201910788111 A CN 201910788111A CN 110492809 B CN110492809 B CN 110492809B
- Authority
- CN
- China
- Prior art keywords
- asynchronous motor
- representing
- function
- neural network
- discrete
- 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
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P21/00—Arrangements or methods for the control of electric machines by vector control, e.g. by control of field orientation
- H02P21/0003—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control
- H02P21/0014—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control using neural networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P21/00—Arrangements or methods for the control of electric machines by vector control, e.g. by control of field orientation
- H02P21/06—Rotor flux based control involving the use of rotor position or rotor speed sensors
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P25/00—Arrangements or methods for the control of AC motors characterised by the kind of AC motor or by structural details
- H02P25/02—Arrangements or methods for the control of AC motors characterised by the kind of AC motor or by structural details characterised by the kind of motor
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P27/00—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage
- H02P27/04—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage using variable-frequency supply voltage, e.g. inverter or converter supply voltage
- H02P27/06—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage using variable-frequency supply voltage, e.g. inverter or converter supply voltage using dc to ac converters or inverters
- H02P27/08—Arrangements or methods for the control of AC motors characterised by the kind of supply voltage using variable-frequency supply voltage, e.g. inverter or converter supply voltage using dc to ac converters or inverters with pulse width modulation
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P2207/00—Indexing scheme relating to controlling arrangements characterised by the type of motor
- H02P2207/01—Asynchronous machines
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Control Of Ac Motors In General (AREA)
- Control Of Electric Motors In General (AREA)
Abstract
The invention belongs to the technical field of asynchronous motor position tracking control, and particularly discloses a neural network approximation-based asynchronous motor dynamic surface discrete fault-tolerant control method. Aiming at the problem of actuator faults which are easy to occur in an asynchronous motor driving system, the method establishes an actuator fault model and simultaneously establishes a discrete fault model of the asynchronous motor system by combining an Euler method; the dynamic surface control technology is introduced into the traditional backstepping method, so that the problems of 'calculation explosion' and 'cause and effect contradiction' in a backstepping control algorithm of a discrete system are solved; the method utilizes a radial basis function neural network to process nonlinear items in an asynchronous motor discrete system, solves the problems of unknown parameters and actuator faults in the system by combining an adaptive control method, and constructs an asynchronous motor dynamic surface discrete fault-tolerant controller based on neural network approximation; the method can overcome the influence of actuator faults, ensure an ideal control effect and realize quick and stable response to the rotating speed.
Description
Technical Field
The invention belongs to the technical field of asynchronous motor position tracking control, and particularly relates to a dynamic surface discrete fault-tolerant control method for an asynchronous motor based on neural network approximation.
Background
Asynchronous motors (IMs) are alternating current motors, also called induction motors, and are mainly used as motors. Asynchronous motors are widely used in industrial and agricultural production, for example: asynchronous motors are used as prime movers for beds, water pumps, metallurgical, mining equipment, and light industrial machinery. However, the mathematical model of the asynchronous motor has the characteristics of high nonlinearity, strong coupling, multivariable and the like, and is easily influenced by uncertain factors such as motor parameter variation and external load disturbance, so that it is a challenging task to realize high-performance control of the asynchronous motor.
In recent years, the research of nonlinear control methods, such as sliding mode control, dynamic surface control, hamiltonian control, backstepping control and other control methods, has made great progress. However, most of these techniques are based on asynchronous motor continuous systems, for which there are fewer control algorithms for discrete systems. Because the actual engineering system mostly adopts the discrete control technology, and the discrete control algorithm is superior to the continuous algorithm in realizability and stability. Therefore, the control method for the discrete system construction of the asynchronous motor has very important practical significance. In addition, the above control methods do not consider the effect of actuator failure during the operation of the asynchronous motor. In the running process of the asynchronous motor, as the asynchronous motor continuously executes control tasks for a long time, the possibility of system execution mechanism failure is increased, once the system execution mechanism fails and is not processed in time, the performance of the control system is reduced, and even serious damage is caused to equipment and personal safety.
The backstepping method is an effective nonlinear system control method, and the core thought is to decompose a complex nonlinear system into a plurality of simple low-order subsystems, design the controller step by introducing a virtual control function, and finally determine a real control law, so that the high-order nonlinear system is effectively controlled. However, when the back-stepping method is applied to a discrete system, the continuous differentiation process of the virtual control function can cause the problems of 'computational explosion' and 'causal contradiction'. In addition, for some high-order systems with high nonlinearity and uncertain parameters, the nonlinear term can cause the controller to become very complex, thus the burden of online calculation is increased, the online control of a computer control system is not facilitated, and the traditional backstepping method is difficult to deal with the problem of uncertain parameters.
Disclosure of Invention
The invention aims to provide a dynamic surface discrete fault-tolerant control method of an asynchronous motor based on neural network approximation, which comprehensively considers the problem of actuator faults easily occurring in the operation of the motor and realizes the rapid and stable position tracking control of the asynchronous motor.
In order to achieve the purpose, the invention adopts the following technical scheme:
the asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation comprises the following steps:
a. establishing discrete dynamic mathematical model of asynchronous motor
Under a synchronous rotating coordinate system, oriented according to the rotor flux linkage, the asynchronous motor driving system model can be expressed as:
in the formula (I), the compound is shown in the specification,nprepresenting the number of pole pairs, T, of an asynchronous motorLRepresenting load torque, J representing moment of inertia, LmRepresenting the mutual inductance, omega the rotor angular velocity, theta the rotor angle, psidRepresenting the rotor flux linkage; i.e. idRepresenting d-axis current, iqRepresenting the q-axis current, udDenotes the d-axis voltage, uqRepresents the q-axis voltage; rsRepresenting the resistance of the stator, LsRepresenting the inductance of the stator; rrRepresenting the resistance of the rotor, LrRepresenting the inductance of the rotor;
to simplify the discrete dynamic mathematical model of an asynchronous motor, new variables are defined as follows:
wherein k is the step number of the discrete system; theta (k), omega (k), iq(k)、ψd(k) And id(k) Respectively representing a rotor angle, a rotor angular speed, a q-axis current, a rotor flux linkage and a d-axis current corresponding to the k step;
the following two actuator fault models are defined respectively:
deviation fault model: u (k) ═ v (k) + b (k) (2)
Wherein u (k) represents the actual control input after the actuator failure, v (k) represents the control input given by the control method, and b (k) is a bounded function;
gain failure fault model: u (k) ═ 1- ρ v (k) (3)
In the formula, rho represents a gain failure parameter, and rho is more than or equal to 0 and less than 1;
the bias and gain failures can be expressed as follows: u (k) ═ 1- ρ v (k) + b (k) (4)
Obtaining a discrete fault model of the asynchronous motor by a new variable and an Euler formula, namely:
wherein, DeltatA sampling period representing a discrete system of asynchronous motors; v. ofq(k) And vd(k) Representing a true control law;
ρ1a gain failure parameter representing a q-axis stator voltage;
ρ2a gain failure parameter representing a d-axis stator voltage;
b1(k) a deviation fault function representing the q-axis stator voltage; b2(k) A deviation fault function representing the d-axis stator voltage;
b. according to the dynamic surface technology and the principle of a back-stepping method, an asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation is designed, and a discrete fault model of an asynchronous motor is simplified into two independent subsystems, namely:
by a state variable x1(k),x2(k),x3(k) And a control input vq(k) Formed subsystem and composed of state variables x4(k),x5(k) And controlInput vd(k) A component subsystem;
for a continuous non-linear function f (Z), there is a radial basis function neural network WTP (Z) is such that: f (Z) ═ WTP (Z) + tau (Z), wherein tau (Z) is an approximation error and satisfies | tau (Z) | less than or equal to any small normal number;
W∈Rlis a weight vector, the number of nodes of the neural network is a positive integer, l is greater than 1, RlA real number vector set;
P(Z)=[p1(Z),...,pl(Z)]T∈Rlis a vector of basis functions; p is a radical ofc(Z) is a Gaussian function, and the expression is as follows:
wherein, c is 1cIs the center of the receiving domain, ηcIs the width of the gaussian function;
defining the dynamic surface filter as:
wherein i is 1,2,3, ζiAs time constant of the filter, αi(k) For virtual control law, make αi(k) The output signal α of the dynamic surface filter is obtained by first-order low-pass filteringid(k),αi(0) Representation αi(k) αid(0) Representation αid(k) Initial value of (c), virtual control law αi(k) The specific structure of (a) will be given in the following control method design;
wherein x is1d(k) For the desired position signal, x4d(k) Is the desired rotor flux linkage signal;
in the control method, a Lyapunov function is selected in each step to construct a control function, and the specific process is as follows:
according to the error variable e2(k)=x2(k)-α1d(k) And is obtained based on the young inequality:
load torque T in actual operation of asynchronous motorLIs a bounded value, set | TLL is less than or equal to d, wherein d is more than 0;
according to the error variable e3(k)=x3(k)-α2d(k) And is obtained based on the young inequality:
from the principle of radial basis function neural network approximation, for a given arbitrary3> 0, a radial basis function neural network W is present3 TP3(z3(k) So that:
wherein, g3(k) Is an unknown non-linear function; z is a radical of3(k)=[x1(k),x2(k),x3(k),x4(k),x5(k)]T,τ3Expressing approximation error and satisfying inequality | tau3|≤3,||W3Is the vector W3Thereby:
wherein, γ3And3is a normal number; definition | | | W3||=η3And η3> 0, define variable η3Is estimated error ofComprises the following steps: is variable η3An estimated value of (d); formula (13) is substituted for formula (12) to obtain:
according to the error variable e5(k)=x5(k)-α3d(k) And is obtained based on the young inequality:
from the radial basis function neural network approximation theorem, it is known that for arbitrary5Not less than 0, a radial basis function neural network W exists5 TP5(z5(k) Make sure that
Wherein, g5(k) Is an unknown non-linear function; z is a radical of5(k)=[x1(k),x2(k),x3(k),x4(k),x5(k)]T,τ5Represents an approximation error and satisfies | τ5|≤5,||W5Is the vector W5Thereby:
wherein, γ5And5is a positive number, and the number of the positive number,is η5The estimated value of (1), define | | W5||=η5And η5> 0, define variable η5Is estimated error ofComprises the following steps:substituting equation (20) into equation (19) yields:
substituting equations (8), (10), (15), and (17) into equation (22) yields:
c. carrying out stability analysis on the constructed asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation;
defining a filtering error mui(k) Comprises the following steps: mu.si(k)=αid(k)-αi(k) I ═ 1,2,3, the Lyapunov function was chosen:
differentiating V (k) to obtain:
definition of vi(k)=αi(k)-αi(k +1) obtained by the formula (6):
further obtaining:
defined by the radial basis function P (Z) | | P3(z3(k))||2<l3,||P5(z5(k))||2<l5,l3And l5Respectively representing a neural network W3 TP3(z3(k) ) and W5 TP5(z5(k) Node number of); based on the young inequality:
by an error variable e3(k)=x3(k)-α2d(k)、e5(k)=x5(k)-α3d(k) Formula (13) and formula (20):
substituting equations (29), (30), (31), (32), and (33) into equation (28) yields:
substituting equations (29), (30), (31), (32), and (34) into equation (28) yields:
during the operation of the motor, the rotor flux linkage is a bounded numerical value, thus definingWhere N is a normal number, the substitution of equations (22), (27), (35) and (36) into equation (25) yields:
The selected parameter satisfiesError of the measurementAndif true, obtaining that delta V (k) is less than or equal to 0;
the true control law v is obtained from the above analysisq(k) And vd(k) Under the action of (2), the tracking error e of the discrete system of the asynchronous motor1(k) And e4(k) It is possible to converge to a sufficiently small neighborhood of the origin and ensure that the other signals are bounded.
The invention has the following advantages:
(1) the method aims at a discrete time system, and has higher stability and realizability compared with a control method of a continuous time system.
(2) The method comprehensively considers the problems of gain failure and deviation of the actuator of the asynchronous motor, eliminates the influence of faults on the control system by utilizing a fault-tolerant control technology, ensures that the designed control system can still stably run after the faults occur, and effectively solves the problem of position tracking control of the asynchronous motor under the condition of the fault of the actuator.
(3) The method adopts the dynamic surface technology, so that the problems of 'calculation explosion' and 'cause and effect contradiction' caused by continuous difference solving of the virtual function in the traditional back step method are effectively avoided; the neural network is used for processing unknown nonlinear terms in the motor system, so that the problem of highly nonlinear control of the asynchronous motor is effectively solved, and finally, more accurate control precision is achieved.
(4) The method of the invention does not need to modify the parameters of the controller according to the difference of the asynchronous motors, can realize the stable control of the asynchronous motors with all models and powers in principle, reduces the measurement of the parameters of the asynchronous motors in the control process, and is beneficial to realizing the quick response of the rotation speed adjustment of the asynchronous motors.
Drawings
Fig. 1 is a schematic diagram of a composite controlled object composed of an asynchronous motor dynamic surface discrete fault-tolerant controller based on neural network approximation, a coordinate transformation unit and an SVPWM inverter in the embodiment of the invention;
FIG. 2 is a rotor angle and rotor angle set value tracking simulation diagram after the control method of the present invention is adopted;
FIG. 3 is a simulation diagram of the rotor angle and the set value tracking error of the rotor angle after the control method of the present invention is adopted;
FIG. 4 is a simulation diagram of rotor flux linkage and rotor flux linkage setpoint tracking after the control method of the present invention is employed;
FIG. 5 is a simulation diagram of the voltage of the stator of the d-axis of the asynchronous motor after the control method of the invention is adopted;
fig. 6 is a simulation diagram of the q-axis stator voltage of the asynchronous motor after the control method of the invention is adopted.
Detailed Description
The basic idea of the invention is as follows:
aiming at the problem of actuator faults which are easy to occur in an asynchronous motor driving system, an actuator fault model is established, and meanwhile, a discrete fault model of the asynchronous motor system is established by combining an Euler method; the dynamic surface control technology is introduced into the traditional back-stepping method, so that the problems of 'calculation explosion' and 'cause and effect contradiction' caused by continuous difference solving in a discrete system are solved; the nonlinear items in the discrete motor model are processed by utilizing the radial basis function neural network, the problems of unknown parameters and actuator faults in the system are solved by combining the self-adaptive control method, and the asynchronous motor dynamic surface discrete fault-tolerant controller based on neural network approximation is constructed.
The invention is described in further detail below with reference to the following figures and detailed description:
FIG. 1 is a schematic diagram of a composite controlled object composed of a neural network approximation-based asynchronous motor dynamic surface discrete fault-tolerant controller, a coordinate transformation unit and an SVPWM inverter in an embodiment of the present invention, where ω represents rotor angular velocity, U representsαAnd UβIndicating the voltage in the two-phase rotating coordinate system, and U, V and W indicating the three-phase ac voltage.
Referring to fig. 1, the asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation adopts components including an asynchronous motor dynamic surface discrete fault-tolerant controller 1 based on neural network approximation, a coordinate transformation unit 2, an SVPWM inverter 3, a rotation speed detection unit 4, and a current detection unit 5.
The rotating speed detection unit 4 and the current detection unit 5 are mainly used for detecting the current value and the rotating speed related variable of the asynchronous motor, the actually measured current and the actually measured rotating speed variable are used as input, voltage control is carried out through the asynchronous motor dynamic surface discrete fault-tolerant controller 1 based on neural network approximation, and finally the three-phase current and rotating speed related variable are converted into the rotating speed of the three-phase electrically controlled asynchronous motor.
In order to design a more efficient controller, it is necessary to build a dynamic model of the asynchronous motor.
The asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation comprises the following steps:
a. establishing discrete dynamic mathematical model of asynchronous motor
Oriented according to the rotor flux linkage (psi) in a synchronous rotating coordinate systemq0), the asynchronous motor drive system model can be expressed as:
in the formula (I), the compound is shown in the specification,nprepresenting the number of pole pairs, T, of an asynchronous motorLRepresenting load torque, J representing moment of inertia, LmRepresenting the mutual inductance, omega the rotor angular velocity, theta the rotor angle, psidRepresenting the rotor flux linkage; i.e. idRepresenting d-axis current, iqRepresenting the q-axis current, udDenotes the d-axis voltage, uqRepresents the q-axis voltage; rsRepresenting the resistance of the stator, LsRepresenting the inductance of the stator; rrRepresenting the resistance of the rotor, LrRepresenting the inductance of the rotor.
To simplify the discrete dynamic mathematical model of an asynchronous motor, new variables are defined as follows:
wherein k is the step number of the discrete system; theta (k), omega (k), iq(k)、ψd(k) And id(k) And respectively representing the corresponding rotor angle, rotor angular speed, q-axis current, rotor flux linkage and d-axis current at the k step.
The actuator faults considered in the embodiment of the invention are deviation faults and gain faults, and the following two fault models are respectively defined:
deviation fault model: u (k) ═ v (k) + b (k) (2)
Where u (k) represents the actual control input after actuator failure, v (k) represents the control input given by the control method, and b (k) is a bounded function.
Gain failure fault model: u (k) ═ 1- ρ v (k) (3)
In the formula, rho represents a gain failure parameter, and rho is more than or equal to 0 and less than 1.
The bias and gain failures can be expressed as follows: u (k) ═ 1- ρ v (k) + b (k) (4)
Obtaining a discrete fault model of the asynchronous motor by a new variable and an Euler formula, namely:
wherein, DeltatA sampling period representing a discrete system of asynchronous motors; v. ofq(k) And vd(k) Representing the true control law.
ρ1A gain failure parameter representing the q-axis stator voltage.
ρ2A gain failure parameter representing the d-axis stator voltage.
b1(k) A deviation fault function representing the q-axis stator voltage; b2(k) A deviation fault function of the d-axis stator voltage is represented.
b. According to the dynamic surface technology and the principle of a back-stepping method, an asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation is designed, and a discrete fault model of an asynchronous motor is simplified into two independent subsystems, namely:
by a state variable x1(k),x2(k),x3(k) And a control input vq(k) Formed subsystem and composed of state variables x4(k),x5(k) And a control input vd(k) And (4) forming a subsystem.
For a continuous non-linear function f (Z), there is a radial basis function neural network WTP (Z) is such that: f (Z) ═ WTP (Z) + tau (Z), wherein tau (Z) is an approximation error and satisfies | tau (Z) | ≦ being any small normal number.
W∈RlIs a weight vector, the number of nodes of the neural network is a positive integer, l is greater than 1, RlIs a set of real vectors.
P(Z)=[p1(Z),...,pl(Z)]T∈RlIs a vector of basis functions. p is a radical ofc(Z) is a Gaussian function, and the expression is as follows:
wherein, c is 1cIs the center of the receiving domain, ηcIs the width of the gaussian function.
Defining the dynamic surface filter as:
wherein i is 1,2,3, ζiAs time constant of the filter, αi(k) For virtual control law, make αi(k) The output signal α of the dynamic surface filter is obtained by first-order low-pass filteringid(k),αi(0) Representation αi(k) αid(0) Representation αid(k) Initial value of (c), virtual control law αi(k) The specific structure of (a) will be given in the following control method design.
wherein x is1d(k) For the desired position signal, x4d(k) Is the desired rotor flux linkage signal.
In the control method, a Lyapunov function is selected in each step to construct a control function, and the specific process is as follows:
according to the error variable e2(k)=x2(k)-α1d(k) And is obtained based on the young inequality:
load torque T in actual operation of asynchronous motorLIs a bounded value, set | TLL is less than or equal to d, wherein d is more than 0.
according to the error variable e3(k)=x3(k)-α2d(k) And is obtained based on the young inequality:
from the principle of radial basis function neural network approximation, for a given arbitrary3> 0, a radial basis function neural network W is present3 TP3(z3(k) So that:
wherein, g3(k) Is an unknown non-linear function; z is a radical of3(k)=[x1(k),x2(k),x3(k),x4(k),x5(k)]T,τ3Expressing approximation error and satisfying inequality | tau3|≤3,||W3Is the vector W3Thereby:
wherein, γ3And3is a normal number; definition | | | W3||=η3And η3> 0, define variable η3Is estimated error ofComprises the following steps: is variable η3An estimated value of (d); formula (13) is substituted for formula (12) to obtain:
according to the error variable e5(k)=x5(k)-α3d(k) And is obtained based on the young inequality:
from the radial basis function neural network approximation theorem, it is known that for arbitrary5Not less than 0, a radial basis function neural network W exists5 TP5(z5(k) Make sure that
Wherein, g5(k) Is an unknown non-linear function;z5(k)=[x1(k),x2(k),x3(k),x4(k),x5(k)]T,τ5represents an approximation error and satisfies | τ5|≤5,||W5Is the vector W5Norm of (d). So that:
wherein, γ5And5is a positive number, and the number of the positive number,is η5The estimated value of (1), define | | W5||=η5And η5> 0, define variable η5Is estimated error ofComprises the following steps:substituting equation (20) into equation (19) yields:
substituting equations (8), (10), (15), and (17) into equation (22) yields:
c. and carrying out stability analysis on the constructed asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation.
Defining a filtering error mui(k) Comprises the following steps: mu.si(k)=αid(k)-αi(k) I ═ 1,2,3, the Lyapunov function was chosen:
differentiating V (k) to obtain:
definition of vi(k)=αi(k)-αi(k +1) obtained by the formula (6):
further obtaining:
defined by the radial basis function P (Z) | | P3(z3(k))||2<l3,||P5(z5(k))||2<l5,l3And l5Respectively representing a neural network W3 TP3(z3(k) ) and W5 TP5(z5(k) Node number of). Based on the young inequality:
by an error variable e3(k)=x3(k)-α2d(k)、e5(k)=x5(k)-α3d(k) Formula (13) and formula (20):
substituting equations (29), (30), (31), (32), and (33) into equation (28) yields:
substituting equations (29), (30), (31), (32), and (34) into equation (28) yields:
during the operation of the motor, the rotor flux linkage is a bounded numerical value, thus definingWhere N is a normal number, the substitution of equations (22), (27), (35) and (36) into equation (25) yields:
The selected parameter satisfiesError of the measurementAndif true, obtaining Δ V (k) is less than or equal to0。
The true control law v is obtained from the above analysisq(k) And vd(k) Under the action of (2), the tracking error e of the discrete system of the asynchronous motor1(k) And e4(k) It is possible to converge to a sufficiently small neighborhood of the origin and ensure that the other signals are bounded.
And simulating the established asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation in a virtual environment to verify the feasibility of the control method.
The motor and load parameters are:
J=0.0586Kg·m2,Rs=0.1Ω,Rr=0.15Ω,Ls=Lr=0.0699H,Lm=0.068H。
the tracking reference signal is: x is the number of1d(k)=sin(Δtk pi/2); the expected rotor flux linkage signal is: x is the number of4d(k)=1。
selecting the control law parameters as follows:
3=1.25,5=1.6,γ3=0.175,γ5=0.25,ζ1=0.0012,ζ2=0.00074,ζ3=0.0012。
the radial basis function neural network is selected as follows: neural network W3 TP3(z3(k) ) and W5 TP5(z5(k) Contains 9 centers evenly distributed in [ -8,8 ]]The width of each internal node is 2.
Considering the failure of the gain of the actuator and the deviation fault of the actuator in the running process of the asynchronous motor, the fault conditions are as follows:
the corresponding simulation results are shown in fig. 2,3, 4, 5 and 6. Wherein:
fig. 2 and fig. 4 are a tracking simulation diagram of the rotor angular position and the set value of the rotor angular position and a tracking simulation diagram of the rotor flux linkage and the set value of the rotor flux linkage, respectively, after being controlled by the method of the present invention, and the simulation results show that:
under the condition that the actuator has a fault, the controller can respond quickly, so that the system can recover the ideal operation effect.
FIG. 3 is a graph showing a simulation of the tracking error of the rotor angular position and the rotor angular position set point after control by the method of the present invention. Fig. 5 and fig. 6 are simulation diagrams of voltages of a d-axis stator and a q-axis stator of an asynchronous motor respectively after being controlled by the method of the invention.
The analog signal clearly shows that the asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation can quickly and stably track the reference signal under the fault condition that the gain of an actuator fails and deviates.
It should be understood, however, that the description herein of specific embodiments is not intended to limit the invention to the particular forms disclosed, but on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention as defined by the appended claims.
Claims (1)
1. The asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation is characterized in that,
the method comprises the following steps:
a. establishing discrete dynamic mathematical model of asynchronous motor
Under a synchronous rotating coordinate system, oriented according to the rotor flux linkage, the asynchronous motor driving system model can be expressed as:
in the formula (I), the compound is shown in the specification,nprepresenting the number of pole pairs, T, of an asynchronous motorLRepresenting load torque, J representing moment of inertia, LmRepresenting the mutual inductance, omega the rotor angular velocity, theta the rotor angle, psidRepresenting the rotor flux linkage; i.e. idRepresenting d-axis current, iqRepresenting the q-axis current, udDenotes the d-axis voltage, uqRepresents the q-axis voltage; rsRepresenting the resistance of the stator, LsRepresenting the inductance of the stator; rrRepresenting the resistance of the rotor, LrRepresenting the inductance of the rotor;
to simplify the discrete dynamic mathematical model of an asynchronous motor, new variables are defined as follows:
wherein k is the step number of the discrete system, theta (k), omega (k), iq(k)、ψd(k) And id(k) Respectively representing a rotor angle, a rotor angular speed, a q-axis current, a rotor flux linkage and a d-axis current corresponding to the k step;
the following two actuator fault models are defined respectively:
deviation fault model: u (k) ═ v (k) + b (k) (2)
Wherein u (k) represents the actual control input after the actuator failure, v (k) represents the control input given by the control method, and b (k) is a bounded function;
gain failure fault model: u (k) ═ 1- ρ v (k) (3)
In the formula, rho represents a gain failure parameter, and rho is more than or equal to 0 and less than 1;
the bias and gain failures can be expressed as follows: u (k) ═ 1- ρ v (k) + b (k) (4)
Obtaining a discrete fault model of the asynchronous motor by a new variable and an Euler formula, namely:
wherein, DeltatA sampling period representing a discrete system of asynchronous motors;
vq(k) and vd(k) Representing a true control law;
ρ1gain failure parameter, ρ, representing the q-axis stator voltage2A gain failure parameter representing a d-axis stator voltage;
b1(k) representing the deviation fault function of the q-axis stator voltage, b2(k) A deviation fault function representing the d-axis stator voltage;
b. according to the dynamic surface technology and the principle of a back-stepping method, an asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation is designed, and a discrete fault model of an asynchronous motor is simplified into two independent subsystems, namely:
by a state variable x1(k),x2(k),x3(k) And a control input vq(k) Formed subsystem and composed of state variables x4(k),x5(k) And a control input vd(k) A component subsystem;
for a continuous non-linear function f (Z), there is a radial basis function neural network WTP (Z) is such that: f (Z) ═ WTP (Z) + tau (Z), wherein tau (Z) is an approximation error and satisfies | tau (Z) | less than or equal to any small normal number;
W∈Rlis a weight vector, the number of nodes of the neural network is a positive integer, l is greater than 1, RlA real number vector set;
P(Z)=[p1(Z),...,pl(Z)]T∈Rlis a vector of basis functions; p is a radical ofc(Z) is a Gaussian function, and the expression is as follows:
wherein, c is 1cIs the center of the receiving domain, ηcIs the width of the gaussian function;
defining the dynamic surface filter as:
wherein i is 1,2,3, ζiAs time constant of the filter, αi(k) For virtual control law, make αi(k) The output signal α of the dynamic surface filter is obtained by first-order low-pass filteringid(k),αi(0) Representation αi(k) αid(0) Representation αid(k) Initial value of (c), virtual control law αi(k) The specific structure of (a) will be given in the following control method design;
wherein x is1d(k) For the desired position signal, x4d(k) Is the desired rotor flux linkage signal;
in the control method, a Lyapunov function is selected in each step to construct a control function, and the specific process is as follows:
according to the error variable e2(k)=x2(k)-α1d(k) And is obtained based on the young inequality:
load torque T in actual operation of asynchronous motorLIs a bounded value, set | TLL is less than or equal to d, wherein d is more than 0;
according to the error variable e3(k)=x3(k)-α2d(k) And is obtained based on the young inequality:
from the principle of radial basis function neural network approximation, for a given arbitrary3> 0, radial basis function neural networks existSuch that:
wherein, g3(k) Is an unknown non-linear function; z is a radical of3(k)=[x1(k),x2(k),x3(k),x4(k),x5(k)]T,τ3Expressing approximation error and satisfying inequality | tau3|≤3,||W3Is the vector W3Thereby:
wherein, γ3And3is a normal number; definition | | | W3||=η3And η3> 0, define variable η3Is estimated error ofComprises the following steps: is variable η3An estimated value of (d); formula (13) is substituted for formula (12) to obtain:
according to the error variable e5(k)=x5(k)-α3d(k) And is obtained based on the young inequality:
from the radial basis function neural network approximation theorem, it is known that for arbitrary5> 0, presence of radial basis function neural networksMake it
Wherein, g5(k) Is an unknown non-linear function; z is a radical of5(k)=[x1(k),x2(k),x3(k),x4(k),x5(k)]T,τ5Represents an approximation error and satisfies | τ5|≤5,||W5Is the vector W5Thereby:
wherein, γ5And5is a positive number, and the number of the positive number,is η5The estimated value of (1), define | | W5||=η5And η5> 0, define variable η5Is estimated error ofComprises the following steps:substituting equation (20) into equation (19) yields:
substituting equations (8), (10), (15), and (17) into equation (22) yields:
c. carrying out stability analysis on the constructed asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation;
defining a filtering error mui(k) Comprises the following steps: mu.si(k)=αid(k)-αi(k) I ═ 1,2,3, the Lyapunov function was chosen:
differentiating V (k) to obtain:
definition of vi(k)=αi(k)-αi(k +1) obtained by the formula (6):
further obtaining:
defined by the radial basis function P (Z) | | P3(z3(k))||2<l3,||P5(z5(k))||2<l5,l3And l5Respectively representing neural networksAndthe number of nodes of (a); based on the young inequality:
by an error variable e3(k)=x3(k)-α2d(k)、e5(k)=x5(k)-α3d(k) Formula (13) and formula (20):
substituting equations (29), (30), (31), (32), and (33) into equation (28) yields:
substituting equations (29), (30), (31), (32), and (34) into equation (28) yields:
during the operation of the motor, the rotor flux linkage is a bounded numerical value, thus definingWhere N is a normal number, the substitution of equations (22), (27), (35) and (36) into equation (25) yields:
The selected parameter satisfiesError of the measurementAndif true, obtaining that delta V (k) is less than or equal to 0;
the true control law v is obtained from the above analysisq(k) And vd(k) Under the action of (2), the tracking error e of the discrete system of the asynchronous motor1(k) And e4(k) It is possible to converge to a sufficiently small neighborhood of the origin and ensure that the other signals are bounded.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910788111.6A CN110492809B (en) | 2019-08-26 | 2019-08-26 | Asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910788111.6A CN110492809B (en) | 2019-08-26 | 2019-08-26 | Asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110492809A CN110492809A (en) | 2019-11-22 |
CN110492809B true CN110492809B (en) | 2020-09-04 |
Family
ID=68553992
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910788111.6A Active CN110492809B (en) | 2019-08-26 | 2019-08-26 | Asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110492809B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111404447B (en) * | 2020-03-23 | 2021-11-19 | 威海海洋职业学院 | Fault-tolerant control method and system for open-phase fault of stator winding of six-phase permanent magnet synchronous motor |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5576632A (en) * | 1994-06-30 | 1996-11-19 | Siemens Corporate Research, Inc. | Neural network auto-associator and method for induction motor monitoring |
KR20110059690A (en) * | 2011-04-18 | 2011-06-03 | 부산대학교 산학협력단 | Method for stator shorted turn fault detection using opposed pair-phase voltage/current of electric motor |
CN106712615A (en) * | 2016-12-23 | 2017-05-24 | 重庆大学 | Control method of gas turbine generator set in hybrid energy power system |
CN107065540A (en) * | 2017-03-15 | 2017-08-18 | 东北电力大学 | A kind of adaptive dynamic surface distribution control method based on neutral net |
CN108964545A (en) * | 2018-07-30 | 2018-12-07 | 青岛大学 | A kind of synchronous motor neural network contragradience Discrete Control Method based on command filtering |
CN110112971A (en) * | 2019-05-15 | 2019-08-09 | 青岛大学 | A kind of asynchronous motor Position Tracking Control method based on finite time dynamic surface |
-
2019
- 2019-08-26 CN CN201910788111.6A patent/CN110492809B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5576632A (en) * | 1994-06-30 | 1996-11-19 | Siemens Corporate Research, Inc. | Neural network auto-associator and method for induction motor monitoring |
KR20110059690A (en) * | 2011-04-18 | 2011-06-03 | 부산대학교 산학협력단 | Method for stator shorted turn fault detection using opposed pair-phase voltage/current of electric motor |
CN106712615A (en) * | 2016-12-23 | 2017-05-24 | 重庆大学 | Control method of gas turbine generator set in hybrid energy power system |
CN107065540A (en) * | 2017-03-15 | 2017-08-18 | 东北电力大学 | A kind of adaptive dynamic surface distribution control method based on neutral net |
CN108964545A (en) * | 2018-07-30 | 2018-12-07 | 青岛大学 | A kind of synchronous motor neural network contragradience Discrete Control Method based on command filtering |
CN110112971A (en) * | 2019-05-15 | 2019-08-09 | 青岛大学 | A kind of asynchronous motor Position Tracking Control method based on finite time dynamic surface |
Non-Patent Citations (1)
Title |
---|
基于神经网络的异步电动机随机自适应动态面控制;刘丽超;《中国优秀硕士学位论文全文数据库(信息科技辑)》;20180215(第2期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN110492809A (en) | 2019-11-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Niu et al. | Adaptive decentralized asymptotic tracking control for large-scale nonlinear systems with unknown strong interconnections | |
Zhang et al. | Fractional order sliding-mode control based on parameters auto-tuning for velocity control of permanent magnet synchronous motor | |
CN101917150B (en) | Robust controller of permanent magnet synchronous motor based on fuzzy-neural network generalized inverse and construction method thereof | |
CN102354107B (en) | On-line identification and control method for parameter of alternating current position servo system model | |
CN110401391B (en) | Fuzzy self-adaptive dynamic surface control method for asynchronous motor stochastic system | |
CN109873582B (en) | Limited time position tracking control method of permanent magnet synchronous motor based on dynamic surface | |
Lin | Nonlinear backstepping control design of LSM drive system using adaptive modified recurrent Laguerre orthogonal polynomial neural network | |
Zhao et al. | Neuroadaptive dynamic surface control for induction motors stochastic system based on reduced-order observer | |
Zerikat et al. | High-performance sensorless vector control of induction motor drives using artificial intelligent technique | |
Yang et al. | Gradient descent algorithm-based adaptive NN control design for an induction motor | |
CN110492809B (en) | Asynchronous motor dynamic surface discrete fault-tolerant control method based on neural network approximation | |
Kılıç et al. | Adaptive controller with RBF neural network for induction motor drive | |
CN113659895B (en) | Permanent magnet synchronous motor full-state constraint finite time control method based on instruction filtering | |
Wang et al. | Online speed controller scheme using adaptive supervisory TSK‐fuzzy CMAC for vector controlled induction motor drive | |
Belhamdi et al. | Direct field-oriented control using fuzzy logic type-2 for induction motor with broken rotor bars | |
CN110378057B (en) | Built-in permanent magnet synchronous motor anti-interference controller and design method thereof | |
Gao et al. | Adaptive fuzzy control with minimal leaning parameters for electric induction motors | |
Delavari et al. | Adaptive reinforcement learning interval type II fuzzy fractional nonlinear observer and controller for a fuzzy model of a wind turbine | |
CN112019111B (en) | State constraint-based fuzzy self-adaptive control method for random system of permanent magnet synchronous motor | |
CN112039374B (en) | Permanent magnet synchronous motor command filtering discrete control method considering input saturation | |
CN111208728B (en) | Linear active disturbance rejection control method, device equipment and storage medium | |
CN114172436B (en) | Asynchronous motor instruction filtering discrete control method based on observer | |
Ganjewar | Novel approaches in sensorless induction motor drive for industrial applications | |
Lin et al. | Sensorless inverter‐fed compressor drive system using back‐EMF estimator with PIDNN torque observer | |
CN113659894B (en) | Asynchronous motor random limited time fuzzy self-adaptive control method based on instruction filtering |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |