CN110829934A - Permanent magnet alternating current servo intelligent control system based on definite learning and mode control - Google Patents
Permanent magnet alternating current servo intelligent control system based on definite learning and mode control Download PDFInfo
- Publication number
- CN110829934A CN110829934A CN201911183121.3A CN201911183121A CN110829934A CN 110829934 A CN110829934 A CN 110829934A CN 201911183121 A CN201911183121 A CN 201911183121A CN 110829934 A CN110829934 A CN 110829934A
- Authority
- CN
- China
- Prior art keywords
- module
- mode
- permanent magnet
- servo
- alternating current
- 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
- 238000005312 nonlinear dynamic Methods 0.000 claims abstract description 27
- 238000003860 storage Methods 0.000 claims abstract description 18
- 238000001514 detection method Methods 0.000 claims abstract description 14
- 238000013528 artificial neural network Methods 0.000 claims description 24
- 238000012549 training Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 7
- 238000013178 mathematical model Methods 0.000 claims description 4
- 229910000831 Steel Inorganic materials 0.000 claims description 3
- 238000013459 approach Methods 0.000 claims description 3
- 239000010959 steel Substances 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 3
- 230000005389 magnetism Effects 0.000 claims description 2
- 230000003068 static effect Effects 0.000 claims description 2
- 238000000034 method Methods 0.000 abstract description 14
- 230000004044 response Effects 0.000 abstract description 5
- 230000008569 process Effects 0.000 description 9
- 230000001360 synchronised effect Effects 0.000 description 7
- 230000008859 change Effects 0.000 description 5
- 230000009471 action Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 239000000047 product Substances 0.000 description 3
- 230000002411 adverse Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000004043 responsiveness Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 239000007795 chemical reaction product Substances 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005288 electromagnetic effect Effects 0.000 description 1
- 238000005530 etching Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009022 nonlinear effect Effects 0.000 description 1
- 238000009527 percussion Methods 0.000 description 1
- 230000010287 polarization Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 235000012431 wafers Nutrition 0.000 description 1
- 238000003466 welding Methods 0.000 description 1
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
- H02P23/00—Arrangements or methods for the control of AC motors characterised by a control method other than vector control
- H02P23/0004—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control
- H02P23/0027—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control using different modes of control depending on a parameter, e.g. the speed
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Feedback Control In General (AREA)
- Control Of Electric Motors In General (AREA)
Abstract
The invention belongs to the field of servo systems, and relates to a permanent magnet alternating current servo intelligent control system based on definite learning and mode control, which comprises: signal detection module, knowledge acquisition and storage module, mode-based control module, wherein: the input end of the signal detection module is connected with the servo motor and is used for detecting the information of the servo motor and outputting the information which is respectively transmitted to the knowledge acquisition and storage module and the mode-based control module; the other input end of the knowledge acquisition and storage module is from the mode-based control module to judge whether to enter the learning-determining module and identify the nonlinear dynamics, and the output is transmitted to the mode-based control module; and the other output end of the mode-based control module is transmitted to the servo motor to drive the servo motor to work. The method can realize local accurate identification on the closed-loop dynamic state of the permanent magnet alternating current servo system under the uncertain nonlinear dynamic influence; the controller can be switched quickly, and the response performance of the servo system is improved.
Description
Technical Field
The invention belongs to the field of servo systems, and relates to a permanent magnet alternating current servo intelligent control system based on definite learning and mode control.
Background
High-grade numerical control machine tools and robots are used as one of key fields of intelligent manufacturing. For a servo system which is one of key parts, the current mainstream scheme is an alternating current servo system adopting a permanent magnet synchronous motor. The alternating current servo system is used as an actuating mechanism of a robot and a numerical control system, and finally finishes the action target by controlling the position, the speed, the torque and the combination of the three. The fields of high-performance numerical control machine tools, automatic assembly of production lines, etching and processing of semiconductor wafers, precision welding of automobile production lines, wing control of remote control airplanes, radar tracking systems, modern precision percussion weapons and the like all rely on servo systems.
A plurality of trades will all carry out structural adjustment and upgrading transformation on a large scale, deep level at present, and is urgent to high technical level's servo product demand, and the focus of market to servo system performance mainly lies in: speed/torque ripple, responsiveness, encoder accuracy/resolution, ease of use, security, and real time bus, among others. Without some pure engineering problems, the scientific problems facing the people are mainly the rotating speed/torque fluctuation and the responsiveness, and on the two key technical indexes, the research level of the current domestic servo system has an obvious gap with the world advanced level: in terms of speed frequency response, the latest sigma-7 series servo products of the japan ansha company can reach 3.1kHz, while the IS620N servo system newly introduced by the domestic advanced level of the shinagawa company has a speed frequency response of 1.2 kHz. In the aspect of speed change rate, the japan anchuan company reaches 0.01% to 0.1%, and the francisco company reaches 0.5%, and it is seen that there is a certain gap between domestic products and international high-end products in a high-performance servo system, and further, the overall competitiveness of high-technology content and high-added-value industries represented by numerically-controlled machine tools, robots, and equipment manufacturing industries is not strong. This situation will have adverse effects on various fields such as national economy, scientific and technological development, military safety, etc.
The factors influencing the key performance indexes of the permanent magnet alternating current servo system mainly comprise the following points: in the actual operation process of a servo system, along with the changes of working environment and operation conditions, the nonlinear elements such as parameter changes, fractional order characteristics, cogging torque, current harmonics, back electromotive force harmonics and the like exist in factors such as nonlinear effects, dielectric anisotropy, hysteresis curves of material polarization, possibly generated electromagnetic effects and the like of power electronic devices, digital controller discretization, switch tube dead zones, sampling quantization errors and the like. The nonlinear dynamic change processes and the possible resonance generated during the operation of the servo system can make the PI controller commonly used in the servo system difficult to completely deal with, finally the reduction of the rotating speed and the torque performance of the permanent magnet synchronous motor is caused, and the requirement of a high-performance servo system by a high-grade numerical control machine tool and a robot can not be met.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a permanent magnet alternating current servo intelligent control system based on definite learning and mode control.
The invention is realized by adopting the following technical scheme:
permanent magnetism exchanges servo intelligent control system based on confirm study and mode control includes: signal detection module, knowledge acquisition and storage module, mode-based control module, wherein:
the input end of the signal detection module is connected with the servo motor and is used for detecting the information of the servo motor and outputting the information which is respectively transmitted to the knowledge acquisition and storage module and the mode-based control module;
the other input end of the knowledge acquisition and storage module is from the mode-based control module to judge whether to enter the learning-determining module and identify the nonlinear dynamics, and the output is transmitted to the mode-based control module;
and the other output end of the mode-based control module is transmitted to the servo motor to drive the servo motor to work, so that the high-performance control of the permanent magnet alternating current servo system is realized.
Preferably, the knowledge acquisition and storage module is used for locally and accurately identifying the nonlinear dynamics of the permanent magnet alternating current servo system through a neural network, constructing a corresponding experience controller, and storing the experience controller into the pattern library so as to rapidly invoke the control module based on the pattern.
Preferably, the knowledge acquisition and storage module comprises: determining a learning module, a dynamic controller module, an RBF network approach unknown dynamic module, an online training module and a mode library module, wherein each module adopts a cascade form; determining the processing results of the inputs of the learning module from the on-line training module, the dynamic estimator of the mode-based control module and the signal detection module respectively; the output of the mode base module is communicated to a dynamic estimator module and an empirical controller module, respectively, of the mode based control module.
Preferably, the mode-based control module is configured to determine an operating mode of the permanent magnet alternating current servo system, and quickly invoke a corresponding experience controller to perform switching control according to information in the mode library, so as to implement intelligent control on the servo motor.
Preferably, the mode-based control module comprises: a dynamic estimator module, a switching control module, and an empirical controller module, wherein: the input of the dynamic estimator module is respectively from the processing results of the mode base and the signal detection module, and the output is respectively transmitted to the determination learning module and the switching control module; the output end of the switching control module is connected with the input end of the experience controller module; the other input of the experience controller module is from a mode library, and the output end of the experience controller module is transmitted to the servo motor to control the servo motor to realize high-performance operation.
Preferably, neglecting the influence of non-linearity and other factors, the mathematical model of the servo motor in the two-phase stationary coordinate system is described as follows:
wherein iαIs the current of the α shaft, and the current of the shaft,
is iαThe differential amount of (a); i.e. iβIs the current of the β shaft, and the current of the shaft,
is iβThe differential amount of (a); theta is an electrical angle of the steel sheet,
is the differential of theta; omega is the electrical angular velocity of the object,a differential amount of ω; u. ofαFor α axis control, uββ axle control quantity, R motor resistance, Lαα axes motor inductanceββ shaft motor inductance, J motor moment of inertia, p motor pole pair number, psifIs a motor magnetic linkage; t isLIs the load torque; b is a friction coefficient; t is electromagnetic torque, and the expression is as follows:
the influence of non-linear factors is considered in practice, including: parameter variation, fractional order characteristics, cogging torque, current harmonics and back electromotive force harmonics, unknown nonlinear dynamics can be introduced into the formula (1), and the formula (1) is corrected to obtain:
wherein, g1Gain of control quantity, g, for α axis current2Gain of control quantity, g, for β axis current31(theta) and g32(θ) is a control quantity gain of the electrical angular velocity; u. ofα、uβController outputs of α axis current and β axis current respectively, gamma1、γ2、γ3Is a nonlinear dynamics.
Preferably, three RBF neural networks are designed as the pair of recognition models gamma1、γ2、γ3Non-linearityAnd dynamically performing identification.
Preferably, the recognition model is described as:
wherein, W1、S1(iα,iβθ, ω) are the weights of the RBF neural network identifying the α axis current and the corresponding Gaussian function, W2、S2(iα,iβθ, ω) are the weights of the RBF neural network identifying the β axis current and the corresponding Gaussian function, W3、S3(iα,iβθ, ω) is the weight of the RBF neural network identifying the electrical angular velocity and the corresponding gaussian function; w1 T、W2 T、W3 TAre respectively W1、W2、W3The transposing of (1).
Preferably, a dynamic controller corresponding to the identification model is constructed, and the dynamic controller is described as:
wherein k is1、k2、k3、k4Are all constants greater than zero;
is the desired value for the α axis current,
is composed of
The differential amount of (a);is the desired value for the β axis current,is composed of
The differential amount of (a); theta*As the desired value of the electrical angle,
is theta*The differential amount of (a); omega*As the desired value of the electrical angular velocity,
is omega*The differential amount of (a).
Preferably, the local accurate approximation of the nonlinear dynamics is realized through a plurality of times of online training, and the online training is described as follows:
wherein the content of the first and second substances,
andα axis current estimated value, β axis current estimated value and electrical angular velocity estimated value respectively, wherein a, b and c are constants larger than zero;
are respectively as
And
when errors between α axis current estimated value, β axis current estimated value and electric angular velocity estimated value and actual system state variables are large, updating weight W of RBF neural network1、W2And W3And continuously performing online training.
Compared with the prior art, the invention has the following advantages and effects:
(1) the invention applies definite learning and is based on a mode control theory, and solves the problems that the identification of a permanent magnet alternating current servo system is difficult and a PI controller commonly used by the permanent magnet synchronous servo system is difficult to completely deal with under dynamic environments such as permanent magnet synchronous motor model parameter change, torque fluctuation, torsional vibration and the like. By the technical scheme, local accurate identification of the closed-loop dynamic state of the permanent magnet alternating current servo system under uncertain nonlinear dynamic influence can be accurately and effectively realized; furthermore, the rapid switching between different dynamic mode-based controllers is realized through rapid dynamic mode identification, so that the rotation speed and torque fluctuation of a servo system are reduced, and the rapid response performance of the system is improved.
(2) Based on a definite learning theory, the invention uses a Radial Basis Function (RBF) neural network to accurately identify and compensate uncertain nonlinear dynamics of the permanent magnet alternating current servo system under different operating conditions. On the basis, the knowledge acquired by identification is stored in a constant neural network form, so that a pattern library is formed, and the knowledge can be conveniently and subsequently utilized to construct a knowledge-based controller. Compared with the traditional permanent magnet alternating current servo control system, the nonlinear dynamic identification convergence condition is easy to meet (part of continuous excitation conditions), the identification precision is high, and the nonlinear dynamic of the servo system can be well fitted.
(3) The invention modifies the traditional permanent magnet alternating current servo control system, adds a knowledge acquisition and storage module and a mode-based control module, dynamically identifies the permanent magnet alternating current servo system based on a definite learning theory, realizes the local accurate neural network identification of the closed loop dynamics of the permanent magnet alternating current servo system under the action of nonlinear factors, and constructs an experience controller based on knowledge to realize high-performance control; and in combination with a dynamic mode identification method, a mode-based control module is used for quickly judging the operation mode of the system and switching the corresponding controller, so that the intelligent control of the alternating current servo system is realized.
Drawings
FIG. 1 is a diagram of a permanent magnet AC servo intelligent control system based on deterministic learning and pattern control according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a pattern library obtained based on a deterministic learning online training in accordance with an embodiment of the present invention;
FIG. 3 is a schematic diagram of a mode-based control module control process according to one embodiment of the present invention;
FIG. 4 is a diagram illustrating dynamic selection of a corresponding empirical controller, in accordance with an embodiment of the present invention.
Detailed Description
The present invention is further illustrated by the following figures and examples, which include, but are not limited to, the following examples.
In the actual operation process of the permanent magnet alternating current servo system, the influence of nonlinear factors such as parameter change, fractional order characteristics, cogging torque, current harmonic waves, back electromotive force harmonic waves and the like exists. The existence of the nonlinear factors causes the problems of the fluctuation of the rotating speed and the torque of the permanent magnet synchronous motor, thereby bringing adverse effects to the stable and efficient operation of the system. Due to the fact that the nonlinear dynamics characteristics of the permanent magnet alternating current servo system are challenging to accurately model, the design initiatives of the intelligent controller are not met. According to the theory of deterministic learning, the permanent magnet AC servo intelligent control system with dynamic identification and mode based shown in FIG. 1 can be used.
The embodiment adopts a Radial Basis Function (RBF) neural network to identify the nonlinear dynamics of the permanent magnet alternating current servo system in different modes, stores the knowledge learned by identification in a constant neural network form, and then constructs a series of mode-based empirical controllers by using the learned knowledge so as to call when the same or similar control tasks are executed to realize high-performance control. When the operation mode of the permanent magnet alternating current servo system changes, in order to quickly identify the change of the mode, an empirical controller in a normal mode is adopted to control all the modes of the permanent magnet alternating current servo system, and a series of dynamic estimators are constructed to evaluate the error of the state variable of the system. When the trained operation mode occurs again, a dynamic mode recognition method is adopted, the operation mode of the current permanent magnet alternating current servo system is judged according to the principle that the error is minimum (and is smaller than a preset value), and an empirical controller corresponding to the mode is selected to control the permanent magnet alternating current servo system so as to realize efficient and stable operation of the permanent magnet alternating current servo system. If the recognition error is larger than the preset value, the operation mode is judged to be a new operation mode which is not trained, at the moment, the operation mode is determined to be learned, and the learning result is added into a previously trained mode library so as to continuously perfect the training mode library.
In one embodiment, as shown in fig. 1, the permanent magnet alternating current servo intelligent control system based on determination learning and mode control comprises: the device comprises a signal detection module, a knowledge acquisition and storage module and a mode-based control module. The knowledge acquisition and storage module may be subdivided into: determining a learning module, a dynamic controller module, an RBF network approach unknown dynamic module, an online training module and a mode library module; while the mode-based control module can also be subdivided into: the dynamic estimator module, the switching control module and the experience controller module.
The input end of the signal detection module is connected with the servo motor and is used for detecting information (including current, rotating speed, position and the like) of the servo motor and outputting the information to the learning determination module and the dynamic estimator module respectively;
the confirming learning module receives the instruction information of the on-line training module or the dynamic estimator module, judges whether confirming learning is needed or not, and transmits the information acquired by the signal detection module to the dynamic controller module if the confirming learning is needed;
the dynamic controller module receives the information for determining the learning module, adjusts the parameters of the controller according to the dynamic performance requirement, and then transmits the corresponding parameters and the processing result to the RBF network approaching unknown dynamic module;
the RBF network approaching unknown dynamic module adopts an RBF neural network to carry out local accurate identification on the nonlinear dynamic of the permanent magnet alternating current servo system under the current control task, and transmits the parameters and the processing result of the network to the online training module;
the online training module calculates the error between the estimated state and the actual system state according to the received information, judges whether to perform the definite learning again, and transmits the parameters of the dynamic controller and the identification result of the RBF network to the mode base module if the expected error is small enough after iteration for multiple times;
the model base module is internally provided with a plurality of registers, the parameters of the dynamic controller and the identification result of the RBF network are stored in a one-to-one correspondence mode, and the stored information is provided for the dynamic estimator and the empirical controller to use.
A mode-based control module comprising: the dynamic estimator module, the switching control module and the experience controller module.
The following further describes embodiments of the present invention.
(1) Knowledge acquisition and storage process:
neglecting the influence of factors such as nonlinearity, the mathematical model of the permanent magnet synchronous motor in the two-phase static coordinate system can be described as follows:
wherein iαIs the current of the α shaft, and the current of the shaft,
is iαThe differential amount of (a); i.e. iβIs the current of the β shaft, and the current of the shaft,
is iβThe differential amount of (a); theta is an electrical angle of the steel sheet,
is the differential of theta; omega is the electrical angular velocity of the object,
a differential amount of ω; u. ofαFor α axis control, uββ axle control quantity, R motor resistance, Lαα axes motor inductanceββ shaft motor inductance, J motor moment of inertia, p motor pole pair number, psifIs a motor magnetic linkage; t isLIs a loadTorque; b is a friction coefficient; t is electromagnetic torque, and the expression is as follows:
in practice, unknown nonlinear dynamics are introduced into the formula (1) by considering the influence of nonlinear factors such as parameter variation, fractional order characteristics, cogging torque, current harmonics and back electromotive force harmonics. Therefore, by correcting equation (1), it is possible to obtain:
wherein, g1Gain of control quantity, g, for α axis current2Gain of control quantity, g, for β axis current31(theta) and g32(θ) is a control quantity gain of the electrical angular velocity; u. ofα、uβController outputs of α axis current and β axis current respectively, gamma1、γ2、γ3Is a nonlinear dynamics. For example: γ 1, γ 2, γ 3 may be nonlinear dynamics of nonlinear factors such as parameter variations, fractional order characteristics, cogging torque, current harmonics, and back-emf harmonics.
In the embodiment, a mathematical model of a permanent magnet synchronous motor of a controlled object of a permanent magnet alternating current servo system is decomposed into two parts, wherein one part is a known time-invariant nominal model, and the other part is an unknown nonlinear dynamic model. The decomposition method can fully utilize known model information, and three RBF neural networks can be respectively designed and identified by unknown nonlinear dynamics. Therefore, the training amount of the neural network can be effectively reduced, and the training speed is further improved. The RBF network comprises the following components: weights and radial basis functions. In this embodiment, the radial basis function is a gaussian function. Let the identification model equation (4) be μ.
Wherein, W1、S1(iα,iβθ, ω) are the weights of the RBF neural network identifying the α axis current and the corresponding Gaussian function, W2、S2(iα,iβθ, ω) are the weights of the RBF neural network identifying the β axis current and the corresponding Gaussian function, W3、S3(iα,iβθ, ω) is the weight of the RBF neural network identifying the electrical angular velocity and the corresponding gaussian function; w1 T、W2 T、W3 TAre respectively W1、W2、W3The transposing of (1).
Further, a dynamic controller identifying the model μ response may be constructed. Let the corresponding dynamic controller equation (5) be denoted as u:
wherein k is1、k2、k3、k4Are all constants greater than zero;
is the desired value for the α axis current,
is composed of
The differential amount of (a);
is the desired value for the β axis current,
is composed of
The differential amount of (a); theta*As the desired value of the electrical angle,
is theta*The differential amount of (a); omega*As the desired value of the electrical angular velocity,
is omega*The differential amount of (a).
And then, local accurate approximation to unknown dynamics can be realized through multiple times of online training. The on-line training can be represented by the following equation:
wherein the content of the first and second substances,
and
the estimated system state variables are α axis current estimated value, β axis current estimated value and electrical angular velocity estimated value respectively;
are respectively as
And
when the error between the three estimated values (α axis current estimated value, β axis current estimated value and electrical angular velocity estimated value) and the actual system state variable is large, the online training is continuously carried out, and the weight W of the RBF neural network is updated1、W2And W3And the updating law of the weight is as follows:
wherein, gamma is1、Γ2、Γ3And σ1、σ2、σ3All are constants greater than zero；The differential quantities of the weights are respectively.
And when the error between the estimated value and the actual system state variable is small enough, exiting the on-line training, and storing the identification model mu of the current running state and the corresponding controller u into a mode library, so that a subsequent mode-based control module can be conveniently and quickly called.
The above steps are a complete knowledge acquisition and storage process. However nonlinear dynamics of servo motors1、γ2、γ3Due to the influence of changes of self parameters and the like, the dynamic performance of the engine under different running states can be different. The above process may be repeated for multiple times to obtain the identification models and corresponding controllers in different operation modes, and the identification models and corresponding controllers are recorded in a pattern library, which is shown in fig. 2.
(2) Mode-based control procedure:
under the action of the same controller u, the dynamic estimator calculates the estimated values of the system state variables under different identification models based on the formula (6) and calling the identification models obtained in the pattern library, and then the estimated values are differenced with the state variables of the actual alternating current servo system, so that an estimated error value is obtained. Obviously, when the identification model is matched with the AC servo system in the current operation mode, the state estimation error is minimal. Therefore, by evaluating the state error, the identification model μ in the operation mode can be quickly found, and the corresponding controller u is obtained from the mode library to perform switching control on the system. The above process is illustrated in fig. 3.
In particular, for the mode-based control process shown in FIG. 3, the present embodiment describes three different operation modes of the AC servo system, and the lower identification model (μ) of the three different operation modes is recorded in the mode library0、μ1、μ2). As shown in FIG. 4, assume the initial operating mode is
At this time, control is adoptedDevice u0The servo motor is controlled by the experience controller to achieve the expected control effect; when the motor is in the operating mode
Become into
In an empirical controller u0Under the action of the dynamic estimator, the state variable of the actual AC servo system is necessarily changed, and the dynamic estimator uses the same controller u0For all the recognition models (μ) in the pattern library0、μ1、μ2) Estimating the system state variable, comparing the estimation result with the actual system state variable acquired by the signal detection module, and judging the operation mode of the current permanent magnet alternating current servo system according to the principle that the error is minimum (and is smaller than a preset value); if all the errors are larger than the preset value after comparison, the information of the current operation mode is not stored in the mode base, and at the moment, the dynamic estimator module sends an instruction to start learning the dynamic state and updates the mode base; if a certain error is smaller than the predetermined value after the comparison, the estimation result (mu in the embodiment) is considered to be the estimation result2) I.e., the current operating mode, the number of the current operating mode (e.g., mode 0, 1, …, n in fig. 2) is then transmitted to the switching control module, and the corresponding controller u is extracted from the mode library2And the experience controller is updated to control the alternating current servo system, so that the high-performance control of the permanent magnet alternating current servo system is realized.
In conclusion, the permanent magnet alternating current servo system dynamic identification and intelligent control system based on the determined learning and mode control can autonomously and accurately identify and control the dynamic state of the servo motor in different operation modes, and further store information into a mode library; when the trained operation mode occurs again, the operation mode can be quickly identified and switched into the experience controller corresponding to the mode, and the intelligent control of the permanent magnet alternating current servo system is realized.
The invention can be advantageously implemented according to the above-described embodiments. It should be noted that, based on the above structural design, in order to solve the same technical problems, even if some insubstantial modifications or colorings are made on the present invention, the adopted technical solution is still the same as the present invention, and therefore, the technical solution should be within the protection scope of the present invention.
Claims (10)
1. Permanent magnetism exchanges servo intelligent control system based on confirm study and mode control, its characterized in that includes: signal detection module, knowledge acquisition and storage module, mode-based control module, wherein:
the input end of the signal detection module is connected with the servo motor and is used for detecting the information of the servo motor and outputting the information which is respectively transmitted to the knowledge acquisition and storage module and the mode-based control module;
the other input end of the knowledge acquisition and storage module is from the mode-based control module to judge whether to enter the learning-determining module and identify the nonlinear dynamics, and the output is transmitted to the mode-based control module;
and the other output end of the mode-based control module is transmitted to the servo motor to drive the servo motor to work.
2. The permanent magnet alternating current servo intelligent control system according to claim 1, wherein the knowledge acquisition and storage module is used for realizing local accurate neural network identification of nonlinear dynamics of the permanent magnet alternating current servo system, constructing a corresponding experience controller, and storing the experience controller into a pattern library so as to facilitate rapid calling of the pattern-based control module.
3. The permanent magnet alternating current servo intelligent control system according to claim 2, wherein the knowledge acquisition and storage module comprises: determining a learning module, a dynamic controller module, an RBF network approach unknown dynamic module, an online training module and a mode library module, wherein each module adopts a cascade form; determining the processing results of the inputs of the learning module from the on-line training module, the dynamic estimator of the mode-based control module and the signal detection module respectively; the output of the mode base module is communicated to a dynamic estimator module and an empirical controller module, respectively, of the mode based control module.
4. The permanent magnet alternating current servo intelligent control system according to claim 1, wherein the mode-based control module is used for judging the operation mode of the permanent magnet alternating current servo system and rapidly calling a corresponding experience controller for switching control according to information in a mode library so as to realize intelligent control of the servo motor.
5. The permanent magnet ac servo intelligent control system of claim 1, wherein the pattern based control module comprises: a dynamic estimator module, a switching control module, and an empirical controller module, wherein: the input of the dynamic estimator module is respectively from the processing results of the mode base and the signal detection module, and the output is respectively transmitted to the determination learning module and the switching control module; the output end of the switching control module is connected with the input end of the experience controller module; the other input of the experience controller module is from a mode library, and the output end of the experience controller module is transmitted to the servo motor to control the servo motor to realize high-performance operation.
6. The permanent magnet alternating current servo intelligent control system according to claim 1, wherein a mathematical model of the servo motor in a two-phase static coordinate system is described by neglecting the influence of factors such as nonlinearity:
wherein iαIs the current of the α shaft, and the current of the shaft,
is iαThe differential amount of (a); i.e. iβIs the current of the β shaft, and the current of the shaft,is iβThe differential amount of (a); theta is an electrical angle of the steel sheet,
is the differential of theta; omega is the electrical angular velocity of the object,
a differential amount of ω; u. ofαFor α axis control, uββ axle control quantity, R motor resistance, Lαα axes motor inductanceββ shaft motor inductance, J motor moment of inertia, p motor pole pair number, psifIs a motor magnetic linkage; t isLIs the load torque; b is a friction coefficient; t is electromagnetic torque, and the expression is as follows:
the influence of non-linear factors is considered in practice, including: parameter variation, fractional order characteristics, cogging torque, current harmonics and back electromotive force harmonics, unknown nonlinear dynamics can be introduced into the formula (1), and the formula (1) is corrected to obtain:
wherein, g1Gain of control quantity, g, for α axis current2Gain of control quantity, g, for β axis current31(theta) and g32(θ) is a control quantity gain of the electrical angular velocity; u. ofα、uβController outputs of α axis current and β axis current respectively, gamma1、γ2、γ3Is a nonlinear dynamics.
7. The permanent magnet alternating current servo intelligent control system according to claim 6, wherein three RBF neural networks are designed as an identification model pair gamma1、γ2、γ3And carrying out nonlinear dynamic identification.
8. The permanent magnet alternating current servo intelligent control system according to claim 7, wherein the identification model is described as:
wherein, W1、S1(iα,iβθ, ω) are the weights of the RBF neural network identifying the α axis current and the corresponding Gaussian function, W2、S2(iα,iβθ, ω) are the weights of the RBF neural network identifying the β axis current and the corresponding Gaussian function, W3、S3(iα,iβθ, ω) is the weight of the RBF neural network identifying the electrical angular velocity and the corresponding gaussian function; w1 T、W2 T、W3 TAre respectively W1、W2、W3The transposing of (1).
9. The permanent magnet alternating current servo intelligent control system according to claim 8, wherein a dynamic controller corresponding to the identification model is constructed, and the dynamic controller is described as:
wherein k is1、k2、k3、k4Are all constants greater than zero;
is the desired value for the α axis current,
is composed of
The differential amount of (a);
is the desired value for the β axis current,
is composed of
The differential amount of (a); theta*As the desired value of the electrical angle,
is theta*The differential amount of (a); omega*As the desired value of the electrical angular velocity,
is omega*The differential amount of (a).
10. The permanent magnet alternating current servo intelligent control system according to claim 9, wherein local accurate approximation to nonlinear dynamics is achieved through a plurality of times of online training, and the online training is described as:
wherein the content of the first and second substances,
and
α axis current estimated value, β axis current estimated value and electrical angular velocity estimated value respectively, wherein a, b and c are constants larger than zero;
are respectively as
And
when errors between α axis current estimated value, β axis current estimated value and electric angular velocity estimated value and actual system state variables are large, updating weight W of RBF neural network1、W2And W3And continuously performing online training.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911183121.3A CN110829934A (en) | 2019-11-27 | 2019-11-27 | Permanent magnet alternating current servo intelligent control system based on definite learning and mode control |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911183121.3A CN110829934A (en) | 2019-11-27 | 2019-11-27 | Permanent magnet alternating current servo intelligent control system based on definite learning and mode control |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110829934A true CN110829934A (en) | 2020-02-21 |
Family
ID=69560075
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911183121.3A Pending CN110829934A (en) | 2019-11-27 | 2019-11-27 | Permanent magnet alternating current servo intelligent control system based on definite learning and mode control |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110829934A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111460732A (en) * | 2020-03-31 | 2020-07-28 | 深圳大学 | Construction method of nonlinear model of planar motor |
| CN111600519A (en) * | 2020-05-26 | 2020-08-28 | 佛山金华信智能科技有限公司 | Servo motor control method, servo motor control device, electronic equipment and storage medium |
| CN113110021A (en) * | 2021-03-17 | 2021-07-13 | 华南理工大学 | Method for identifying servo system and designing controller |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102497156A (en) * | 2011-12-27 | 2012-06-13 | 东南大学 | Neural-network self-correcting control method of permanent magnet synchronous motor speed loop |
| CN106357186A (en) * | 2016-09-05 | 2017-01-25 | 桂林电子科技大学 | Method and system for controlling constant torque of switched reluctance motor by use of composite control current |
| CN106788028A (en) * | 2016-12-20 | 2017-05-31 | 江苏大学 | Bearing-free permanent magnet synchronous motor intensified learning controller and its building method |
| CN109227543A (en) * | 2018-10-19 | 2019-01-18 | 华南理工大学 | A kind of limited pattern-based intelligent control method of flexible joint robot in position |
| CN109946975A (en) * | 2019-04-12 | 2019-06-28 | 北京理工大学 | A kind of intensified learning optimal track control method of unknown servo-system |
-
2019
- 2019-11-27 CN CN201911183121.3A patent/CN110829934A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102497156A (en) * | 2011-12-27 | 2012-06-13 | 东南大学 | Neural-network self-correcting control method of permanent magnet synchronous motor speed loop |
| CN106357186A (en) * | 2016-09-05 | 2017-01-25 | 桂林电子科技大学 | Method and system for controlling constant torque of switched reluctance motor by use of composite control current |
| CN106788028A (en) * | 2016-12-20 | 2017-05-31 | 江苏大学 | Bearing-free permanent magnet synchronous motor intensified learning controller and its building method |
| CN109227543A (en) * | 2018-10-19 | 2019-01-18 | 华南理工大学 | A kind of limited pattern-based intelligent control method of flexible joint robot in position |
| CN109946975A (en) * | 2019-04-12 | 2019-06-28 | 北京理工大学 | A kind of intensified learning optimal track control method of unknown servo-system |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111460732A (en) * | 2020-03-31 | 2020-07-28 | 深圳大学 | Construction method of nonlinear model of planar motor |
| CN111460732B (en) * | 2020-03-31 | 2023-05-16 | 深圳大学 | Construction method of nonlinear model of planar motor |
| CN111600519A (en) * | 2020-05-26 | 2020-08-28 | 佛山金华信智能科技有限公司 | Servo motor control method, servo motor control device, electronic equipment and storage medium |
| CN111600519B (en) * | 2020-05-26 | 2024-02-23 | 佛山金华信智能科技有限公司 | Servo motor control method and device, electronic equipment and storage medium |
| CN113110021A (en) * | 2021-03-17 | 2021-07-13 | 华南理工大学 | Method for identifying servo system and designing controller |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Rigatos | Adaptive fuzzy control of DC motors using state and output feedback | |
| CN110829934A (en) | Permanent magnet alternating current servo intelligent control system based on definite learning and mode control | |
| Liu et al. | High precision robust control for periodic tasks of linear motor via B-spline wavelet neural network observer | |
| Zhang et al. | Adaptive robust control for servo mechanisms with partially unknown states via dynamic surface control approach | |
| Nguyen et al. | Recurrent neural network-based robust adaptive model predictive speed control for PMSM with parameter mismatch | |
| Dang et al. | Neuro-fuzzy control of interior permanent magnet synchronous motors: Stability analysis and implementation | |
| Luo et al. | Anti-oscillation and chaos control of the fractional-order brushless DC motor system via adaptive echo state networks | |
| Szczepanski et al. | PMSM drive with adaptive state feedback speed controller | |
| Liu et al. | Iterative learning based neural network sliding mode control for repetitive tasks: With application to a PMLSM with uncertainties and external disturbances | |
| Yuan et al. | Adaptive jerk control and modified parameter estimation for PMLSM servo system with disturbance attenuation ability | |
| Cortés-Romero et al. | Algebraic parameter identification for induction motors | |
| Yin et al. | A speed estimation method for induction motors based on strong tracking extended Kalman filter | |
| Chaouch et al. | Optimized torque control via backstepping using genetic algorithm of induction motor | |
| Prasad et al. | Fuzzy sliding mode control of a Permanent Magnet Synchronous Motor with two different fuzzy membership functions | |
| Lin et al. | Robust recurrent-neural-network sliding-mode control for the XY table of a CNC machine | |
| Salem et al. | Model predictive control for deadbeat performance of induction motor drives | |
| Li et al. | Adaptive backstepping sliding mode control for the oscillation displacement system of continuous casting mold with mismatched disturbances | |
| Perdukova et al. | A model-based fuzzy control of an induction motor | |
| CN114977928A (en) | Parameter self-tuning method for speed ring and position ring of permanent magnet synchronous servo system | |
| Qin et al. | New control strategy for PMSM driven bucket wheel reclaimers using GA-RBF neural network and sliding mode control | |
| Wang et al. | A high performance permanent magnet synchronous motor servo system using predictive functional control and Kalman filter | |
| Vo et al. | Pulse-width modulation direct torque control induction motor drive with Kalman filter | |
| Kaminski et al. | Neural speed controller based on two state variables applied for a drive with elastic connection | |
| Aguilar Mejía et al. | Adaptive Speed Controller for a Permanent Magnet Synchronous Motor | |
| Ltifi et al. | A sliding mode integrated control technique and the PI regulator for induction machine |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20211105 Address after: 510635 No. 315, 3 / F, building 8, compound 663, Tianhe North Road, Tianhe District, Guangzhou, Guangdong Applicant after: Guangzhou Hongwei Technology Co.,Ltd. Address before: 510640 No. five, 381 mountain road, Guangzhou, Guangdong, Tianhe District Applicant before: SOUTH CHINA University OF TECHNOLOGY |
|
| TA01 | Transfer of patent application right |