CN1421545A - Variable-discourse universe adaptive fuzzy control method and device for pre-roasted electrolytic aluminium production - Google Patents

Abstract

The present invention is variable-discourse universe adaptive fuzzy control method and device for pre-roasted electrolytic aluminium production. The device includes voltage detecting unit to detect the voltage variance of electrolytic tank; upper computer to set alumina concentration and perform concentrated control of technological operation and material feeding of the electrolytic tank; electrolytic tank controlling computer to receive the set alumina concentration value from the upper computer, perform variable-discourse universe fuzzy control calculation based on the voltage variance to obtain the current alumina feeding interval time; and executing mechanism to control the alumina feeding amount based on the received alumina feeding interval time.

Description

The change domain adaptive fuzzy control method and the device of prebake Aluminum Electrolysis Production

Technical field

The present invention relates to the production control method and the device of electrolytic aluminum, relating to a kind of particularly is that raw material uses prebaked cell to produce the change domain adaptive fuzzy control method and the device of electrolytic aluminum with the aluminum oxide.

Background technology

Since the eighties, the electrolytic aluminum industrial realization extra normal development, the range of application of aluminium constantly enlarges, and the demand of aluminium is also enlarged day by day.The main raw material of Aluminum Electrolysis Production is aluminum oxide, anode block.It is few that at present China's electrolytic aluminum is taked advanced large-scale anode prebaked cell factory, and the market competitiveness is very weak, therefore takes modern technique, expands the scale of production to reduce pollution to environment, and the market outlook of building large-scale electrolysis aluminium manufacturer are very wide.

Generally, prebake electrolytic aluminum and a high current efficiency in order to obtain a stably manufactured need be controlled at 1.2%～3.5% with alumina concentration in the electrolyzer.The method of alumina concentration generally is to utilize the electric current and the voltage of the electrolyzer that can survey in traditional control electrolyzer, revise the blanking timed interval (NB) of aluminum oxide, the concentration that makes aluminum oxide is in a lower level, also the purpose of anode effect (AE) does not take place in control to reach.Yet pre-calcining electrolytic cell is a kind of nonlinear and time-varying system of complexity, its model has uncertainty, and because the on-line parameter that can measure has only the electric current and the voltage of electrolyzer, so remain in a lower level for the concentration that will control, the effect that traditional conventional control method does not all obtain with aluminum oxide.

The control problem that solves above-mentioned prebaked cell that develops into of intelligent fuzzy control technique provides favourable theoretical basis.General fuzzy control does not need the mathematical model of controll plant, by gathering skilled labor's control experience, forms fuzzy rule base, and the simulating human expert controls prebaked cell fully.But the fuzzy interpolation control method under the general domain, fuzzy rule base complexity, fuzzy rule even can reach more than 1000, therefore computing is complicated, the hysteresis delay of control is bigger, and bigger to human expert's experience dependence, thereby lower to the adaptability of electrolyzer.

Summary of the invention

Therefore, the purpose of this invention is to provide control method and device that a kind of employing that is used for the prebake Aluminum Electrolysis Production becomes the Adaptive Fuzzy Control of domain, by this control method and device, can simplify fuzzy rule base, operand is descended greatly, control hysteresis postpones to reduce simultaneously, and dependence expertise within reason, thereby has improved the adaptability to electrolyzer.

For realizing above-mentioned purpose of the present invention, a kind of change domain Adaptive Fuzzy Control device of prebake Aluminum Electrolysis Production is provided, comprising: voltage check device, the voltage change of detection electrolyzer; Upper computer is set the set(ting)value of alumina concentration, and the technological operation and the raw material dispensing of electrolyzer carried out centralized Control; Slot control machine, reception is from the set(ting)value of the alumina concentration of upper computer, and the voltage change Δ U that uses electrolyzer becomes domain fuzzy control calculating, draw the time t of the blanking interval NB of current aluminum oxide, topworks, according to the blanking that calculates by slot control machine that the is received time t of NB at interval, the blanking amount of controlled oxidation aluminium.

For realizing above-mentioned purpose of the present invention, a kind of change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production is provided, comprise step: upper computer carries out centralized Control to electrolyzer, and is alumina concentration set(ting)value of electrolyzer setting; Slot control machine receives the alumina concentration set(ting)value set, and the voltage change Δ U that gathers from the electrolyzer cycle, and the span of Δ U is [U _a, U _b]; Reception is from the set(ting)value of the alumina concentration of upper computer, and the voltage change Δ U that uses electrolyzer becomes the calculating of domain Adaptive Fuzzy Control, draw the time t of the blanking interval NB of current aluminum oxide, the blanking time of actuating mechanism controls electrolyzer is to adjust the concentration of aluminum oxide; Slot control machine feeds back to upper computer with the state of electrolyzer such as voltage, strength of current and the temperature of superheat etc. of electrolyzer by the cycle.

Brief Description Of Drawings

To the detailed description of the preferred embodiments of the present invention, it is clearer that above-mentioned purpose of the present invention and advantage will become by with reference to the accompanying drawings, wherein:

Fig. 1 is the change domain Adaptive Fuzzy Control device according to prebake Aluminum Electrolysis Production of the present invention;

Fig. 2 is the detailed composition structure iron of slot control machine 2 among Fig. 1;

Fig. 3 is the oscillogram of the membership function that adopts of present embodiment; And

Fig. 4 is the schema according to the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production of the present invention.

Embodiment

Specifically describe the preferred embodiments of the present invention below by accompanying drawing.

Fig. 1 is the change domain Adaptive Fuzzy Control device according to prebake Aluminum Electrolysis Production of the present invention.As shown in Figure 1, the voltage change Δ U of electrolyzer 1 is sent to slot control machine 2 through the voltage check device (not shown).The set(ting)value that slot control machine 2 receives from the alumina concentration of upper computer 3, by seeing (flame color, liquid level sharpness), survey (electrolysis temperature, molecular ratio, two levels), touch (thickness of furnace wall), calculating (five degree/day trend) etc., the set(ting)value of alumina concentration is done in the choosing of the alumina concentration that determines to upper computer 3 with the technologist.Slot control machine 2 uses the voltage change Δ U of electrolyzer 1 to become the domain fuzzy control and calculates, draw the time t of the blanking interval NB of current aluminum oxide, and the time t that calculates is sent to topworks 4, topworks 4 can adopt such as super concentrated phase and carry aluminum oxide device, pneumatic conveying aluminum oxide device or motor-driven carrier to carry the aluminum oxide device.Topworks 4 is according to the blanking amount of the time t controlled oxidation aluminium of blanking interval NB.The technological operation and the raw material dispensing of 3 pairs of electrolyzers 1 of upper computer carry out centralized Control.

Fig. 2 is the detailed composition structure iron of slot control machine 2.Slot control machine 2 comprises former piece contraction-expansion factor calculating device 21, control calculating device 22, consequent contraction-expansion factor calculating device 23.Wherein former piece contraction-expansion factor calculating device 21 calculates the former piece contraction-expansion factor that becomes in the domain fuzzy control according to current voltage change Δ U, and the former piece contraction-expansion factor that is calculated is sent in the control calculating device 22.Control calculating device 22 calculates the input variable that becomes the interpolation fuzzy control under the domain according to the former piece contraction-expansion factor, and calculates the preliminary blanking time t of NB at interval.With the preliminary blanking of being calculated at interval the time t of NB be provided to consequent contraction-expansion factor calculating device 23, consequent contraction-expansion factor calculating device 23 calculates the consequent contraction-expansion factor that becomes in the domain fuzzy control according to current voltage change Δ U and the preceding blanking time interval value that once calculates, and the consequent contraction-expansion factor be multiply by preliminary blanking t pitch time that is calculated, obtain the revised blanking time t of NB at interval.

The change domain fuzzy control that is described in detail in the slot control machine below to be carried out is calculated.

The input variable that becomes the domain adaptive fuzzy controller in slot control machine 2 according to the present invention only is the voltage change Δ U of electrolyzer, and output variable is the time t of the blanking interval NB of aluminum oxide.If [U _a, U _b] be the span of Δ U (be called and import domain), A ₁, A ₂..., A ₇Be input domain [U _a, U _b] on fuzzy division, fuzzy set can be chosen membership functions such as " choppy sea ", " square wave ", " Gaussian " here.The fuzzy set of present embodiment is taked is as shown in Figure 3 " choppy sea " membership function, and the corresponding membership function on each fuzzy division is can value as follows particularly:

Above-mentioned is the fuzzy division of former piece domain, perhaps is called the former piece fuzzy set of fuzzy inference rule.Here save in traditional fuzzy inference rule storehouse, and only the above-mentioned of appearance is " the fuzzy control function " that is equivalent to " fuzzy inference rule storehouse ".

Wherein remember [t _a, t _b] be the span of output variable NB (be called and export domain), t _iBe corresponding fuzzy set A in the output domain _iThe peak dot value, t in calculating _iInitial value determine mainly to depend on the knowledge of experience and domain expert in the past, t _iValue generally do not carry out real-time renewal; (as 3～5 week) adjusts t according to practical situation after only needing for some time _iValue get final product.

Draw as after becoming domain thought, the change domain Adaptive Fuzzy Control function that obtains into: $t\left(k\right)=\mathrm{\β}(\mathrm{\ΔU}\left(k\right),t(k-1))\underset{i=1}{\overset{7}{\mathrm{\Σ}}}{A}_i\left(\frac{\mathrm{\ΔU}\left(k\right)}{\mathrm{\α}\left(\mathrm{\ΔU}\left(k\right)\right)}\right)t_i......\left(8\right)$ Wherein α (x) is the former piece contraction-expansion factor, and (x y) is the consequent contraction-expansion factor to β.Compare with the fuzzy control interpolating function under the general domain, promptly with $t\left(k\right)=\underset{i=1}{\overset{7}{\mathrm{\Σ}}}{A}_i\left(\mathrm{\ΔU}\left(k\right)\right)t_i$ Compare, become domain Adaptive Fuzzy Control function and increased former piece contraction-expansion factor and consequent contraction-expansion factor.

Usually we advise that former piece contraction-expansion factor value is as follows: $\mathrm{\α}\left(x\right)={\left(\frac{\left|x\right|}{E}\right)}^{\mathrm{\τ}},\mathrm{\τ}>0......\left(9\right)$

Wherein in the prebake Aluminum Electrolysis Production, the fuzzy controller of utilization is simply singly to go into singly to go out fuzzy controller; Therefore there is no need to calculate overall error, E directly value is $E\underset{=}{\mathrm{\Δ}}\max \{U_a,U_b\}$

The initial value of τ wherein is to determine according to expertise, the value of τ be one can online adjusting parameter, different electrolyzer τ values may be different slightly, general value is 0.7～2.

Usually all right value of former piece contraction-expansion factor is as follows:

α(x)＝1-λesp(-kx ²)，λ∈(0，1)，k＞0……(10)

Wherein the initial value of λ, k is to determine that according to expertise λ, k also are the parameters of online adjusting, and different electrolyzer values may be different slightly, general λ value 0.5～0.99, k value 0.5～1.

Usually we advise that consequent contraction-expansion factor value is as follows:

β(x，y)＝1-λesp(-k ₁x ²-k ₂y ²)，λ∈(0，1)，k ₁，k ₂＞0.......(11)

Wherein λ, k ₁, k ₂Initial value be to determine λ, k according to expertise ₁, k ₂Also be the parameter of online adjusting, different electrolyzer values is different slightly, and general λ value is 0.5～0.99, k ₁, k ₂Value 0.5～1.

Usually all right value of consequent contraction-expansion factor is as follows:

Successive type $\mathrm{\β}\left(\mathrm{\ω}\right)=1+K_I{\∫}_0^{\mathrm{\ω}}\mathrm{\ΔU}\left(\mathrm{\τ}\right)\mathrm{d\τ},K_I>0......\left(12\right)$

Discrete type $\mathrm{\β}\left(k\right)=1+K_{I}T\underset{\mathrm{\λ}=1}{\overset{k}{\mathrm{\Σ}}}\mathrm{\ΔU}\left(\mathrm{\λ}\right),K_I,T>0......\left(13\right)$

Generally get integration constant K in formula (12), (13) _IBe 0.5～2, parameter T is the discrete control system sampling time in the formula (13), and size is set according to Controlling System.

Fig. 4 is the schema according to the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production of the present invention.At step S1 upper computer the technological operation and the raw material dispensing of electrolyzer carried out centralized Control, and be that this electrolyzer is set an alumina concentration set(ting)value.At step S2, slot control machine receives the alumina concentration set(ting)value set, and the voltage change Δ U that gathers from the electrolyzer cycle.At step S3, calculate the former piece contraction-expansion factor, and calculate the input that becomes domain interpolation fuzzy control calculating.At step S4, calculate the input variable that becomes interpolation fuzzy control under the domain according to the former piece contraction-expansion factor, and calculate the preliminary blanking time t of NB at interval.At step S5, calculate the consequent contraction-expansion factor, and the consequent contraction-expansion factor be multiply by preliminary blanking t pitch time that is calculated, obtain the revised blanking time t of NB at interval, the time of NB offers topworks 4 at interval with revised blanking then.At step S6, the blanking time of topworks's 4 control electrolyzers is to adjust the concentration of aluminum oxide.At step S7, slot control machine feeds back to upper computer with the state of electrolyzer such as voltage, strength of current and the temperature of superheat etc. of electrolyzer by the cycle.

Abovely at an electrolyzer change domain adaptive fuzzy control method of the present invention and device have been described with reference to Fig. 1-4, in actual applications, can use a slot control machine, upper computer to control a plurality of topworkies respectively to blanking pitch time of a plurality of electrolyzer blankings.

By above-mentioned explanation as can be known, compare with common fuzzy interpolation controller, change domain adaptive fuzzy interpolation control device of the present invention and method have following advantage: when the fuzzy division of domain, the shape of membership function is not had special requirement, such as selecting simple " choppy sea " membership function; When setting up the inference rule storehouse, do not need too many inference rule, operand descends greatly, such as preferred embodiment of the present invention 7 inference rules is only arranged; Depend on intuition and set up inference rule, within reasonly depend on the expert, improved adaptability electrolyzer; The control effect that becomes the domain adaptive fuzzy controller is far superior to the traditional fuzzy controller, and control hysteresis postpones to reduce obviously; Because change domain adaptive fuzzy controller simplicity of design of the present invention is more convenient for emulation and adopts software to realize, save production cost simultaneously.

The invention is not restricted to the foregoing description, any those skilled in the art can obtain various distortion and change under the enlightenment of this specification sheets, and therefore, protection scope of the present invention should limit with appending claims.

Claims (10)

1. the change domain Adaptive Fuzzy Control device of a prebake Aluminum Electrolysis Production comprises:

Voltage check device, the voltage change of detection electrolyzer;

Upper computer is set the set(ting)value of alumina concentration, and the technological operation and the raw material dispensing of electrolyzer carried out centralized Control;

Slot control machine receives the set(ting)value from the alumina concentration of upper computer, and uses the voltage change Δ U of electrolyzer to become domain fuzzy control calculating, draws the time t of the blanking interval NB of current aluminum oxide;

Topworks is according to the blanking that is calculated by slot control machine that the is received time t of NB at interval, the blanking amount of controlled oxidation aluminium.

2. the change domain Adaptive Fuzzy Control device of prebake Aluminum Electrolysis Production as claimed in claim 1, wherein slot control machine comprises:

Former piece contraction-expansion factor calculating device calculates the former piece contraction-expansion factor that becomes in the domain fuzzy control according to current voltage change Δ U;

The control calculating device calculates the input variable that becomes the interpolation fuzzy control under the domain according to the former piece contraction-expansion factor, and calculates the preliminary blanking time t of NB at interval;

Consequent contraction-expansion factor calculating device, calculate the consequent contraction-expansion factor that becomes in the domain fuzzy control according to current voltage change Δ U and the preceding blanking time interval value that once calculates, and the consequent contraction-expansion factor be multiply by preliminary blanking t pitch time that is calculated, obtain the revised blanking time t of NB at interval.

3. the change domain adaptive fuzzy control method of a prebake Aluminum Electrolysis Production comprises step:

Upper computer carries out centralized Control to electrolyzer, and is alumina concentration set(ting)value of electrolyzer setting;

Slot control machine receives the alumina concentration set(ting)value set, and the voltage change Δ U that gathers from the electrolyzer cycle, and the span of Δ U is [U _a, U _b];

Reception is from the set(ting)value of the alumina concentration of upper computer, and uses the voltage change Δ U of electrolyzer to become the domain Adaptive Fuzzy Control and calculate, and the blanking that draws current aluminum oxide is the time t of NB at interval;

The blanking time of actuating mechanism controls electrolyzer is to adjust the concentration of aluminum oxide;

Slot control machine feeds back to upper computer with the state of electrolyzer such as voltage, strength of current and the temperature of superheat etc. of electrolyzer by the cycle.

4. the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production as claimed in claim 3,

Wherein becoming domain Adaptive Fuzzy Control calculation procedure comprises:

Calculate former piece contraction-expansion factor α (x), and calculate the input that becomes domain interpolation fuzzy control calculating;

Calculate the input variable that becomes interpolation fuzzy control under the domain according to former piece contraction-expansion factor α (x), and calculate the preliminary blanking time t of NB at interval;

Calculate the consequent contraction-expansion factor, and the consequent contraction-expansion factor be multiply by preliminary blanking t pitch time that is calculated, obtain the revised blanking time t of NB at interval, the time with revised blanking interval NB offers topworks then, wherein

In the time t calculation procedure of blanking interval NB, adopt $t\left(k\right)=\mathrm{\β}(\mathrm{\ΔU}\left(k\right),t(k-1))\underset{i-1}{\overset{7}{\mathrm{\Σ}}}{A}_i\left(\frac{\mathrm{\ΔU}\left(k\right)}{\mathrm{\α}\left(\mathrm{\ΔU}\left(k\right)\right)}\right)t_i$

A wherein ₁, A ₂..., A ₇Be input domain [U _a, U _b] on the fuzzy set of fuzzy division.

5. the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production as claimed in claim 4,

Wherein said fuzzy set can be chosen in the membership functions such as " choppy sea ", " square wave ", " Gaussian ".

6. the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production as claimed in claim 4 wherein in calculating former piece contraction-expansion factor step, adopts $\mathrm{\α}\left(x\right)={\left(\frac{\left|x\right|}{E}\right)}^{\mathrm{\τ}},\mathrm{\τ}>0$

Wherein α (x) is the former piece contraction-expansion factor, and E directly value is $E\underset{=}{\mathrm{\Δ}}\max \{U_a,U_b\},$ The initial value of τ is to determine according to expertise, the value of τ be one can online adjusting parameter, different electrolyzer τ values is different slightly, general value is 0.7～2.

7. the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production as claimed in claim 4 wherein in calculating former piece contraction-expansion factor step, adopts

α(x)＝1-λesp(-kx ²)，λ∈(0，1)，k＞0

Wherein α (x) is the former piece contraction-expansion factor, and the initial value of λ, k is to determine that according to expertise λ, k also are the parameters of online adjusting, and different electrolyzer values is different slightly, general λ value 0.5～0.99, k value 0.5～1.

8. the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production as claimed in claim 4, wherein in calculating consequent contraction-expansion factor step, but consequent contraction-expansion factor value is as follows:

β(x，y)＝1-λesp(-k ₁x ²-k ₂y ²)，λ∈(0，1)，k ₁，k ₂＞0

Wherein λ, k ₁, k ₂Initial value be to determine λ, k according to expertise ₁, k ₂Also be the parameter of online adjusting, different electrolyzer values is different slightly, and general λ value is 0.5～0.99, k ₁, k ₂Value 0.5～1.

9. the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production as claimed in claim 4, wherein in calculating consequent contraction-expansion factor step, but consequent contraction-expansion factor value is as follows:

Successive type $\mathrm{\β}\left(\mathrm{\ω}\right)=1+K_I{\∫}_0^{\mathrm{\ω}}\mathrm{\ΔU}\left(\mathrm{\τ}\right)\mathrm{d\τ},K_I>0$

Generally get integration constant K _IBe 0.5～2.

10. the change domain adaptive fuzzy control method of prebake Aluminum Electrolysis Production as claimed in claim 4, wherein in calculating consequent contraction-expansion factor step, but consequent contraction-expansion factor value is as follows:

Discrete type $\mathrm{\β}\left(k\right)=1+K_{I}T\underset{\mathrm{\λ}=1}{\overset{k}{\mathrm{\Σ}}}\mathrm{\ΔU}\left(\mathrm{\λ}\right),K_I,T>0$

Wherein parameter T is the discrete control system sampling time, and size is set according to Controlling System; Generally get integration constant K _IBe 0.5～2.

Images