CN1895809A - Controlling system of distributor of large moulded forging hydraulic press - Google Patents

Abstract

A control system for the distributor of large die-forging hydraulic press features that the absolute photoelectric encoder is used as the rotation angle setting up and tracking sensor. Its control method includes such steps as converting the setting value of rotation angle of handle to binary digital signals, converting the setting value of rotation angle of distributor to binary digital signals, sending them to PLC binary input module of programmable controller, and outputting them to relative servo electromagnetic valves for controlling hydraulic cylinder, driving rack-gear driver and driving the distributor to rotate.

Description

Controlling system of distributor of large moulded forging hydraulic press

[technical field] the present invention relates to the control system of hydraulic die press, particularly the control system of hydraulic die press distributor.

[background technology] large-scale die forging water press is a kind of important plant equipment in the machine building industry, is used to forge common alloy forged piece or complex forging with holes.The die forging process by four moveable parts in order co-operating finish, they are vertical crossbeams, and left and right horizontal component and travelling table, the movement velocity and the stroke of each several part all adopt " oil control water " mode to realize control, oil hydraulic system drives distributor, distributor control water valve.One group of cam is installed on the distributor axis, the shape of cam is different with the installation site, when distributor turns to different angle initialization values, the corresponding push rod of cam driven rises or descends, thereby open or close corresponding water valve, water under high pressure enters different working cylinders and drives removable working portion and finish the die forging processing action.At present, the control system of distributor of large moulded forging hydraulic press adopts selsyn to set and tracking transducer as corner, constitutes analog controller by polar relay and phase-sensitive amplifier.Mainly there are the following problems for this control system: position control accuracy is low, and the corner tracking error is big, reaches ± 9 °, and under-voltage amount is big; The position tracing process produces vibration, causes the die forging process pressure oscillation, makes the inner texture of forging inhomogeneous, influences the forging part quality.Under-voltage amount is the physical quantity that is used for describing the die forging machining accuracy, the influence of factors such as it is stressed, corner and corner tracking error, and having the greatest impact of angular errors wherein, reducing the effective measures of under-voltage amount is the precision that improve the distributor angle position; Yet distributor cam face curve complexity, rotary inertia is bigger, the rotary inertia that mechanical system is converted on the distributor axis is the nonlinear function of corner, reducing angle error in tracking can make it rotate the generation vibration, if rotating, distributor produces vibration, will cause the die forging pressure oscillation, make the reliability variation of inner texture of forging and system.

[summary of the invention] the objective of the invention is to improve the angle position control accuracy of distributor of large moulded forging hydraulic press, reduces the under-voltage amount of distributor and eliminates the corner vibration.The present invention adopts absolute optical encoder and programmable controller to constitute control system and drives the distributor rotation.Adopt segmentation to adjust control strategy, reduce the tracking error of distributor, eliminate the vibration of distributor corner, improve the machining accuracy and the inner texture quality of forging ' s block dimension.The present invention mainly comprises following several respects content:

At first replace selsyn to set and tracking transducer as corner with absolute optical encoder.The given photoelectric encoder of corner is installed on the operating desk, and the turning cylinder by elasticity shaft coupling and operating grip links to each other, and corner detects photoelectric encoder and is installed in the distributor axle head, also is connected with the distributor turning cylinder with the elasticity shaft coupling.Operating grip provides the setting value of distributor corner, and encoder converts the corner of operating grip and distributor to the multidigit binary digital signal, delivers to the PLC switching input module.The numerical control system that constitutes with programmable controller replaces being made of polar relay and phase-sensitive amplifier the analog control system of controller again.The switching value output module output control signal of programmable controller is given corresponding servo electromagnetic valve, the control hydraulic cylinder, drive rack and pinion gearing, the driving distributor rotates, the corresponding push rod of cam driven on the distributor axis rises or descends, and opens or closes the pressure that corresponding water valve changes water under high pressure.The intelligent control technology that application is adjusted based on segmentation is divided into two stages with distributor corner tracing process and adjusts.Phase I is that the distributor corner is fast to the setting value motion stage, adopt the artificial neural network self-learning algorithm, size according to pressure, distributor corner tracking error and the corner tracking error rate of change of water under high pressure calculates suitable power-off time lead, make the magnetic valve outage, drive system is owing to effect of inertia, and distributor continues to stop after angle of rotation; Second stage is the accurate adjusting stage of distributor corner, judge whether the corner tracking error satisfies the scope of setting, if do not satisfy, then according to size and symbol, the pressure of water under high pressure and the control accuracy of requirement of corner tracking error, adopt fuzzy inference rule to determine the time of switching on once more to corresponding solenoid valve, allow distributor be rotated further a smaller angle, the error that its corner is followed the tracks of reaches requirement.

By above-mentioned control device and control algolithm, the tracking error of distributor by ± 9 ° be reduced to ± 3 °, reduced under-voltage amount, solved large-scale die forging water press because of the big difficult problem that influences forging part dimensioned precision of under-voltage amount, improve the inside texture and the quality of forging part, improved the functional reliability of control system of water press simultaneously.

The present invention is described in further detail below in conjunction with accompanying drawing.

Fig. 1 controlling system of distributor hardware configuration schematic diagram;

Fig. 2 adjusts the control system theory diagram for segmentation;

Fig. 3 ε _i-NN structural representation.

[embodiment] controlling system of distributor hardware configuration as shown in Figure 1, the given photoelectric encoder 4 of corner is installed on the operating desk 1, link to each other by the turning cylinder of elasticity shaft coupling 3 with operating grip 2, corner detects photoelectric encoder 6 and is installed in the distributor axle head, also is connected with distributor turning cylinder 7 with elasticity shaft coupling 3.Operating grip 2 provides the setting value of distributor corner, and encoder converts the corner of operating grip and distributor to the multidigit binary digital signal, delivers to the PLC switching input module by cable 5.Calculate the controlled quentity controlled variable signal by programmable controller and export to corresponding servo electromagnetic valve 11 through the switching value output module of programmable controller, control hydraulic cylinder 10 drives rack and pinion gearing 12,9, drives distributor and rotates.One group of cam 13 is installed on the distributor axis, and the shape of cam is different with the installation site, and when distributor turned to different angle initialization values, the corresponding push rod 8 of cam driven rose or descends, thereby opens or closes the pressure that corresponding water valve changes water under high pressure.

As Fig. 2, hydraulic press distributor corner can be divided into two stages, and the phase I is that the distributor corner is fast to setting value motion stage, employing artificial neural network self study ε _i-NN algorithm goes out suitable power-off time lead ε according to the size of pressure, distributor corner tracking error and the corner tracking error rate of change of water under high pressure this moment by The Fitting Calculation _i, making the magnetic valve outage, drive system is owing to effect of inertia, and distributor continues to stop after angle of rotation; Second stage is the accurate adjusting stage of distributor corner, judges corner tracking error Δ θ _iWhether satisfy-δ _i＜Δ θ _i＜δ _i, if do not satisfy, then according to Δ θ _iSize and symbol, the pressure of water under high pressure and the control accuracy δ of requirement _iAdopt fuzzy inference rule to determine the time τ that switches on once more to corresponding solenoid valve _i, allow distributor be rotated further a smaller angle, thus the error that its corner is followed the tracks of reach ± 3 ° in.For example, operating grip can move in 0 °～180 ° scopes, and " idle running ", " work ", " backhaul " and " stopping " four positions are set respectively, the angle value that they are corresponding certain respectively.During die forging work, pull operating grip, provide the angle value of the distributor rotation of requirement, be assumed to " backhaul ": 50 °, the control accuracy δ of requirement _iBe ± 3 °, then elder generation turns to about 40 ° position by magnetic valve power-off time lead of neutral net match output by distributor according to the size of pressure, distributor corner tracking error and the corner tracking error rate of change of water under high pressure this moment, because effect of inertia, it can be rotated further an angle, suppose to be parked in 43 ° position, at this moment corner tracking error Δ θ _iBe 50 °-43 °=+ 7 °, and control accuracy δ _iBe ± 3 °, do not satisfy-δ _i＜Δ θ _i＜δ _iBecause Δ θ _iSymbol for just, according to Δ θ _iSize, the pressure of water under high pressure and the control accuracy δ of requirement _iAdopt fuzzy inference rule to calculate the time τ that switches on once more to corresponding servo valve _i, allow distributor continue to rotate 5 ° in the same way, add that the distributor effect of inertia is rotated further 1 °, then distributor has turned over 49 ° altogether in whole process, the required precision of 50 ± 3 ° of scopes of foot, and the vibration situation can not appear.

ε _iThe specific implementation apportion of-NN algorithm and fuzzy reasoning control control algolithm is as follows:

A. ε _i-NN algorithm

Adopt three layers of BP neutral net to realize outage lead ε _iMatch.3 of input layers are respectively corner tracking error Δ θ _iAnd rate of change Δ θ _i ^﹠amp;And pressure P _iDeng, carry out normalized; Because the training time is subjected to the restriction of production actual conditions, hidden layer node Q gets 5; 1 of output layer node is lead ε _iBecause ε _iCan not be negative value, so the neuronic activation functions of output layer is got non-negative Sigmoid function, and the desirable positive and negative Sigmoid function of the activation functions of hidden layer neuron.

I BP neutral net NN is input as:

The input and output of neutral net hidden layer are:

$\left.\begin{array}{c}{\mathrm{net}}_n^2\left(k\right)=\underset{m=0}{\overset{3}{\mathrm{\Σ}}}{w}_{\mathrm{mn}}^2{o}_m^1\left(k\right)\\ o_n^2\left(k\right)=f\left[{\mathrm{net}}_n^2\left(k\right)\right](n=\mathrm{0,1},,L4)\\ O_Q^2\left(k\right)=1\end{array}\right\}---\left(2\right)$

In the formula: w _Mn ²Be the hidden layer weight coefficient; w _3n ²Be threshold value, $w_{3n}^2=1;$ F () is an activation functions, and f ()=tanh (x) is for surmounting tan; Superscript 1,2,3 is represented input layer, hidden layer and output layer respectively.

The input and output of neutral net output layer are:

$\left.\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="net"\; filenames="A20051003186500083.png"\; state="\{\{state\}\}">net</figure-callout>}}_1^3\left(k\right)=\underset{n=0}{\overset{\mathrm{<figure-callout\; id="1"\; label="Q\; w\; n"\; filenames="A20051003186500083.png"\; state="\{\{state\}\}">Q</figure-callout>}}{\mathrm{\&Sigma;}}}{w}_n^{}{1}_3^{}{o}_n^2\left(k\right)\\ o_1^3\left(k\right)=g\left[{\mathrm{net}}_1^3\left(k\right)\right]={\mathrm{\&epsiv;}}_i\left(k\right)\end{array}\right\}---\left(3\right)$

In the formula: w _N1 ³Be the output layer weight coefficient; w ₅₁ ³Be threshold value, $w_{51}^3=1;$ G () is an activation functions, g ()=[1+tanh (x)]/2.

Getting performance index function is:

J ^(k)＝[Δθ _ie(k)] ²/2 (4)

In the formula: Δ θ _Ie(k)=Δ θ _i(k)-δ _i, Δ θ _Ie(k) be switch electromagnetic valve again turn-on time τ _iBack residue tracking error Δ θ _i(k) and δ _iDifference relatively.

According to the weight coefficient of steepest descent method corrective networks, promptly press J the negative gradient direction search of weight coefficient is adjusted, and the minimum inertia item of the additional accelerating ated test overall situation, then have:

${\mathrm{\&Delta;w}}_{\mathrm{mn}}^2(k+1)=-\mathrm{\&eta;}\frac{{\&PartialD;J}^{\left(k\right)}}{{\&PartialD;w}_{\mathrm{mn}}^2}+{\mathrm{\&alpha;\&Delta;w}}_{\mathrm{mn}}^2\left(k\right)---\left(5\right)$

${\mathrm{\&Delta;w}}_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="A20051003186500083.png"\; state="\{\{state\}\}">n</figure-callout>}1}^3(k+1)=-\mathrm{\&eta;}\frac{{\&PartialD;J}^{\left(k\right)}}{{\&PartialD;w}_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="A20051003186500083.png"\; state="\{\{state\}\}">n</figure-callout>}1}^3}+{\mathrm{\&alpha;\&Delta;w}}_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="A20051003186500083.png"\; state="\{\{state\}\}">n</figure-callout>}1}^3\left(k\right)---\left(6\right)$

In the formula: η is a learning rate; α is an inertia coeffeicent.

Self study ε _i-NN algorithm network training adopts off-line and the online method that combines, and gathers Δ θ by field controller between limber up period _i(k), Δ θ _i ^﹠amp;And P _i, be stored in the computer, train to obtain suitable lead ε _i, try out again, manufacture suitable fine setting by learning machine.Self-learning algorithm is by carrying out in the off time between two processes of operation supervise and control computer in die forging process, and the amount trimmed of calculating is judged, optimizes lead ε _iThe pressure limit of water under high pressure is at 2000～10000 ton hours, and the scope of power-off time lead is 10%～15% of a given corner.

B. fuzzy reasoning control algolithm

In the design of system ambiguous controller, take into full account expertise; Adopt the method that quantizes to set up Triangleshape grade of membership function; IF...THEN... sets up fuzzy control rule by the experience induction; Adopt the Mamdani Fuzzy Logic Reasoning Algorithm, carry out approximate reasoning; Obtain once more the fuzzy quantity of conduction time τ _i, behind the ambiguity solution, obtain once more τ conduction time _i

According to the requirement of native system control accuracy, corner tracking error Δ θ _iAnd θ _RiThe fuzzy variable Δ θ _iBe taken as negative big, negative in, negative little, zero, just little, the center, honest }, i.e. { NB, NM, NS, ZO, PS, PM, PB}.Pressure P _iSet a standard value in advance, the deviation of detected value and standard value is divided by the fuzzy variable of the ratio of standard value P _iBe taken as negative big, negative in, negative little, zero, just little, the center, honest }, i.e. { NB, NM, NS, ZO, PS, PM, PB}.Conduction time τ _iFuzzy variable τ _iBe taken as negative big, negative little in bearing, zero, just little, the center, honest, promptly NB, and NM, NS, ZO, PS, PM, PB}, be 0 the conduction time again of ZO correspondence, and N represents counter-rotating, and P represents just to change.

According to expertise and operating personnel's experience accumulation, servo valve once more conduction time τ _iFuzzy inference rule as shown in table 1.

Table 1 fuzzy control rule table

If distributor rotational angle deficiency or excessive appears in the accurate adjustment in second stage, surpassed the scope of control accuracy regulation, τ conduction time then is described once more _iScope that excessive or above-mentioned fuzzy control rule table is set is unreasonable, should reset τ _iScope or set up new rational fuzzy control rule.The pressure limit of water under high pressure is at 2000～10000 ton hours, and the time range that adopts fuzzy inference rule to switch on for corresponding solenoid valve once more is 1～1.5 second.

Claims (2)

1. a controlling system of distributor of large moulded forging hydraulic press comprises operating desk, and operating grip and distributor is characterized in that:

(1) sets and tracking transducer as corner with absolute optical encoder, be installed in the given photoelectric encoder of corner on the operating desk, turning cylinder by elasticity shaft coupling and operating grip links to each other, convert the multidigit binary digital signal to for the setting value of corner operating grip, the corner that is installed in the distributor axle head detects photoelectric encoder, be connected with the distributor turning cylinder with the elasticity shaft coupling, the corner of distributor is converted to the multidigit binary digital signal, binary signal is sent to the PLC switching input module, the input signal of the numerical control system that constitutes as programmable controller, the switching value output module output control signal of programmable controller is given corresponding servo electromagnetic valve, the control hydraulic cylinder, drive rack and pinion gearing, the driving distributor rotates, the corresponding push rod of cam driven on the distributor axis rises or descends, and opens or closes the pressure that corresponding water valve changes water under high pressure;

(2) intelligent control technology of adjusting with segmentation, distributor corner tracing process is divided into two stages to be adjusted, phase I is that the distributor corner is fast to the setting value motion stage, size according to pressure, distributor corner tracking error and the corner tracking error rate of change of water under high pressure, adopt the artificial neural network self-learning algorithm, calculate the power-off time lead; Second stage is the accurate adjusting stage of distributor corner, when corner tracking error during greater than setting range, the time of adopting fuzzy inference rule to switch on for corresponding solenoid valve once more, allows distributor be rotated further a smaller angle.

2. controlling system of distributor according to claim 1, it is characterized in that: the pressure limit of described water under high pressure is at 2000～10000 ton hours, the scope of power-off time lead is 10%～15% of a given corner, and the time range that adopts fuzzy inference rule to switch on for corresponding solenoid valve once more is 1～1.5 second.

Images