CN106546697A - A kind of on-line prediction method of bauxite flotation process mash acid alkalinity - Google Patents
A kind of on-line prediction method of bauxite flotation process mash acid alkalinity Download PDFInfo
- Publication number
- CN106546697A CN106546697A CN201610968376.0A CN201610968376A CN106546697A CN 106546697 A CN106546697 A CN 106546697A CN 201610968376 A CN201610968376 A CN 201610968376A CN 106546697 A CN106546697 A CN 106546697A
- Authority
- CN
- China
- Prior art keywords
- value
- ore
- error
- alkali
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N31/00—Investigating or analysing non-biological materials by the use of the chemical methods specified in the subgroup; Apparatus specially adapted for such methods
- G01N31/22—Investigating or analysing non-biological materials by the use of the chemical methods specified in the subgroup; Apparatus specially adapted for such methods using chemical indicators
- G01N31/221—Investigating or analysing non-biological materials by the use of the chemical methods specified in the subgroup; Apparatus specially adapted for such methods using chemical indicators for investigating pH value
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Analytical Chemistry (AREA)
- Molecular Biology (AREA)
- Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Biophysics (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Manufacture And Refinement Of Metals (AREA)
Abstract
A kind of bauxite flotation process mash acid alkalinity on-line prediction method, existing for bauxite flotation mash acid alkalinity regulator addition point to acid-base value test point makes acid-base value control delayed compared with large dead time, and the low problem of acid-base value manual detection efficiency, the present invention initially sets up ore and alkali reacts the ore alkali consumption regression model for causing;Hydrolyze in water further according to the alkali not reacted with ore, and alkaline impact of the recirculated water to hydrolysising balance, set up acid-base value mechanism prediction model;Error compensation model is set up according to the error time sequence that acid-base value measured value is constituted with mechanism model predicted value;Error compensation direction is modified based on Expert Rules according to working conditions change, mechanism model is compensated using revised offset and obtain mash acid alkalinity predicted value.Prediction of the present invention for actual production process mash acid alkalinity, root-mean-square error is 0.0935, and maximum relative error is 2.83%, and 90% test sample relative error is within ± 2%.
Description
Technical field
The present invention relates to a kind of on-line prediction method of bauxite flotation process mash acid alkalinity.
Background technology
Liquor of Ore Dressing Bayer process alumina producing makes the substantial amounts of low-grade bauxite of China obtain for visiing by ore concentration of bauxite
Ear method is produced, and is greatlyd save energy consumption, is reduced discharge.The target of ore concentration of bauxite is in the premise for ensureing concentrate grade
Under, the rate of recovery of aluminium is improved as far as possible, and the rate of recovery directly determines the utilization rate of ore.
Ore concentration of bauxite adopts foam flotation method, by adding flotation medicine in the ore pulp that the mineral grain for grinding is constituted
Agent adjusts the surface nature of mineral grain, and air is passed through in ore pulp, forms bubble, makes valuable mineral particle be attached to bubble
Surface forms froth bed with bubble floating, reclaims foam enrichment valuable mineral, and gangue mineral is stayed in water and drained from bottom land.Bubble
Foam floatation process is related to the physical-chemical reaction of series of complex, and influence factor is a lot, wherein, the acid-base value of flotation pulp (is used
PH value is represented) it is a very crucial variable, mash acid alkalinity directly affects the activity of the composition of ore pulp ion, floating agent
And the floatability of mineral, decide that can floating agent effectively be ionized into required ion, thus any mineral only have
Best flotation effect can be just obtained under suitable acid-base value environment.But, in actual production process, for grinding aid, pH
Value regulator (Na2CO3Solution) before grinding machine add ore in, Jing ore grindings and two-stage classification and two buffer tanks after ability
The test point of pH value is reached, and the detection of pH value is obtained by artificial offline chemical examination, and adjust the addition of pH regulators manually,
This delayed control for causing pH value is delayed, reagent consumption is big, fluctuation of operating conditions is frequent, and the big high labor intensive of workman's workload.
Therefore, real-time and accurate prediction is carried out to mash acid alkalinity, the efficiency to improving ore concentration of bauxite process is significant.
At present, neutral net, SVMs etc. be based on the modeling method of data, the modeling method based on mechanism, and
Mixed method based on data and mechanism is all used for the prediction of process variable, and achieves good effect.Meanwhile, though pH value
All critically important amount in numerous production processes, but during ore concentration of bauxite, pH regulators in addition to it there is hydrolysis in water,
Some can be reacted with ore and be consumed;Meanwhile, it is flotation that in grinding process, water used is most of
Recirculated water, pH value itself are strong basicity, increased the complexity of floatation process pH regulations.So far, to bauxite flotation process
And some other floatation process, all there is no from pH adjuster addition, working condition and pH adjuster and ore and
The research that the reaction angle of water is predicted to pH value.The hard measurement of the pH value prediction inherently pH value based on foam characteristics, nothing
Method is used to solve the pH value for causing control lag issues are stopped during grind grading because of ore pulp.And the control of floatation process pH
Artificial experience, no effective On-Line Control Method are completely dependent on still.
For this purpose, how to be predicted to mash acid alkalinity according to real-time working condition, it is that pH value control is solved based on PREDICTIVE CONTROL
The basis of lag issues, the economic and technical index, mitigation labor strength to raising floatation process suffer from particularly significant
Meaning.
The content of the invention
It is an object of the invention to provide a kind of on-line prediction method of bauxite flotation process mash acid alkalinity.
The present invention principle be:In floatation process determine mash acid alkalinity process mainly include hydrolysis of the alkali in water and
The reaction of mineral and alkali.It is primarily based on experiment and industrial production data and ore is set up with pH value adjustment using multilinear fitting method
Agent Na2CO3The regression model of the ore alkali consumption that alkali reaction is caused:
O=b0+b1X1+b2X2+b3c+b4pHw (1)
In formula, o (g ton ore deposit) represents the alkali number that alkali and 1 ton of mineral reaction are consumed, X1For alundum (Al2O3) in raw ore
Percentage composition, X2For the percentage composition of silica in raw ore, c (g ton ore deposit) is the addition of sodium carbonate in 1 ton of mineral,
pHwFor the pH value of recirculated water, b0,...,b4For regression parameter.
Ore pulp is made in alkalescence further according to hydrolysis is not produced in water with the alkali of ore reaction, while, it is considered to alkaline recirculated water
Addition affect the hydrolysis of alkali, write out the equilibrium equation of hydroxide ion according to hydrolysis mechanism, set up the mechanism of acid-base value
Forecast model.
PH adjuster (Na in floatation process2CO3Solution) mass concentration be designated as w (g/L), flow is designated as L (L/min);
Discharge quantity flow is designated as F (ton hour);Between pH adjuster addition point and pH value test point, the flow of recirculated water is designated as Q1
(L/min), pH value pHwRepresent, the flow of new water is designated as Q2(L/min), then in recirculated water hydroxide ion concentration (mol/
L) it is:
Represented with x (mol/L) in the hydroxide ion concentration of pH value test point, according to the alkali not reacted with ore in water
One-stage hydrolysis is produced, is had according to hydrolysising balance:
Solution formula (3):
In formula, k1For Na2CO3The one-stage hydrolysis equilibrium constant, its value be 1.8 × 10-4, c1For alkali not with mineral reaction
Alkali concn (mol/L) of equal value after the water of all additions is dissolved in, its computational methods is:
In formula, 106 is the molal weight (g/mol) of sodium carbonate, and 60 for being scaled g/h by unit g/min.
According to hydroxide ion concentration total in solution, can try to achieve mash acid alkalinity mechanism model prediction value expression is:
pr=14+lg (x+c2) (6)
In formula, prFor mechanism model predicted value.
Set up according to the history error time sequence constituted with mechanism model predicted value by mash acid alkalinity measured value and be based on
The Error Compensated Prediction model of autoregressive moving average obtains the error compensation value of subsequent time.
The actual measurement sample value of bauxite flotation mash acid alkalinity is designated as rt(t=1,2 ..., n), the mechanism mould at correspondence moment
Type predicted value is prt(t=1,2 ..., n), the two subtracts each other and obtains error et=rt-prt, remember that its error time sequence is { e1,
e2,...,et,...,en};According to error time sequence { e1,e2,...,et,...,en, set up based on autoregressive moving average
Error Compensated Prediction model:
In formula, error terms of the ε for model,And θj(j=1,2 ..., it is q) undetermined coefficient.Using
Unit root test method to modeling use time series carry out stationarity identification, do not exist unit root then be stationary sequence, there is list
Position root illustrates the sequence non-stationary, non-stationary series is first carried out with difference processing and is used further to modeling;Based on auto-correlation function (ACF)
Model is carried out determining rank with the Method of determining the optimum of partial autocorrelation function (PACF), according to the hangover and truncation of ACF and PACF come
Determine model order, that is, determine the value of p and q, the sequence when seasonal effect in time series ACF and PACF have p ranks and q ranks hangover property respectively
It is classified as ARMA (p, q) sequence;According to history error sequence, identification is carried out to parameter using least square method and is obtained And θj(j=1,2 ..., q).
Ore properties relatively stablize when, process working conditions change be cause pH value change main cause, meanwhile, because ARMA errors it is pre-
Survey model only relevant with the error amount at front several moment, it is impossible to reflect the unexpected fluctuation of operating mode, compensating error time sequence can
Can there is a problem of it is in opposite direction with what is should actually compensated, and experienced workman working condition change it is more apparent when can standard
Really judge the change direction of pH value.Therefore, when working conditions change is larger, Expert Rules are set up according to working conditions change and Heuristics
Error in judgement compensation direction, Expert Rules compensation direction withAdopt when direction is consistentDirectly mechanism model is compensated, no
It is defined by Expert Rules compensation direction when consistent, offset is adoptedAbsolute value, so as to realize the prediction of mash acid alkalinity.
Rule of thumb, Na2CO3Flow increment, discharge quantity increment, inflow increment, the fluctuation of pH in Circulating Water increment is to ore deposit
The impact that material consumption alkali number is produced is larger, therefore according to the situation of change of this tittle, sets up Expert Rules to determine error compensation side
To incremental representation current sample time variable uses Δ β relative to the variable quantity of previous hour sampling instanttTon represents t
Relative to the discharge quantity increment of previous sampling instant, Δ ptRepresent that t increases relative to the pH in Circulating Water of previous sampling instant
Amount, Δ mtG ton ore deposit represents the mineral facies per ton of t for the Na in the mineral per ton of previous sampling instant2CO3Addition
Increment, Δ ωtTon represents recirculated water and new water total increment of the t relative to previous sampling instant, ktRepresent the expert of t
Rule output, 1 and 0 corresponds to pH value respectively increases and subtracts, and obtaining error compensation rule is:
IF Δβt>=3.4, THEN kt=1;
IF Δβt<=-4.7, THEN kt=0;
IF -4.7<Δβt<3.4AND Δpt<=-0.21, THEN kt=1;
IF -4.7<Δβt<3.4AND Δpt>=0.23, THEN kt=0;
IF -4.7<Δβt<3.4AND -0.21<Δpt<0.23AND Δmt<=-100, THEN kt=1;
IF -4.7<Δβt<3.4AND -0.21<Δpt<0.23AND Δmt>=100, THEN kt=0;
IF -4.7<Δβt<3.4AND 0.21<Δpt<0.23AND -100<Δmt<100AND
Δωt>=12,
THEN kt=1;
IF -4.7<Δβt<3.4AND 0.21<Δpt<0.23AND -100<Δmt<100AND
Δωt<=-6, THEN kt=0;
ELSE ktWithDirection is consistent.
The predicted value for obtaining mash acid alkalinity is compensated with revised error prediction value to mechanism model predicted value, is counted
Calculating formula is:
When being predicted to mash acid alkalinity using this programme, real-time monitoring bauxite Milling classification process adds the new water yield, circulation
The water yield, alkali number and mine-supplying quantity information, analysis tcrude ore composition information, tries to achieve ore alkali consumption according to formula (1), according to formula in real time
(8) mash acid alkalinity mechanism model predicted value is tried to achieve, and according to formula (7) calculation error offset, and mistake is determined according to Expert Rules
Difference compensation direction, obtains the final predicted value of model by formula (8), realizes the on-line prediction of mash acid alkalinity.The beneficial effect of the present invention
It is really:The alkali consumption that the present invention is caused to ore and alkali reaction carries out quantification and models by designing a kind of simple experiment,
And then using the Error Compensated Prediction model based on autoregressive moving average and the error compensation adjustment in direction based on Expert Rules
Method carries out online compensation to mechanism model, predicts the acid-base value of ore pulp solution, adapts to the change of working condition, can be effectively
Actual production process is applied to, the root-mean-square error obtained using the method prediction is 0.0935, maximum relative error is
2.83%.
Description of the drawings
Forecast model structures of the Fig. 1 for mash acid alkalinity;
Fig. 2 predicts the outcome to mash acid alkalinity for modelling by mechanism;
Fig. 3 is the Error Compensated Prediction result based on autoregressive moving-average model and Expert Rules;
Fig. 4 finally predicts the outcome for mash acid alkalinity.
The present invention is described in further detail with reference to the accompanying drawings and detailed description.
Specific embodiment
As shown in figure 1, the present invention is using based on the ore alkali consumption regression model of industrial production data, based on flotation solution
The mechanism model of the principles of chemistry and the Error Compensated Prediction model realization based on autoregressive moving average are to flotation pulp soda acid
The prediction of degree.Case is done using 285 groups of continuous creation datas of certain ore concentration of bauxite factory herein to be analyzed (not comprising improper work
Data under condition), wherein front 215 groups of data are used for parameter identification, 70 groups of data are used for compliance test result afterwards.
1. ore alkali consumption regression model
Ore can consume a part of alkali with alkali reaction, so that the alkali in adding ore pulp can not be completely used for adjusting ore pulp
PH value;Meanwhile, because many comprising diaspore, chlorite, iron ore, titanium ore, illite etc. in diaspore ore
Mineral are planted, composition is sufficiently complex, and ore is also sufficiently complex with the effect of alkali, designed and grind grading actual production process condition
The experiment that (including slurry fineness, concentration and temperature) is close to as far as possible determines the alkali number that ore and alkali reaction are consumed.Experimentation is:
Collection actual production ore is first crushed, and is then ground to (- 200 mesh identical with the slurry fineness that flotation is entered in actual production
Content is more than 85%), a certain amount of ore (500g) is weighed, adds water to form ore pulp (concentration 45%) in ore, while plus
Enter different alkali numbers (2200g/t ore deposits, 2500g/t ore deposits, 2800g/t ore deposits), and keep solution temperature and actual production slurry temperature
(38 DEG C) being close to, 1 pH value being detected every 5 minutes, until 1 hour after alkali is added, note pH value is Pn(n=1,2 ...,
12)。
It is 20 minutes according to actual production lag time, the alkali number according to needed for the pH value that following formula is calculated when reacting 20 minutes
o2(g ton ore deposit):
In formula, k1For the one-stage hydrolysis equilibrium constant of alkali, its value is 1.8 × 10-4, V is calculated to obtain according to pulp densityWaterFor
0.61 liter.
The alkali number that note is initially added is o1(g ton ore deposit), then ore and alkali react the alkali number for consuming and are:
oh=o1-o2 (2)
So as to the ore according to above-mentioned experiment acquisition under corresponding with actual production grind grading condition reacts what is caused with alkali
Quantity of alkali consumption data.
Collection actual production corresponding with experiment ore is added water, adds alkali and composition of ores data, based on these experiments and work
Ore alkali consumption regression model is tried to achieve in the fitting of industry creation data.
The chemical composition of ore pulp is detected, and carries out contrast discovery with raw ore chemical composition, bauxite it is main into
In point, before and after grind grading, some produce change into branch, and have some components unchangeds, wherein aluminum oxide, silica and water meeting
React with alkali, the potassium oxide and sodium oxide molybdena for having part can be dissolved in water, but the content of potassium oxide and sodium oxide molybdena is very little, so can
Not consider, separately it is believed that di-iron trioxide, titanium oxide and magnesia it is water insoluble also not with alkali react, calcium oxide is dissolved in aquatic
Into calcium hydroxide precipitation.Meanwhile, caustic dosage is more, and the alkali number that ore is consumed is also more, and the addition of the recirculated water in alkalescence
Equivalent to the addition of more polybase;Therefore, in the case where other chemical reactions are not considered, obtained according to experiment and actual production data
Regression equation to ore alkali consumption is:
O=25.917+0.0947X1-0.107X2+0.36c-1.91pHw(R2=0.867) (3)
In formula, o (g ton ore deposit) represents the alkali number that alkali and 1 ton of mineral reaction are consumed, X1For alundum (Al2O3) in raw ore
Percentage composition, X2For the percentage composition of silica in raw ore, c (g ton ore deposit) is the addition of sodium carbonate in 1 ton of mineral, pHw
For the acid-base value of recirculated water.
2. mash acid alkalinity mechanism prediction model
After alkali adds ore pulp, there is chemical reaction in some of a part and bauxite composition, complete according to another part
Portion produces one-stage hydrolysis in water and is calculated.And as the addition of recirculated water makes ore pulp in alkalescence, the hydrolysis of alkali is affected, thus
Hydrolysising balance sets up the mechanism prediction model of acid-base value.
PH adjuster (Na in note floatation process2CO3Solution) mass concentration be w (g/L), flow be L (L/min);It is short
In time, the average discharge quantity flow of (20 minutes) is F (ton hour);Follow between pH adjuster addition point and pH value test point
The flow of ring water is Q1(L/min), its pH value pHwRepresent, the flow of new water is Q2(L/min), hydroxide ion in recirculated water
Concentration (mol/L) be:
According to one-stage hydrolysis is not all produced in water with the alkali of ore reaction, remember dense in pH value test point hydroxide ion
Spend for x (mol/L), had according to hydrolysising balance:
Solution formula (5) can be obtained:
In formula, k1For the one-stage hydrolysis equilibrium constant of alkali, its value is 1.8 × 10-4,c1For not with mineral reaction alkali soluble in
Equivalent concentration (mol/L) after the water of all additions, its computational methods is:
In formula, 106 is the molal weight (g/mol) of sodium carbonate, and 60 for being scaled g/h. by unit g/min
According to hydroxide ion concentration total in solution, can try to achieve mash acid alkalinity mechanism model prediction value expression is:
pr=14+lg (x+c2) (8)
In formula, prFor mechanism model predicted value.
As shown in Fig. 2 wherein, root-mean-square error is 0.1488 to mash acid alkalinity mechanism model prediction effect, maximum relative
Error is 3.28%.
3. the error time sequence prediction based on autoregressive moving-average model is compensated
Mash acid alkalinity measured value has certain correlation with the error time sequence constituted by mechanism model predicted value,
Error time sequence is predicted using autoregressive moving-average model for this.Using the error time sequence of front 215 groups of data
Row are modeled, and wherein corresponding to the measured value of mash acid alkalinity, time series is rt(t=1,2 ..., n), each time point institute
Corresponding mechanism model predicted value isThe two subtracts each other and obtains corresponding error and be
Remember that its error time sequence is { e1,e2,...,et,...,en}。
According to error time sequence { e1,e2,...,et,...,en, set up the error compensation based on autoregressive moving average
Forecast model:
In formula, error terms of the ε for model,And θj(j=1,2 ..., it is q) undetermined coefficient.Using
Unit root test method carries out stationary test and obtains no presence of unit root, the time in error time sequence to error time sequence
Sequence meets the stationarity condition of sequence;Using determining rank side based on auto-correlation function (ACF) and partial autocorrelation function (PACF)
Method carries out determining rank to model, and the auto-correlation function of error time sequence trails for 6 ranks, and partial autocorrelation function trails for 1 rank, therefore from
Regression order is 6, moving average exponent number be 1, i.e. p=6, q=1.Identification is carried out to parameter using least square method to obtainAnd θj(j=1 ..., q), finally obtaining Error Compensated Prediction model is:
4. the error compensation adjustment in direction based on Expert Rules
Mechanism model reflects the OH that the hydrolysis of alkali in floatation process and the addition of recirculated water cause-Shadow of the change to pH value
Ring.With the proviso that chemistry of Flotation ambient stable.But whole floatation process is extremely complex, an alkali part and the ore deposit being added in ore pulp
Stone reacts, and another part produces hydrolysis, and the discharge quantity of the addition and ore of alkali directly affects the alkali number for producing hydrolysis, new water and
OH in the addition and recirculated water of recirculated water-Concentration can affect slurry pH.When ore properties are relatively stablized, process work
Condition change is the main cause for causing pH value to change, meanwhile, because ARMA error predictions model it is only relevant with the error amount at front several moment,
The unexpected fluctuation of operating mode can not be reflected, be there may be the compensation of error time sequence and in opposite direction asked with what is should actually compensated
Topic, and experienced workman can accurately judge the change direction of pH value when working condition change is more apparent.Therefore, according to work
Condition changes and Heuristics sets up Expert Rules error in judgement compensation direction, Expert Rules compensation direction withWhen direction is consistent
Directly mechanism model is compensated, is defined by Expert Rules compensation direction when inconsistent, offset is adoptedAbsolute value, from
And realize the prediction of mash acid alkalinity.
Rule of thumb, Na2CO3Increment, discharge quantity increment, inflow increment, the fluctuation of pH in Circulating Water increment are consumed to mineral
The impact that alkali number is produced is larger, therefore according to the situation of change of this tittle, sets up Expert Rules to determine error compensation direction, increases
Amount represents that current sample time variable, relative to the variable quantity of previous hour sampling instant, uses Δ βtTon represents that t is relative
In the discharge quantity increment of previous sampling instant, Δ ptRepresent pH in Circulating Water increment of the t relative to previous sampling instant, Δ
mtG ton ore deposit represents the mineral facies per ton of t for the Na in the mineral per ton of previous sampling instant2CO3Addition increment,
ΔωtTon represents recirculated water and new water total increment of the t relative to previous sampling instant, ktRepresent that the Expert Rules of t are defeated
Go out, 1 and 0 corresponds to pH value respectively increases and subtract, and obtaining error compensation rule is:
IF Δβt>=3.4, THEN kt=1;
IF Δβt<=-4.7, THEN kt=0;
IF -4.7<Δβt<3.4AND Δpt<=-0.21, THEN kt=1;
IF -4.7<Δβt<3.4AND Δpt>=0.23, THEN kt=0;
IF -4.7<Δβt<3.4AND -0.21<Δpt<0.23AND Δmt<=-100, THEN kt=1;
IF -4.7<Δβt<3.4AND -0.21<Δpt<0.23AND Δmt>=100, THEN kt=0;
IF -4.7<Δβt<3.4AND 0.21<Δpt<0.23AND -100<Δmt<100AND
Δωt>=12,
THEN kt=1;
IF -4.7<Δβt<3.4AND 0.21<Δpt<0.23AND -100<Δmt<100AND
Δωt<=-6, THEN kt=0;
ELSE ktWithDirection is consistent.
Mechanism model is compensated with revised error prediction value and obtain the predicted value of mash acid alkalinity and be:
When predicting to mash acid alkalinity using this programme, ore alkali consumption is tried to achieve according to formula (3), further according to not anti-with ore
There is hydrolysis in the alkali answered in water, try to achieve mash acid alkalinity mechanism model predicted value by formula (8), according to the actual measurement of mash acid alkalinity
Value is set up Error Compensated Prediction model with the history error time sequence constituted by mechanism model value and obtains error compensation value such as formula
(10), shown in, based on Expert Rules round-off error compensation direction, both directions are directly compensated to mechanism model when consistent, no
It is defined by Expert Rules compensation direction when consistent, i.e., mechanism model predicted value is compensated by formula (11) and obtain mash acid alkalinity
Predicted value.
The production mash acid alkalinity of 3 days continuous to factory predicts the outcome as shown in Figure 4.Wherein, root-mean-square error is
0.0935,90% test sample relative error shows that forecast model has degree of precision, disclosure satisfy that reality in ± 2%
Production requirement.On-line checking result is provided for production, the real-time of detection is improve and is alleviated the labour intensity of workman.
Claims (1)
1. a kind of on-line prediction method of bauxite flotation process mash acid alkalinity, it is characterised in that first set up ore and adjust with pH value
Whole dose of Na2CO3The ore alkali consumption regression model that reaction is caused:
O=b0+b1X1+b2X2+b3c+b4pHw (1)
In formula, o g ton ore deposits represent the alkali number that alkali and 1 ton of mineral reaction are consumed, X1For the percentage of alundum (Al2O3) in raw ore
Content, X2For the percentage composition of silica in raw ore, c g ton ore deposits are the addition of alkali in 1 ton of mineral, pHwFor recirculated water
PH value, b0,...,b4For regression parameter;
Hydrolyze further according to ionization is not produced in water with the alkali of ore reaction, and consider that the addition of alkaline recirculated water affects ore pulp acid
Basicity, sets up acid-base value mechanism prediction model, pH adjuster Na in floatation process according to hydrolysis2CO3The quality of solution is dense
Degree is designated as w g/L, and its flow is designated as L L/min;Discharge quantity flow is designated as F ton hours;PH adjuster addition point is examined with pH value
Between measuring point, the flow of recirculated water is designated as Q1L/min, its pH value pHwRepresent, the flow of new water is designated as Q2L/min, recirculated water
Concentration mol/L of middle hydroxide ion is:
Represented with xmol/L in the hydroxide ion concentration of pH value test point, one is produced in water according to the alkali not reacted with ore
Level hydrolysis, has according to hydrolysising balance:
Solution formula (3):
In formula, k1For Na2CO3The one-stage hydrolysis equilibrium constant, its value be 1.8 × 10-4, c1For not with mineral reaction alkali soluble in
Alkali concn mol/L of equal value after the water of all additions, its computational methods is:
In formula, 106 is molal weight g/mol of sodium carbonate, and 60 for being scaled g/h by unit g/min;
According to hydroxide ion concentration total in ore pulp solution, trying to achieve mash acid alkalinity mechanism model prediction value expression is:
pr=14+lg (x+c2) (6)
In formula, prFor mechanism model predicted value;
Set up according to the error time sequence constituted by mash acid alkalinity measured value and mechanism model predicted value and slided based on autoregression
Dynamic average error time sequence compensation model;
The actual measurement sample value of note bauxite flotation mash acid alkalinity is rt, t=1,2 ..., n, the corresponding mechanism model of each time point
Predicted value isT=1,2 ..., n, the two subtracts each other and obtains corresponding error and beAccording to error time sequence
{e1,e2,...,et,...,en, set up autoregressive moving average error compensation model:
In formula, error terms of the ε for model,I=1,2 ..., p and θj, j=1,2 ..., q is undetermined coefficient, using unit root
Method of inspection carries out stationarity identification to time series, using the Method of determining the optimum based on auto-correlation function and partial autocorrelation function to mould
Type carries out determining rank, that is, determine the value of p and q, according to history error sequence, carry out identification to parameter using least square method and obtainI=1,2 ..., p and θj, j=1,2 ..., q;
Expert Rules error in judgement compensation direction is set up according to working conditions change and Heuristics, incremental representation current sample time is used
Certain variable uses Δ β relative to the variable quantity of previous hour sampling instanttTon represents the blanking of the relatively previous sampling instant of t
Amount increment, Δ ptRepresent the pH in Circulating Water increment of the relatively previous sampling instant of t, Δ mtG ton ore deposit represents the every of t
Ton mineral facies are to the Na in the mineral per ton of previous sampling instant2CO3Addition increment, Δ ωtTon represents that t is relatively previous
The recirculated water of sampling instant and new water total increment, ktRepresent t Expert Rules output, 1 and 0 respectively correspond to pH value increase and
Subtract, obtaining error compensation rule is:
IF Δβt>=3.4, THEN kt=1;
IF Δβt<=-4.7, THEN kt=0;
IF -4.7<Δβt<3.4ANDΔpt<=-0.21, THEN kt=1;
IF -4.7<Δβt<3.4ANDΔpt>=0.23, THEN kt=0;
IF -4.7<Δβt<3.4AND-0.21<Δpt<0.23ANDΔmt<=-100, THEN kt=1;
IF -4.7<Δβt<3.4AND-0.21<Δpt<0.23ANDΔmt>=100, THEN kt=0;
IF -4.7<Δβt<3.4AND 0.21<Δpt<0.23AND-100<Δmt<100AND
Δωt>=12,
THEN kt=1;
IF-4.7<Δβt<3.4AND 0.21<Δpt<0.23AND-100<Δmt<100AND
Δωt<=-6, THEN kt=0;
ELSE ktWithDirection is consistent;
Determine it is just to compensate or negative compensation according to offset rule, Expert Rules compensation direction withAdopt when direction is consistentDirectly
Connect and mechanism model is compensated, be then defined by Expert Rules compensation direction when inconsistent, offset is adoptedAbsolute value, use
Revised error prediction value compensates the predicted value for obtaining mash acid alkalinity to mechanism model predicted value, obtains final predicted value
For:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610968376.0A CN106546697B (en) | 2016-10-28 | 2016-10-28 | A kind of on-line prediction method of bauxite flotation process mash acid alkalinity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610968376.0A CN106546697B (en) | 2016-10-28 | 2016-10-28 | A kind of on-line prediction method of bauxite flotation process mash acid alkalinity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106546697A true CN106546697A (en) | 2017-03-29 |
CN106546697B CN106546697B (en) | 2018-11-23 |
Family
ID=58394426
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610968376.0A Active CN106546697B (en) | 2016-10-28 | 2016-10-28 | A kind of on-line prediction method of bauxite flotation process mash acid alkalinity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106546697B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108519770A (en) * | 2018-04-27 | 2018-09-11 | 东北大学 | A kind of experiment porch of floatation process operation control |
WO2023221473A1 (en) * | 2022-05-18 | 2023-11-23 | 中国矿业大学 | Online soft measurement method for clean coal ash content during lump coal shallow trough sorting process |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS54149304A (en) * | 1978-05-16 | 1979-11-22 | Ajinomoto Kk | Froth floatation of alumina and hematite |
CN103894292A (en) * | 2014-04-03 | 2014-07-02 | 河南东大矿业股份有限公司 | Method for supplementing soda after direct flotation of bauxite |
CN104331622A (en) * | 2014-11-06 | 2015-02-04 | 中南大学 | Online prediction method for RSA (Ratio of Soda to Aluminate) of outlet solution of two-stream-process alumina dissolution process |
CN104991038A (en) * | 2015-06-09 | 2015-10-21 | 河南东大矿业股份有限公司 | Distinguishing method for pH value of bauxite direct flotation ore |
CN105271335A (en) * | 2015-11-16 | 2016-01-27 | 中南大学 | Proportion optimizing method for inlet ore pulp and alkali liquor in aluminum oxide double-stream dissolution process |
-
2016
- 2016-10-28 CN CN201610968376.0A patent/CN106546697B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS54149304A (en) * | 1978-05-16 | 1979-11-22 | Ajinomoto Kk | Froth floatation of alumina and hematite |
CN103894292A (en) * | 2014-04-03 | 2014-07-02 | 河南东大矿业股份有限公司 | Method for supplementing soda after direct flotation of bauxite |
CN104331622A (en) * | 2014-11-06 | 2015-02-04 | 中南大学 | Online prediction method for RSA (Ratio of Soda to Aluminate) of outlet solution of two-stream-process alumina dissolution process |
CN104991038A (en) * | 2015-06-09 | 2015-10-21 | 河南东大矿业股份有限公司 | Distinguishing method for pH value of bauxite direct flotation ore |
CN105271335A (en) * | 2015-11-16 | 2016-01-27 | 中南大学 | Proportion optimizing method for inlet ore pulp and alkali liquor in aluminum oxide double-stream dissolution process |
Non-Patent Citations (1)
Title |
---|
任会峰等: "基于最小二乘支持向量回归机的矿浆酸碱度鲁棒软测量", 《上海交通大学学报》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108519770A (en) * | 2018-04-27 | 2018-09-11 | 东北大学 | A kind of experiment porch of floatation process operation control |
CN108519770B (en) * | 2018-04-27 | 2020-10-27 | 东北大学 | Experimental platform for flotation process operation control |
WO2023221473A1 (en) * | 2022-05-18 | 2023-11-23 | 中国矿业大学 | Online soft measurement method for clean coal ash content during lump coal shallow trough sorting process |
Also Published As
Publication number | Publication date |
---|---|
CN106546697B (en) | 2018-11-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Nakhaei et al. | Recovery and grade accurate prediction of pilot plant flotation column concentrate: Neural network and statistical techniques | |
Kaartinen et al. | Machine-vision-based control of zinc flotation—a case study | |
Shean et al. | A review of froth flotation control | |
Zhu et al. | Probability density function of bubble size based reagent dosage predictive control for copper roughing flotation | |
CN101620217B (en) | Device for controlling concentration of erosion solution in concrete durability experiment | |
CN101526814A (en) | Leaching rate prediction and optimization operation method in wet metallurgical leaching process | |
CN106546697A (en) | A kind of on-line prediction method of bauxite flotation process mash acid alkalinity | |
CN110866659A (en) | On-line prediction method for caustic ratio in dissolution process of high-pressure pipeline in Bayer-process alumina production | |
Rojas et al. | Model based predictive control of a rougher flotation circuit considering grade estimation in intermediate cells | |
CN104062953A (en) | Gold hydrometallurgy process running state online evaluation method | |
Oosthuizen et al. | On the current state of flotation modelling for process control | |
CN108762238B (en) | DCD-based hydrometallurgy leaching process fault diagnosis method | |
Nakhaeie et al. | Concentrate grade prediction in an industrial flotation column using artificial neural network | |
CN108647722A (en) | A kind of zinc ore grade flexible measurement method of Kernel-based methods size characteristic | |
CN108037747A (en) | Golden hydrometallurgy whole process hierarchical block operating status on-line evaluation method | |
Venkatesan et al. | Optimisation of air rate and froth depth in flotation using a CCRD factorial design–PGM case study | |
CN103241713B (en) | Preparation method of sodium hypochlorite | |
Rosenkranz et al. | Sustainable processing of mineral resources | |
Little | The development and demonstration of a practical methodology for fine particle shape characterisation in minerals processing | |
CN114512197A (en) | Method for predicting sinter drum index based on ore blending technology | |
Popli | Real-time process monitoring for froth flotation processes using image processing and dynamic fundamental models | |
CN208944364U (en) | A kind of liquid level detection device and control system for flotation cell | |
Ramlall | An investigation into the effects of UG2 Ore variability on froth flotation. | |
Zhang | Factors affecting the kinetics of froth flotation | |
Jing et al. | Prediction of concentrate grade and recovery rate of tailings in the process of production based on chaotic ant colony algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |