CN203896241U - Dual linear motor contour compensation device based on fuzzy RBF network integral sliding-mode - Google Patents

Abstract

The utility model discloses a dual linear motor contour compensation device based on a fuzzy RBF network integral sliding-mode by aiming at a dual linear motor numerical control feeding system capable of moving vertically toward each other along an axial direction. A real-time contour error can be used as input of a controller, and the contour error can approach zero in a limited time by the strong self-learning ability, and then the contour processing precision can be improved. The device comprises a rectification and voltage stabilization unit, an IPM inverter, a digital signal processor DSP, a Hall sensor, a grating ruler, a current detection unit, a position speed detection unit, an optical coupler isolation circuit, a driving protection circuit, and a fault detection and protection circuit. The DSP comprises a quadrature encoding pulse QEP circuit of an event manager EVA, an ADC module, a PWM unit, a Flash memory unit, a program memory, a timer, and a PDPINT pin. A position signal setter, a linear motor position ring, a PI controller of a speed ring, a PI controller of a current ring, a contour error calculator, a fuzzy RBF network integral sliding-mode contour compensator, and a driver device are disposed in the DSP. The dual linear motor contour compensation device is advantageous in that robustness performance is good, the compensation device can be used for the contour processing tasks of any tracks, and the high-precision contour control can be realized.

Description

Dual linear motor outline compensation device based on fuzzy RBF neural network Integral Sliding Mode

Technical field

The invention belongs to automation control and numerical control field, particularly a kind of dual linear motor outline compensation device and control method based on based on fuzzy RBF neural network Integral Sliding Mode.

Background technology

Along with scientific and technological development, under the strong support of automatic control system, servo system development rapidly, is almost all applied in every field to some extent.This also makes the today of accelerating in process of industrialization, and the required precision of production equipment is also uprised more thereupon, can say that it is weighing a national overall national strength and scientific and technological level to a great extent for it.Therefore Digit Control Machine Tool also becomes the processing machine tool of contemporary machinery manufacturing industry.It provides required equipment can to each department of country, it is the commanding elevation of industrial production contention technical advantage, also be the key component of using new high-tech industry transformation and promoting traditional industries, the competition that realizes enterprise's high efficiency production and have is by force significant, impels industrial production fast-developing.Core as advanced manufacturing technology---Numeric Control Technology, the plateform system directly driving with dual linear motor is the most representative, and its profile is followed the tracks of control ability and is had a very important role to improving system machining accuracy and performance.

Direct drive mode has saved the transmission link in the middle of all, although brought fast response time, efficiency advantages of higher, has also increased the difficulty in electrical control.In order to improve the contour accuracy of dual linear motor feed control system, a lot of methods lay particular emphasis on the design of linear electric motors single shaft being carried out to controller, by reducing tracking error and then reducing profile errors, but be only to can't resolve dynamically not mating of existing between twin shaft and the problem such as track following profile errors model complexity arbitrarily like this.

Summary of the invention

For the deficiency existing in existing control technology, the invention provides a kind of dual linear motor outline compensation device based on fuzzy RBF neural network Integral Sliding Mode, by Based Intelligent Control and modern control program controller designed in conjunction and by DSP, realize and controlling, thereby reach high-precision processing request.

A dual linear motor outline compensation device based on fuzzy RBF neural network Integral Sliding Mode, this device comprises rectification stable pressuring unit, IPM inverter, digital signal processor DSP, Hall element, grating scale, current detecting unit, position and speed detecting unit, optical coupling isolation circuit, Drive Protecting Circuit and fault detect and protective circuit; Rectification stable pressuring unit connects IPM inverter, and IPM inverter is connected to dual linear motor platform by a Drive Protecting Circuit; Hall element is connected between IPM inverter and Drive Protecting Circuit on the one hand, and Hall element is connected to digital signal processor DSP by current detecting unit on the other hand; Position and speed detecting unit is connected to dual linear motor platform by grating scale on the one hand, is connected on the other hand digital signal processor DSP; IPM inverter is connected to optical coupling isolation circuit by second Drive Protecting Circuit, and optical coupling isolation circuit is connected to digital signal processor DSP on the one hand, is connected between rectification stable pressuring unit and IPM inverter on the other hand by fault detect and protective circuit.

Digital signal processor DSP comprises quadrature coding pulse circuit QEP, ADC module, PWM unit, Flash memory cell, program storage, timer and the PDPINT pin of task manager EVA; QEP link position speed detection unit, ADC module connects current detecting unit, and PWM unit and PDPINT pin are all connected to optical coupling isolation circuit.

The dual linear motor outline compensation device of employing based on fuzzy RBF neural network Integral Sliding Mode carries out the method for contour machining, comprises that step is as follows:

Step 1: offer the corresponding position signalling instruction of dual linear motor diaxon by the position signalling control point adjustment in TMS320F2812 digital signal processor, as the Position Control amount input of diaxon;

Step 2: by electric current, position and speed detection unit, position, speed and current signal are sampled and relatively acquisition position error signal given by diaxon and actual output;

Step 3: the position error signal that regulates single shaft by PI positioner, employing profile errors calculator calculates the profile errors amount of any track and by compensating control based on fuzzy RBF neural network Integral Sliding Mode outline compensation device, the output of then both being controlled is as the input of position device;

Step 4: Negotiation speed detecting signal unit is determined the speed of linear electric motors;

Step 5: by grating scale sample rate signal, in TMS320F2812 digital processing unit relatively after, execution speed ring PI controller;

Step 6: by built-in current sampling device carry out current sample and in DSP relatively after, carry out electric current loop PI controller;

Step 7: current value is carried out to 3/2 conversion;

Step 8: utilize the q axle of rotation to calculate torque;

Step 9: output calculating torque is carried out to 2/3 conversion;

Step 10: as carrier wave and triangular modulation, allow digital signal processor DSP produce six road pwm pulse signals to the current value after conversion, drive the diaxon of dual linear motor to carry out given profile traces working motion according to the size of current-order;

Regulator rectifier circuit converts direct current to three-phase alternating current and powers to IPM inversion unit, IPM inversion unit produces Liu road pwm pulse signal according to DSP conducting and the shutoff of six IGBT switch elements in IPM inversion unit is controlled, and drives linear electric motors operation.

Employing profile errors calculator described in step 3 calculates the profile errors amount of any track and compensates control by built-in profile errors compensation arrangement, and the output of then both being controlled, as the input of position device, comprises that step is as follows:

Step 3-1: calculate profile errors value and the profile errors value of track can be by ε=-e arbitrarily according to real-time position, speed and current signal and position error signal _xsin φ+e _ycos φ represents and meets x axle tracking error is e _x; Y axle tracking error is e _y; On x axle and given path, the angle of line is φ a bit and between desired locations.

Step 3-2: when profile errors exists, the input of the dual linear motor outline compensation device based on fuzzy RBF neural network Integral Sliding Mode is the sliding formwork switching function s (t) relevant to profile errors, can be determined by following formula

$s\left(t\right)=(\frac{d}{d_t}+\mathrm{\λ})\underset{0}{\overset{t}{\∫}}\mathrm{\ϵ}\left(t\right)\mathrm{dt}$

Wherein: λ is greater than zero constant; The object of controlling is to make it is profile errors ε → 0

Step 3-3: adopt the fuzzy RBF neural network control with self-learning capability to approach study and fuzzy output to target function, described target function is

Learning algorithm is chosen as

$w_{\mathrm{ko}}^4=-\mathrm{\η}\frac{\∂E}{\∂w}=-\mathrm{\η}\frac{\∂S\stackrel{\·}{S}}{{\∂U}_{\mathrm{FNN}}}\frac{{\∂U}_{\mathrm{FNN}}}{\∂w_{\mathrm{ko}}^4}$

The error term of definition rule layer and membership function layer is

${\mathrm{\δ}}_k^3\stackrel{\mathrm{\Δ}}{=}-\mathrm{\η}\frac{\∂S\stackrel{\·}{S}}{{\∂U}_{\mathrm{FNN}}}\frac{\∂U_{\mathrm{FNN}}{\∂y}_o^4}{\∂y_o^4{\∂y}_k^3}\frac{{\∂y}_k^3}{\∂{\mathrm{net}}_k^3}=-\mathrm{\η}\left|\frac{K_f}{M}\right|{\mathrm{Sw}}_{\mathrm{ko}}^4$

${\mathrm{\δ}}_j^2=-\mathrm{\η}\frac{\∂S\stackrel{\·}{S}}{{\∂U}_{\mathrm{FNN}}}\frac{\∂U_{\mathrm{FNN}}{\∂y}_o^4}{{\∂y}_o^4{\∂y}_k^3}\frac{{\∂y}_k^3}{\∂{\mathrm{net}}_k^3}\frac{{\∂\mathrm{net}}_k^3}{{\∂y}_j^2}\frac{{\∂y}_j^2}{\∂{\mathrm{net}}_j^2}=\underset{k}{\mathrm{\Σ}}{\mathrm{\δ}}_k^3{y}_k^3$

Adjust in the following way membership function parameter

$\mathrm{\Δ}{C}_{\mathrm{ij}}=-\mathrm{\η}\frac{\∂E}{\∂c_{\mathrm{ij}}}=-\mathrm{\η}\frac{\∂E}{\∂\mathrm{ne}{t}_j^2}\frac{\∂{\mathrm{net}}_j^2}{\∂c_{\mathrm{ij}}}=-\mathrm{\η}{\mathrm{\δ}}_j^2\frac{2(x_i-c_{\mathrm{ij}})}{{\mathrm{\σ}}_{\mathrm{ij}}^2}$

$\mathrm{\Δ}{\mathrm{\σ}}_{\mathrm{ij}}=-\mathrm{\η}\frac{\∂E}{\∂{\mathrm{\σ}}_{\mathrm{ij}}}=-\mathrm{\η}\frac{\∂E}{\∂{\mathrm{net}}_j^2}\frac{\∂{\mathrm{net}}_j^2}{\∂{\mathrm{\σ}}_{\mathrm{ij}}}=-\mathrm{\η}{\mathrm{\δ}}_j^2\frac{2(x_i-c_{\mathrm{ij}})}{{\mathrm{\σ}}_{\mathrm{ij}}^3}$

Wherein: the average of Gaussian membership function and standard deviation are respectively C _ijand σ _ij; Neural net connection weight matrix is

Fuzzy output valve is

$y_i^1=f_i^1{x}_i^1,i=\mathrm{1,2}$

$y_j^2=f_j^2\left({\mathrm{net}}_j^2\right)=\exp (-\frac{(x_i^2-C_{\mathrm{ij}})}{{\left({\mathrm{\σ}}_{\mathrm{ij}}\right)}^2}),j=\mathrm{1,2}......n$

$y_k^3={\mathrm{net}}_k^3=\underset{j=1}{\overset{N}{\mathrm{\Π}}}{y}_j^2$

Wherein: being input as of outline compensation device output for fuzzy control;

Step 3-4: the connection weight matrix by fuzzy output and neural net multiplies each other, can obtain the position compensation signal of profile errors compensation arrangement

$U_{\mathrm{FNN}}=y_o^4=W_{\mathrm{ko}}^4\·y_k^3=\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{w}_{\mathrm{ko}}^4\·y_k^3,o=1$

Advantage of the present invention: the dual linear motor numerical control feeding system for axially mutually moving both vertically, has proposed fuzzy RBF neural network Integral Sliding Mode outline compensation apparatus and method.This control method can be followed the tracks of operation to track arbitrarily, and modern sliding formwork is controlled to combine with Based Intelligent Control and effectively eliminated the buffeting problem of sliding formwork self existence.Do not lose under the prerequisite of Sliding Mode Robust, the profile errors of dual linear motor is directly controlled, effectively improved the contour accuracy of whole control system.The present invention obtains whole profile errors the tracking error of each single shaft according to profile errors computational methods, and then the design by controller makes the margin of error level off to gradually zero.

Accompanying drawing explanation

Fig. 1 is the designed outline compensation control device of the present invention and the system block diagram of method

Fig. 2 is the designed any track profile error model figure of the present invention

Fig. 3 is the structural representation of the designed fuzzy RBF neural network Integral Sliding Mode of the present invention outline compensation method

Fig. 4 is the overall flow figure of the designed device and method of the present invention

Fig. 5 is the designed outline compensation device of the present invention hardware chart

Fig. 5-(1) is current detection circuit figure

Fig. 5-(2) are position detecting circuit figure

Fig. 5-(3) are for controlling power circuit diagram

Fig. 5-(4) are drive circuit figure

Fig. 5-(5) are IPM isolation inversion and Drive Protecting Circuit figure.

Embodiment:

Below in conjunction with accompanying drawing, technical scheme of the present invention is specifically described:

Dual linear motor outline compensation apparatus and method based on fuzzy RBF neural network Integral Sliding Mode, this device comprises rectification stable pressuring unit, IPM inverter, digital signal processor DSP, Hall element, grating scale, current detecting unit, position and speed detecting unit, optical coupling isolation circuit, Drive Protecting Circuit and fault detect and protective circuit.Rectification stable pressuring unit connects IPM inverter, and IPM inverter is connected to dual linear motor platform by a Drive Protecting Circuit; Hall element is connected between IPM inverter and Drive Protecting Circuit on the one hand, and Hall element is connected to digital signal processor DSP by current detecting unit on the other hand; Position and speed detecting unit is connected to dual linear motor platform by grating scale on the one hand, is connected on the other hand digital signal processor DSP; IPM inverter is connected to optical coupling isolation circuit by second Drive Protecting Circuit, and optical coupling isolation circuit is connected to digital signal processor DSP on the one hand, is connected between rectification stable pressuring unit and IPM inverter on the other hand by fault detect and protective circuit.Digital signal processor DSP comprises quadrature coding pulse circuit QEP, ADC module, PWM unit, Flash memory cell, program storage, timer and the PDPINT pin of task manager EVA; QEP link position speed detection unit, ADC module connects current detecting unit, and PWM unit and PDPINT pin are all connected to optical coupling isolation circuit.

In DSP, be also equipped with PI controller, profile errors calculator, fuzzy RBF neural network Integral Sliding Mode outline compensation device and the drive assembly of position signalling control point adjustment, linear electric motors position signalling, rate signal and current signal.The realize operation principle of these control devices in DSP be to interrupt by Open Timer device, call and be stored in the corresponding control device in program storage and complete corresponding control, first idiographic flow is following sets the initial position of each axle of dual linear motor by position signalling control point adjustment according to any track of planning processing, secondly through the PI control device of linear electric motors position signalling, dual linear motor platform is controlled, obtained an output signal as a part for drive assembly control signal; Then according to the single shaft position error signal of each axle of dual linear motor of control system gained, after the calculating of profile errors calculator, obtain the input of fuzzy RBF neural network Integral Sliding Mode outline compensation apparatus, after controlling by this device, obtain an output signal and as another part of drive assembly control signal, the finally input using the output of drive assembly as speed ring PI control device, by obtaining the single shaft control amount of linear electric motors after controlling, structured flowchart as shown in Figure 1.

Fig. 1 is the structured flowchart of the designed dual linear motor outline compensation apparatus and method based on based on fuzzy RBF neural network Integral Sliding Mode of the present invention, wherein X _d, Y _dand X _a, Y _arepresent the input and output of each axle, ξ _xand ξ _yfor the disturbance of the corresponding diaxon of dual linear motor, e _xand e _yfor the site error of the corresponding diaxon of dual linear motor, u _xand u _yfor the control inputs amount of the corresponding diaxon of dual linear motor, C _xand C _ycompensating gain value for the corresponding diaxon of dual linear motor, real-time profile errors value when ε is any track following, s (t) is the sliding formwork diverter surface relevant with profile errors, w, σ, c is weights and the membership function parameter of Fuzzy RBF Neural Network, and the present invention utilizes the self-learning capability of Based Intelligent Control to approach the sliding-mode surface function relevant with profile errors it is minimized at finite time, and profile errors minimizes.Reaching high-precision outline compensation controls.Finally realized the design of fuzzy RBF neural network Integral Sliding Mode outline compensation control device, its structure principle chart is as shown in the dotted line frame Ι in Fig. 1.The profile errors amount of any track of mentioning can geometrical relationship as shown in Figure 2 obtain.The effect of the fuzzy RBF neural network Integral Sliding Mode outline compensation device that the present invention is designed is when improving system robustness, to eliminate buffet and reach high accuracy processing request.

The profile errors expression formula of the known any track of Fig. 2 is ε=-e _xsin φ+e _ycos φ wherein x axle tracking error is e _x; Y axle tracking error is e _y; On x axle and given path, the angle of line is φ a bit and between desired locations.

Dual linear motor outline compensation device design procedure based on fuzzy RBF neural network Integral Sliding Mode is as follows:

Step 1: offer the corresponding position signalling instruction of dual linear motor diaxon by the position signalling control point adjustment in TMS320F2812 digital signal processor, as the Position Control amount input of diaxon;

Step 2: by signal picker, position, speed and current signal are sampled and relatively acquisition position error signal given by diaxon and actual output;

Step 3: the position error signal that regulates single shaft by the built-in PI positioner of DSP, adopt profile errors calculator calculate the profile errors amount of any track and compensate control by built-in profile errors compensation arrangement, the output of then both being controlled is as the input of position device;

Step 4: Negotiation speed detecting signal unit is determined the speed of linear electric motors;

Step 5: by grating scale, sample, in TMS320F2812 digital processing unit relatively after, carry out PI controller;

Step 6: by built-in current detecting unit carry out current sample and in DSP relatively after, carry out PI controller;

Step 7: current value is carried out to 3/2 conversion;

Step 8: utilize the q axle of rotation to calculate torque;

Step 9: output calculating torque is carried out to 2/3 conversion

Step 10: as carrier wave and triangular modulation, allow digital signal processor DSP produce six road pwm pulse signals to the current value after conversion, drive the diaxon of dual linear motor to carry out given profile traces working motion according to the size of current-order;

Regulator rectifier circuit converts direct current to three-phase alternating current and powers to IPM inversion unit, IPM inversion unit produces Liu road pwm pulse signal according to DSP conducting and the shutoff of six IGBT switch elements in IPM inversion unit is controlled, and drives linear electric motors operation.

Employing profile errors calculator described in step 3 calculates the profile errors amount of any track and compensates control by built-in profile errors compensation arrangement, and the output of then both being controlled, as the input of position device, comprises that step is as follows:

Step 3-1: calculate profile errors value and the profile errors value of track can be by ε=-e arbitrarily according to real-time position, speed and current signal and position error signal _xsin φ+e _ycos φ represents and meets x axle tracking error is e _x; Y axle tracking error is e _y; On x axle and given path, the angle of line is φ a bit and between desired locations.

Step 3-2: when profile errors exists, the input of the dual linear motor outline compensation device based on fuzzy RBF neural network Integral Sliding Mode is the sliding formwork switching function s (t) relevant to profile errors, can be determined by following formula

$s\left(t\right)=(\frac{d}{d_t}+\mathrm{\λ})\underset{0}{\overset{t}{\∫}}\mathrm{\ϵ}\left(t\right)\mathrm{dt}$

Wherein: λ is greater than zero constant; The object of controlling is to make it is profile errors ε → 0

Step 3-3: adopt the fuzzy RBF neural network control with self-learning capability to approach study and fuzzy output to target function, described target function is

Learning algorithm is chosen as

$w_{\mathrm{ko}}^4=-\mathrm{\η}\frac{\∂E}{\∂w}=-\mathrm{\η}\frac{\∂S\stackrel{\·}{S}}{{\∂U}_{\mathrm{FNN}}}\frac{{\∂U}_{\mathrm{FNN}}}{\∂w_{\mathrm{ko}}^4}$

The error term of definition rule layer and membership function layer is

${\mathrm{\δ}}_k^3\stackrel{\mathrm{\Δ}}{=}-\mathrm{\η}\frac{\∂S\stackrel{\·}{S}}{{\∂U}_{\mathrm{FNN}}}\frac{\∂U_{\mathrm{FNN}}{\∂y}_o^4}{\∂y_o^4{\∂y}_k^3}\frac{{\∂y}_k^3}{\∂{\mathrm{net}}_k^3}=-\mathrm{\η}\left|\frac{K_f}{M}\right|{\mathrm{Sw}}_{\mathrm{ko}}^4$

${\mathrm{\δ}}_j^2=-\mathrm{\η}\frac{\∂S\stackrel{\·}{S}}{{\∂U}_{\mathrm{FNN}}}\frac{\∂U_{\mathrm{FNN}}{\∂y}_o^4}{{\∂y}_o^4{\∂y}_k^3}\frac{{\∂y}_k^3}{\∂{\mathrm{net}}_k^3}\frac{{\∂\mathrm{net}}_k^3}{{\∂y}_j^2}\frac{{\∂y}_j^2}{\∂{\mathrm{net}}_j^2}=\underset{k}{\mathrm{\Σ}}{\mathrm{\δ}}_k^3{y}_k^3$

Adjust in the following way membership function parameter

$\mathrm{\Δ}{C}_{\mathrm{ij}}=-\mathrm{\η}\frac{\∂E}{\∂c_{\mathrm{ij}}}=-\mathrm{\η}\frac{\∂E}{\∂\mathrm{ne}{t}_j^2}\frac{\∂{\mathrm{net}}_j^2}{\∂c_{\mathrm{ij}}}=-\mathrm{\η}{\mathrm{\δ}}_j^2\frac{2(x_i-c_{\mathrm{ij}})}{{\mathrm{\σ}}_{\mathrm{ij}}^2}$

$\mathrm{\Δ}{\mathrm{\σ}}_{\mathrm{ij}}=-\mathrm{\η}\frac{\∂E}{\∂{\mathrm{\σ}}_{\mathrm{ij}}}=-\mathrm{\η}\frac{\∂E}{\∂{\mathrm{net}}_j^2}\frac{\∂{\mathrm{net}}_j^2}{\∂{\mathrm{\σ}}_{\mathrm{ij}}}=-\mathrm{\η}{\mathrm{\δ}}_j^2\frac{2(x_i-c_{\mathrm{ij}})}{{\mathrm{\σ}}_{\mathrm{ij}}^3}$

Wherein: the average of Gaussian membership function and standard deviation are respectively C _ijand σ _ij; Neural net connection weight matrix is

Fuzzy output valve is

$y_i^1=f_i^1{x}_i^1,i=\mathrm{1,2}$

$y_j^2=f_j^2\left({\mathrm{net}}_j^2\right)=\exp (-\frac{(x_i^2-C_{\mathrm{ij}})}{{\left({\mathrm{\σ}}_{\mathrm{ij}}\right)}^2}),j=\mathrm{1,2}......n$

$y_k^3={\mathrm{net}}_k^3=\underset{j=1}{\overset{N}{\mathrm{\Π}}}{y}_j^2$

Wherein: being input as of outline compensation device output for fuzzy control;

Step 3-4: the connection weight matrix by fuzzy output and neural net multiplies each other, can obtain the position compensation signal of profile errors compensation arrangement

$U_{\mathrm{FNN}}=y_o^4=W_{\mathrm{ko}}^4\·y_k^3=\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{w}_{\mathrm{ko}}^4\·y_k^3,o=1$

Fig. 3 is the structural representation of institute's invention fuzzy RBF neural network Integral Sliding Mode outline compensation method, and wherein neural network structure is chosen as 3 layers of RBF feedforward neural network, and its advantage is that speed does not exist local minimum problem soon; Fuzzy control contains input layer, membership function layer, rules layer and output layer, and membership function is in order to be chosen as Gaussian with neural net hidden layer action function structure matching, and rules layer is product reasoning; And wherein represent respectively the output valve of every layer.

In the inventive method, system program flow chart as shown in Figure 4.The main program of software comprises system initialization; Opening INT1, INT2 interrupts; Permission timer interrupts; Timer Interrupt Subroutine.Wherein initialize routine comprises and closes all interruptions, dsp system initialization, initialization of variable, task manager initialization, AD initialization and quadrature coding pulse QEP initialization.Interrupt service subroutine comprises protection interruption subroutine and T1 underflow interrupt service subroutine.Other parts are as mover initialization location, and the PI of position, speed and current signal regulates, and fuzzy RBF neural network Integral Sliding Mode outline compensation adjusting etc. is all carried out in timer TI underflow Interrupt Subroutine.

The protection interrupt response that IPM guard signal produces belongs to external interrupt, and INT1 interrupt priority level is than the height of timer T1.IPM can send guard signal automatically in abnormal conditions such as overcurrent, overvoltages, and this signal is connected to the power drive protection pin of DSP through conversion once have abnormal conditions to occur, DSP can enter protection interruption subroutine, first forbids all interruptions, then block PWM output and make motor stall at once, play the effect of protection motor and IPM.

Control method of the present invention is finally realized by dsp processor, and step is as follows

Step 1 system initialization

Step 2 DSP initialization

Step 3 initialization register and variable

Step 4 initialization interrupt vector

Step 5 is opened interruption

Whether step 6 has general purpose timer underflow to interrupt producing

Step 7 TN1 interrupts processing sub-control program

Step 8 protection is interrupted processing

Step 9 finishes

Wherein: TN1 interrupts processing sub-control program according to the following steps:

Step 1 TN1 interrupts sub-control program;

Step 2 keeps the scene intact;

Step 3 pair institute's pursuit path is planned and the given signal in position of definite dual linear motor

Step 4 is calculated motor speed and the direction of motion

Step 5 sample rate current and voltage

Step 6 is called the PI controller of displacement

Step 7 is called speed and current PI controller

Step 8 sampling desired value rear and each axle compares acquisition position deviation

Step 9 judges whether to exist profile errors; Be to enter step 10, otherwise enter step 12

Step 10 is called real-time profile errors model calculator

Step 11 is called real-time profile errors compensating controller

The new current value of step 12 sampling

Step 13 pair current sampling data carries out 3/2 conversion

Step 14 utilizes the q axle of rotation to calculate torque

The electric current of step 15 pair output carries out 2/3 conversion

Step 16 generates PWM waveform with triangular wave as carrier wave electric machine phase current is controlled, and then dual linear motor is carried out to profile control

Step 17 interrupts returning

Fig. 5 is the hardware elementary diagram of invented outline compensation device, and this device comprises rectification stable pressuring unit, IPM inverter, digital signal processor DSP, Hall element, grating scale, current detecting unit, position and speed detecting unit, optical coupling isolation circuit, Drive Protecting Circuit and fault detect and protective circuit.Wherein DSP comprises the quadrature coding pulse circuit QEP of task manager EVA, ADC module, PWM unit, Flash memory cell, program storage, in timer and PDPINT pin and program storage, be also equipped with the position signalling control point adjustment of being realized by software program, linear electric motors position signalling, the PI controller of rate signal and current signal, profile errors calculator, fuzzy RBF neural network Integral Sliding Mode outline compensation device and drive assembly, once there is overvoltage in system, overcurrent, the faults such as under voltage, DSP will block PWM output signal, with protection IPM module.

Fig. 5-(1) is the current detection circuit of invented hardware system, according to the design of whole system, in control system, there is current feedback ring, the current signal of the motor of therefore need to sampling, what in the design, measure is the two-phase in linear electric motors three-phase current.In native system, adopt CSM025PTS series Hall current sensor, the current range that it can gather is-16A～+ l6A that the voltage range of output is 0～5V.Because the AD module samples of dsp chip is voltage signal, current signal need to be converted into voltage signal.AD module input voltage range with DSP is 0～3V, and samples to such an extent that the voltage range of Hall element is 0～5V, need to do suitable conversion, and the voltage signal of sampling is converted into suitable input range.The AD module of 16 passages need to be used wherein three, and what in the design, select is ADCIN08 and two passages of ADCIN09, and the current signal that the voltage signal after sampling is converted into two-phase again deposits in corresponding register, carries out electric current adjusting.Through Hall element, the proportional voltage signal that is converted into of electric current detected to obtain.The voltage signal of Hall element output is not suitable for the voltage range that AD module gathers, and need to voltage signal, carry out convergent-divergent through amplifier.

Fig. 5-(2) are the position detecting circuit of invented hardware system, the detection of the position signalling of linear electric motors is to realize the accurate very important factor of controlling of linear electric motors and is exactly rate signal by its differential, the design adopts the RGH24X series grating scale of Britain RENISHAW company, its resolution is 1um, and maximum speed can reach 5m/s.The quadrature coding pulse circuit (QEP) of TMS320F2812 task manager EVA is connected with grating scale, after the QEP of task manager is enabled, just can count corresponding pin output pulse, determine the positional information of motor movement, according to the positional information of measuring, adjust accordingly control strategy, control motor movement.Install the linear electric motors of grating scale in the situation that of motion, the 6 road square-wave signals that read head output is relevant with positional information, wherein they are three pairs of reverse RS422A square-wave signals, after differential received, produce three road pulse signals and comprise two-way quadrature coding pulse signal A and B, and a road reference point signal Z.Quadrature coding pulse signal A and B are used for the positional information of detection of straight lines motor movement, and Z signal is used for doing the reference signal of clear point position in rectilinear motion process.The 3 road pulse signals of exporting due to grating ruler reading head are 5V, and the I/O mouth voltage of DSP is 3.3V, therefore need a level conversion core SN74LVC4245DW realize 5V to the level conversion of 3.3V.Enable quadrature coding pulse circuit and the capturing unit CAP3 of dsp chip task manager EVA, QEP1 and QEP2 receive respectively A and B two-way orthogonal pulses simultaneously, and CAP3 catches reference burst signal Z.

The quadrature coding pulse of the read head output of grating scale is that two-way frequency can change mutually orthogonal pulse train.When linear electric motors move, the read head of grating scale can produce quadrature coding pulse signal, by QEP1 and QEP2 interface, received respectively, according to the phase difference of A and B pulse signal, be+90 ° or-90 ° of directions of motion of determining linear electric motors, can determine the counting direction of the counter of general purpose timer simultaneously, if the pulse that the pulse advance QEP2 that QEPI receives receives, counter is to increase progressively counting so, on the contrary countdown.Because the quadrate encode module of DSP is all counted input orthogonal pulses rising edge and trailing edge, therefore via the frequency of exporting pulse after QEP circuit, become four times of input, and task manager can offer this clock its general purpose timer as the clock frequency of general purpose timer.In the design, general purpose timer TZ is set to orientation and increases/subtract counting, and the quadrature coding pulse circuit of task manager EVA is not only for it provides clock frequency but also counting direction is provided.The resolution of grating scale is 1um, the every mobile lum of linear electric motors general purpose timer to quadruple later pulse once count, by count values different in the counter of twice of front and back, can determine the distance of linear electric motors motion, the counting direction that simultaneously has special register-stored counter is also the direction of motion of linear electric motors.In the interrupt service subroutine of DSP, can obtain the movable information of linear electric motors.

Fig. 5-(3) are that the linear electric motors of invented hardware system are controlled power circuit, and the designed power circuit of the present invention will obtain direct voltage exactly.First by a rectifier bridge, interchange is converted into direct current, then by filtering, obtains good direct current and be stored in large electric capacity, at the delivery outlet using electric capacity two ends as DC power supply, offer power model.Wherein add a fuse, excessive in order to prevent circuital current, play protective circuit effect.What relay played is the effect of a Based Intelligent Control, in program when DSP initialization, capacitor charging, after initialization completes, send a high level in program Kl port, a switch motion occurs relay makes electric capacity two ends directly connect supply voltage, at this time makes the voltage at electric capacity two ends reach maximum, during experiment, maximum can arrive 320V, is reached for motor power reguirements.

Fig. 5-(4) are the linear electric motors drive circuit of invented hardware system, linear electric motors drive circuit mainly comprises an Intelligent Power Module, that the present invention selects is IRAMSl0UP60B, it is applicable in the motor of relatively high power, and the power of motor scope that it can drive is 400W～750W; The main three-phase bridge circuit being formed by 6 IGBT, the pwm control signal that on control board, dsp chip produces is input to power model, control the shutoff of 3 brachium pontis, produce appropriate drive voltage, driving HIN1 and LIN1 in linear electric motors motion diagram is respectively the control signal of the upper and lower bridge arm of first-phase, and they are all Low level effectives.The operating voltage VDD of IRAMSl0UP60B is 15V, and VSS is earth terminal, in order to reach good decoupling effect, adds two decoupling capacitors in parallel at these two ends.Because the PWM ripple signal of inputting is digital signal, and IRAMS10UP60B does not possess the function of digital signal and power signal isolation, therefore before the input control signal of IRAMS10UP60B, need to add the isolation of glazing misfortune, in figure, TLPll3 has realized the function that the pwm signal of input is converted into analog signal, is then input to the control signal input of corresponding brachium pontis.At output u, the v of three-phase voltage, the bootstrap capacitor that w adds respectively a 2.2uF.When Itrip port is low level, chip is normally worked, and the control signal of inputting when the upper part of brachium pontis is low level, and when lower part is high level, this has output voltage mutually; When upper be high level, lower for low level time output voltage be zero; Two is all that low level situation is not allow appearance, can cause short circuit like that, burns chip.Therefore when Itrip port is high level, chip is not worked, and there is no Voltage-output, and in circuit, adding a pull down resistor, to make Itrip port be low level, and power model can normally be worked like this.Power chip self had gentle overcurrent protection, when the effect of self-protection appears playing when abnormal in circuit.

Fig. 5-(5) are the IPM isolation drive protective circuit of invented hardware system; IPM is that logical circuit, drive circuit protective circuit and the testing circuit of power device and a control action is integrated or fits together; mainly complete functions such as driving signal amplification, power amplification, various protection (comprising overcurrent protection, short-circuit protection, overtemperature protection, under-voltage protection), on device property, there is the switching characteristic of IGBT.The IPM that native system is selected is Mitsubishi third generation Intelligent Power Module PM20CSJ060.Its nominal parameter is 600V, 20A, and applicable power of motor is 1.5KW, and switching frequency reaches as high as 20KHz.The pwm signal that DSP module produces is input to the respective pin of IPM module through light-coupled isolation.

Claims (2)

1. the dual linear motor outline compensation device based on fuzzy RBF neural network Integral Sliding Mode, is characterized in that: this device comprises rectification stable pressuring unit, IPM inverter, digital signal processor DSP, Hall element, grating scale, current detecting unit, position and speed detecting unit, optical coupling isolation circuit, Drive Protecting Circuit and fault detect and protective circuit; Rectification stable pressuring unit connects IPM inverter, and IPM inverter is connected to dual linear motor platform by a Drive Protecting Circuit; Hall element is connected between IPM inverter and Drive Protecting Circuit on the one hand, and Hall element is connected to digital signal processor DSP by current detecting unit on the other hand; Position and speed detecting unit is connected to dual linear motor platform by grating scale on the one hand, is connected on the other hand digital signal processor DSP; IPM inverter is connected to optical coupling isolation circuit by second Drive Protecting Circuit, and optical coupling isolation circuit is connected to digital signal processor DSP on the one hand, is connected between rectification stable pressuring unit and IPM inverter on the other hand by fault detect and protective circuit.

2. the dual linear motor outline compensation device based on fuzzy RBF neural network Integral Sliding Mode according to claim 1, is characterized in that: digital signal processor DSP comprises quadrature coding pulse circuit QEP, ADC module, PWM unit, Flash memory cell, program storage, timer and the PDPINT pin of task manager EVA; QEP link position speed detection unit, ADC module connects current detecting unit, and PWM unit and PDPINT pin are all connected to optical coupling isolation circuit.