CN102540883A - Intelligent optimized energy saving and emission reducing controlling system - Google Patents

Abstract

The invention relates to an intelligent optimized energy saving and emission reducing controlling system. The system suitable for controlling the combustion of a calcinator for anode and cathode calcination, graphite electrode calcination, carbon electrode calcination, special carbon calcination and shashlik graphitization is mainly used for detecting the temperature control. The carbon calcinating temperature control precision decides the carbon electrode quality. Lots of control methods can be used for controlling the temperature so that the temperature control precision reaches the designated temperature control precision. The temperature control methods require that less fuel is consumed on the basis of meeting the requirements of the temperature control precision. The carbon calcinating temperature is difficult to control because the furnace chamber is large, the working conditions are complex and the interference factors are more. The design of the temperature control methods lies in the elimination of interference factors in data in the calcinating site.

Description

Intelligent optimization energy-saving and emission-reduction control system

Technical field:

The present invention is suitable for anode-baking, the negative electrode roasting, and graphite electrode, carbon resistance rod, extraordinary charcoal, the graphited calciner combustion control system of skewer reaches the effect of optimizing energy-saving and emission-reduction.

Background technology:

China is electrolyzing aluminum industry big producing country, and along with increasing substantially of electrolytic aluminium, the demand of antianode carbon piece also grows with each passing day; Present most of producer has all adopted manual control model, has the temperature control that differs, and firing cycle is long; Production efficiency is hanged down inferior problem, and is especially uncontrollable to the fugitive constituent that in the roasting of charcoal piece constitutes, produces, and causes fugitive constituent to run off in a large number; Waste of fuel, contaminated environment.

Summary of the invention:

In view of manually with based on the PID calciner combustion control system of traditional control theory, the sintering temperature control of existence is poor, consume energy, and contaminated environment, production efficiency is hanged down inferior problem, has invented a kind of new calciner method for controlling combustion and system and has replaced.The temperature controlled precision of carbon baking has determined the quality of carbon electrode.There are a lot of control methods can make temperature control reach specified accuracy.Requirement to temperature-controlled process is on the basis that the satisfied temperature control accuracy requires, and consumes fuel still less as far as possible.The temperature controlled difficult point of carbon baking is that furnace chamber is huge, operating mode is complicated, disturbing factor is many.The design of control method, key are to reject the disturbing factor in the roasting field data.

Technical scheme provided by the invention:

The method that we adopt based on the problems referred to above is based on the degree of association notion in the data mining technology, promptly rejects in the data data with model interaction degree low (receive disturb bigger).

One. set up Temperature Control Model.Because its temperature control of whole range request excessively of roasting is a curve rather than straight line; And conventional instrument can't be adjusted setting value automatically; This just requires operating personnel will go to revise setting value at any time, and this also is that the calciner design of being done in the past can't reach the reason that ideal temperature is controlled.On the other hand, because field environmental condition is abominable, factors such as the movability of burner, fire path temperature control is difficult to meet the demands always.

Temperature control comprises three aspects: at first, and the control of negative pressure.Action of negative pressure has two, and the one, the oxygen supply of assurance burning makes the abundant burning of the fuel and the asphalt volatile constituents of burning rack; The one, the temperature of control smoke evacuation frame negative pressure frame and the difference of target temperature are within allowed band.And, also require accomplish above-mentioned task with the little fuel consumption of trying one's best.The elementary tactics of control is when the smoke evacuation frame, reduces negative pressure when negative pressure frame temperature is too high; Temperature is crossed increases negative pressure when hanging down, so that the high-temperature flue gas of burning rack flow to the negative pressure frame as early as possible.In order to reach with above-mentioned purpose, the mathematical model that we adopt is following:

Δv _i＝α(O _i+1-T _i)+β(T _i-T _i-1)+γ(ΔT _i-ΔT _i-1) (1)

Wherein,

Δ v _i: i (next moment) fuel aperture increment constantly;

O _I+1: i+1 target temperature constantly

T _i: i actual temperature constantly

Δ T _i: be carved into the thermograde of i during constantly during i-1, equal T _i-T _I-1

α, beta, gamma: be 3 model parameters

Secondly, burning rack temperature control.The synoptic diagram of a flame system of Fig. 1.Wherein the flow direction of flue gas is to flow to the 1st burning rack from the 3rd burning rack.So, the fuel supply of front burning rack, the conversion of temperature all can have direct influence to the rear wall firing frame, and this also is more difficult reason of control of burning rack.So, when setting up the model of rear wall firing frame, need to consider the relevant variable of front burning rack that it is following that the aperture of 3 burning racks is estimated mathematical model:

Frame 3 apertures estimate=f (temperature, temperature difference, thermograde, aperture, the aperture gradient.。。)(1.1)

Frame 2 apertures estimate=f (temperature, temperature difference, thermograde, aperture, the aperture gradient,

Frame 3 thermogrades, frame 3 is estimated aperture.。。) (1.2)

Frame 1 aperture estimates=f (temperature, temperature difference, thermograde, aperture, the aperture gradient,

Frame 3 thermogrades, frame 3 is estimated aperture

Frame 2 thermogrades, frame 2 is estimated aperture.。。) (1.3)

The same with estimating of target negative pressure, the prediction model of burning rack aperture also is similar to.In roasting process, also need constantly to the model correction.

Once more, the detection of overfueling, overfueling is meant that generally fuel is imperfect combustion because oxygen supply is not enough.Even at this moment increasing fuel can not elevate the temperature.These unburned fuels have been wasted fully.Crucial problem is the generation that how to detect this situation.Our way is to set up a mathematical model, and this model provides a value according to the data of system acquisition, that is:

Y=f (negative pressure, suction gradient, burning rack temperature, gradient.。。) (1.4)

The scope of model output valve is 0～1.Set in advance a threshold value Tg, when output valve during less than threshold value Tg,

Explain and the fuel surplus occurred.At this moment should increase negative pressure, guarantee the sufficient supplies of oxygen.After increasing negative pressure, excessive supplied fuel burning can elevate the temperature, and the supply that at this moment needs the controlling models of burning rack to come fuel metering is to guarantee that temperature error is within allowed band.

Two. image data: inquire about situation such as logout in the past, crash analysis for making things convenient for the staff, software provides functions such as record data, the inquiry of historical data; Offer curves pattern video data during inquiry, the simple enlarging function of offer curves, the concrete numerical value of concrete time point is checked in permission under the situation of not amplifying.System acquisition and preserve a period of time (for example 2 hours) since furnace chamber temperature, fuel aperture, negative pressure data, as historical data.Be designated as:

(T _i, v _i, p _i, Δ T _i), (i=1,2 ..., N) (wherein N is counting of the data of preserving) (2.1) these data to have implied between each amount and to have concerned (in order narrating conveniently, abbreviating " roasting rule " as) really.

Three. the programming rate that data denoising, roasting technique require is slow, and at short notice and do not receiving under the extraneous strongly disturbing situation, the roasting environmental change is also not obvious.In the historical data of consequently preserving, should there were significant differences between the data of closing on mutually, and implied " roasting rule " in the data.But as than strong jamming the time, no longer close association is arranged in the data, must from historical data, reject such data with " roasting rule ".So that through System Discrimination acquisition and roasting environment matched model parameter more.Otherwise the model through Model Distinguish obtains from historical data differs bigger with actual " roasting rule ", causes control accuracy to reduce, even causes vibration significantly.In order from historical data (2), to reject the data that receive larger interference, the degree of association of definition of data and model:

${\mathrm{\λ}}_i=\frac{\mathrm{\Δ}{v}_i-\mathrm{\Δ}{\tilde{v}}_i}{\max \{\mathrm{\Δ}{v}_i,-\mathrm{\Δ}{\tilde{v}}_i\}},(i=\mathrm{1,2},\·\·\·,N)---\left(3.1\right)$

Wherein:

Δ v _i: be historical data item (T _i, v _i, p _i, Δ T _i) in fuel aperture gradient

Be the fuel aperture gradient degree of association λ that from i, the historical data item in two moment of i-1, obtains with current model _iWeighed the historical data in the i moment and the correlation degree of current model, span is [1,1].Degree of association absolute value is more little, and historical data item and model interaction degree are high more, and the sharp degree of association is rejected the data step that receives larger interference from historical data (2):

Calculate each sampled point (T _i, v _i, p _i, Δ T _i) degree of association λ of data _i

Set rejecting ratio θ (number) less than 1

Ask λ _i(i=1,2 ..., maximal value N) and minimum M ax, Min, and calculate intermediate value Mean=(Max+Min)/2

Calculate and reject threshold value TH=(Max-Mean) * (1-θ)

Four. Model Distinguish because the property complicated and changeable of roasting environment, is difficult to theoretically to provide the relation between each factor, the variable with the form of mathematic(al) representation.The way that solves is, in roasting process with historical data constantly revise, Optimization Model, make model more approach actual conditions.

That so-called historical data is meant is that control system is gathered, current time and before roasting process in some data.Along with the carrying out of roasting process, historical data is also constantly increasing.Relation between each variable that historical data implies is also constantly changing.Near the data of gathering the current time have implied the model of current time.With these data-optimized, adaptive models, model just should be more accurate so.In order to use the historical data Optimization Model, also need a self-learning algorithm, or be called adaptive algorithm.We adopt an optimized Algorithm of Model Distinguish, with the historical data of rejecting the interfering data item, adopt the parameter alpha in the least square method estimation model (1), beta, gamma.Promptly under the given model structure, and historical data under, obtain the best model parameter.

Five. control temperature with model, process is following: at first refresh historical data; Secondly, receive strongly disturbing sampled point in the rejecting historical data; Once more, with remaining historical data correction model; At last, estimate fuel aperture gradient, revise aperture with this with model

Beneficial effect of the present invention is following:

1. adopt the method for mathematical model identification, confirm control algolithm, realize controlling models online adaptive property, find the parameter and the factor of the quirk optimizing controller that sets each other off.

2. based on the mathematics digging technology, reject the sampled point that is disturbed in the historical data, for certain target temperature curve; The control actual temperature is as far as possible near the target temperature curve; Make itself and surplus residual quantity minimum, and do not make the control system vibration, realize the dynamic optimization function of control system.

3. adopt the negative pressure system optimizing control, make fugitive constituent perfect combustion, improve the fume emission qualification rate, thereby reach energy savings and reduce environmental pollution.

4. has the network intelligence recognition function; According to the characteristics of roasting technique, smoke evacuation frame, negative pressure frame and burning rack move next cover furnace chamber of frame and carry out the product processing of next cycle after machining one-period; Can discern certain support body, satisfy the needs of control purpose.Wherein carry out the switching of mathematical model and the identification of variables corresponding and demonstration.

Description of drawings:

Fig. 1 control system network constitutes synoptic diagram

Fig. 2 flame system synoptic diagram, the target temperature of negative pressure frame and actual temperature

The adaptive process synoptic diagram of Fig. 3 model

Concrete embodiment:

Intelligent optimization energy-saving and emission-reduction control system is divided into host service function management system and scene control station on the spot; System is divided into three-layer network; Ground floor is the execute-in-place control station, and major function is temperature monitoring, control gas electromagnetic valve switching frequency and smoke evacuation adjustment doors aperture, to main regulation station transmission total data with accept main regulation station transmission data, handle field data general data is carried out chain control etc.; Can also select various operator schemes, manual mode and self-action.The second layer is that master-control room is regulated active station; All visual informations in the centralized displaying roasting process; Comprise operator scheme, set-point value, alarming value, flame system run location state, variable and order, can move warning, flame, operator scheme etc. managed; The 3rd layer for higher management station and data base administration station, can generate history report, storing history, the system burst incident handled etc.Key step is following, in the 1st step, sets up Temperature Control Model; The 2nd step image data: comprise furnace chamber temperature (T), fuel aperture (V)/negative pressure (P); In the 3rd step, data are carried out denoising; The 5th goes on foot, and adopts the eastern control algolithm of method of Model Distinguish, finds the parameter and the factor of optimizing controller; In the 6th step, control temperature with model.

Claims (4)

1. intelligent optimization energy-saving and emission-reduction control system, its characteristic is following: Temperature Control Model (2) image data is set up in (1): comprise furnace chamber temperature (T), fuel aperture (v), negative pressure (P), (3) data denoising, (4) Model Distinguish, temperature is controlled with model in (5).

2. like the described intelligent optimization energy-saving and emission-reduction of right 1 system, it is characterized in that: this mathematical model has second order difference item backward.

3. like the described intelligent optimization energy-saving and emission-reduction of right 1 system, it is characterized in that:, reject in the historical data and receive strongly disturbing sampled point based on data mining technology.

4. like the described intelligent optimization energy-saving and emission-reduction of right 1 system, it is characterized in that: data are carried out denoising, in order from historical data, to reject the data that receive larger interference, the degree of association of definition of data and model:

${\mathrm{\λ}}_i=\frac{{\mathrm{\Δv}}_i-\mathrm{\Δ}{\tilde{v}}_i}{\max \{\mathrm{\Δ}{v}_i,-\mathrm{\Δ}{\tilde{v}}_i\}},(i=\mathrm{1,2},\·\·\·,N)$

Wherein:

Δ v _i: be historical data item (T _i, v _i, p _i, Δ T _i) in fuel aperture gradient

Be the fuel aperture gradient degree of association λ that from i, the historical data item in two moment of i-1, obtains with current model _iWeighed the historical data in the i moment and the correlation degree of current model, span is [1,1].Degree of association absolute value is more little, and historical data item and model interaction degree are high more.

Images