AU2020200691A1 - Methods and devices for detecting abnormal alumina concentration and for monitoring electrolyte temperature in aluminum electrolysis driven by process mechanism knowledge - Google Patents

Abstract

The embodiments of the present disclosure provide a method and a device for detecting abnormal alumina concentration and a method and a device for monitoring electrolyte temperature in aluminum electrolysis driven by process mechanism knowledge. The method for detecting abnormal alumina concentration in aluminum electrolysis driven by process mechanism knowledge, comprising: determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum values of the pseudo alumina concentration in an integral feed period which is a feed period before anode effect occurs, and includes an overfeed period and an underfeed period; determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution of the maximum and minimum values of the pseudo alumina concentration in the integral feed period; determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period; and detecting whether the alumina concentration is abnormal or not, according to the values of the pseudo dissolution hysteresis coefficient in the integral feed period. 2/12 S301 determining a relationship between a pseudo alumina concentration and a feeding state S302 determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period determining values of the pseudo dissolution hysteresis coefficient in the integral feed period Q303 detecting whether the alumina concentration is abnormal or not, according to the values of the pseudo dissolution hysteresisQ-0 coefficient in the integral feed period 9 0 end Figure 3

Description

2/12

S301 determining a relationship between a pseudo alumina concentration and a feeding state

S302 determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period

determining values of the pseudo dissolution hysteresis coefficient in the integral feed period Q303

detecting whether the alumina concentration is abnormal or not, according to the values of the pseudo dissolution hysteresisQ-0 coefficient in the integral feed period 9 0

end

Figure 3

I METHODS AND DEVICES FOR DETECTING ABNORMAL ALUMINA CONCENTRATION AND FOR MONITORING ELECTROLYTE TEMPERATURE IN ALUMINUM ELECTROLYSIS DRIVEN BY PROCESS MECHANISM KNOWLEDGE TECHNICAL FIELD

[0001] The present disclosure relates to the field of aluminum electrolysis control technology, and in particularly to a method and a device for detecting abnormal alumina

concentration in aluminum electrolysis driven by process mechanism knowledge, and a

method and a device for monitoring electrolyte temperature in aluminum electrolysis driven

by process mechanism knowledge.

BACKGROUND

[0002] Aluminum electrolysis is a process in which alumina is used to obtain molten

aluminum (i.e. metal) on a cathode by passing direct current through a carbon anode in a

molten cryolite (i.e. bath) in an electrolysis cell, with carbon dioxide and carbon monoxide

being emitted at the carbon anode. Due to complex composition, high temperature and strong

corrosion of bath, it is highly difficult to obtain alumina dissolution rate, bath temperature

electrolyte temperature and alumina concentration.

[0003] Most research results on alumina dissolution property, bath temperature and alumina concentration are based on a mechanism model or an experimental cell for specified process

conditions, and many key parameters are fixed. In industrial cells, these key parameters

change in real time and cannot be measured. Therefore, it is difficult to establish a

mechanism model that adapts to frequent changes in working conditions, which limits the

applicability and accuracy of mechanism-model-based methods.

[0004] Therefore, it is urgent to provide methods to quantify solubility of alumina and to

detect abnormal alumina concentration or monitor electrolyte temperature.

SUMMARY

[0005] According to embodiments of the present disclosure, a method and a device for

detecting abnormal alumina concentration in aluminum electrolysis driven by process mechanism knowledge, and a method and a device for monitoring electrolyte temperature in aluminum electrolysis driven by process mechanism knowledge are provided, to improve accuracy of detecting and monitoring.

[0006] According to one aspect of the present disclosure, a method for detecting abnormal alumina concentration in aluminum electrolysis driven by process mechanism knowledge includes: determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum values of the pseudo alumina concentration in an integral feed period which includes an overfeed period and an underfeed period; determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution of the maximum and minimum values of the pseudo alumina concentration in the integral feed period; determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period; and detecting whether the alumina concentration is abnormal or not, according to the values of the pseudo dissolution hysteresis coefficient in the integral feed period.

[0007] Furthermore, the detecting whether the alumina concentration is abnormal or not, according to the pseudo dissolution hysteresis coefficient in each integral feed period includes: determining the alumina concentration is abnormal low, when the values of the pseudo dissolution hysteresis coefficient in the integral feed period are positive and beyond a preset normal range; determining the alumina concentration is abnormal high, when the values of the pseudo dissolution hysteresis coefficient in the integral feed period are negative and beyond the preset normal range.

[0008] Furthermore, the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period includes: determining the relation between the pseudo alumina concentration and the feed state is a first type, when the pseudo alumina concentration only has a local maximum value in the overfeed period, the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period include: when the relation between the pseudo alumina concentration and the feed state is the first type, determines a sequence position of the local maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period

[0009] Furthermore, the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period includes: determining the relation between the pseudo alumina concentration and the feed state is a second type, when the pseudo alumina concentration has a local maximum value and a global maximum value in the overfeed period, the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period include: when the relation between the pseudo alumina concentration and the feed state is the second type, determines a sequence position of the global maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

[0010] Furthermore, the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period includes: determining the relation between the pseudo alumina concentration and the feed state is a third type, when the pseudo alumina concentration only has a minimum value in the overfeed period, the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period include: when the relation between the pseudo alumina concentration and the feed state is the third type, determines a sequence position of the end of overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

[0011] Furthermore, the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina

concentration in each integral feed period includes: determining the relation between the

pseudo alumina concentration and the feed state is a forth type, when the pseudo alumina

concentration monotonically increase in underfeed period, and has multiple local maximum

values in the overfeed period, the determining a sequence position of a pseudo dissolution

hysteresis coefficient in each integral feed period, according to the relation between the

pseudo alumina concentration and the feed state, and the distribution state of the pseudo

alumina concentration in each integral feed period include: when the relation between the

pseudo alumina concentration and the feed state is the forth type, determines a sequence

position of the last local maximum value in the overfeed period as the sequence position of

the pseudo dissolution hysteresis coefficient in the integral feed period.

[0012] Furthermore, the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina

concentration in each integral feed period includes: determining the relation between the

pseudo alumina concentration and the feed state is a fifth type, when the pseudo alumina

concentration monotonically increase in the integral feed period, the determining a sequence

position of a pseudo dissolution hysteresis coefficient in each integral feed period, according

to the relation between the pseudo alumina concentration and the feed state, and the

distribution state of the pseudo alumina concentration in each integral feed period include:

when the relation between the pseudo alumina concentration and the feed state is the fifth

type, determines a sequence position of the end of overfeed period as the sequence position of

the pseudo dissolution hysteresis coefficient in the integral feed period.

[0013] Furthermore, the determining a relation between a pseudo alumina concentration and

a feed state, according to a distribution of the maximum and minimum of the pseudo alumina

concentration in each integral feed period includes: determining the relation between the

pseudo alumina concentration and the feed state is a sixth type, when the pseudo alumina

concentration only has a local maximum value in the underfeed period, the determining a

sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period,

according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period include: when the relation between the pseudo alumina concentration and the feed state is the sixth type, determines a sequence position of the beginning of the underfeed period as the sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period.

[0014] Furthermore, the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period includes: determining the relation between the pseudo alumina concentration and the feed state is a seventh type, when the pseudo alumina concentration only has a local minimum value in the underfeed period, the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period include: when the relation between the pseudo alumina concentration and the feed state is the seventh type, determines a sequence position of the beginning of the underfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

[0015] Furthermore, the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period includes: determining the relation between the pseudo alumina concentration and the feed state is an eighth type, when the pseudo alumina concentration has multiple local maximum value in the underfeed period and overfeed period, the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period include: when the relation between the pseudo alumina concentration and the feed state is the eighth type, determines a sequence position of the first local maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

[0016] Furthermore, the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period includes: determining the relation between the pseudo alumina concentration and the feed state is a ninth type, when the pseudo alumina concentration has multiple local maximum values in the underfeed period, and monotonically increase in overfeed period, the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period include: when the relation between the pseudo alumina concentration and the feed state is the ninth type, determines a sequence position of the first local maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

[0017] According to another aspect of the present disclosure, a device for detecting abnormal alumina concentration in aluminum electrolysis driven by process mechanism

knowledge includes: a relation determination unit is used for determining a relation between

a pseudo alumina concentration and a feed state, according to a distribution of the maximum

and minimum values of the pseudo alumina concentration in an integral feed period which

includes an overfeed period and an underfeed period; a coefficient position determination unit

is used for determining a sequence position of a pseudo dissolution hysteresis coefficient in

the integral feed period, according to the relation between the pseudo alumina concentration

and the feed state, and the distribution of the maximum and minimum values of the pseudo

alumina concentration in the integral feed period; a hysteresis coefficient determination unit

is used for determining values of the pseudo dissolution hysteresis coefficient in the integral

feed period, according to a difference between the sequence position of the pseudo

dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed

period; and a concentration abnormal detect unit is used for detecting whether the alumina

concentration is abnormal or not, according to the values of the pseudo dissolution hysteresis

coefficient in the integral feed period.

[0018] According to still another aspect of the present disclosure, a method for monitoring

electrolyte temperature in aluminum electrolysis driven by process mechanism knowledge

includes: determining a relation between the pseudo alumina concentration and the feed state

and the relation indicates that the alumina concentration is normal, when the pseudo alumina

concentration only has a local maximum value in an overfeed period of an integral feed period; determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period; determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period; and monitoring electrolyte temperature in the integral feed period, according to the values of the pseudo dissolution hysteresis coefficient when the alumina concentration is normal in the integral feed period.

[0019] According to still another aspect of the present disclosure, a device for monitoring electrolyte temperature in aluminum electrolysis driven by process mechanism knowledge includes: a relationship determination unit is used for determining a relation between the pseudo alumina concentration and the feed state and the relation indicates that the alumina concentration is normal, when the pseudo alumina concentration only has a local maximum value in an overfeed period of an integral feed period; a position determination unit is used for determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period; a coefficient determination unit is used for determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period; and a temperature monitor unit is used for monitoring electrolyte temperature in the integral feed period, according to the values of the pseudo dissolution hysteresis coefficient when the alumina concentration is normal in the integral feed period.

[0020] The embodiments of the present disclosure use pseudo dissolution hysteresis coefficient (PDHC for short) to quantify alumina solubility with alumina concentration changing, so the value of PDHC can be used to automatically detect abnormal state of alumina concentration in industrial cells.

[0021] By quantifying alumina dissolution property, changes of electrolyte temperature can be obtained by tracking long-term changes of PDHC, and then electrolyte temperature can be monitored in industrial cells.

BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is a schematic diagram of NCV (normalized cell voltage)ui lagging behind feed state when alumina concentration is normal.

[0023] Figure 2 is a schematic diagram of pseudo alumina concentration(PAC) P and

feed state(FDS) Fui corresponding to NCV Ul when alumina concentration is normal.

[0024] Figure 3 is a flowchart of a method for detecting abnormal alumina concentration in aluminum electrolysis driven by process mechanism knowledge according to a first

embodiment of the present disclosure.

[0025] Figure 4 is a schematic diagram of pseudo alumina concentration P4 and feed

state FU4 corresponding to NCV U4 when alumina concentration is below a normal range

(i.e. a preset range, Figure 4 is corresponding to the first type of relation between pseudo

alumina concentration P4 and feed state).

[0026] Figure 5 is a schematic diagram of pseudo alumina concentration Ps and feed

state FU5 corresponding to NCV U5 when alumina concentration is lower than the

normal range (Figure 5 is corresponding to the second type of relation between pseudo

alumina concentration P4 and feed state).

[0027] Figure 6 is a schematic diagram of pseudo alumina concentration PU6 and feed

state FU 6 corresponding to NCV U6 when alumina concentration is lower than the

normal range (Figure 6 is corresponding to the third and the forth type of relation between

pseudo alumina concentration P4 and feed state).

[0028] Figure 7 is a schematic diagram of pseudo alumina concentration P7 and feed

state FU 7 corresponding to NCV U7 when alumina concentration is lower than the

normal range (Figure 7 is corresponding to the fifth type of relation between pseudo alumina

concentration P4 and feed state).

[0029] Figure 8 is a schematic diagram of pseudo alumina concentration Ps and feed state Fs corresponding to NCV U8 when alumina concentration is higher than the normal range (Figure 8 is corresponding to the sixth type of relation between pseudo alumina concentration P4 and feed state).

[0030] Figure 9 is a schematic diagram of pseudo alumina concentration Pg and feed

state Fug corresponding to NCV U9 when alumina concentration is higher than the

normal range (Figure 9 is corresponding to the seventh type of relation between pseudo

alumina concentration P4 and feed state).

[0031] Figure 10 is a schematic diagram of pseudo alumina concentration Plo and feed

state Fuio corresponding to NCV U1O when alumina concentration is higher than the

normal range (Figure 10 is corresponding to the eighth type of relation between pseudo

alumina concentration P4 and feed state).

[0032] Figure 11 is a schematic diagram of pseudo alumina concentration Pu1 and feed

state Ful corresponding to NCV Ull when alumina concentration is higher than the

normal range (Figure 11 is corresponding to the ninth type of relation between pseudo

alumina concentration P4 andfeedstate).

[0033] Figure 12 is a diagram of normal distribution probability of a total of 800 PDHCs

divided into four groups.

[0034] Figure 13 is a schematic diagram of analysis Anomaly I according to the first

embodiment of the present disclosure.

[0035] Figure 14 is a histogram of advanced time when Anomaly I is detected.

[0036] Figure15 (a) -(d) are schematic diagrams of analysis Anomaly II according to the

first embodiment of the present disclosure.

[0037] Figure 16 is a histogram of advanced time when Anomaly II is detected.

[0038] Figure 17 is a schematic structural diagram of a device is used to detecting abnormal

alumina concentration in aluminum electrolysis driven by the knowledge of process

mechanism knowledge according to a second embodiment of the present disclosure.

[0039] Figure 18 is a flowchart of a method for monitoring electrolyte temperature of aluminum electrolysis driven by process mechanism knowledge provided by a third

embodiment of the present disclosure.

[0040] Figure 19 are diagrams of PDHC at different electrolyte temperatures, wherein:

(a) curves of normalized cell voltage U2,pseudo alumina concentration and feed state

(b) curves of normalized cell voltage U3 , pseudo alumina concentration and feed state (c) diagram of PDHC with normalized cell voltage U2 and U3.

[0041] Figure 20 is a box diagram of different electrolyte temperatures.

[0042] Figure 21 is a schematic structural diagram of a device is used to monitoring electrolyte temperature of aluminum electrolysis driven by process mechanism knowledge

provided by a forth embodiment of the present disclosure.

DETAILED DESCRIPTION

[0043] The inventive idea of the present disclosure is to control the time at which the

corresponding controlled components may complete the control instructions by controlling

the sending time of the control instructions of the controlled components, such that the

multiple controlled components that work cooperatively may be able to complete the

operations corresponding to the control instructions synchronously, and thus a mechanical or

electrical overload failure of the controlled components may be avoided, and a damage rate

of the controlled components may be reduced.

[0044] Exemplary embodiments of the present disclosure will be described in detail below

with reference to the accompanying drawings.

[0045] Before embodiments of the present disclosure are described, relationship between

alumina dissolution property and melt motion, electrolyte temperature and alumina

concentration will be described.

[0046] Besides chemical and physical properties, alumina dissolution rate is also affected by

the combined two driving forces of melt motion. One is the metal fluctuation and flow that

motivated by the electromagnetic force. The other is the circulating movement with vortices

driven by gas bubbles in bath. The alumina powder is fed per unit time and distributed throughout the cell with the melt motion, in which the well-dispersed particles dissolve rapidly, while those with poor dispersing will form agglomeration of alumina and result in the increase of dissolution time. Therefore, short-term changes in the dissolution performance of alumina are affected by the intensity of the melt motion in the cell.

[0047] According to the Arrhenius formula and the alumina dissolution rate curve in cryolite solution obtained by the experiment of Gerlach et al., the relation between

temperature T and dissolution rate K can be calculated out. The alumina dissolution rate of

a-Al 2 03 and y/-A 2 03 are denoted as follows respectively.

SK -. 71xi0 +2.59 7: (1)

InK= - 0.88X 104 +4.12 (2)

[0048] From formulas (1) and (2), it is known that the alumina dissolution rate of a-Al 03 2

and y-Al 203 both increase as the temperature increases. Generally, the electrolyte

temperature in industrial cell changes slowly, so medium-term (i.e. mid-term) and long-term changes of alumina dissolution property are affected by changes in electrolyte temperature, which indicate that increase of dissolution rate can reflect increase of electrolyte temperature.

[0049] Dissolution rate is independent on the existing alumina concentration in bath. When alumina concentration is below 5%, alumina dissolution rate obeys a zero-order reaction law, i.e., alumina dissolution rate is independent on the amount of alumina that is already dissolved. The target control range of alumina concentration in an industrial cell is typically approximately 1.5%-3.5%. Hence, alumina dissolution rate in an industrial cell has little or

no relation with alumina concentration in the bath. Therefore, in an industrial cell, the effect of change in alumina concentration on alumina dissolution properties can be negligible. However, the speed of alumina dissolution directly affects that variation in alumina concentration will be fast or slowly. Especially for large-sized and current-intensity cells, with reduction in the amount of bath that is used per unit of aluminium, variation in alumina concentration is obviously more sensitive to alumina dissolution rate.

[0050] Based on the above-described process mechanism analysis, process mechanism knowledge for alumina dissolution properties in industrial cells is derived and can be described as follows.

[0051] M1: Short-term changes in alumina dissolution properties are affected by melt motion, and the alumina dissolution property changes with intensity of the melt motion.

[0052] M2: Mid-term and long-term changes in the alumina dissolution properties are affected by electrolyte temperature. Alumina dissolution rate increases with increasing bath temperature.

[0053] M3: Alumina concentration in an industrial cell has a negligible effect on alumina dissolution properties. However, alumina dissolution properties can directly affect variation in the alumina concentration will be fast or slowly, and that the larger the industrial cell is, the stronger the effect will be.

[0054] Relation between the alumina concentration and cell voltage is described as follows. Both cell voltage and line current can be collected by a cell control system. The cell control system can record feed state, anode movement and anode effect. Electrolyte temperature is measured daily by manually inserting consumable thermocouples into the bath and is recorded. At present, it is difficult to achieve detection of alumina concentration and electrolyte temperature. Cell voltage is a widely used signal in automatic control of aluminum electrolysis.

[0055] A U-shaped curve relation between cell voltage and alumina concentration makes it possible to detect changes in alumina concentration through changes in cell voltage. Alumina concentration has a U -shaped curve (U- curve for short) relation with theoretical cell voltage (or cell resistance). The U- curve with an extremely low value of theoretical cell voltage

UA, in a medium concentration region, corresponding to concentration CL.The position

of the concentration CL_ corresponding to the cell voltage UhL fluctuates within a range

of 3% to 5% depending on electrolyte composition, electrolyte temperature, and other process

conditions. When alumina concentration is higher or lower than CL, cell resistance will

increase. At the left of an extreme point of alumina concentration, cell voltage increases significantly with decreasing concentration. When alumina concentration is reduced to an effect critical concentration CA, cell voltage will rise sharply, and anode effect will occur.

The alumina concentration interval (cE,c,_) is called an alumina concentration

controllable area, and also called a normal alumina concentration area. Cell voltage

components that change with alumina concentration mainly include anodic over potential,

ohmic potential drop in the melt, and reaction potential. Correspondingly, mechanism

knowledge is described as follows.

[0056] M4: On the left side ofC,,,, (low concentration side), the main reason of the cell

resistance increases significantly with decreasing concentration is that anode overvoltage

increases significantly with decreasing concentration. Among the anode overvoltage, the one

that has the greatest effect on the cell resistance is the overvoltage ri, caused by gas film

resistance.

[0057] M5: On the right side ofC,,, (high concentration side), overvoltage is little

affected by changes in alumina concentration. The main reason of cell resistance increases

with increasing concentration is that electrolyte resistivity increases accordingly with alumina

concentration. When alumina concentration rises to a certain level, the precipitation will also

become an important reason for increasing resistance in the high concentration region). In

order to eliminate interference caused by a series of current changes, NCV(normal cell

voltage) UN(k) is used as a main signal of automatic control of the cell in industrial

production, i.e.,

UN(k)= R, (k),, + B = U(k)-B i,, + B . Where Ro (k)= U(k)-B is the pseudo cell resistance at time 1(k) 1(k)

tk , U(k) is sampling cell voltage, 1(k) is sampling cell current, Ih is reference series

current, and B is pseudo reaction potential.

[0058] Relationship between NCV and feed state is described as follows. The relation

between NCV and feed state can be expressed by the following mechanism knowledge.

[0059] M6: If alumina concentration is normal (in the low concentration area), the NCV and

the feed state conform to an under-rise and over-fall relation, which means the NCV increases during underfeed and decreases during overfeed.

[0060] M7: If alumina concentration is normal, change trend of NCV switches from rising to falling, corresponding to alumina concentration in electrolyte switching from decreasing to increasing, which means the peak value of NCV corresponds to the lowest value of alumina concentration.

[0061] M8: If alumina concentration is abnormal, the under-rise and over-fall relation between NCV and feed state breaks down.

[0062] Based on the above-mentioned mechanism knowledge, through long-term industrial field investigation and analysis of a large number of historical data, the inventor of this application found that the time of the NCV changing from an uptrend to a downtrend slightly lags behind a transition time of underfeed period to overfeed period. Lag time is different in each integrated feed period(IFP). In a cell, alumina feed directly affects alumina concentration in bath. To further obtain alumina concentration information, this embodiment defines an integrated feed frequency (IFF), and constructs a formal representation of a practical integrated feed frequency (practical-IFF) for feed state. Generally, the end of underfeed period and pre-overfeed period in a IFP are the periods when alumina concentration in bath is the lowest. At this time, NCV is very sensitive to changes in alumina concentration in bath. To use this sensitive period as an analysis object, this embodiment defines the following.

[0063] FO is a feed state sequence. The i th underfeed period of FO and a overfeed period

appearing nearest i th underfeed period of FO are define as the i th practical integrated

feed period (practical-IFP) of FO . The i th practical-IFP is denoted by a quadruple

TF(i)p, (t(i)t(i)t(i)t(i)). t (i) is the start position of the i th underfeed period.

t,,(i) is the end position of the i th underfeed period, t(i) is the start position of the i th

overfeed period, and te(i) is a vector of the end position of the i th overfeed period.

t,(i)= t (i+ k, and t,(i + 1)= t (i)+ k2 ,k,k 2 i N+ . k, is a first conversion factor,

and indicates the number of sampling points from underfeed t,(i) to overfeed t(i) in the i th practical-IFP. k 2 is a second conversion factor and indicates the number of sampling points from the end of overfeed to(i) in the i th practical-IFP to the start of underfeed in the (i + 1) th practical-IFP, where t,(i)< t(i) < t,(i) < te(i) . Generally, alumina concentration is lowest in the middle period of each practical-IFP which is a period between the end of the underfeed period and the pre-overfeed period in a practical-IFP.

[0064] The practical-IFP length is defined as the difference between the end position of an

overfeed period and the start position of a underfeed period, i.e.'F(i) (t(i)- t (I)+1)

Dt is sampling time interval, and the i th practical-IFP duration is

Difp,FOi)0 kifP,F (i) xDt= (t(i)- t,(i)+ 1)xDt . The ith practical integrated feed frequency

(practical-IFF) ffF (i ifp,F 0

[0065] Furthermore, FO is a feed state sequence corresponding to NCV UO recorded by a

cell controller in time interval Dt . There are n IFPs in F . A practical integrated feed

matrix (practical-IFM) of the feed state FO is defined as follows.

M ° ifF ifFifFip

where Tp,F)= (tu,F ueF osF oeF i nf if I N .

[0066] Define practical-IFP duration vector T ed, and practical integrated feed frequency

vector ffeedF, for the feed stateFO .

,T Tfeed,F ° Dt ' , pF'(2), ip

-Dt'(M '1 0 0 1'+ IT =Dt'( feed,Fo

'T ffeed,F 0 fp,Ffi , fpFo(2),L ,ff,F (ni

1 , 1 1 ,L 1 L , u Dt if, f, ipF f

Specifically, Figure 1 shows NCV ui (corresponding to the curve in Figure 1, referred to as

NCV for short) and feed state Fu (corresponding to the square wave in Figure 1, referred to

as FDS) when the alumina concentration is normal. Fml has six practical-IFPs Txu(i),

i=1,---,6. The practical integrated feed matrix Mfeed,Ul is denoted as following.

S54 171 172 2999 9300 408 409 540 Q541 629 630 7629 Mfeed,U1 Q763 8 83 14 941~ Q942 1018 1019 1147V a148 1241 1242 13690 e u

[0067] Then a start position ofTju(i) is t" = 72 409 630 814 1019 1242V.

It can be seen from Figure 1 that ui changes from an uptrend to a downtrend around

tttr,U1 89 435 660 834 1058 1253(. In all IFPs, t, lags behind t_,, and lag P,1lgbhndada 18

time varies in different IFPs. Based on the above-mentioned mechanism knowledge M1, M2

and M3, it can be inferred that:

(1) Lag time can be used to characterize the solubility of bath to alumina particles in

industrial cells;

(2) Change in alumina concentration caused by transition between underfeed and overfeed

has little effect on the solubility of bath to alumina particles at the transition and the effect

can be ignored;

(3) Lag time varies in different IFPS, which is because of short-term changes in alumina

dissolution property caused by t melt movement in electrolytic cell.

[0068] Because NCV contains a variety of information such as metal fluctuations, which ia

denoted as ui (curve waveform) oscillates violently in Figure 1. It is not possible to directly

determine the precise position of the transiting of alumina concentration change trend by ui.

[0069] In Figure 2, a pseudo alumina concentration (PAC) curve is pseudo alumina

concentration P, corresponding to ui and Fui, which have the following characteristics:

(1) Pu and ui have same change trend; (2) P and Fu conform to the under-rise and over-fall relation; (3) in each IFP , (i) transition time of change trend of P, lags transition time from underfeed period to overfeed period of Fm; (4) Pm has only one maximum value, and the position of the maximum value position is not far fromt(i) in an overfeed period.

[0070] To be able to determine the precise position of the transition of alumina concentration change trend, NCV is not used directly, and pseudo alumina concentration which characterizes alumina concentration trend change is used to calculate the lag time instead. For explanation about pseudo alumina concentration, please refer to exist disclosures in aluminum electrolysis.

[0071] Figure 3 is a flowchart of a method for detecting an abnormality of alumina concentration in aluminum electrolysis based on process mechanism knowledge provided by the first embodiment of this invention. Referring to Figure 3, the method for detecting an abnormality of alumina concentration in aluminum electrolysis based on process mechanism knowledge includes S301 to S304.

[0072] In S301, a relation between a pseudo alumina concentration and a feed state is determined, according to a distribution of the maximum and minimum values of the pseudo alumina concentration in an integral feed period which includes an overfeed period and an underfeed period.

[0073] In S302, a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period is determined, according to the relation between the pseudo alumina concentration and the feed state, and the distribution of the maximum and minimum values of the pseudo alumina concentration in the integral feed period.

[0074] Specifically, this embodiment of the present disclosure defines a pseudo dissolution

hysteresis coefficient (PDHC)position td (P,F) as a sequence position used to calculate

PDHC in relation to NCV UN , which is a function of pseudo alumina concentration P and

feed state F. When the alumina concentration is normal, the PDHC position is the sequence position when the pseudo alumina concentration P is transited from an upward trend to a downward trend and where the global maximum value of P within an IFP appears. PDHC position characterizes the change trend of alumina concentration. The sequence position from a gradual decrease trend to a gradual increase trend means that alumina concentration has reached the lowest value of this IFP, as shown by te (i) in Figure 1 and Figure 2.

[0075] In S303, values of the pseudo dissolution hysteresis coefficient in the integral feed period are determined, according to a difference between the sequence position of the pseudo

dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed

period.

[0076] Specifically, this embodiment defines a PDHC D, pseudo dissolution

hysteresis coefficient, PDHC for short) as the difference between the PDHC position in a IFP

and the sequence position at the beginning of the overfeed in the corresponding IFP, i.e.

Dpdhu (P,F)= dhcu (P,F)-t jgN (F)

[0077] Specifically, this embodiment defines PDHT DpdhtuN (pseudo dissolution hysteresis

DI~dhIU\,D hC\(P, F ) t,,dhU ( P, F))- tO (F time, PDHT for short). Dh,,u' (P, F, f )- ' , where

) N N N

is sampling frequency.

tpdhcUl ttri=[189 435 660 834 1058 1253]T

[0078] In Figure 2,pdcU trU

DPdhc,Ul= tpIhc,U - 1tU I=tphcu I -Med,i -* e3 =[17 26 30 20 39 11]I

1 D when £ 0.1Hz, Dpdht,U pdhc, _[170 260T30020039011 (seconds). At

[0079] In S304, detecting whether the alumina concentration is abnormal or not, according to the values of the pseudo dissolution hysteresis coefficient in the integral feed period.

[0080] The PDHT of each IFP is approximately consistent with the results obtained in a

field experiments, which basically accords with the results of chemical experiments.

Electrolyte temperature and composition change slowly, and their short-term effects on

dissolution performance are negligible. Therefore, although the PDHC value of each IFP is

different, the range of the difference is small. It can be known from M1 that the intensity of

melt motion is the main influencing factor of dissolution lag time. The smaller PDHC and

PDHT indicate that the melt fluctuation in a cell in a IFP is more intense, and alumina

particles in a bath diffuse well, and the solubility is strong. The bigger PDHC and PDHT

indicate that the melt fluctuation in a cell in a IFP is more slowly, and alumina particles in a

bath do not diffuse well, and the solubility is weak.

[0081] In summary, the alumina dissolution property is quantified by PDHC, which can characterize relative solubility of alumina in industrial cells. As PDHC is obtained from NCV

collected and feed state recorded, it is also available. NCV is affected by process conditions

such as the current intensity, electrolyte composition, and cell temperature. PDHC also is

affected by these corresponding process conditions. The process conditions of the same cell

are relatively stable, and so as PDHC. The process conditions of different cells are different,

and so as PDHC.

[0082] Specific detection methods can refer to following description.

[0083] There is the under-rise and over-fall relation between pseudo alumina concentration

and feed state when alumina concentration is normal, which means pseudo alumina

concentration in each IFP has only one maximum value, and this maximum value appears not

farfrom t,(i) during an overfeed period, as shown in Figure 2.

[0084] In practical production, alumina concentration in a cell may be abnormal, and the

under-rise and over-fall relation between NCV and feed state will accordingly break down.

After a long-term field investigation and mechanism process analysis, it was found that there

are a variety of abnormal relations between pseudo alumina concentration and feed state. This

embodiment divides the abnormal relations related to alumina concentration abnormal

relations into nine types, and analyses underlying process semantics based on the nine types.

[0085] The following five types indicated abnormal situation of alumina concentration

below the normal range. Anode effect is usually related to alumina concentration, wettability

between the electrolyte and carbon anode, anode gas film, anode overvoltage, and critical

current density. It is generally believed that decrease in alumina concentration is the main

cause of the anode effect. Therefore, aiming at abnormal relation between NCV and feed

state before the anode effect, the acquisition of PDHC in a IFP when the alumina

concentration is low, which is discussed as follows.

[0086] For IFP2 (the corresponding relation is the first type), its characteristics is that NCV increase significantly and pseudo alumina concentration only had a maximum value in an

overfeed. NCV U4 has anode effect at tAEU4 = 1464, P 4 (as shown by the curve in

Figure 4) and FU 4 (as shown by the square wave in Figure 4). In third IFP of FU 4

TU 4(3)= (1010 1114 1115 1396) (labeled IFP2), there is a typical anomaly before the

anode effect between P 4 and FU 4 . During the overfeed period, P 4 suddenly increases

significantly. At t ... 4 (3,1) =1380, P4 has a maximum value of 4534mV. The sequence

position of the maximum value seriously lags behind t, 4 (3)= 1115 at the beginning of

the overfeed period. At t = 1330, when the rate of change of P4 is 0.3095mV / s, an

abnormality is detected and the anode is ordered to move. IFP matrix of U4 is:

6243 422 423 645 MeedU 4 u4 ,TU 4 (2),TU 4 (3)] = 646 845 846 1009. 010 1114 1115 1396

The position of PDHC is at tPDHCU4 =[517 938 1380]T.Then PDHC is denoted by

DPDHC,U4 =[ 9 4 92 265]T when f = =0.1 Hz , PDHT is denoted by At

DPDHTU 4 =[940 920 2650]T (s) .

[0087] In this embodiment, abnormal alumina concentration before the anode effect

corresponding to a phenomenon that the average value of cell resistance (or cell voltage) and

the slope of the cell resistance (or cell voltage) suddenly increases significantly is recorded as

an abnormally low alumina concentration I, denoted as Anomaly I.

[0088] In the first two IFPs T 4 (1) and T 4 (2) of U4, the relation between P4 and

Fu4 conforms to the under-rise and over-fall rule ( also called relation), but a transition

position of change trend of alumina concentration is seriously lags behind a time point when

underfeed period switches to overfeed period. Therefore, the DPDHCU4 Of U4 is

significantly greater than DPDHC,Ul of u1 with normal concentrations. So Anomaly I can be detected by DPDHCU4 in three IFPs before the anode effect occurs. It isgenerally considered that detecting an abnormality 15 minutes before the anode effect occurs is effective. It can be seen from Figure 4 that if PDHC is used, Anomaly I can be detected (1464-645)/6=136.5 minutes earlier, and if slope method is used, Anomaly I can be detected

(1464-1330)/6=22.3 minutesearlier.

[0089] Therefore, not only can the Anomaly I be detected by PDHC with a positive integer beyond a preset range, but also Anomaly I can be detected in advance by multiple IFPs.

[0090] For IFP3 (the corresponding relation is the second type), its characteristics that NCV slowly rises, and pseudo alumina concentration has only a local maximum in overall period.

[0091] As shown in Figure 5, U5 has an anode effect at tAE,U 5 = 984. In the second IFP

T1 5 (2) = (535 607 608 890) (labeled IFP3), Ps has following properties. In the early

period of T 5(2), U5 conforms to the under-rise and over-fall rule. In an overfeed period,

P5 has only a local maximum. P5 slowly rises after t = 726 , and has a global

maximum value of 4253mV at t= 890. The maximum change rate of P5 appears at

t = 863, and is only 0.0232mV / s. The main reason of P5 slowly rising after t = 726 in

overfeed period is that in early stage of overfeed period of Ts(2), alumina concentration is

very low, and wettability between electrolyte and carbon anode becomes poor, and evolved

gas easily enters into interface between anode and electrolyte. With gradual generation,

attachment and aggregation of anode bubbles, a continuous gas film is formed. The

over-voltage caused by film resistance of the continuous gas film constantly increase and

eventually leads to P5 rising after trough,U5(2,1=726.

[0092] P4 and P5 have the under-rise and over-fall relation in first n IFPs of anode

effect, but transition time of change trend of alumina concentration lags heavily behind

switching time from underfeed period to overfeed period. T 5 (2) differs to T 4 (3) in that

P does not increase suddenly and significantly, but slowly increases after

2 ttroughU3( ) =726 before anode effect. The critical concentration of the effect is approached,

but the average value and slope of cell voltage have not increased significantly before anode effect. Here, abnormally low alumina concentration when pseudo alumina concentration changes slowly is recorded as abnormally low alumina concentration II, noted as Anomaly II.

Anomaly II cannot be detected using slope method. PDHC of T,(2) calculated by using

the local maximum value t.. (2,1)= 646 is DPDHCU 5 (2)= 646-608= 38. Anomaly II

cannot be distinguished from the situation when the concentration is normal based on PDHC

of T5(2). In order to enable PDHC to detect Anomaly II, the position of the global

maximum value of pseudo alumina concentration in overfeed period of a IFP is set to be

PDHC position. As shown in Figure 5, the global maximum value position 890 of Ps is

the PDHC position during the overfeed period of T 5 (2). Corresponding, PDHC is

DPDHC,5( 2 )= 2 82 , i.e., DPDHCU 5 =[103 282T

In this way, DPDHCU5 can detect Anomaly II shown in Figure 5 in two IFPs before anode

effect, and Anomaly II cannot be detected by the slope method.

[0093] For IFP4 (the corresponding relation is the third type), its characteristics is that during the overfeed period, NCV first decreases and then rises, and pseudo alumina concentration has only a minimum value during overfeed period.

[0094] The NCV U6 is shown in Figure 6, and the anode effect occurs at tAEU 6 = 882.

IFP TU 6 (4) and TU 5 (2) are similar and both of them are Anomaly II. TU 6 (3) and

TJ6(1) are also similar. T16 (1)=(18 40 41 195) (labeled IFP4) as an example is

described as follows. Within TUX1), Pve falls first and then rises. When ttrough,U(1,1)= 98,

there is a minimum value.

[0095] Within te [40,98], Pd ecreases with increasing concentration. According to the

U-curve, it can be inferred that alumina concentration is in a low concentration region. When

te [197,287], P 6 conforms to the under-rise and over-fall relation which indicates that alumina concentration is still in low concentration region. It can be inferred that: (1) after

[t" U1) - ts U(1/ f, = (195-41) /0.1=1540 seconds overfeed, T 6 (1) is still in low

alumina concentration region; (2) after t troughU(1,1)=98 , the leading factor for P6

increase is not high concentration region (the right side of CL, ), but increasing of

overvoltage ci7, caused by film resistance. A positive integer with a large absolute value is

used to represent low concentration in TJ6(1) and TU6 (3) , and the end position of

tpDCU (1) = toe (1) =195 overfeed is used to represent the PDHC position. PDHCU6 oeU6

tPDHC,U6(3) oe,U6(3)= 465, then the PDHCs of T 6 (1) and TU6(3) are denoted as follows

respectively.

DPDHCU 6 (1) tPDHCU 6 (1)-s,U6(1) =195- 41=154.

DPDHCU 6 (3) tPDHCU 6 (3)-,U6(3)= 465- 400 = 65.

[00961 For IFP5 (the corresponding relation is the fourth type), its characteristics is that the

NCV rises and falls repeatedly during the overfeed period, and pseudo alumina concentration

monotonically increases during underfeed period, and has multiple local maxima during

overfeed period.

[0097] The second IFP TU6 (2)=(197 219 220 375) of U6 (labeled IFP5) has the

following characteristics: during the underfeed period, PU6 monotonically increases without

local extreme points; there is the under-rise and over-fall relation when te [197,287] p6

increases again after ttrough,U6( 2 , 1) = 287, so there are two local maxima in overfeed period,

i.e. ert,U6(2,1)=233 andterest,U6(2,2)=350.

[0098] The under-rise and over-fall relation in early stage of TU6 (2) means that

concentration of TU6 (2) is in low concentration region. In the following TU6 (3) ,

concentration is still in low concentration region. It can be inferred that: (1) overfeed period lasts [te (2)-t U6 (2)]/f =(375-220)/0.1 =1550 seconds and does not make alumina concentration in TU 6(2) enter into high concentration region; (2)a main factor for P6 rising and falling after tu( 2) =287 is generation, growth and departure of anode gas.

After teret,U6(2,2) =350 , the decrease of the film resistance overvoltage pf, with

departure of the gas film leads to the decrease of P6. Until underfeed period of TU 6 (3),

PU6 is still affected by the main factor. To use a positive integer with a large absolute value

to indicate low concentration in TU 6 (2), the last local maximum position in overfeed period

is used to indicate the PDHC position, i.e., tPDHCU6(2) cretU6(2,2)= 350, and then PDHC is

DPDHCU6 (2) 130

[0099] In summary, anomalies in four IFPs of U6 are classified into abnormal situations labeled IFP4, IFP5, IFP4, and IFP3, respectively. In this way, abnormal concentration can be

detected on DPDHC,U6=[154 130 65 150]' in the fourth IFP before the anode effect.

[00100] For IFP6 (the corresponding relation is the fifth type), its characteristics is that the NCV rises slowly throughout entire IFP, and pseudo alumina concentration monotonically increases.

[00101] As shown in Figure 7, the anode effect occurs at tAEU 7 = 442. Before the anode

effect, P7 rises slowly and has a maximum value 4385mV at t = 425 and a maximum

change rate 0.1010mV / s at t = 287. The main factors leading to the rise of P7 are poor

wettability of anode surface, accumulation of anode gas, and continuous increase of film resistance overvoltage. The slope method also cannot detect such anomalies. As this type of anomaly usually occurs before the anode effect. In order to continue to use PDHC with a positive value to characterize the type of anomaly, the end position of overfeed period of IFP

is set to the PDHC position, i.e., tPDHCU7(1) t,U7(1)= 425 , and then PDHC

DPDHCU 7 (1)= 151.

[00102] The relation between pseudo alumina concentration and feed state in IFPs labeled IFP2 to IFP6 indicates process semanteme of that the under-rise and over-fall relation which

is caused by poor wettability of anode surface due to low alumina concentration,

accumulation of anode gas, and continuous increase of over-voltage of film resistance. In

summary, the setting of PDHC position driven by process semanteme makes positive value of

PDHC exactly reflect cell condition when the under-rise and over-fall relation is broken

down due to alumina concentration lower than the normal range.

[00103] Four types of abnormal alumina concentration higher than the normal range are described as follows.

[00104] For IFP7 (the corresponding relation is the sixth type), the characteristics can be described as: the relation between NCV and feed state is the under-fall and over-rise relation,

which means the NCV decreases during underfeed and increases during overfeed, and pseudo

alumina concentration has only a maximum value during underfeed period.

[00105] As shown in Figure 8, within Ts(1)= (225 985 986 1118), U8 conforms to

the under-fall and over-rise relation when the concentration is normal. P8 has only one

maximum value in underfeed period. P8 changes from a downtrend to an uptrend at

t = 864 in underfeed period.

[00106] The pseudo alumina concentration in Tus(1) and T (1) conforms to a relation

which alumina concentration decrease first and then increase. The difference is that the

minimum value of P6 appears in overfeed period for T 6 (1), while the minimum value of

PJ appears in underfeed period for Tus(). The difference indicates that pseudo alumina

concentration and feed state for Tus(1) have completely different semanteme as for TU(1).

[00107] At beginning of Tus(1) , the relation between PJ and Fus is a under-fall

relation. It can be inferred from the U-curve that concentration at beginning of Tu (1) is

high. After underfeed lasting for (1)u- troughU8 sU8 = (864 - 225)/0.1 = 6390

seconds, concentration returns to low region at t =864. Therefore, in underfeed period after t =864, the relation between PJ and Fus is the under-rise relation which means the

NCV increases during underfeed. At t (l)=985, the maximum underfeed duration is

reached. Subsequently, after overfeed with lasting

toU(l)8Jf= (oe,U81)- (1118-100/0.1= 1180seconds,concentrationgraduallyenters

into high area, and a relation between PJ and Fus is the under-rise relation. Therefore,

the pseudo alumina concentration and feed state conform to the under-fall and over-rise

relation as shown by Tus(1) in Figure 8, it means that concentration enters into high

concentration region in the later stage of IFP.

[00108] Set PDHC position of Tus(l) in Figure 8 to tPDHCU 8(1) c t,U8(11)=237, then

PDHC DPDHCU8(1) =237 -986= -749. In this way, a negative integer with a large absolute

value (referred to as "negative large") is used to indicate abnormal situation of concentration

in high concentration region, which is referred to as Abnormally High.

[00109] For IFP8 (the corresponding relation is the seventh type), the characteristics is that

the relation between NCV and the feed state conform to an under-fall and over-rise relation,

which means the NCV decreases during underfeed and increases during overfeed. Pseudo

alumina concentration has no local maximum and has only one minimum value during the

underfeed period.

[00110] As shown in Figure 9, within Tv9 (1) = (233 992 993 1125), the relation

between U9 and F9 is a under-fall and over-rise relation, which is opposite to the

under-rise and over-fall relation when alumina concentration is normal. There is no local

maximum in entire IFP. Pg changes from a downtrend to an uptrend at t = 909 in

underfeed period. Tu,(l) is similar to Tu,(1) , as shown in Figure 8. The pseudo alumina

concentration conforms to an under-fall and over-rise relation, which indicates that

concentration is in a high region and belongs to an abnormally high type. Similar to T (1)

the starting position of underfeed of Tu9 (1) represents the position of PDHC, i.e., tPDHC,U9(1) U9 (1)= 233.Then PDHC is

DPDHCU9(1) tPDHCU99(1)-to9(1) t u9(1)-tob(1) 233-993 o,-760.

[00111] For IFP9 (the corresponding relation is the eighth type), the characteristics is that NCV is relatively flat in underfeed period, and pseudo alumina concentration has multiple

local maxima in underfeed and overfeed periods.TU 0 (2)= (1126 1777 1778 1904) as

shown in Figure 1.When te [1126,1557], the overall trend of Ul0 is flat, but there are

slight fluctuations, and Pio has multiple local maximums. After t",ughUio(2,3)=1557,

PEvo gradually rises until the first local maximum in overfeed period. When

te [1557,1904] , the relation between the NCV PNio and feed state Foo conforms to the

under-rise and over-fall relation, and the trend transition point is at tJ5(2,1) =1811.

[00112] From the above analysis, it can be known that the concentration in Fio(I) is high.

Within Fmo(2),when te[1126,1557], the overall trend of Pio is flat. From the U-curve,

it can be known that the concentration is in middle concentration range. When

te[1557,1904], the under-rise and over-fall relation of Pmo indicates that concentration

returns back to low concentration area at t=1557 after 4310 seconds of underfeed (The

underfeed interval is increased at t=1331). The. There is a similar situation at Fto(3),

there are two local maxima at tU1(3,1)=2712 and C ,m(3,2)=2858 in overfeed

period, but the difference is overall trend for Fmo(3) isdownward.

[00113] In summary, in early stage of T 10 (2) and Tlo(3) , concentration is in middle

concentration area and returns back to the normal range (low concentration area) in the late end of the underfeed( t=1557 and t= 2591)after along period of underfeed.

[00114] In order to use PDHC to represent process semanteme in T 10 (2) and Tlo(3) in

which concentration returns back to the normal range, the first local maximum value position

in overfeed period of T 10 (2) and Tul(3) is set to the PDHC peak position.

3 tPDHC,UO( 2 ) -et,Ul(0 18 11tPDHCUlO(3 ) tO( ,1)=2712

Then PDHC is DPDHC,UO(2) tPDHC,UO(2)-tUl 0 (2) 1811 1778 33.

DPDHCUl(0 tPDHC,Ulo(3)-s,Ulo(3) 2712-2668 44.

DPDHC,U=[-760 33 44]. DPDHCUI0(1)= -760 indicates that concentration in the first IFP

is high. For the following two IFPs, DPDHC,Ul (2)=33 and DPDHC,UO(3) =44 indicate that

concentration gradually decreases to the normal range. Thus, a negative value of PDHC and a subsequent normal value can be used to represent the semanteme of alumina concentration gradually returning from a higher level to a normal level.

[00115] For IFP1O (the corresponding relation is the ninth type), the characteristics is that NCV Ull is relatively flat in underfeed period. The pseudo alumina concentration has multiple local maximums in underfeed period, and monotonically increase in overfeed period

[00116] As shown in Figure 11, in Tu(1)= (511 1269 1270 1554) the following

phenomena occurs. In underfeed period of Tu(1), the overall trend of Ul1 is relatively

flat, and PufI has 4 local maxima. t"estul(1,1)=668 , test,Ui(1, 2 )=810

t:ret,Ull(1,3)=1011, t" tUll(1,4)=1142. In the later period of underfeed and the entire

overfeed period, Pul conforms to the under-fall and over-rise relation. During overfeed

period, Pu monotonically increases.

[00117] Tu(1 ) differs toT10(2) in that Pu conforms to the under-fall and over-rise

relation when t e [1142,1554]. From the U-curve, it can be known that the concentration in

underfeed period is in middle concentration range, and does not return back to low concentration range until the end of underfeed. In overfeed period, concentration continues to increase because feed amount is higher than electrolytic consumption. During overfeed

period, monotonic increasing of Pl indicates that alumina concentration is in high

concentration region.

[00118] To use a negative large value of PDHC to characterize high concentration in

TU,,(1), the first local maximum value position in underfeed period is set to the PDHC peak

position, i.e., tPDHC,Ul11)=tu ut 1 (1,l 668•

Then PDHC is DPDHC,Ul(1) tPDHC,Ul1(1) ,U(1) 668-1270 -632.

[00119] The procedure CalculatePDHC(P,F,f) calculates PDHCDPDHC and PDHT of

pseudo alumina concentration curve P and feed state F with a sampling frequency off.

[00120] It can be known that when alumina concentration in electrolyte is normal, cell voltage and feed state conform to the under-rise and over-fall relation, and the PDHC value changes within the normal range. When concentration is abnormal, the under-rise and over-fall relation breaks down, and the PDHC value also changes. Therefore, the range of alumina concentration can be detected according to the change range of PDHC value.

[00121] Figure 12 is a normal distribution probability chart of a total of 800 PDHCs divided into four groups. The first group of PDHCs is represented by a third distribution from left to right in figure 12, which is the normal distribution probability chart of 200 PDHCs (represented by LT) at a lower electrolyte temperature with normal alumina concentration. The second group of PDHCs is represented by a second distribution from left to right in Figure 12, which is the normal distribution probability chart of 200 PDHCs (represented by HT) at higher electrolyte temperature with normal alumina concentration. The third group of PDHCs is represented by the first distribution from left to right in Figure 12, which is the normal distribution probability chart of 200 negative PDHCs value divided by 10 and rounded when the under-rise and over-fall rule break down (represented by ALC). The fourth group of PDHCs is represented by the fourth distribution from left to right in Figure 12, which is a normal distribution probability chart of 200 abnormally high PDHCs (represented by AHC) before the anode effect. The fourth group of PDHCs is based on NCV (the anode effect occurred at the 2883th sampling point) of the first 8 hours (a total of 2880 sampling points) of 200 anode effects that have occurred in a total of 20 cells in multiple factories.

[00122] From the T test, the above four groups of PDHC all obey the normal distribution. Table 1 is the statistical characteristics of the four groups of PDHC in Figure 12. When alumina concentration is normal, Skewness coefficient of PDHCs at lower electrolyte temperature and higher electrolyte temperature are both greater than 0, and the distribution is slightly to the right side. Kurtosis coefficients are all less than 0, and the absolute value is small. Therefore, it can be considered that PDHCs obey a thin-tailed distribution when alumina concentration is normal. When alumina concentration is abnormal, Kurtosis coefficients are all greater than 0, and the absolute values are relatively large. PDHCs obey a fat-tailed distribution. The distribution of abnormal low value (ALV) is slightly to the left side, and the distribution of abnormal high value (AHV) is slightly to the right side. In addition, the mean, the standard deviation, the range, and the coefficient of variation of the four groups of PDHC are also different each other, which indicates that PDHC can be used not only for electrolyte temperature monitoring, but also for low alumina concentration. The mean, standard deviation, range, coefficient of variation, skewness, and kurtosis in Table 1 all can be calculated with common methods.

[00123] Table 1 Statistical characteristics of PDHC in different cell conditions Group Classification Mean Standard Range Variation Skewness Kurtosis Alumina

deviation coefficient concentration

1 LT: Low 35.71 15.8157 78 44.2893 0.28585 -0.0962 Normal

temperature

2 HT: High 21.025 9.9841 55 47.4869 0.29337 -0.0229 Normal

temperature

3 ALV: -15.95 18.3121 89 -114.8096 -0.85948 0.4324 Anomaly-High

Abnormally

low value

4 AHV: 138.095 44.3425 219 32.1101 0.98229 0.5122 Anomaly-Low

Abnormally

high value

[00124] Detection of anomaly-low alumina concentration is described as follows. In

industrial production, the detection of abnormal alumina concentration is of great significance.

The earlier an abnormal concentration is detected, the sooner measures can be taken to prevent anode effects, and the sooner the cell conditions are stabilized, thus increase productivity. It is generally believed that anomaly-low alumina concentration is detected at minutes before anode effect can effectively prevent anode effect to occur.

[00125] Further analysis of the above 200 NCV samples in the fourth group, it is found that: samples belongs to Anomaly I, 102 samples belongs to Anomaly II, and 13 anode effects has nothing to do with alumina concentration. For 187 samples related to alumina concentration, the difference between the sequence position at the end of IFP where PDHC first appeared abnormally and the sequence position when anode effect occurs is the earliest

time to detect concentration anomaly, i.e., when DPDHC,U ()3 qPDHC

' t = (tAEU oe,U(i) , in which there are slight changes for qPDHC with different cell

process conditions such as electrolyte temperature and composition.

[00126] Analysis of Anomaly I,it can be seen from Figure 4 that t Anomaly I can be detected 13 minutes earlier based on PDHC method than the slope method. For the worst case shown in Figure 13, PDHC method and the slope method detected Anomaly I at the same time. Figure 14 is a histogram of the time when Anomaly I was detected for the first time before the occurrence of anode effect using PDHC. Among 85 samples, only 8 samples are detected within 15 minutes before anode effect, and 90.6% of the 85 samples are able to detect low alumina concentrations 15 minutes earlier than anode effect. Of all samples, the average time of abnormally detected for the first time is 126 minutes earlier.

[00127] Analysis of Anomaly II, Table 2 is a list of NCV instances in which Anomaly-Low II is detected by 1 to 6 actual-IFPs in advance using the PDHC method.

Table 2 PDHC of Anomaly II and its position (before the anode effects occur) NCVNO. U12 U13 U6 U14 U5 Ul5 Electrolyte temperature (Celsius) 970 964 964 977 973 967 Number of advanced IFPs 6 5 4 3 2 1 Sequence position when effects occur : tA E,U 2386 1464 882 1340 984 1274 Sequence position when abnormality is first detected:tu(i) 519 396 195 649 534 1154

First detection of abnormal time (minutes) 319.7 177.8 114.5 115.7 75 20

Figure 15 shows pseudo alumina concentration, feed state, and PDHC position of U12, U13,

U14, and Ul5 in the table. Using CalculatePDHC () algorithm to get their PDHC.

DPDHC,U2 =[20 134 158 70 191 251 226] DPDHC,U23 8 118 173 182 T

DPDHC,U14 170 106 143]TD = [33 62 26 160]T .Figure 15 (a) shows PU12

Fu12 and tPDHC,U12. Figure 15 (b) shows P13 , Fu13 , and tPDHC,U13. Figure 15 (c) shows

P 14 , F 14 , and tPDHC,U14. Figure 15 (d) shows P 1 51 , F 5 , and tPDHC,U15•

[00128] Figure 16 is a histogram of the time when the Anomaly II is first detected prior to the occurrence of anode effect using PDHC. For the 102 samples with Anomaly II, they cannot

be detected by the slope method, 94.1% of the 102 samples can be detected 15 minutes earlier

based on PDHC, and the average advanced time was 144 minutes earlier based on PDHC.

[00129] It can be known that, for all 200 samples, 86.5% of the 200 samples can be detected minutes earlier using the PDHC method; the rest 6.5% of the 200 samples can be detected

by neither the PDHC method nor the slope method. Compared with the slope method, the

PDHC method can detect Anomaly I in advance. The PDHC method can detect Anomaly II

that cannot be detected by the slope method; and also detect abnormalities such as

Anomaly-High alumina concentration.

[00130] In this embodiment, the alumina dissolution property is quantified by PDHC, and

relative solubility of alumina in industrial cells can be characterized. Observing the

short-term changes of PDHC can help understanding alumina solubility and concentration in

electrolytic cell. By tracking long-term changes of PDHC, electrolyte temperature change in

industrial cells can be monitored. PDHC can provide a basis for detection of alumina

concentration and monitoring of electrolyte temperature. As PDHC value has a specific

meaning, it is an explicit representation of process mechanism knowledge implicit in NCV

and feed state. PDHC for detection of alumina concentration has the advantages of high

versatility, low calculation cost, and good interpretability.

[00131] In this embodiment, alumina concentration is quantified by PDHC and PDHT. By

calculating PDHC when alumina concentration is normal or abnormal, it can be know that

PDHC value changes within the normal range when concentration is normal. PDHC is a

large positive value when concentration is Anomaly-Low, and PDHC is negative with a large absolute value when concentration is Anomaly-High. In this way, alumina concentration state of industrial cells is automatically detected by PDHC.

[00132] Figure 17 is a structural block diagram of an alumina concentration abnormality detection device for aluminum electrolysis driven by the process mechanism knowledge provided by the second embodiment of the present disclosure, which may be used to execute operations of the method of the first embodiment shown in Figure 3, and the explanation for Figure 1 to Figure 16 may be applied to the second embodiment. As shown in Figure 17, the alumina concentration abnormality detection device includes a relation determination unit 1701, a coefficient position determination unit 1702, a hysteresis coefficient determination unit 1703, and a concentration abnormal detect unit 1704.

[00133] The relation determination unit 1701 is used for determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum values of the pseudo alumina concentration in an integral feed period which includes an overfeed period and an underfeed period.

[00134] The coefficient position determination unit 1702 is used for determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution of the maximum and minimum values of the pseudo alumina concentration in the integral feed period.

[00135] The hysteresis coefficient determination unit 1703 is used for determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period.

[00136] The concentration abnormal detect unit 1704 is used for detecting whether the alumina concentration is abnormal or not, according to the values of the pseudo dissolution hysteresis coefficient in the integral feed period.

[00137] Specifically, the concentration abnormal detect unit1704 is further used for determining the alumina concentration is abnormal low, when the values of the pseudo dissolution hysteresis coefficient in the integral feed period are positive and beyond a preset normal range; and determining the alumina concentration is abnormal high, when the values of the pseudo dissolution hysteresis coefficient in the integral feed period are negative and beyond the preset normal range.

[00138] The embodiments of the present disclosure, process mechanism knowledge is combined, and the relation between NCV and feed state is explored. An implicit essential relation between alumina dissolution property, alumina concentration, electrolyte temperature, and cell voltage is found. It is explicitly expressed by PDHC, which reflects a profound relation between cell voltage and alumina dissolution property, alumina concentration, and electrolyte temperature. PDHC quantifies the alumina dissolution property in industrial cells. PDHC not only solves the problem of monitoring of electrolyte temperature, but also detects abnormal alumina concentration. Compared with the slope method, PDHC can not only detect abnormal alumina concentration earlier multiple IFPs, but also detect Anomaly II that cannot be detected by the slope method. Therefore, electrolyte temperature monitoring and abnormal detection of alumina concentration based on PDHC has the advantages of low cost, high accuracy, low calculation cost, good interpretability and reproducibility.

[00139] Figure 18 is a flowchart of a method for monitoring electrolyte temperature of aluminum electrolysis driven by process mechanism knowledge provided by an embodiment of the present disclosure. As shown in Figure 18, the method for monitoring electrolyte temperature of aluminum electrolysis includes S1801 to S1804.

[00140] In S1801, a relation between the pseudo alumina concentration and the feed state and the relation indicates that the alumina concentration is determined normal, when the pseudo alumina concentration only has a local maximum value in an overfeed period of an integral feed period.

[00141] In S1802, a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period is determined.

[00142] In S1803, values of the pseudo dissolution hysteresis coefficient in the integral feed period is determined, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period.

[00143] In S1804, electrolyte temperature is monitored in the integral feed period, according to the values of the pseudo dissolution hysteresis coefficient when the alumina

concentration is normal in the integral feed period.

[00144] The specific monitoring principle and method is described as follows. The electrolyte temperature in a cell changes slowly. It can be known from process mechanism

knowledge M2 that if PDHC and PDHT are tracked for a long time, temperature information

can be obtained. To analyze change of PDHC value under different electrolyte temperatures,

selecting NCV U2 and U3 with the under-rise and over-fall relation when the

electrolyte temperature is K =965C (Figure 19 (a)) and K =973C (Figure 19 (b)) when

alumina concentration is normal, and calculate PDHCs as below.

DPDHC,U2 =[32 38 31 41 44 53 33 55 26 42]T

DPDHC,U3 =[18 40 41 25 33 25 37 30 27 30]T

[00145] Figure 19 (c) is plotted after sorting DPDHCU2 and DPDHCU3 in ascending order. In

Figure 19 (c), "." indicates sorted DPDHCU2, and "*" indicates sorted DPDHCU3. PDHCs at

lower electrolyte temperature are significantly higher than those at higher temperature, which

is consistent with qualitative relation between alumina dissolution rate and increase in

temperature based on mechanism analysis. Therefore, tracking PDHC for a longer period of

time can help obtain the information of electrolyte temperature change.

[00146] To further analyze relation between electrolyte temperature and PDHC, a large

number of NCV data in continuous electrolytic cells with lower temperature (955°C-965°C)

and higher temperature (966°C-978C) are selected based on electrolyte temperature data

measured manually. In order to avoid impact of manual operation and special cell conditions

on analysis of PDHC properties, the PDHCs are selected when alumina concentration is

normal in first three consecutive IFPs without special cell conditions such as metal tapping,

anode changing, beam raising, and no anode effect. Box plots of 200 selected PDHCs at

higher and lower electrolyte temperatures and corresponding order statistics are drawn, as

shown in Figure 20 and Table 3.

Table 3 Ordinal statistics of PDHC in different electrolyte temperatures

Temperature Median Lower Upper Quartile Triple Lower Upper

type quartile quartile deviation mean outlier outlier

cut-offs cut-offs

LT: Low 36 24 45 21 35.25 -7.5 76.5

temperature

HT: High 21 14 27.5 13.5 20.875 -6.25 47.75

temperature

[00147] In Figure 20, the upper dashed line is the median of PDHCs at 200 low temperatures, and the lower dashed line is the median of PDHCs at 200 high temperatures. For other

ordinal statistics, see Table 3. Figure 19 (c), Figure 20, and Table 3 show that PDHC is more

distinguishable between higher and lower electrolyte temperatures, and absolute ordinal

statistics of PDHC at lower temperatures are significantly larger than those at higher

temperatures, which means dissolution rate of the alumina in low temperature electrolytes is

significantly slower than that of high temperature electrolytes.

[00148] Specifically, the electrolyte temperature is determined abnormal high, when the values of the pseudo dissolution hysteresis coefficient when the alumina concentration is

normal in the integral feed period are lower than a first preset coefficient value, and the

electrolyte temperature is determined abnormal low, when the values of the pseudo

dissolution hysteresis coefficient are higher than a second preset coefficient value, when the

alumina concentration is normal in the integral feed period.

[00149] With other process conditions unchanged, electrolyte temperature changes slowly,

and alumina solubility is affected by the long-term influence of electrolyte temperature. If

change of PDHCs is tracked for a long time, temperature change of electrolytes in industrial

cells can be obtained. This property of PDHC can be used for monitoring of electrolyte

temperature, which can change the current situation of measuring electrolyte temperature

once a day by manual use of thermocouples, and alleviate the difficulty of measuring

electrolyte temperature changes. In addition, PDHC can also be used as a feature to recognize various types of working conditions that are related to electrolyte temperature and superheating, or be used as a basis for automatic labelling in an intelligent algorithm.

[00150] Figure 21 is a structural block diagram of an electrolyte temperature monitoring device driven by the mechanism and process knowledge provided by an embodiment of this

invention, which may be used to execute operations of the method of the embodiment shown

in Figure 18, and the explanation for Figure 18 to Figure 20 may be applied to the forth

embodiment. As shown in Figure 21, the electrolyte temperature monitoring device includes a

relationship determination unit 2101, a position determination unit 2102, a coefficient

determination unit 2103, and a temperature monitor unit 2104.

[00151] The relation determination unit 2101 is used for determining a relation between the pseudo alumina concentration and the feed state and the relation indicates that the alumina

concentration is normal, when the pseudo alumina concentration only has a local maximum

value in an overfeed period of an integral feed period.

[00152] The position determination unit 2102 is used for determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period, according to the

relation between the pseudo alumina concentration and the feed state, and the distribution of

the maximum and minimum values of the pseudo alumina concentration in the integral feed

period.

[00153] The coefficient determination unit 2103 is used for determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference

between the sequence position of the pseudo dissolution hysteresis coefficient and a start

sequence position of overfeed in the integral feed period.

[00154] The temperature monitor unit 2104 is used for monitoring electrolyte temperature in

the integral feed period, according to the values of the pseudo dissolution hysteresis

coefficient when the alumina concentration is normal in the integral feed period.

[00155] In this embodiment, through exploratory analysis, it is found that an implicit

essential relation between alumina dissolution property, alumina concentration, electrolyte

temperature, and cell voltage, which can be explicated by PDHC. The PDHC reflects the

profound relation between cell voltage and alumina dissolution property, alumina

concentration, and electrolyte temperature, which quantifies the alumina dissolution property in industrial cells. PDHC can solve the problem of monitoring of electrolyte temperature that cannot be achieved in the aluminum electrolysis industry.

Claims (18)

WHAT IS CLAIMED IS:

1. A method for detecting abnormal alumina concentration in aluminum electrolysis

driven by process mechanism knowledge, comprising:

determining a relation between a pseudo alumina concentration and a feed state,

according to a distribution of the maximum and minimum values of the pseudo alumina

concentration in an integral feed period which comprises an overfeed period and an

underfeed period;

determining a sequence position of a pseudo dissolution hysteresis coefficient in the

integral feed period, according to the relation between the pseudo alumina concentration and

the feed state, and the distribution of the maximum and minimum values of the pseudo

alumina concentration in the integral feed period;

determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period; and detecting whether the alumina concentration is abnormal or not, according to the values of the pseudo dissolution hysteresis coefficient in the integral feed period.

2. The method according to claim 1, wherein the detecting whether the alumina concentration is abnormal or not, according to the pseudo dissolution hysteresis coefficient in each integral feed period comprises: determining the alumina concentration is abnormal low, when the values of the pseudo dissolution hysteresis coefficient in the integral feed period are positive and beyond a preset normal range; determining the alumina concentration is abnormal high, when the values of the pseudo dissolution hysteresis coefficient in the integral feed period are negative and beyond the preset normal range.

3. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is a first type, when the pseudo alumina concentration only has a local maximum value in the overfeed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the first type, determines a sequence position of the local maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

4. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is a second type, when the pseudo alumina concentration has a local maximum value and a global maximum value in the overfeed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the second type, determines a sequence position of the global maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

5. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is a third type, when the pseudo alumina concentration only has a minimum value in the overfeed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the third type, determines a sequence position of the end of overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

6. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is a forth type, when the pseudo alumina concentration monotonically increase in underfeed period, and has multiple local maximum values in the overfeed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the forth type, determines a sequence position of the last local maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

7. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is a fifth type, when the pseudo alumina concentration monotonically increase in the integral feed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the fifth type, determines a sequence position of the end of overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

8. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is a sixth type, when the pseudo alumina concentration only has a local maximum value in the underfeed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the sixth type, determines a sequence position of the beginning of the underfeed period as the sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period.

9. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is a seventh type, when the pseudo alumina concentration only has a local minimum value in the underfeed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the seventh type, determines a sequence position of the beginning of the underfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

10. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is an eighth type, when the pseudo alumina concentration has multiple local maximum value in the underfeed period and overfeed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the eighth type, determines a sequence position of the first local maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

11. The method according to claim 2, wherein the determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum of the pseudo alumina concentration in each integral feed period comprises: determining the relation between the pseudo alumina concentration and the feed state is a ninth type, when the pseudo alumina concentration has multiple local maximum values in the underfeed period, and monotonically increase in overfeed period; the determining a sequence position of a pseudo dissolution hysteresis coefficient in each integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution state of the pseudo alumina concentration in each integral feed period comprises: when the relation between the pseudo alumina concentration and the feed state is the ninth type, determines a sequence position of the first local maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

12. A device for detecting abnormal alumina concentration in aluminum electrolysis driven by process mechanism knowledge, comprising: a relation determination unit, is used for determining a relation between a pseudo alumina concentration and a feed state, according to a distribution of the maximum and minimum values of the pseudo alumina concentration in an integral feed period which comprises an overfeed period and an underfeed period; a coefficient position determination unit, is used for determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period, according to the relation between the pseudo alumina concentration and the feed state, and the distribution of the maximum and minimum values of the pseudo alumina concentration in the integral feed period; a hysteresis coefficient determination unit, is used for determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period; and a concentration abnormal detect unit, is used for detecting whether the alumina concentration is abnormal or not, according to the values of the pseudo dissolution hysteresis coefficient in the integral feed period.

13. The device according to claim 12, wherein the concentration abnormal detect unit is further used for determining the alumina concentration is abnormal low, when the values of the pseudo dissolution hysteresis coefficient in the integral feed period are positive and beyond a preset normal range; and determining the alumina concentration is abnormal high, when the values of the pseudo dissolution hysteresis coefficient in the integral feed period are negative and beyond the preset normal range.

14. A method for monitoring electrolyte temperature in aluminum electrolysis driven by process mechanism knowledge, comprising: determining a relation between the pseudo alumina concentration and the feed state and the relation indicates that the alumina concentration is normal, when the pseudo alumina concentration only has a local maximum value in an overfeed period of an integral feed period; determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period; determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period; and monitoring electrolyte temperature in the integral feed period, according to the values of the pseudo dissolution hysteresis coefficient when the alumina concentration is normal in the integral feed period.

15. The method according to claim 14, wherein the monitoring electrolyte temperature in the integral feed period, according to the values of the pseudo dissolution hysteresis coefficient when the alumina concentration is normal in the integral feed period comprises: determining the electrolyte temperature is abnormal high, when the values of the pseudo dissolution hysteresis coefficient are lower than a first preset coefficient value and the alumina concentration is normal in the integral feed period, and determining the electrolyte temperature is abnormal low, when the values of the pseudo dissolution hysteresis coefficient are higher than a second preset coefficient value and the alumina concentration is normal in the integral feed period.

16. The method according to claim 15, wherein the determining a sequence position of a pseudo dissolution hysteresis coefficient in an integral feed period comprises: determining a sequence position of a local maximum value in the overfeed period as the sequence position of the pseudo dissolution hysteresis coefficient in the integral feed period.

17. A device for monitoring electrolyte temperature in aluminum electrolysis driven by process mechanism knowledge, comprising: a relationship determination unit, is used for determining a relation between the pseudo alumina concentration and the feed state and the relation indicates that the alumina concentration is normal, when the pseudo alumina concentration only has a local maximum value in an overfeed period of an integral feed period; a position determination unit, is used for determining a sequence position of a pseudo dissolution hysteresis coefficient in the integral feed period; a coefficient determination unit, is used for determining values of the pseudo dissolution hysteresis coefficient in the integral feed period, according to a difference between the sequence position of the pseudo dissolution hysteresis coefficient and a start sequence position of overfeed in the integral feed period; a temperature monitor unit, is used for monitoring electrolyte temperature in the integral feed period, according to the values of the pseudo dissolution hysteresis coefficient when the alumina concentration is normal in the integral feed period.

18. The device according to claim 17, wherein the temperature monitor unit comprises: a first monitor module, is used for determining the electrolyte temperature is abnormal high, when the values of the pseudo dissolution hysteresis coefficient are lower than a first preset coefficient value and the alumina concentration is normal in the integral feed period, and a second monitor module, is used for determining the electrolyte temperature is abnormal low, when the values of the pseudo dissolution hysteresis coefficient are higher than a second preset coefficient value and the alumina concentration is normal in the integral feed period.