CN1105302C - Viscosity and dispersity estimator for banburying mixer mixing rubber and its mathematical model establishing method - Google Patents

Abstract

The present invention relates to a prediction device for the viscosity and the dispersity of rubber compounds of a banbury mixer and a mathematical model establishing method thereof. The device is composed of a cabinet, a CPU main board, a passive mother board, a PC104 bus input board, a signal input conditioning board, a hard disk drive controller, a floppy disk drive controller, a display and a printer, wherein the PC104 input board is arranged on the CPU main board fixed on the passive mother board, the main board is then electrically connected with a hard disk drive and a floppy disk drive, and the signal input conditioning board is fixed on the passive mother board fixed in the cabinet. Circuits are formed by the electrical connection of a 24-path switching value input isolating circuit, a 24-path switching value output isolation driving circuit, an 8-path analog quantity input buffer circuit and a 2-path V/F converting circuit. The present invention also comprises the steps of the mathematical model establishing method for the prediction of the viscosity and the dispersity of rubber compounds. The present invention has the advantages of online detection and control in real time, high product qualification ratio and greatly reduced cost.

Description

Viscosity and dispersity estimator for banburying mixer mixing rubber and mathematical model establishing method thereof

The present invention is viscosity and dispersity estimator for banburying mixer mixing rubber and mathematical model establishing method thereof, belongs to mixer mixing glue automatic detection technology of parameter control technology.

Banbury is softened rubber increasing sizing material plasticity, to reduce a kind of equipment of compound viscosity, and it is a kind of equipment of rubber or plastics and solid particulate filler or liquid mixed milling, plasticizing, dispersion. Banbury to the materials such as rubber plastic and solid particulate filler or liquid material mixed milling, plasticizing, dispersing technology after, generally product is had the requirement that reaches certain viscosity and decentralization. Whether reality reaches requirement, and existing method is after the banbury processing technology, from the product of producing, get certain sample and carry out the quality testing of viscosity and decentralization in the laboratory, as nonconforming, the just process reprocessing that can reprocess, sampling detects again, and whether see qualified again. Can not reprocess, just can only process or scrap. For this reason, product need be parked a period of time general about 24 hours in the workshop toward contact. The artificial the problems referred to above that solve the prior art existence of invention, once set up the Mathematical Modeling of prediction compound viscosity and decentralization by every kind of sizing material at banbury, predict by the quality automatic on-line detecting system, obtain in actual applications good effect, when just the mixing sizing material kind of banbury is a lot, at this moment the workload of setting up prediction compound viscosity and decentralization Mathematical Modeling is larger, thereby has influence on applying of it.

Purpose of the present invention is exactly in order to overcome and to solve existing banbury product compound viscosity, dispersion parameter detection need in the laboratory sampling Detection, both loaded down with trivial details, time-consuming, make again product percent of pass low, need often do over again or scrap, shortcoming and the problem of product cost increase etc., a kind of material to banbury production of research invention need not be sampled to the laboratory and detect, and can detect, control by real-time online, make product percent of pass is high, cost reduces greatly viscosity and dispersity estimator for banburying mixer mixing rubber and mathematical model establishing method thereof.

The present invention realizes by following technical proposals: the structural representation of viscosity and dispersity estimator for banburying mixer mixing rubber as shown in Figure 1, its circuit block diagram as shown in Figure 2, its circuit theory diagrams are as shown in Figure 3. It is connected and composed jointly by Industry Control machine enclosure 1, Industry Control level cpu motherboard 2, passive mother board 3, PC104 bus signals tablet 4, signal input conditioning plate 5, hard disk 6, disk drive controller 7, display 8 and printer 9, its mutual alignment and annexation are: PC104 bus signals tablet 4 is installed on the Industry Control level cpu motherboard 2 by the PC104 bus hub, 2 of cpu motherboards are fixed on the passive mother board 3 by the isa bus slot, be electrically connected mutually with hard disk 6 and floppy disk 7 by flat signal cable again, signal input conditioning plate 5 is fixed on the passive mother board 3 by the isa bus slot, and passive mother board 3 is fixed in the Industry Control machine enclosure 1; The display socket, printer socket, the analog input socket that expand respectively by above-mentioned each plate are electrically connected mutually with display, printer and banbury main motor current transmitter, elastomeric compound sizing material glue temperature transmitter; Its circuit (the high reason of signal input plate) is made of 24 way switch amount input isolation circuits, 24 way switch amounts output isolated drive circuit, 8 road analog input buffer circuits, the common electrical connection of 2 road V/F change-over circuits, and its interconnected relationship is: 24 way switch amount input isolation circuits are electrically connected mutually with industry spot switching value input circuit and PC104 bus tablet respectively; 24 way switch amounts output isolated drive circuit is electrically connected mutually with industrial field control object and PC104 bus signals tablet respectively, 8 road analog input buffer circuits are electrically connected mutually with industry spot simulated measurement input circuit, PC104 bus signals tablet respectively, and 2 road V/F change-over circuits are electrically connected mutually with industry spot simulated measurement input circuit, PC104 bus signals tablet respectively. Its action principle is: signal input conditioning plate 5 is with the switching value signal in the banbury production process, the mixing active power signal of doing of banburying owner motor, sizing material variations in temperature signal, behind photoelectricity isolation, Hyblid Buffer Amplifier, be input to PC104 bus signals tablet 4, be input to cpu motherboard through PC104 bus signals tablet 4 again. Computer carries out record analysis according to these input signals to the banburying process of banbury, foundation returns out corresponding Mathematical Modeling, calculate viscosity, decentralization predicted value, and mixing process is controlled, to reach the purpose of pre-viscosimetric, decentralization and control mix quality.

In the circuit theory diagrams, wherein a certain way switch amount input isolation circuit is connected and composed jointly by resistance R 19～R20, photoelectric coupled device GDQ2 as shown in Figure 3; A certain way switch amount output isolated drive circuit is by resistance R 21～R23, and photoelectric coupled device GDQ3, transistor T1～T2, diode D4, relay J are electrically connected formation jointly; A certain road analog input buffer circuit is by resistance R 1～R7, operational amplifier U1A, U2A, and adjustable resistance RW1, two utmost point voltage-stabiliser tube DW1～DW2 are electrically connected formation jointly; A certain road V/F change-over circuit is by resistance R 8～R18, operational amplifier U3A～U4A, V/F switching device U5, adjustable resistance RW2～RW4, two utmost point voltage-stabiliser tube DW3～DW4, diode D1～D3, capacitor C 1～C3, photoelectric coupled device GDQ1 is electrically connected formation jointly. The simple style of work principle of circuit is as follows: its action principle of switching value input isolation circuit is when the on-the-spot switching value of industrial production is connected, infrared photodiode among the photoelectric coupled device GDQ2 just has electric current to flow through, infrared phototriode conducting, its colelctor electrode is to computer export one low level; The action principle of switching value output isolated drive circuit is: in the time will exporting a control signal, computer export one low level, infrared light diode among the LMDS Light Coupled Device GDQ3 just has electric current to flow through, infrared phototriode conducting, colelctor electrode is exported a low level, and this low level is so that transistor T1 cut-off, the colelctor electrode of T1 is exported a high level, this level is so that transistor T2 conducting, and control relay gets electric, exports a control signal; Its action principle of analog input buffer circuit is the analog voltage signal at banbury scene inputs to operational amplifier through resistance R 1 negative terminal, because R4=R1, R5=R7, first utmost point of this amplifier, the second utmost point multiplication factor are 1, adjusted by adjustable resistance RW1 the zero point of analog quantity, to voltage signal of 1: 1 of computer export; Its action principle of V/F change-over circuit is that the analog voltage signal at banbury scene becomes 0～10V to this voltage through OP amplifier U3A, and adjusted by adjustable resistance RW4 the zero point of analog quantity, and the V-F conversion is finished by U5. Linear in order to change, increased OP amplifier U4A, it is that electric current is imported, full scale is 0～100MA, in order to guarantee the stable capacitor C 2 that increased of backfeed loop, the frequency of U5 is decided by resistance R 16, capacitor C 3 or adjustable resistance RW3, and it is up to the isolated frequency signal of 10KHZ to computer exportable.

The method for building up of banbury prediction compound viscosity decentralization Mathematical Modeling: with parameters such as the temperature in the viscosity and dispersity estimator for banburying mixer mixing rubber collection mixing process of the present invention, time, power, energy, when setting up Mathematical Modeling, above-mentioned parameters in 2～6 kinds of rubber compounding processes of minimum collection, the sample lot number that each sizing material gathers is more than at least 15 batches, total is between 40～150 batches, and the pressure and other parameters of the activity coefficient when gathering every kind of rubber compounding, closed-smelting machine rotor rotating speed, ram; Again above-mentioned sampling sample is detected viscosity and the decentralization of material in the laboratory by national standard, and testing result is inputted compound viscosity decentralization prediction unit of the present invention; It is as follows that it sets up the Mathematical Modeling step: (1) gets rid of the exceptional value of sample data: from the statistics Xue Zhi, such as the variable X Normal Distribution, then 99% value should drop in [μ-3 σ, μ+3 σ], if drop on outside this scope, then these data worked as outlier exclusion; (2) carry out correlation analysis: the relation of seeing viscosity or decentralization and relevant parameters from coefficient correlation; (3) set up Mathematical Modeling: take viscosity or decentralization as dependent variable, return take activity coefficient, rotating speed etc. as independent variable and to set up Mathematical Modeling; (4) carry out relevant check: comprising the check of conspicuousness F, the check of coefficient R, residual test: if by check, then regression equation is set up; If do not pass through, then revise Mathematical Modeling, continue regression testing, until mathematical prediction model is set up; (5) prediction compound viscosity and decentralization: after setting up mathematical prediction model, under at monitoring condition, during the every mixing a collection of end of banbury, can provide viscosity and the decentralization numerical value of this batch elastomeric compound; As under controlled conditions, the viscosity of this batch elastomeric compound and decentralization value reach in the scope of required value, and banbury just finishes the mixing of this batch. The Mathematical Models flow diagram of banbury prediction compound viscosity, decentralization as shown in Figure 4.

The principle of prediction compound viscosity is: from the rheology theory analysis of banbury, the expression formula of closed-smelting machine rotor torque T is: $T=\frac{\mathrm{\πN\μ}}{2}[1+\frac{x}{a}(\frac{\mathrm{\πx}}{e}-\frac{1}{D_c}-\frac{3\mathrm{\π}(x^2{a}^2-1)}{1+\frac{a^{3}e}{x{D}_t}}]----\left(1\right)$ In the formula: N is rotor speed, and μ is compound viscosity, and a is that lug, nib is risen trench gap and compared h/g; X is that rotor axle body diameter/rotor rib top diameter compares D_c/D _t E is the lug, nib top width. For specific banbury, a, x, e, D_c、D _tDeng being constant, if replace numerical value in the formula bracket with A, then 1. formula can be reduced to: $T=\frac{\mathrm{\πN\μ}}{2}A----\left(2\right)$ And the armature spindle power P is: $P=\mathrm{TN}=\frac{\mathrm{\π}{N}^2\mathrm{\μ}}{2}A$ $\mathrm{\μ}=\frac{2P}{\mathrm{\π}{N}^{2}A}----\left(3\right)$ From then on 3. in the formula as seen, viscosity, mu is only relevant with P, N, A. In addition, the different not direct relations of viscosity and every kind of prescription. From above analysis, we think that from the angle of every banbury the Mathematical Modeling that adopts Multiple Regression Analysis Method to set up the prediction compound viscosity is possible. Unsatisfactory at first in the practical application. After further study, consider in the hypothesis of derivation following formula, material is full of whole banburying chamber, and actual and underfill, therefore should introduce the coefficient of fullness of banburying chamber, be the material filling coefficient in the practical application, after introducing activity coefficient, return the Mathematical Modeling of setting up and obtained satisfied result.

The principle of prediction elastomeric compound decentralization: according to plastic fluid mixes with the solids filler dispersion mechanism and etc. the deflection theory, suppose that plastic fluid and solids are by certain thickness minute multilayer placement, under external force plastic fluid distortion, solids are corresponding displacement also, larger such as distortion, then the thickness of plastic fluid and solids is more and more less, be that distance between plastic fluid and between solids is more and more less, that is to say that deflection is larger, it is better to mix dispersion, and certain such as plastic fluid viscosity, deflection is larger, the external force work done is just larger, and consumed energy is more. Therefore, how many sizes of decentralization should be directly proportional with catabiotic. But because rubber is not plastic fluid, but viscoelastic fluid, this solids of carbon black are not loose existence, exist but form carbon black agglomerate with certain agglomeration power each other, under external force, the viscoelastic fluid rubber deformation is after external force is removed, exist elasticity in various degree to recover, but carbon black agglomerate not necessarily produce corresponding displacement. So decentralization is only undesirable with the model that energy consumption is set up, according to the rubber milling of internal mixer rheological theory, the distortion that only produces greater than carbon black agglomerate agglomeration power, could be to being dispersed with effect, the energy in therefore consuming should deduct the part energy that consumes less than carbon black agglomeration power. The decentralization of carbon black should be relevant with the watt level in a period of time, and power is large, and then decentralization is all right. According to this, the realization that just is successful on the whole of the Mathematical Modeling of setting up prediction elastomeric compound carbon black dispension degree.

The present invention has following advantage and useful effect compared with prior art: 1. the material of existing banbury production can be qualified, also can be underproof, or even waste product. And to detect just through instrument additional on laboratory or the production line and know. Be exactly the technology by every kind of sizing material modeling and forecasting that the inventor once invented, also still will set up the sizing material of forecast model and could predict, and workload be larger. And the banbury of employing the technology of the present invention, the material of its institute's processing is just known viscosity and the decentralization of the material that it produces when end is mixing. Under controlled conditions, after mixing material reaches its quality requirement, just can finish mixing; 2. use the banbury of the technology of the present invention, it is all measurable to need only any sizing material mixing on this banbury. The present invention has overcome described shortcoming and problem that prior art exists fully, the material of the banbury production of adopting the technology of the present invention is not needed to be sampled to the laboratory again to be detected, and can detect, control by real-time online, therefore, product percent of pass improves, cost reduces greatly, and applying of it will bring very large economic benefit and social benefit.

The below further specifies as follows to Figure of description: Fig. 1 is the contour structures schematic diagram that banbury is set up mathematical model prediction compound viscosity, decentralization device; Fig. 2 is its circuit working functional-block diagram; Fig. 3 is its circuit theory diagrams; Fig. 4 is the Mathematical Models program circuit block diagram of banbury prediction compound viscosity, decentralization; Fig. 5 is the Normal P-P of residual error during mixer mixing adhesiveness mathematical prediction model is set up.

Embodiments of the present invention can be divided into the enforcement of compound viscosity of the present invention, decentralization prediction unit and the enforcement of Mathematical Models: the enforcement of (one) compound viscosity, decentralization prediction unit: (1) is by shown in Figure 3, draw the printed circuit board (PCB) of signal input conditioning plate, then screen components and parts and install and simple debugging. For example: U1, U2 can select LM324 type operational amplifier; The optional TL082 type of U3, U4 operational amplifier; The optional LM331 type of U5 V/F changes integrated device; The optional 4N25 type of GDQ1 photoelectric coupled device; The optional TL117 type of GDQ2 photoelectric coupled device; (2) by Fig. 1, design shown in Figure 2 and adopt machine-tooled method processing process industry controller cabinet 1 processed or choose the 1PC-65 type; Choose again required plate, part, for example: the optional PCA-6153 type of Industry Control cpu motherboard; The optional PCA-6108 type of passive mother board; The optional ADT200 type of PC104 bus signals tablet; The optional Quantum 3.2AT of hard disk; The optional SONY1.44 type of disk drive controller; The optional PH1L1PS105A type of display; Optional National 1121 types of printer; Then by the described interconnected relationship of top specification connection is installed, just can implements preferably compound viscosity of the present invention, decentralization prediction unit; (2) enforcement of the method for building up of the Mathematical Modeling of mixer mixing adhesiveness, decentralization prediction as long as undertaken by the described method step of top specification, just can be implemented preferably. The inventor is through in a few years research and the test of production practices, and existing a lot of successful embodiment, and having implemented a period of time in production practices prove very successfully, and larger economic benefit is arranged. The below is only for 3 embodiment:

The foundation of embodiment 1:BB370 type banbury viscosity mathematical prediction model; (1) BB370 type banbury gathers the relevant parameters of every batch of mixing process by its micro computer monitoring or control device, and present embodiment only gathers the instantaneous power of every batch of binder removal time; Rotating speed and activity coefficient when (2) searching every kind of rubber compounding from prediction unit of the present invention, present embodiment have three kinds of sizing materials to gather altogether 86 batch samples: 27 samples of the first sizing material, and rotating speed is 35 rev/mins, activity coefficient is 0.6813; 36 samples of the second sizing material, rotating speed is 35 rev/mins, activity coefficient is 0.6640: 23 samples of the third sizing material, rotating speed is 40 rev/mins, activity coefficient is 0.6933; (3) to above-mentioned 86 batch samples, detect, and input in the prediction unit of the present invention by standard GB/T/T1232-92 " half finished rubber Mooney viscosity mensuration " standard; (4) data to collecting check whether there is abnormal data. From statistics as can be known, if the variable X Normal Distribution, then 99% value should drop in [μ-3 σ, μ+3 σ], if drop on outside this scope, then suspects it is abnormal data, should get rid of; (5) by the predicting model of following structural form $\mathrm{\μ}=\frac{P}{N^2}\×1000\×(C_1+C_2/\sqrt{r}+C_3/r^2)$

In the formula: μ is Mooney Viscosity of Rubber Mix; P is binder removal point instantaneous power; N is rotor speed; R is activity coefficient; Following formula is carried out the nonlinear regression process, will obtain constant C₁、C ₂、C ₃Estimated result tabulation as shown in table 1. Therefore the forecast model of compound viscosity is: $\mathrm{\μ}=\frac{P}{N^2}\×1000\×(-138729+15144/\sqrt{r}-20681.1/r)$ Table 1: the result of nonlinear regression

	The loss of μ: (observation-predicted value)2Total losses=209.50950, R=0.87014, the variance of explanation=75.715%
		Sample size N=86	C 1                    C 2                   C 3
Standard error of estimate t (83) P-value	-138729             151444                 -20681.1 20237.13             22133.54               3053.493 -6.85519             6.842285               -6.77294  0.00000             0.000000                0.00000

(6) coefficient of multiple correlation R of this model=0.87014 represents that our model and the effect of 86 sample data matches are relatively good. And it can also be seen that P/N from table 1², r is all more remarkable on the impact of μ. Can pass through conspicuousness F, coefficient R and residual test; (7) for the compound viscosity mathematical prediction model of setting up is tested, in BB370 type mixer mixing process, gather again 18 batch samples, in this 18 batch sample: 9 batches in the first glue, 6 batches in the second glue, 3 batches in the third glue, and handle is surveyed Mooney viscosity value and predicted value compares, and is shown in side by side table 2:

Table 2: compound viscosity prediction and residual analysis

Catalogue number(Cat.No.)	Binder removal power activity coefficient rotary speed measure value prediction value residual error P γ N
		1     2     3     4     5     6     7     8     9     10     11     12     13     14     15     16     17     18	0.368    0.6813      35      58.5     58.124     0.376     0.370    0.6813      35      60.8     58.440     2.360     0.376    0.6813      35      59.5     59.338     0.112     0.376    0.6813      35      58.6     59.388    -0.778     0.374    0.6813      35      61.5     59.072     2.428     0.372    0.6813      35      60.5     58.756     1.744     0.368    0.6813      35      58.4     58.124     0.276     0.374    0.6813      35      59.0     59.072    -0.072     0.382    0.6813      35      60.8     60.336     0.464     0.342    0.6640      35      57.7     60.404    -2.704     0.344    0.6640      35      58.1     60.757    -2.657     0.352    0.6640      35      59.6     62.170    -2.570     0.358    0.6640      35      60.1     63.230    -3.130     0.362    0.6640      35      61.5     63.936    -2.436     0.344    0.6640      35      58.3     60.757    -2.457     0.412    0.6933      40      56.4     56.389     0.011     0.418    0.6933      40      55.7     56.488    -0.788     0.420    0.6933      40      57.5     57.484     0.016

As seen from Table 2: the mean difference of predicted value and measured value comparison is 0.4105 Mooney point, and the predicted value of difference in 2.5 Mooney points has 14, accounts for 77.78%; Difference has 3 within 2.5～3 Mooney points, account for 16.67%; Difference has 1 at 3～3.5 Mooney points, accounts for 5.56%; Standard-required is within ± 3 Mooney points, and as seen predicting the outcome of this model can formally be dropped into utilization.

The foundation of embodiment 2:BB270 type banbury viscosity mathematical prediction model: (1) BB270 type banbury gathers the mixing process parameter by its microprocessor-based monitoring device; 47 batches of binder removal point instantaneous powers, wherein: 28 batches in the first glue, 19 batches in the second glue; (2) detect, the activity coefficient of two kinds of glue is changed in the prediction unit of the present invention in the lump by standard GB/T/T1232-92 " half finished rubber Mooney viscosity mensuration " standard; (3) obey normal state by data and announce, should have 99% value should drop in [μ-3 σ, μ+3 σ] scope, get rid of this extraneous exceptional value; (4) determine coefficient correlation, its correlation matrix is as shown in table 3 below, exists obvious linear relationship between the visible Mooney viscosity μ of coefficient correlation shown in the table 3 and power P, activity coefficient γ;

Table 3: correlation matrix table

Variable	P        γ                   μ
		Power P activity coefficient γ Mooney viscosity μ	1.000    -0.842     0.925   -0.842     1.000    -0.892    0.925    -0.892     1.000

(5) Mathematical Modeling is set up in recurrence:

Table 4: regression analysis

Sample size N=47	R＝0.9486 R 2=0.8998 adjusts rear R=0.8953 R2The standard error that=0.8016 F (2,44)=197.66 estimates: 1.4840
							The Beta value	The standard error of Beta value	Estimates of parameters B	The standard error of B value	T value (44)	P value level
	Intercept			370.509	86.7071	4.2731							0.00102
P							0.5996	0.0885	95.776	14.1401	6.7734	0.00000
	FA	-0.3868	0.0885	-464.455	106.2859	-4.36986							0.0075

Can get μ=370.509+95.77P-464.455 γ from regression analysis: the inspection of (6) regression mathematical model: from variance analysis as seen from Table 5, regression model is very remarkable. Polynary coefficient of determination R²=0.8998, R after adjusting²=0.8016, very strong linear relationship is arranged between Mathematical Modeling independent variable and the dependent variable;

Table 5: variance analysis

	Quadratic sum	All square	The free degree	The F value	Critical value F (2,44)	Conspicuousness
							Regression error	870.6287   96.9053	435.3143   2.2024	2    44	197.66	5.18	**
Summation	967.5340

The simultaneously check of visible regression parameter from table 4, from the P value level of T statistic as seen, each regression parameter is all very remarkable, and the Normal P-P of residual error (being that Q-Q schemes) is as shown in Figure 5. As seen from Figure 5, this figure almost point-blank, show the basic Normal Distribution of residual error, meet to return and suppose, relevant check all can be passed through: (7) forecast test: 15 batches of BB270 type banbury prediction samplings, wherein: 8 batches in the first glue, 7 batches in the second glue, and detect its Mooney viscosity, and compare with predicted value, the result is shown in shown in the table 6:

Table 6: predicted value and measured value are relatively

Catalogue number(Cat.No.)	P	FA	REV	Measured value	Predicted value	Residual error	Relative error
								1	0.4768	0.7599	40	62.000	63.236	-1.236	0.01994
2	0.4736	0.7599	40	63.100	62.930	0.170	0.11269
								3	0.4592	0.7599	40	61.800	61.550	0.250	0.00405
4	0.4896	0.7599	40	62.800	64.462	-1.662	0.02646
								5	0.5152	0.7599	40	67.600	66.914	0.686	0.01015
6	0.5072	0.7599	40	65.400	66.148	-0.748	0.01144
								7	0.4992	0.7599	40	63.200	65.381	-2.181	0.03451
8	0.4704	0.7599	40	61.200	62.623	-1.423	0.02325
								9	0.4144	0.7676	40	55.540	57.260	-1.720	0.03097
10	0.4688	0.7599	40	62.100	62.470	-0.370	0.00596
								11	0.4208	0.7676	40	51.830	54.296	-2.466	0.04758
12	0.4384	0.7676	40	56.060	55.982	0.076	0.00136
								13	0.4336	0.7676	40	52.570	55.522	-2.952	0.05615
14	0.4432	0.7676	40	55.770	56.442	-0.672	0.01205
								15	0.4544	0.7676	40	59.260	57.514	1.746	0.02946

As seen from Table 6, the difference of predicted value and measured value with interior have 7, accounts for 46.67% at 1 Mooney point; Difference accounts for 33.33% have 5 of 1～2 Mooney point; Difference accounts for 20% have 3 of 2～3 Mooney points; Mean difference is 2.1%, and difference is 4.9% of pre-viscosimetric average, and visible this mathematics model of setting up can come into operation.

Embodiment 3: the foundation of elastomeric compound decentralization mathematical prediction model: (1) adopts microcomputer control or supervising device, gather the relevant parameters of mixer mixing process: sizing material temperature K amounts to 78 batch samples when binder removal point instantaneous power P, mixing total time T, gross energy E, binder removal, and finds mixing rotating speed V, filling remainder FA and the decentralization FSD that records by national standard in the lump the input control device from control device; (2) get rid of exceptional value: because negative appears in the 38th sample, and decentralization is divided into 1 to 10 grade by standard, so think that these data are exceptional value, should leave out: (3) carry out correlation analysis: correlation matrix is as shown in table 7, because 78 batch samples that gather, rotating speed is 60 rev/mins, so put aside the impact of the rotating speed factor in this mathematics model. From table 7 coefficient correlation can, the correlation between decentralization FSD and mixing gross energy E, total time T, the activity coefficient FA is larger, comparatively speaking, the correlation between decentralization FSD and power and the temperature is then poorer.

Table 7: correlation matrix

Variable	FSD	E	T	K	P	FA
							FSD	1.000	-0.699	-0.520	-0.482	-0.415	-0.776
E	-0.699	1.000	0.621	0.718	0.491	0.844
							T	-0.520	0.621	1.000	0.001	-0.001	0.356
K	-0.482	0.718	0.001	1.000	0.676	0.751
							P	-0.415	0.491	-0.100	0.676	1.000	0.692
FA	-0.775	0.844	0.356	0.751	0.692	1.000

(4) foundation of mathematical prediction model: to decentralization FSD, do linear regression with 5 independents variable such as above P, T, E, K, FA, can obtain the result shown in following table table 8:

Table 8: decentralization regression model result

Sample size N=74	R＝0.8562 R 2=0.7332 adjusts rear R=0.8447 R2The standard error that=0.7135 F (5,68)=37.365 estimates: 0.6710
							The Beta value	The standard error of Beta value	Estimates of parameters B	The standard error of B value	T value (44)	P value level
	Cut square			66.3215	7.5648	8.7672							0.00000
E							0.3288	0.2264	0.1308	0.0901	1.4523	0.15101
	T	-0.3714	0.1483	-0.0378	0.0151	-2.5039							0.01468
K							-0.0900	0.1562	-0.0172	0.0298	-0.5762	0.56637
	P	0.0601	0.1013	2.8064	4.7338	0.5928							0.55526
FA							-0.9308	0.1447	-94.3048	14.6678	-6.4264	0.00000

As seen from Table 8, decentralization FSD and factor E, K, P are all not remarkable, and therefore, for the prediction of decentralization, this mathematics model is false; (5) revise the new independent variable of establishment, set up new Mathematical Modeling: through relatively, factor E and factor T are done with down conversion, obtain new independent variable AP, AP=E/T

Take AP and FA as independent variable, do the curve regression analysis, obtain result as shown in table 9 below:

Table 9: the decentralization regression model result after the conversion

Sample size N=74	R＝0.8534 R 2=0.7282 adjusts rear R=0.8489 R2The standard error that=0.7206 F (2,71)=95.118 estimates: 0.6627
							The Beta value	The standard error of Beta value	Estimates of parameters B	The standard error of B value	T value (44)	P value level
	Cut square			60.8380	4.1865	14.5319							0.00000
FA							-0.9641	0.7213	-97.7178	7.3102	-13.3673	0.00000
	AP	0.2853	0.7326	22.0627	5.5781	3.9553							0.00018

The gained regression equation is: $\mathrm{FSD}=60.838-97.7178\mathrm{FA}+22.0627\frac{E}{T}$

As seen from Table 9: the polynary coefficient of determination R that obtains after the recurrence²=0.7282, F statistic and t statistic show that equation and regression coefficient are all very remarkable; (6) this mathematics model has passed through check, can drop into application.

Claims (2)

1, a kind of display that comprises, the viscosity and dispersity estimator for banburying mixer mixing rubber that printer forms, it is characterized in that: it is by Industry Control machine enclosure (1), Industry Control level cpu motherboard (2), passive mother board (3), PC104 bus signals tablet (4), signal input conditioning plate (5), hard disk (6), disk drive controller (7), display (8) and printer (9) connect and compose jointly, its mutual alignment and annexation are: PC104 general line signal tablet (4) is installed on the Industry Control level cpu motherboard (2) by the PC104 bus hub, cpu motherboard (2) then is fixed on the passive mother board (3) by the isa bus slot, be electrically connected mutually with hard disk (6) and floppy disk (7) by flat signal cable again, signal input conditioning plate (5) is fixed on the passive mother board (3) by the isa bus slot, and passive mother board (3) is fixed in the Industry Control machine enclosure (1); The display socket, printer socket, the analog input socket that expand respectively by above-mentioned each plate are electrically connected mutually with display, printer and banbury main motor current transmitter, elastomeric compound sizing material glue temperature transmitter: its circuit is made of 24 way switch amount input isolation circuits, 24 way switch amounts output isolated drive circuit, 8 road analog input buffer circuits, the common electrical connection of 2 road V/F change-over circuits, and its interconnected relationship is: 24 way switch amount input isolation circuits are electrically connected mutually with industry spot switching value input circuit and PC104 bus tablet respectively; 24 way switch amounts output isolated drive circuit is electrically connected mutually with industrial field control object and PC104 bus signals tablet respectively, 8 road analog input buffer circuits are electrically connected mutually with industry spot simulation volume input circuit, PC104 bus signals tablet respectively, and 2 road V/F change-over circuits are electrically connected mutually with industry spot simulated measurement input circuit, PC104 bus signals tablet respectively.

2, a kind of method of Mathematical Models of banbury prediction compound viscosity decentralization: it is characterized in that: with temperature, time, power, the energy parameter in the viscosity and dispersity estimator for banburying mixer mixing rubber collection mixing process claimed in claim 1, when setting up Mathematical Modeling, above-mentioned parameters in 2～6 kinds of rubber compounding processes of minimum collection, the sample lot number that each sizing material gathers is more than at least 15 batches, total is between 40～150 batches, and the pressure parameter of the activity coefficient when gathering every kind of rubber compounding, closed-smelting machine rotor rotating speed, ram; Again above-mentioned sampling sample is detected viscosity and the decentralization of special material in the laboratory by national standard, and testing result is inputted compound viscosity of the present invention disperse prediction unit; It is as follows that it sets up the Mathematical Modeling step: (1) gets rid of the exceptional value of sample data: from the statistics Xue Zhi, such as the variable X Normal Distribution, then 99% value should drop in [μ-3 σ, μ+3 σ], if drop on outside this scope, then these data worked as outlier exclusion; (2) carry out correlation analysis: the relation of seeing viscosity or decentralization and relevant parameters from coefficient correlation; (3) set up Mathematical Modeling: take viscosity or decentralization as dependent variable, set up Mathematical Modeling take activity coefficient, rotating speed as independent variable returns; (4) carry out relevant check: comprising the check of conspicuousness F, the check of coefficient R, residual test; If by check, then regression equation is set up; If do not pass through, then revise Mathematical Modeling, continue regression testing, until mathematical prediction model is set up; (5) prediction compound viscosity and decentralization; After setting up mathematical prediction model, under at monitoring condition, during the every mixing a collection of end of banbury, can provide viscosity and the decentralization numerical value of this batch elastomeric compound; As under controlled conditions, the viscosity of this batch elastomeric compound and decentralization numerical value reach in the scope of required value, and banbury just finishes the mixing of this batch.