CN104965967B - A kind of yield real-time predicting method of atmospheric and vacuum distillation unit - Google Patents
A kind of yield real-time predicting method of atmospheric and vacuum distillation unit Download PDFInfo
- Publication number
- CN104965967B CN104965967B CN201510136660.7A CN201510136660A CN104965967B CN 104965967 B CN104965967 B CN 104965967B CN 201510136660 A CN201510136660 A CN 201510136660A CN 104965967 B CN104965967 B CN 104965967B
- Authority
- CN
- China
- Prior art keywords
- tower
- atmospheric
- yield
- oil
- subtract
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The invention discloses a kind of yield real-time predicting method of atmospheric and vacuum distillation unit, live real time data is handled using data reconciliation techniques, with reference to the differential evolution algorithm after improvement, model parameter real time correction is carried out to distillation process, ensure that established model can adapt to different production status, realize the flowsheeting of actual condition.On the basis of model after calibration, analysis of key operation/process conditions, the influence to destilling tower Side Product Yield for Atmosphere and property.According to influence of these performance variables to product yield, with reference to neural net model establishing technology, operating condition and atmospheric and vacuum distillation unit critical product yield are associated, produce the corresponding atmospheric and vacuum distillation unit critical product yield for meeting production operation, realization device yield real-time estimate.This invention can quickly update other Optimized model (such as planning optimization PIMS models, optimizing scheduling ORION models) Crude Oil distilling apparatus yield datas in real time, improve the optimization precision of planning model such as PIMS models.
Description
Technical field
The present invention relates to a kind of atmospheric and vacuum distillation unit yield real-time predicting method based on mechanism and operation characteristic, the party
Method can be used for atmospheric vacuum distillation process modeling and simulating, model real time correction and with PIMS production planning optimizations model/Orion
Scheduling Optimization Model is built in real time.
Background technology
Atmospheric and vacuum distillation unit is the primary production link of whole petroleum refining industry, wherein primary distillation tower, atmospheric tower and vacuum tower conduct
The key equipment of atmospheric and vacuum distillation unit, it is responsible for completing the main process of crude oil time processing.After atmospheric and vacuum distillation unit directly handles desalination
Crude oil, it is therefore an objective to crude oil is cut into the product of various different fractions.Substantial amounts of energy is consumed in the process.These cuts produce
Product or the charging as follow-up workshop section, or sold after mediation as product oil, there is certain quality requirement to them.
Atmospheric and vacuum distillation unit is typically made up of first furnace, primary distillation tower, atmospheric pressure kiln, atmospheric tower, vacuum furnace and vacuum tower, such as
Fig. 1.The main production task of primary distillation tower is to extract the cut before 180 DEG C of major part in crude oil, is split to reform with ethene
Solve device and quality raw materials are provided.The load of atmospheric pressure kiln can also be mitigated simultaneously, so as to realize the purpose of energy-conservation.The production of atmospheric tower
Task is to extract the cut before in crude oil 350 DEG C, and original is provided for reformation, ethylene cracker and Product Oil Blending device
Material.Be evaporated under reduced pressure mainly by before in normal bottom heavy oil 500 DEG C cut extract, be catalytic cracking, hydrocracking unit and
The secondary devices such as residual oil weight-lightening provide raw material.
Various crude oil rectifying columns in crude oil distillation system have identical process characteristic, and essential characteristics are：The object of processing
It is a kind of complex mixture --- crude oil, the product of production is also complex mixture --- various cut products, production process
Operating condition is complicated and changeable.These factors can all make each Side Product Yield for Atmosphere of atmospheric and vacuum distillation unit and property that significant change occur.
By taking atmospheric tower as an example, in the air-distillation system of atmospheric and vacuum distillation unit, crude oil is cut into Chang Dingshi cerebrols, the line of Chang Yi, two, three
5 kinds of products such as oil and atmospheric residue.According to polynary principles of rectification, it is necessary to there is 4 rectifying columns these products could to be isolated
Come, atmospheric tower uses composition operation, and centre sets side take-off, and equivalent to 4 rectifying column overlapped in series form.First base oil is normal
Enter atmospheric tower after being heated to feeding temperature in pressure stove, and feeding temperature is property (such as the true boiling point of crude oil curve by crude oil
Deng), the quality requirement (such as doing) of each cut product, overflash degree requirement and vaporization section oil gas partial pressure determine.Enter material temperature
Du Taigao can cause energy dissipation, can also increase the vapour phase load of tower, reduce the treating capacity of tower；The too low then light oil of feeding temperature is received
Rate declines.In order to ensure the quality of each cut product, stripper is provided with by atmospheric tower, is stripped using superheated vapour, with drop
The content of light components in low sideline product；Enter superheated vapour in atmospheric tower rising pouring, so as to reduce oil and gas partial pressure, help bottom of towe
The vaporization of light oil in heavy oil, while charging vaporization rate in vaporization section will be also improved, improve the yield of distillate.Due to normal pressure
Tower bottom of towe without reboiler, mainly by charging brought into by the heat required for distillation operation, excessive if the capacity of returns of tower top is excessive
Liquid-phase reflux the steam at each position in tower can be made to condense too much, cause the temperature in tower to be everywhere generally reduced, product group
Into generally lightening, distillate oil yield and decline, the content of contained light oil increases in atmospheric residue；On the contrary, each point temperature meeting in tower
Rise comprehensively, degree of fractionation declines, and the fractional composition of product becomes weight, and product is unqualified.
It can be seen that in actual production process, the side line critical product of atmospheric and vacuum distillation unit push up as before oil, often top oil,
A normal line is oily, normal two wires oil, atmosphere 3rd side cut are oily, normal four lines oil, first vacuum side stream, second line of distillation is oily, subtracts third fractional oil, the receipts for the line oil etc. that subtracts four
Rate is had obvious difference by the change of crude quality and operating condition.The actual production and the production schedule of these critical products
Certain deviation often be present in desired value.It is to find out its cause, crucial in such as PIMS in current production activity such as production schedule layout
Model, production scheduling model such as ORION scheduling models are using fixed yield, it is difficult to accurate description device actual moving process.
Therefore, flowsheeting is carried out to atmospheric and vacuum distillation unit, establishes that an energy is accurate, the yield model of quantitative description atmospheric and vacuum distillation unit is to carry
The key of the high current oil refining apparatus production schedule/scheduling model accuracy.
The core of accurate yield model is accurate mechanism model.Atmospheric and vacuum distillation unit is mainly by primary distillation tower, atmospheric tower
Formed with vacuum tower, the operation principle of these towers is all principles of rectification, by rectifying so as to the various products to be meeted the requirements.
Distillation is a kind of unit operation of separation of liquid mixtures, and the basic foundation of separation is the volatile difference of each component.Rectifying is
Most widely used distillation procedure, by the engineering means of backflow, high purity product can be obtained.To the tower of a steady state operation,
The mathematical modeling established by strict method includes equation below：Component materials equilibrium equation, phase equilibrium equation, enthalpy balance equation etc..
Further, since petroleum distillate is a kind of extremely complex mixture of composition, the simulation of petroleum distillate is carried out by component one by one
Calculate, it is clear that can not possibly also be not necessarily to.Solve the method for this problem and oil is exactly cut into boiling range as ten degree extremely
The narrow fraction of tens of degree, i.e. pseudocomponent (or virtual component), its property then is described with a series of numeral of determinations, such as than
Weight, boiling point, molecular weight, critical constant and eccentric factor etc..When can not directly obtain some property datas, it can borrow
, can also be by specific with the data of the same or like pure hydrocarbon of the mean boiling point of boiling point and narrow fraction (usually with n-alkane)
Associated diagram or calculating formula ask it is fixed it.
The present invention considers model accuracy and computational efficiency, using the data gathered from actual production process, in essence
On the basis of evaporating principle, tower model is established by principle of math equation model, determines the independent variable of tower, and based on crude data
Storehouse and actual measurement properities optimization crude oil virtual component, by the optimized algorithm of intelligence, foundation can well describe commercial plant characteristic
Atmospheric vacuum distillation process mechanism model, for real-time estimate atmospheric and vacuum distillation unit critical product yield.
The content of the invention
In view of the above problems, the invention provides a kind of real based on the atmospheric and vacuum distillation unit yield of mechanism and operation characteristic
When Forecasting Methodology.Atmospheric vacuum rectifying model and adiabatic destilling tower model based on principles of rectification, according to actual industrial data, application
Improve difference algorithm and real time correction is carried out to mechanism model, and based on this, carry out operation for key operation/process conditions
Specificity analysis, with reference to nerual network technique, agent model is established, oil property and operating condition are closed with Assay data
Connection, the real-time estimate of atmospheric and vacuum distillation unit yield is realized, to establish accurate planning optimization PIMS models/Scheduling Optimization Model
Theory support is provided.
Concrete technical scheme is as follows：
A kind of yield real-time predicting method of atmospheric and vacuum distillation unit, comprises the following steps：
1) data established using .net interfacings between live real-time data base and atmospheric and vacuum distillation unit are communicated, real
The real-time collection of existing atmospheric and vacuum distillation unit service data；
Data mediation standard is established with reference to field condition and knowhow, rejects useless and wrong real time data.According to
The routine evaluations data of oil property and narrow component hydro carbons composition and its property data, as crude oil classification, true boiling point distillation are bent
The analyze data (but not limited to)s such as line, global density curve, sulfur content, acidity, condensation point, viscosity, mean molecule quantity, generation are former
Oily virtual component, and the macroscopic property of virtual component is estimated by correlation and calculates other physico-chemical properties accordingly, such as pass through
The data such as enthalpy, specific volume, viscosity calculate the characteristic of current oil product, as virtual component content (being divided by boiling point), virtual component are related
Property for example octane number, anti-knock index, Lei Shi vapour pressures, Lei Shi vapour pressure indexs, alkene, alkane, alkane, cycloalkane and
Aromatic hydrocarbons equal size；
2) according to the atmospheric and vacuum distillation unit real time data of collection, with the model predication value and actual value of sideline product flow
The difference of two squares it is minimum be used as optimization aim, solved, fitted model parameters, realized often using the differential evolution algorithm after improvement
Decompressor mechanism model real time correction；
3) based on the model after correction, for different crude oils, the yield of the critical product under the conditions of different operating is analyzed, is built
Vertical product yield analytical database；
4) the neutral net agent model of actual condition can be accurately reflected using the training of product yield analytical database, and
Oil property and ordinary decompression column operating condition are associated with Assay data, realize the real-time estimate of atmospheric and vacuum distillation unit yield.
The real-time estimate can be in other Optimized models (such as planning optimization PIMS models, optimizing scheduling ORION models)
Crude(oil)unit real-time update yield data, improve the optimization accuracy of model.
Real time data described in step (2) is selected from primary distillation tower tower top pressure, vaporization section pressure, feed zone pressure, heating furnace
Temperature, overhead condenser temperature, bottom of towe stripped vapor, atmospheric tower tower top pressure, vaporization section pressure, feed zone pressure, tower top are cold
Condenser temperature, furnace temp, bottom of towe vapor flow of stripper, normal line quantity of steam, normal two wires quantity of steam, atmosphere 3rd side cut quantity of steam, often
One extracts flow out, Chang Yizhong returns tower temperature degree, Chang Erzhong extracts flow out, Chang Erzhong returns tower temperature degree, Chang Sanzhong extracts flow, Chang Sanzhong out
Return tower temperature degree, vacuum tower tower top pressure, vaporization section pressure, feed zone pressure, heating furnace load, bottom of towe stripped vapor, subtract one
Extract flow out, subtract one in return tower thermic load, subtract two in extract flow out, subtract two in return tower temperature degree, subtract three in extract flow out, subtract three
Return tower temperature degree, subtract four in extract flow out, subtract four in return the one or more in tower temperature degree and product information.
Differential evolution algorithm after being improved described in step (2) is the differential evolution algorithm with trigonometric mutation, and is defined
Optimization aim is：
Product i yield actual values；Product i yield predicted values.
Wherein, decision variable X includes the sensing factor and activation energy of each reaction.
Critical product described in step (3) is selected from just top is oily, often pushing up an oily, normal line, oily, normal two wires is oily, atmosphere 3rd side cut is oily, often
Four line oil, first vacuum side stream, second line of distillation oil, subtract third fractional oil, the line that subtracts four oil and decompression residuum.
Atmospheric and vacuum distillation unit is in reply production change in step (4), change of properties or operating condition change for crude oil,
Carry out operating characteristic analysis, obtain influences data to the yield and property of ordinary decompression column side line critical product.
The property of the crude oil is selected from classification, density, sulfur content, nitrogen content, acid number, pour point and the true boiling point distillation of crude oil
Data；
The operating condition is selected from primary distillation tower tower top pressure, vaporization section pressure, feed zone pressure, furnace temp, tower top
Condenser temperature, bottom of towe stripped vapor, atmospheric tower tower top pressure, vaporization section pressure, feed zone pressure, overhead condenser temperature,
Furnace temp, bottom of towe vapor flow of stripper, normal line quantity of steam, normal two wires quantity of steam, atmosphere 3rd side cut quantity of steam, a normal extraction stream
Amount, Chang Yizhong return tower temperature degree, Chang Erzhong extracts flow out, Chang Erzhong returns tower temperature degree, Chang Sanzhong extracts flow out, Chang Sanzhong returns tower temperature
Degree, vacuum tower tower top pressure, vaporization section pressure, feed zone pressure, heating furnace load, bottom of towe stripped vapor, subtract one in extraction stream
Measure, subtract one in return tower thermic load, subtract two in extract out flow, subtract two in return tower temperature degree, subtract three in extract out flow, subtract three in return tower temperature
Spend, subtract four in extract out flow, subtract four in return tower temperature degree.
Neutral net agent model described in step 4 uses backpropagation (Back propagation, BP) neutral net,
Choose 10 input variables：Material name (correspond to oil property), feed loading, primary distillation tower furnace outlet temperature, normal pressure
Tower furnace outlet temperature, vacuum tower furnace outlet temperature, atmospheric tower vapor flow of stripper, vacuum tower vapor flow of stripper, fore-running
Overhead condenser temperature, Atmospheric Tower condenser temperature, Top of Vacuum Tower pressure；11 output variables：Just top is oily, often pushes up oil, often
One line is oily, normal two wires is oily, atmosphere 3rd side cut is oily, normal four lines oil, first vacuum side stream, second line of distillation are oily, subtract third fractional oil, subtracting four, line is oily, depressurizes slag
Oil；Hidden layer is arranged to 6 layers.
Beneficial effects of the present invention are as follows：Based on principles of rectification, the modeling of primary distillation tower, atmospheric tower and vacuum tower is completed with imitating
Very, distillation model parameter is corrected with reference to actual operating data, establishes the Atmospheric vacuum that can well describe actual operating mode
Still-process mechanism model.
Brief description of the drawings
Fig. 1 is existing atmospheric and vacuum distillation unit simplified flowchart；
Fig. 2 is the model tower of rectifying；
Fig. 3 is Plate model；
Fig. 4 is model real time correction simplified flowchart
Fig. 5 is multilayer feedforward neural network structural representation；
Fig. 6 is neutral net agent model structure chart.
Embodiment
The present invention is specifically described below by embodiment.It is necessarily pointed out that following examples are only used
In the invention will be further described, it is impossible to be interpreted as limiting the scope of the invention, professional and technical personnel in the field
Some the nonessential modifications and adaptations made according to present disclosure, still fall within protection scope of the present invention.
Embodiment 1
The implementation method of the present invention is specifically introduced with reference to chart：
1st, atmospheric and vacuum distillation modelling by mechanism
Atmospheric and vacuum distillation unit is mainly made up of primary distillation tower, atmospheric tower and vacuum tower, and the operation principle of these towers is all gas-liquid
Equilibrium separation, the product to be meeted the requirements by rectifying.Distillation is to utilize the volatile difference of each component in liquid mixture
Realize a kind of unit operation of key component separation.Rectifier unit is made up of rectifying column, condenser and reboiler etc., such as Fig. 2 institutes
Show.
When establishing the model of rectifying column, king-tower condenser is set to first piece of theoretical plate, counted from top to down, king-tower bottom
Most next piece of theoretical plate is reboiler (if any), NsFor the theoretical plate sum in tower system.Fig. 3 represents either plate in tower
Thing (heat) flows into artificial situation.
In figure 3, F is inlet amount, and Q is heat, LnTo leave the amount of liquid of n-layer plate,For to the amount of liquid of lower plywood,
WLnFor the amount of liquid extracted out from n-layer plate,For from top plate come amount of liquid,For the gas amount come from lower plywood, VnFor
The gas amount of n-layer plate is left,For to the gas amount of top plate, WVnFor the gas amount extracted out from n-layer plate,For from associated column
The amount of liquid at bottom,For the gas amount from related tower top, WLiFor the amount of liquid from associated column side line, WVmTo carry out auto-correlation
The gas amount of tower side line, subscript j, k, l, m are the plate number in Tower for Simulation.Wherein, associated column refers to that stripper etc. and king-tower have thing
The connected tower of material.
To the tower of a steady state operation, the mathematical modeling established by strict method includes following 5 equations：
Component materials equilibrium equation：
Phase equilibrium equation：
Yi,n=Ki,nXi,nI=1 ..., NC；N=1 ..., NS (2)
Mole fraction adds and equation：
Total bag material balance equation：
Enthalpy balance equation：
In above equation, i marks for component, riMarked for column plate, NCFor total number of components, NSFor total number of plates, K is equal
Weigh constant, and X is that liquid phase forms mole fraction, and Y is that vapour phase forms mole fraction, HLFor liquid phase enthalpy, HVFor vapour phase enthalpy.
Further, since petroleum distillate is a kind of extremely complex mixture of composition, to be evaporated by component one by one to carry out oil
The simulation divided calculates, it is clear that can not possibly also be not necessarily to.The method for solving this problem herein exactly cuts into oil
Boiling range is ten degree of virtual narrow fraction, determines various nature parameters according to lab analysis system data and crude oil storage data, such as
Proportion, boiling point, molecular weight, critical constant and eccentric factor etc..When can not directly obtain some property datas, it can borrow
, can also be by specific with the data of the same or like pure hydrocarbon of the mean boiling point of boiling point and narrow fraction (conventional n-alkane)
Associated diagram or calculating formula obtain.
The present invention takes the number of plates and its operating pressure of each tower, feed entrance point, feed flow rates, composition and phase, heating
The position of hot device and load, sideline product position and flow rate are considered as datum, by T, L of each column plate, V, X, Y, Q, overhead reflux
Variable is used as than, product population and vapour-liquid ratio etc..On this basis, with reference to design conditions data, foundation can be described well often
Tower and the process modeling of vacuum tower distillation process operating characteristic are pressed, according to real-time running data, model ginseng is carried out to distillation process
Number on-line correction, it is ensured that the model established can adapt to different production status, realize the flowsheeting of actual condition.
2nd, data acquisition and mediation
Because industry spot situation is complicated, production process is affected by various factors, completely the machine according to Experimental report
Managing the atmospheric and vacuum distillation unit model established can not accurate simulation actual device, it is therefore desirable to comes with reference to field device actual motion characteristic
Calibration model.It is collection and the reconciliation process process for realizing field data first.
1) on-site data gathering：In actual production process, most of factories all can record dress using real-time data base
The operation conditions put, and the position number at respective counts strong point is provided so as to gathered data.The present invention is developed using VB.net interfacings
Catalytic cracking unit field data real-time acquisition system, it is possible to achieve by the reading of live real time data, and be stored into local number
According in storehouse.Need the data that gather mainly include primary distillation tower tower top pressure, vaporization section pressure, feed zone pressure, furnace temp,
Overhead condenser temperature, bottom of towe stripped vapor, atmospheric tower tower top pressure, vaporization section pressure, feed zone pressure, overhead condenser temperature
Degree, furnace temp, bottom of towe vapor flow of stripper, normal line quantity of steam, normal two wires quantity of steam, atmosphere 3rd side cut quantity of steam, normal one extract out
Flow, Chang Yizhong return tower temperature degree, Chang Erzhong extracts flow out, Chang Erzhong returns tower temperature degree, Chang Sanzhong extracts flow out, Chang Sanzhong returns tower temperature
Degree, vacuum tower tower top pressure, vaporization section pressure, feed zone pressure, heating furnace load, bottom of towe stripped vapor, subtract one in extraction stream
Measure, subtract one in return tower thermic load, subtract two in extract out flow, subtract two in return tower temperature degree, subtract three in extract out flow, subtract three in return tower temperature
Spend, subtract four in extract out flow, subtract four in return the information such as tower temperature degree and product flow and property.
2) data reconciliation process：Limited to by Site Detection instrument reliability, the data directly got from DCS are often
The problems such as uneven material, heat imbalance be present, therefore cannot be directly used to establish mounted cast.In order to ensure model sample
The accuracy of data, it is necessary to which the data gathered in real time are established with mediation standard, specifically used following several method：(1) use
Day average carrys out calibration model；(2) codomain of data is determined according to statistics and knowhow, judges the accurate of data according to this
Property, wrong data is deleted from local data base；(3) for the data that can not be gathered within specific period, redundancy is established
Calculation formula, this point is derived by gathering other data.
3rd, atmospheric and vacuum distillation unit mechanism model real time correction
The step for the real time correction implementation process of atmospheric and vacuum distillation unit mechanism model is introduced with reference to real case.
Model real time correction can be classified as Parameter Estimation Problem, and Parameter Estimation Problem is converted into optimization by the present invention first
Problem, i.e.,：
Wherein, decision variable X includes the sensing factor and activation energy of each reaction,WithProduct oil is represented respectively
The actual measurement of each component and the mass percent of prediction.For such optimization aim, the present invention uses improved difference
Algorithm solves to problem.
Difference algorithm (differential evolution, DE) is a kind of random search algorithm based on population, it
Have the characteristics that simple in construction, fast convergence rate, robustness are high.The Variation mechanism of algorithm, that is, the method for generating filial generation are：
R '=r1+F*(r2-r3) (7)
Wherein, r ' is newly-generated offspring individual, r1,r2,r3It is the three different parents randomly selected in population
Body, F are differential evolution operator, generally a constant.
Due to the objective decision numerous variables, amount of calculation is very big when causing Algorithm for Solving, it is therefore desirable to which algorithm is entered
Row improves, and accelerates its convergence rate.The present invention have selected the improvement difference algorithm with trigonometric mutation, and this method is proved to carrying
There is remarkable effect in terms of high algorithm the convergence speed, its improved Mutation Strategy can be expressed as：
R '=(r1+r2+r3)/3+(p2-p1)(r1-r2)+(p3-p2)(r2-r3)+(p1-p3)(r3-r1) (8)
Wherein p1=| f (r1) |/p ', p2=| f (r2) |/p ', p3=| f (r3) |/p ', p '=| f (r1)|+|f(r2)|+|f
(r3)|
The simplified flowchart of model real time correction is as shown in Figure 4.
This example chooses the atmospheric and vacuum distillation unit of 3,500,000 tons/year of certain refinery, and mould is carried out with the working condition average value of 10 days
Intend.The operation operating mode Part load of device is：Primary distillation tower：Tower top pressure is 3bar, vaporization section pressure 3.6bar, feed zone pressure
3.8bar, 338 DEG C of furnace temp, 132 DEG C of overhead condenser temperature, bottom of towe stripped vapor 20kg/h；Atmospheric tower：Tower top pressure
1.04bar, vaporization section pressure 1.3bar, feed zone pressure 1.7bar, 89 DEG C of overhead condenser temperature, 375 DEG C of furnace temp,
Bottom of towe stripped vapor 4600kg/h, normal line steam 1000kg/h, normal two wires steam 600kg/h, atmosphere 3rd side cut steam 1200kg/h,
Chang Yizhong returns 119 DEG C of tower temperature degree, and Chang Erzhong returns 165 DEG C of tower temperature degree, and Chang Sanzhong returns 235 DEG C of tower temperature degree, vacuum tower：Tower top pressure
0.029bar, vaporization section pressure 0.02919bar, feed zone pressure 0.03300bar, heating furnace load 10606560w, bottom of towe vapour
Steam 2000kg/h is carried, tower thermic load -5600589.3w is returned in subtracting one, 161.029 DEG C of tower temperature degree is returned in subtracting two, tower is returned in subtracting three
220.538 DEG C of temperature, 287.385 DEG C of tower temperature degree is returned in subtracting four.
4th, atmospheric and vacuum distillation unit product yield is analyzed
By the realization of above step, atmospheric and vacuum distillation unit mechanism model by the correction of real time execution performance data
Through the actual conditions that can accurately reflect atmospheric and vacuum distillation unit.
Using the atmospheric and vacuum distillation unit mechanism model of correction (such as feed loading, each for different crude oils and operating condition
Each section of heat-exchange temperature of tower furnace temp, raw material, destilling tower operation temperature, rectifying column vapor flow of stripper, side take-off amount, side line
Vapor flow of stripper, backflow take thermic load etc., the critical product yield of acquisition device, and wherein critical product mainly includes：Push up as before
Oil, often push up oily an oily oily, normal line, normal two wires oil, atmosphere 3rd side cut, normal four lines oil, first vacuum side stream, second line of distillation is oily, subtracts third fractional oil, subtracts four
Line oil, decompression residuum etc..And local data library storage yield data is established, and the read-write of data can be realized.Table one is certain work
Under condition, atmospheric and vacuum distillation unit model critical product predicted value and actual contrast.
Table 2 and table 3 are the influence of normal pressure heater outlet temperature offside line yield and tower reactor stripping vapor to the shadow of side line yield
Ring.From table two, it can be seen that, with the increase of normal pressure heater outlet temperature, often top naphtha makes are almost unchanged, normal line oil yield
It is increased slightly；Normal two wires oil and atmosphere 3rd side cut oil yield are as normal pressure heater outlet temperature increase has different degrees of increase.Normal pressure
Heater outlet temperature raising is the most obvious for improving atmosphere 3rd side cut oil yield, that is, improves normal pressure furnace temperature and cause what is dissolved in normal base oil
A large amount of light groups are gasified and entered in atmosphere 3rd side cut oil product.
Atmospheric and vacuum distillation unit model critical product predicted value and actual contrast under certain operating mode of table 1
| Operational factor | Actual value, % | Model predication value, % | Error (%) |
| Just top oil | 7.00 | 7.06 | 0.8 |
| Often top oil | 1.55 | 1.56 | 0.6 |
| Normal line oil | 8.06 | 8.15 | 1.1 |
| Normal two wires oil | 13.62 | 13.97 | 2.6 |
| Atmosphere 3rd side cut oil | 3.38 | 3.33 | 1.5 |
| Normal four lines oil | 0.44 | 0.45 | 2.01 |
| First vacuum side stream | 6.89 | 7.06 | 2.5 |
| Second line of distillation oil | 7.44 | 7.72 | 3.7 |
| Subtract third fractional oil | 6.28 | 6.32 | 0.7 |
| The line that subtracts four is oily | 4.12 | 4.21 | 2.3 |
| Decompression residuum | 40.65 | 39.43 | 3.0 |
The influence of the normal pressure heater outlet temperature offside line yield of table 2
Influence of the atmospheric tower tower reactor stripping vapor of table 3 to side line yield
As can be seen from Table 3, with the increase of tower bottom steam amount, Chang Dingshi cerebrols, a normal line are oily, normal two wires is oily and atmosphere 3rd side cut
Oil product yield has different degrees of increase.Naphtha and the more normal two wires oil of normal line oil product incrementss and atmosphere 3rd side cut oil increase
Dosage is few.
The separation accuracy for improving tower is using energy expenditure as cost.Normal pressure heater outlet temperature is improved, is increased in tower
Gas-liquid load, strengthen gas-liquid mass transfer in tower；Further, it is also possible to using increase tower reactor in the disposal ability allowed band of tower
Stripping vapor dosage, using reduces oil gas partial pressure, increases the relative volatility of weight component in crude oil, and weight component is become
It must be easier to separate.It follows that the model that the present invention establishes can preferably reflect industrial production actual conditions.
5th, atmospheric and vacuum distillation unit product yield real-time estimate
Also to model while atmospheric and vacuum distillation unit yield real-time estimate technical requirements mechanism model has high accuracy
Computational efficiency propose very high requirement.Calibration model mechanism described above is sufficiently complex, and solving speed is very slow, Er Qierong
Do not restrain easily, it is clear that the needs of real-time estimate can not be met, therefore need badly and establish agent model using nerual network technique come real
The real-time estimate of existing plant yield.
The present invention uses backpropagation (Back propagation, BP) neutral net, and this is one kind in process control
Most widely used neural network structure.BP neural network structure is as shown in Figure 5 and Figure 6.Total is by L layer nerve tuples
Into first layer is input layer, and last layer is output layer, and other layers are hidden layer, and the model that can obtain each neuron is：
In formula：
ypjk：In j layers each neuron of kth under pth group sample state and output；
xpjk：K-th of neuron being exported with function under pth group sample state in j layers；
wpjk：The link weights of i-th of neuron, p-th of neuron into j layers in j-1 layers.Wherein wpjoIt is defined as j layers
The threshold values of k neuron.
f(y)：The nonlinear activation function of neuron.
The purpose of neural network learning is to find out a series of weights, after every group of input vector of sample is acted on network,
The reality output vector of its network is consistent with the desired output vector of sample, and whole learning process is each neuron in adjustment network
Between connection weight, the error energy function of following networks is reached minimum：
Wherein：
opk：After pth group sample input vector acts on network, the function output of k-th of neuron in network output layer
Desired value.
BP algorithm is used for the problem concerning study for solving above-mentioned Multilayer Feedforward Neural Networks, and the learning algorithm is by signal forward-propagating and mistake
Poor backpropagation composition.Traditional BP algorithm can be briefly summarized as follows.
Wherein：
In formula：
δpjk：After pth group sample input vector acts on network, the control information of k-th of neuron function output in j layers；
t：Learning time；
f′(y)：The single order derived function of neuron activation functions.
Formula (12)-(13) are the learning rules of weights in BP networks, formula learning speed and situation term coefficient be usually by
Empirically determined, they can not be with network structure, network state and external learning environment Auto-matching, net in traditional BP algorithm
Need artificially to be adjusted during network training.
According to above-mentioned neural network structure, choose neutral net and input 10：Material name (correspond to oil property), enter
Expect load, primary distillation tower furnace outlet temperature, atmospheric tower furnace outlet temperature, vacuum tower furnace outlet temperature, atmospheric tower
Vapor flow of stripper, vacuum tower vapor flow of stripper, fore-running tower top condenser temperature, Atmospheric Tower condenser temperature, vacuum tower top pressure
Power；Output 11：Just top is oily, often pushing up an oily, normal line, oily, normal two wires is oily, atmosphere 3rd side cut is oily, normal four lines oil, first vacuum side stream, second line of distillation
Oil, subtract third fractional oil, the line that subtracts four oil, decompression residuum；Hidden layer is arranged to 6 layers.Neutral net agent model is trained with this, sees Fig. 5.
Shown according to analyze data above, the atmospheric and vacuum distillation unit critical product yield under the influence of performance variable and operation
Relationship between variables are nonlinear, and have certain difference with crude oil Assay data used in Production planning model.Therefore, originally
Invention is acted on behalf of forecast model on the basis of device mechanism model, using neutral net and calculated, and non-linear object is linearized
Processing, produce the corresponding atmospheric and vacuum distillation unit yield for meeting production operation, realization device yield real-time estimate.This invention can
Real-time conveniently and efficiently renewal planning model Crude Oil distilling apparatus data, raising plan/scheduling model such as PIMS models/
The optimization precision of Orion models.
By the progress of above step, the present invention can realize to be received based on the atmospheric and vacuum distillation unit of mechanism and operation characteristic
The real-time estimate of rate.This method is using rectifying column model as theoretical foundation, using actual industrial data to atmospheric and vacuum distillation unit machine
Manage model and carry out real time correction, and neutral net agent model is trained by calculating the analyze data of calibration model acquisition, gram
The slow limitation of atmospheric and vacuum distillation unit model calculating speed is taken.Atmospheric and vacuum distillation unit is calculated using neutral net agent model
The yield of critical product, associating for operating condition and crude oil distillation process critical product yield is realized, reach atmospheric and vacuum distillation dress
The effect of yield real-time estimate is put, is the accurate optimization of other models, such as planning optimization PIMS models, optimizing scheduling ORION moulds
Type provides theory support.
Claims (7)
1. the yield real-time predicting method of a kind of atmospheric and vacuum distillation unit, it is characterised in that comprise the following steps：
1) data established using .net interfacings between live real-time data base and atmospheric and vacuum distillation unit are communicated, and are realized normal
The real-time collection of vacuum distillation apparatus service data；
2) according to the atmospheric and vacuum distillation unit real time data of collection, with the flat of the model predication value of sideline product flow and actual value
Variance minimum is used as optimization aim, is solved using the differential evolution algorithm after improvement, fitted model parameters, realizes based on essence
Evaporate the real time correction of the atmospheric and vacuum distillation unit mechanism model of principle；
3) based on the model after correction, for different crude oils, the yield of the critical product under the conditions of different operating is analyzed, establishes production
Product yield analysis database；
4) using the training of product yield analytical database the neutral net agent model of actual condition can be accurately reflected, and by original
Oil nature and ordinary decompression column operating condition are associated with Assay data, realize the real-time estimate of atmospheric and vacuum distillation unit yield.
2. yield real-time predicting method according to claim 1, it is characterised in that real time data described in step (2) is selected
Steamed from primary distillation tower tower top pressure, vaporization section pressure, feed zone pressure, furnace temp, overhead condenser temperature, bottom of towe stripping
Vapour, atmospheric tower tower top pressure, vaporization section pressure, feed zone pressure, overhead condenser temperature, furnace temp, bottom of towe stripping steam
Vapour amount, normal line quantity of steam, normal two wires quantity of steam, atmosphere 3rd side cut quantity of steam, normal one extract flow, Chang Yizhong time tower temperature degree, normal two out
Middle extraction flow, Chang Erzhong return tower temperature degree, Chang Sanzhong extracts flow out, Chang Sanzhong returns tower temperature degree, vacuum tower tower top pressure, vaporization section
Pressure, feed zone pressure, heating furnace load, bottom of towe stripped vapor, subtract one in extract out flow, subtract one in return tower thermic load, subtract two
Middle extraction flow, return tower temperature degree in subtracting two, subtract three in extract flow out, subtract three in return tower temperature degree, subtract four in extract flow out, subtract four
The one or more returned in tower temperature degree and product information.
3. yield real-time predicting method according to claim 1, it is characterised in that the difference after being improved described in step (2)
It is the differential evolution algorithm with trigonometric mutation to divide evolution algorithm, and defines optimization aim and be：
<mrow>
<mi>min</mi>
<mi> </mi>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
</munder>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>C</mi>
<mrow>
<mi>a</mi>
<mi>c</mi>
<mi>t</mi>
<mi>u</mi>
<mi>a</mi>
<mi>l</mi>
</mrow>
<mi>i</mi>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>C</mi>
<mrow>
<mi>p</mi>
<mi>r</mi>
<mi>e</mi>
<mi>d</mi>
<mi>i</mi>
<mi>c</mi>
<mi>t</mi>
</mrow>
<mi>i</mi>
</msubsup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>;</mo>
</mrow>
Wherein, decision variable X includes the sensing factor and activation energy of each reaction,Product i yield actual values；Production
Product i yield predicted values.
4. yield real-time predicting method according to claim 1, it is characterised in that critical product described in step (3) selects
From first top oil, often push up oily an oily oily, normal line, normal two wires oil, atmosphere 3rd side cut, normal four lines oil, first vacuum side stream, second line of distillation is oily, the line that subtracts three
Oil, the line that subtracts four oil and decompression residuum.
5. yield real-time predicting method according to claim 1, it is characterised in that atmospheric and vacuum distillation unit is being answered in step (4)
When changing to production, change of properties or operating condition change for crude oil, carry out operating characteristic analysis, obtain to ordinary decompression column
The yield and property of side line critical product influence data.
6. yield real-time predicting method according to claim 5, it is characterised in that the property of the crude oil is selected from crude oil
Classification, density, sulfur content, nitrogen content, acid number, pour point and true boiling point distillation data；
The operating condition is selected from primary distillation tower tower top pressure, vaporization section pressure, feed zone pressure, furnace temp, overhead condensation
Device temperature, bottom of towe stripped vapor, atmospheric tower tower top pressure, vaporization section pressure, feed zone pressure, overhead condenser temperature, heating
Furnace temperature, bottom of towe vapor flow of stripper, normal line quantity of steam, normal two wires quantity of steam, atmosphere 3rd side cut quantity of steam, it is normal one extract out flow, often
Tower temperature degree is returned in one, Chang Erzhong extracts flow out, Chang Erzhong returns tower temperature degree, Chang Sanzhong extracts flow out, Chang Sanzhong returns tower temperature degree, decompression
Column overhead pressure, vaporization section pressure, feed zone pressure, heating furnace load, bottom of towe stripped vapor, subtract one in extract out flow, subtract one
In return tower thermic load, subtract two in extract out flow, subtract two in return tower temperature degree, subtract three in extract out flow, subtract three in return tower temperature degree, subtract four
Middle extraction flow, return tower temperature degree in subtracting four.
7. yield real-time predicting method according to claim 1, it is characterised in that neutral net described in step 4 is acted on behalf of
Model uses reverse transmittance nerve network, chooses 10 input variables：Material name, feed loading, primary distillation tower furnace outlet
Temperature, atmospheric tower furnace outlet temperature, vacuum tower furnace outlet temperature, atmospheric tower vapor flow of stripper, vacuum tower stripping steam
Vapour amount, fore-running tower top condenser temperature, Atmospheric Tower condenser temperature, Top of Vacuum Tower pressure, wherein the material name is corresponding
In oil property；11 output variables：Just top is oily, often pushing up an oily, normal line, oily, normal two wires is oily, atmosphere 3rd side cut is oily, normal four line is oily, subtracts
One line oil, second line of distillation oil, subtract third fractional oil, the line that subtracts four oil, decompression residuum；Hidden layer is arranged to 6 layers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510136660.7A CN104965967B (en) | 2015-03-26 | 2015-03-26 | A kind of yield real-time predicting method of atmospheric and vacuum distillation unit |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510136660.7A CN104965967B (en) | 2015-03-26 | 2015-03-26 | A kind of yield real-time predicting method of atmospheric and vacuum distillation unit |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104965967A CN104965967A (en) | 2015-10-07 |
| CN104965967B true CN104965967B (en) | 2018-02-13 |
Family
ID=54220005
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510136660.7A Active CN104965967B (en) | 2015-03-26 | 2015-03-26 | A kind of yield real-time predicting method of atmospheric and vacuum distillation unit |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104965967B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107220392B (en) * | 2016-03-22 | 2020-07-28 | 中国石油化工股份有限公司 | Normal pressure tower and normal line 10% point prediction method for atmospheric and vacuum device |
| CN107220393B (en) * | 2016-03-22 | 2020-07-28 | 中国石油化工股份有限公司 | Atmospheric tower common line dry point prediction method for atmospheric and vacuum device |
| CN107220705B (en) * | 2016-03-22 | 2020-10-09 | 中国石油化工股份有限公司 | Atmospheric tower top dry point prediction method for atmospheric and vacuum device |
| CN105938092B (en) * | 2016-04-12 | 2019-02-26 | 南京富岛信息工程有限公司 | A kind of true boiling point curve bearing calibration based on crude oil real-time |
| CN106202910A (en) * | 2016-07-07 | 2016-12-07 | 华东理工大学 | The yield real-time predicting method of a kind of residual hydrocracking device and application thereof |
| CN106709169A (en) * | 2016-12-12 | 2017-05-24 | 南京富岛信息工程有限公司 | Property estimation method for crude oil processing process |
| CN109214012B (en) * | 2017-06-29 | 2022-03-01 | 中国石油天然气股份有限公司 | Energy efficiency obtaining method and device for atmospheric and vacuum distillation unit |
| CN109993358B (en) * | 2019-03-25 | 2021-07-16 | 联想(北京)有限公司 | Method and device for training yield prediction model |
| CN111914381B (en) * | 2019-05-07 | 2023-04-25 | 宁波大学 | Operation optimization method of atmospheric and vacuum device based on KPLSR model |
| CN110987862A (en) * | 2019-11-06 | 2020-04-10 | 汉谷云智(武汉)科技有限公司 | Diesel oil on-line blending method |
| CN110796318B (en) * | 2020-01-06 | 2020-05-05 | 汉谷云智(武汉)科技有限公司 | Real-time operation optimization method and device for catalytic fractionation device |
| CN111241677A (en) * | 2020-01-09 | 2020-06-05 | 浙江中控技术股份有限公司 | Atmospheric and vacuum device production simulation method and system based on machine learning |
| CN113343404B (en) * | 2020-03-02 | 2022-11-08 | 中国石油化工股份有限公司 | Optimization method and system for fractionation absorption stabilization system model of catalytic cracking unit |
| CN111860938B (en) * | 2020-06-01 | 2023-05-30 | 浙江中控技术股份有限公司 | Global blending scheduling optimization method for crude oil storage and transportation system |
| CN113241127B (en) * | 2021-03-19 | 2023-06-02 | 中国石油大学(北京) | Phase equilibrium model construction method, apparatus, device and storage medium |
| CN115079572B (en) * | 2022-06-30 | 2023-02-03 | 福建省龙德新能源有限公司 | Energy management control system for preparing lithium hexafluorophosphate and control method thereof |
| CN115430344B (en) * | 2022-08-31 | 2023-04-28 | 福建省龙德新能源有限公司 | Automatic batching system and batching method for lithium hexafluorophosphate preparation |
| CN116798534B (en) * | 2023-08-28 | 2023-11-07 | 山东鲁扬新材料科技有限公司 | Data acquisition and processing method for acetic acid propionic acid rectification process |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101955426A (en) * | 2010-08-23 | 2011-01-26 | 华东理工大学 | Industrial purified terephthalic acid device azeotropic rectification solvent dehydrating process optimization operating method |
| CN101986320A (en) * | 2010-08-23 | 2011-03-16 | 华东理工大学 | Modeling method for heterogeneous azeotropic rectification solvent dehydrating tower of industrial purified terephthalic acid (PTA) device |
| CN103310123A (en) * | 2013-07-10 | 2013-09-18 | 华东理工大学 | Coupling modeling method for verifying and optimizing design of industrial ethylene steam cracking furnace |
| CN103605821A (en) * | 2013-09-16 | 2014-02-26 | 华东理工大学 | Ethylene cracking furnace group load distribution optimization method |
| CN103605325A (en) * | 2013-09-16 | 2014-02-26 | 华东理工大学 | Complete period dynamic optimization method for industrial ethylene cracking furnace and based on surrogate model |
| CN103729695A (en) * | 2014-01-06 | 2014-04-16 | 国家电网公司 | Short-term power load forecasting method based on particle swarm and BP neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9140679B2 (en) * | 2010-12-28 | 2015-09-22 | Chevron U.S.A. Inc. | Process for characterizing corrosivity of refinery feedstocks |
-
2015
- 2015-03-26 CN CN201510136660.7A patent/CN104965967B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101955426A (en) * | 2010-08-23 | 2011-01-26 | 华东理工大学 | Industrial purified terephthalic acid device azeotropic rectification solvent dehydrating process optimization operating method |
| CN101986320A (en) * | 2010-08-23 | 2011-03-16 | 华东理工大学 | Modeling method for heterogeneous azeotropic rectification solvent dehydrating tower of industrial purified terephthalic acid (PTA) device |
| CN103310123A (en) * | 2013-07-10 | 2013-09-18 | 华东理工大学 | Coupling modeling method for verifying and optimizing design of industrial ethylene steam cracking furnace |
| CN103605821A (en) * | 2013-09-16 | 2014-02-26 | 华东理工大学 | Ethylene cracking furnace group load distribution optimization method |
| CN103605325A (en) * | 2013-09-16 | 2014-02-26 | 华东理工大学 | Complete period dynamic optimization method for industrial ethylene cracking furnace and based on surrogate model |
| CN103729695A (en) * | 2014-01-06 | 2014-04-16 | 国家电网公司 | Short-term power load forecasting method based on particle swarm and BP neural network |
Non-Patent Citations (1)
| Title |
|---|
| "原油蒸馏装置产品质量指标软测量方法研究与应用";王峰;《中国优秀硕士学位论文全文数据库 信息科技辑》;20120215(第2期);第14-16页，第37-38页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104965967A (en) | 2015-10-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104965967B (en) | A kind of yield real-time predicting method of atmospheric and vacuum distillation unit | |
| CN104765346B (en) | A kind of oil refining process whole process modeling method | |
| CN104789256B (en) | A kind of yield real-time predicting method of catalytic cracking unit | |
| CN100409232C (en) | Method and system for operating a hydrocarbon production facility | |
| Mahalec et al. | Inferential monitoring and optimization of crude separation units via hybrid models | |
| CN103524284B (en) | Forecasting and optimizing method for ethylene cracking material configuration | |
| CN104804761B (en) | A kind of yield real-time predicting method of hydrocracking unit | |
| CN108279251B (en) | A kind of method and device thereof of petroleum molecules level separation process simulation | |
| CN101201331B (en) | Soft measuring method for on-line determining petroleum naphtha quality index on top of primary tower | |
| CN101169387B (en) | Soft sensing method for on-line determination of atmospheric tower top naphtha quality index | |
| CN106444672A (en) | Molecular-level real time optimization (RTO) method for oil refining and petrochemical device | |
| CN104765347B (en) | Yield real-time predicting method in a kind of residual oil delayed coking | |
| CN101414158A (en) | Method for optimizing cracking reaction operating condition of ethylene cracking furnace | |
| CN102626557A (en) | Molecular distillation process parameter optimizing method based on GA-BP (Genetic Algorithm-Back Propagation) algorithm | |
| CN104484714A (en) | Real-time prediction method for catalytic reforming device | |
| SA519410799B1 (en) | A Hybrid Machine Learning Approach Towards Olefins Plant Optimization | |
| CN102998013A (en) | Soft sensing method for true temperature of pyrolysis mixed products at outlet of ethylene cracking furnace | |
| CN105701267A (en) | Method for modelling oil catalytic cracking reaction regeneration part | |
| CN104361153A (en) | Method for predicting coking amount of heavy oil catalytic cracking settler | |
| Fouladvand et al. | Simulation and optimization of aromatic extraction from lube oil cuts by liquid-liquid extraction | |
| CN106202910A (en) | The yield real-time predicting method of a kind of residual hydrocracking device and application thereof | |
| CN105740960A (en) | Optimization method of industrial hydrocracking reaction condition | |
| CN110288197A (en) | Plan production optimization method based on molecule trend | |
| CN115938502A (en) | Chemical product characteristic prediction method and system based on molecular-level reaction mechanism | |
| CN100441552C (en) | Process for advanced controlling rectifying apparatus of butadiene |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |