CN106709087A - Random seed number preprocessing, least square postprocessing and parallel lumped kinetics method - Google Patents
Random seed number preprocessing, least square postprocessing and parallel lumped kinetics method Download PDFInfo
- Publication number
- CN106709087A CN106709087A CN201510771667.6A CN201510771667A CN106709087A CN 106709087 A CN106709087 A CN 106709087A CN 201510771667 A CN201510771667 A CN 201510771667A CN 106709087 A CN106709087 A CN 106709087A
- Authority
- CN
- China
- Prior art keywords
- fraction oil
- matrix
- rate data
- oil
- quality point
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/30—Circuit design
- G06F30/36—Circuit design at the analogue level
- G06F30/367—Design verification, e.g. using simulation, simulation program with integrated circuit emphasis [SPICE], direct methods or relaxation methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/18—Manufacturability analysis or optimisation for manufacturability
Abstract
The invention discloses a random seed number preprocessing, least square postprocessing and parallel genetic lumped kinetics method and system. The method comprises the steps of obtaining distillate oil mass fraction data and a lumped kinetics equation through simulated distillation, and determining matrix metadata of a hydrogenation reaction rate matrix; then performing preprocessing, genetic algorithm processing and high-precision finishing processing on the determined matrix metadata through a random seed number preprocessing method, a genetic algorithm and a nonlinear least square method in sequence to obtain final optimized matrix metadata; and determining a model of the lumped kinetics equation according to the final optimized matrix metadata. According to the method, a plurality of lumped components can be defined for distillate oil raw materials and products, so that the flexibility requirement of product cutting in industrial production is met; the built lumped kinetics model enhances the flexibility for different fraction cutting schemes; and the degree of matching between a distribution curve of the distillate oil mass fraction data calculated through the lumped kinetics model and an experimental result is relatively high.
Description
Technical field
The present invention relates to fraction oil hydrocracking process studying technological domain, and in particular to Yi Zhongsui
Machine subnumber pretreatment least square post processing parallel genetic lumping kinetics method and system.
Background technology
One of vital task of petrochemical industry is by low-quality, high impurity content by hydrogenation reaction
The crude oil of the macromolecular that height is done or its pretreatment fraction oil be processed, with generate high-quality,
Low impurity content, all kinds of fraction oil products of high added value and the original of downstream petrochemicals
Material.Due to world's crude oil price, the price of product oil price and downstream petroleum and petrochemical industry product and
Demand constantly jumbo fluctuation, therefore, oil refining enterprise allows for the work to petroleum refining process
Skill parameter carries out effective adjustment in real time, to adapt to crude oil, product oil and downstream petrochemical industry
The change requirement of the price and demand of product.
Oil refining enterprise carries out in real time the effectively premise of adjustment to the technological parameter of petroleum refining process
Depend on the heightened awareness and the relatively accurate mathematics related to the process to hydrogenation process
The establishment and solution of pattern.
Mentioned lumped reaction kinetics divide relatively easy for the fraction of oil product at present, difficult
With raw material fraction and product involved by the reaction of high-precision description actual industrial and technological experiment
Complex distributions of the product fraction for cutting temperature.And then cause evaporating using different cutting schemes
The oily cutting result of part is very big with the error of result of calculation.
The content of the invention
For defect of the prior art, it is minimum that the present invention provides a kind of random seed number pretreatment
Two multiply post processing parallel genetic lumping kinetics method and system, greatly reduce using different
The fraction oil cutting result of cutting scheme and the error of result of calculation.
In a first aspect, the present invention provides a kind of post processing of random seed number pretreatment least square simultaneously
Row heredity lumping kinetics method, including：
S1, according to fraction oil quality point rate data and lumping kinetics equation, it is determined that hydrogenation splits
Change all M matrix metadata of reaction rate matrix, the fraction oil quality point rate packet
Include and be simulated the product fraction oil that distillation test is obtained to feedstock oil under different technology conditions
Fraction oil quality point rate data and the feedstock oil is carried out digital simulation product fraction oil
Fraction oil quality point rate data；
S2, M matrix metadata in step S1 is entered by random seed number preprocess method
Row optimization；
S3, in the step S2 optimize after M matrix metadata by genetic algorithm continuation
Optimize；
S4, to optimizing in step S3 after M matrix metadata by nonlinear least square method
Optimize, and determine M matrix metadata after optimization；
S5, the M matrix metadata determined according to step S4, determine lumping kinetics equation
Model；
The model of S6, the lumping kinetics equation determined according to the step S5, with to feedstock oil
The fraction oil quality point rate data of the product fraction oil that distillation test is obtained are simulated as first
Beginning condition, calculates the distillate mass fraction number of the product fraction oil corresponding to differential responses air speed
According to；
S7, step S2-S6 is performed into the lumping kinetics by polycaryon processor parallel computation
The simulation work of equation.
Optionally, the step S1 includes：
S11, the division virtual lump component of hydrocracking reaction；
S12, assume being hydrocracked lumped reaction kinetics；
S13, structure hydrocracking reaction network；
It is S14, true according to the lumped reaction kinetics of the hydrocracking reaction network and hypothesis
Determine lumping kinetics equation；
S15, the fraction according to the product fraction oil that distillation test acquisition is simulated to feedstock oil
Oil quality point rate data, the feedstock oil is carried out digital simulation product fraction oil fraction oil
Mass fraction data and lumping kinetics equation, determine the institute of hydrocracking reaction rate matrix
There is M matrix metadata.
Optionally, the step S11 includes：
S111, determine the feedstock oil be simulated under different technology conditions distillation test acquisition
Product fraction oil fraction oil quality point rate data and the mean boiling point of product fraction oil；
S112, the product for being simulated distillation test under different technology conditions according to the feedstock oil
The fraction oil quality point rate data of product fraction oil and the mean boiling point of product fraction oil, divide and add
The virtual lump component of hydrogen cracking reaction.
Optionally, the step S13 includes：
In N number of virtual lump component after division, the 1st product fraction of virtual lump component
The mean boiling point highest of oil, the mean boiling point of the product fraction oil of the virtual lump component of n-th is most
It is low；
The virtual lump components of i-th (1≤i≤N) are the i-th node, and i-th node includes i-1
In-degree and N-i out-degree；
Wherein, i represents the i-th node of virtual lump component；N represents the number of virtual lump component
Mesh, and each one node of virtual lump component correspondence, common N number of node.
Optionally, lumping kinetics equation is in the step S14：
Wherein, CiAnd CjRepresent the fraction oil quality point rate data of different virtual lump components；γi
The dynamics stoichiometric number of different virtual lump components is represented, different values represent different virtual respectively
The reaction of formation and consumption reaction of lump component；N represents the number of virtual lump component；I and
J represents different virtual lump components respectively；kaiRepresent matrix metadata.
Optionally, the matrix metadata is to include the lower triangular matrix of diagonal element.
Optionally, the step S2 includes：
S21, the multiple matrix element numbers of exhaustive acquisition that preset times are carried out to M matrix metadata
According to group；
S22, M matrix metadata in each matrix element data group substituted into respectively it is described
Lumping kinetics equation calculates the fraction oil quality point rate data of product fraction oil；
S23, the fraction oil quality point rate data of the product fraction oil that will be calculated with corresponding technique
Under the conditions of according to experiment obtain product fraction oil fraction oil quality point rate data contrasted,
Obtain the calculating product fraction oil fraction oil quality point rate data with corresponding technique bar
The residual error of the fraction oil quality point rate data of the product fraction oil obtained according to experiment under part is minimum
When corresponding matrix metadata, redefine M matrix metadata.
Optionally, in the step S23, calculated value is with the residual error err of experiment value：
Wherein, CCal, iRepresent the fraction oil by calculating the i-th virtual lump component for obtaining
Amount point rate data, and CTest, iRepresent by testing evaporating for the i-th virtual lump component for obtaining
Part oil quality point rate data, N is the number of virtual lump component, p and q be 0,1,2 or
It is infinitely great.
Optionally, the step S3 includes：
S31, the error function expressed by the residual error err are used as object function to be optimized；
S32, in the step S2 optimize after M matrix metadata in each matrix
Metadata carries out the multiple populations of disturbance generation in the range of default value；
S33, according to disturbance after multiple populations obtain the residual error err respectively；
S34, using the reciprocal function of the residual error err as the fitness function of genetic algorithm, choosing
Take population at individual corresponding during fitness maximum；
S35, the population at individual is carried out into population duplication, it is individual as population male parent；
S36, the population male parent individuality is intersected and is made a variation and produced new population at individual,
Using the new population at individual as M matrix metadata after optimization.
Second aspect, present invention also offers place after a kind of random seed number pretreatment least square
Reason parallel genetic lumping kinetics system, including：
Parameter primarily determines that module, for according to fraction oil quality point rate data and collection total output
Equation is learned, all M matrix metadata of hydrocracking reaction rate matrix are determined, it is described to evaporate
Part oil quality point rate data are simulated distillation to feedstock oil in fact under being included in different technology conditions
Test the fraction oil quality point rate data of the product fraction oil of acquisition and the feedstock oil is counted
Calculate the fraction oil quality point rate data of the product fraction oil of fitting；
Parameter pretreatment module, for primarily determining that M matrix metadata in module to parameter
Optimized by random seed number preprocess method；
Parameter optimization module, for M matrix metadata after the optimization of parameter pretreatment module
Optimization is proceeded by genetic algorithm；
Parameter determination module, for leading to M matrix metadata after the optimization of parameter optimization module
Cross nonlinear least square method to optimize, and determine M matrix metadata after optimization；
Model building module, for the M matrix metadata determined according to parameter determination module,
Determine the model of lumping kinetics equation；
Mass fraction computing module, for the collection total output determined according to the model building module
The model of equation is learned, to be simulated evaporating for the product fraction oil that distillation test is obtained to feedstock oil
Part oil quality point rate data calculate the product corresponding to differential responses air speed and evaporate as primary condition
The fraction oil quality point rate data of part oil；
Parallel computation module, for performing the collection total output by polycaryon processor parallel computation
Learn the simulation work of equation.
As shown from the above technical solution, a kind of random seed number pretreatment that the present invention is provided is minimum
Two multiply post processing parallel genetic lumping kinetics method and system, are obtained by randomly choosing mechanism
Fraction oil quality point rate data and lumping kinetics equation, determine hydrogenation reaction rate matrix
Matrix metadata, then passes sequentially through random seed number preprocess method, genetic algorithm, non-thread
Property the matrix metadata that determines of least square method pair pre-processed, genetic algorithm treatment and high
Precision refine treatment obtain final optimization pass after matrix metadata, and according to the final optimization pass after
Matrix metadata determines the model of lumping kinetics equation.The method divides multiple lump components,
Meet the requirement on flexibility of industrial production product cutting, and treated by three kinds of algorithms and set up mould
Parameter in type is optimized, and the lumped reaction kinetics after the foundation enhance different fractions and cut
The flexibility of scheme is cut, and the fraction oil quality point rate calculated by the lumped reaction kinetics
The fraction oil quality point of the product fraction oil that the distribution curve of data is obtained with simulation distillation test
The distribution curve goodness of fit of rate data is higher.
Brief description of the drawings
Fig. 1 pre-processes least square for a kind of random seed number that one embodiment of the invention is provided
Post-process the schematic flow sheet of parallel genetic lumping kinetics method；
Fig. 2 distills figure for the simulation of the raw material fraction oil that one embodiment of the invention is provided；
Fig. 3 distills figure for the simulation of the product fraction oil that one embodiment of the invention is provided；
The hydrocracking reaction network topological diagram that Fig. 4 is provided for one embodiment of the invention；
I-th virtual component of hydrocracking reaction network that Fig. 5 is provided for one embodiment of the invention
Node digraph；
The structure letter of the hydrocracking reaction rate matrix that Fig. 6 is provided for one embodiment of the invention
Figure；
The flow chart of the parallel algorithms that Fig. 7 is provided for one embodiment of the invention；
The use random seed number that Fig. 8 is provided for one embodiment of the invention is calculated, genetic algorithm is excellent
The product that the lumping kinetics equation group that change and nonlinear least square method refine are calculated is obtained
The distribution map of product fraction oil quality point rate data and the product fraction oil quality point acquired in experiment
The comparison diagram of the distribution map of rate data；
Fig. 9 divides calculating acquisition product fraction for the different lumps that one embodiment of the invention is provided
The comparison diagram of the distribution map of oil quality point rate data；
Figure 10 pre-processes least square for a kind of random seed number that one embodiment of the invention is provided
Post-process the structural representation of parallel genetic lumping kinetics system.
Specific embodiment
Below in conjunction with the accompanying drawings, the specific embodiment invented is further described.Hereinafter implement
Example is only used for clearly illustrating technical scheme, and can not limit this hair with this
Bright protection domain.
Lumping kinetics method is exactly applied to be hydrocracked or the relative of catalytic cracking process has
One of effect and feasible mathematic(al) mode.The basic thought of the method is will to react all kinds of oil being related to
Product simplify according to certain principle and are divided into specific virtual lumped component, and dividing mode can be according to
The boiling point of fraction oil, boiling range are divided.Also can be according to the component of fraction oil (especially with four components more
For common) divided.Splitting scheme has larger flexibility in itself.Then lump is established
The virtual reaction network of component, and according to the reaction network and the kinetics of correlation established
Relation establishes the equation of lumped reaction, and the form of equation is general with corresponding with first order reaction
Linear differential equation system it is most commonly seen.Then on the premise of reaction primary condition is brought into, lead to
The equation of the lumped reaction that the parsing and Numerical Methods Solve for crossing correlation are established, so as to obtain each
Parsing or numerical function form of the virtual lump fraction oil product of group with Annual distribution.And then realize
For being hydrocracked or catalytic cracking reaction product is for reaction velocity (soaking periods) relation
Dynamic detailed description.
After Fig. 1 is for a kind of random seed number pretreatment least square provided in an embodiment of the present invention
The schematic flow sheet of parallel genetic lumping kinetics method is processed, as shown in figure 1, the method bag
Include following steps：
S1, according to fraction oil quality point rate data and lumping kinetics equation, it is determined that hydrogenation splits
Change all M matrix metadata of reaction rate matrix, the fraction oil quality point rate packet
Include and be simulated the product fraction oil that distillation test is obtained to feedstock oil under different technology conditions
Fraction oil quality point rate data and the feedstock oil is carried out digital simulation product fraction oil
Fraction oil quality point rate data；
S2, M matrix metadata in step S1 is entered by random seed number preprocess method
Row optimization；
S3, in the step S2 optimize after M matrix metadata by genetic algorithm continuation
Optimize；
S4, to optimizing in step S3 after M matrix metadata by nonlinear least square method
Optimize, and determine M matrix metadata after optimization；
It will be appreciated that in above-mentioned steps S4, M matrix metadata after optimizing to step S3
The generalized error function for being constituted passes through nonlinear least square method, successively to each matrix element
Data optimize calculating, and M square when redefining error function value within a preset range
Array element data, wherein, the generalized error function by with M matrix metadata as independent variable,
Steamed with by simulation with the fraction oil quality point rate data of the product fraction oil according to digital simulation
Evaporate the absolute value conduct of the fraction oil quality point rate data difference of the product fraction oil that experiment is obtained
What functional value determined.
S5, the M matrix metadata determined according to step S4, determine lumping kinetics equation
Model.
The model of S6, the lumping kinetics equation determined according to the step S5, with to feedstock oil
The fraction oil quality point rate data of the product fraction oil that distillation test is obtained are simulated as first
Beginning condition, calculates the fraction oil quality point rate number of the product fraction oil corresponding to differential responses air speed
According to.
That is it is empty with the difference corresponding to the simulation distillation figure of the raw material fraction oil shown in Fig. 2
The fraction oil quality point rate data for intending component react differential equation group as lumping kinetics is solved
Initial value, using Runge-Kutta methods calculate solve the differential equation group.The interval of integration
From 0 to reaction velocity inverse.
S7, step S2-S6 is performed into the lumping kinetics by polycaryon processor parallel computation
The simulation work of equation.
The above method obtains fraction oil quality point rate data and lump by randomly choosing mechanism
Kinetics equation, determines the matrix metadata of hydrogenation reaction rate matrix, then pass sequentially through with
The matrix element that machine subnumber preprocess method, genetic algorithm, nonlinear least square method pair determine
After data are pre-processed, genetic algorithm is processed and high accuracy refine treatment obtains final optimization pass
Matrix metadata, and lumping kinetics side is determined according to the matrix metadata after the final optimization pass
The model of journey.The method divides multiple lump components, meets the flexible of industrial production product cutting
Property require, and treat the parameter set up in model by three kinds of algorithms and optimize, the foundation
Lumped reaction kinetics afterwards enhance the flexibility of different faction cut schemes, and by this
The distribution curve and experimental result of the fraction oil quality point rate data that lumped reaction kinetics are calculated
The goodness of fit is higher.
Describe the above method in detail below by specific embodiment, following examples are only used for
The method of the present invention is illustrated, but is not used to limit protection scope of the present invention.
Above-mentioned steps S1 is specifically included：
S11, the division virtual lump component of hydrocracking reaction；
Step S11 specifically includes following steps：
S111, determine the feedstock oil be simulated under different technology conditions distillation test acquisition
Product fraction oil fraction oil quality point rate data and the mean boiling point of product fraction oil；
S112, the product for being simulated distillation test under different technology conditions according to the feedstock oil
The fraction oil quality point rate data of product fraction oil and the mean boiling point of product fraction oil, divide and add
The virtual lump component of hydrogen cracking reaction.
First from 40ml small hydrogenation devices according to different technology conditions successively Extracting temperature, treatment
Amount and the different corresponding fraction oil product of 27 groups of experiments of hydrogen dividing potential drop power, gather 27 kinds of works respectively
The fraction oil quality point rate number that distillation test is obtained is simulated under skill operating condition to feedstock oil
According to the fraction oil quality point rate data of digital simulation are carried out according to the feedstock oil as this mould
The parameter fitting data set that type method is calculated.By the fraction oil mould corresponding to raw material and checking test
Intend distillation data to draw as shown in Figures 2 and 3.In fig. 2, wherein X-coordinate representative simulation is steamed
Evaporate cutting temperature/degree Celsius, Y-coordinate represents raw material fraction oil quality point rate data, in Fig. 3,
X-coordinate representative products fraction oil cutting temperature/degree Celsius, Y-coordinate representative products fraction oil quality
Divide rate data.27 groups of simulation distillation test data are divided, by all oil product initial boiling points
Minimum value is made as the maximum that the initial boiling point of model of fit system does all oil products simultaneously
It is doing for model of fit system, so all fraction oil boiling ranges being calculated in model of fit
Boiling point range in.Then to the boiling range involved by calculating according to experiment and calculating the need for carry out
Lump is divided, and carrying out 100 lumps to the scope at this divides.
S12, assume being hydrocracked lumped reaction kinetics；
Step S12 comprises the following steps：
S121, react fraction and generation fraction belong to same boiling range when, then the reaction is refused to examine
Consider；
S122, hydrocracking reaction have one-way and irreversibility；
S123, the speed constant of the hydrocracking reaction are influenced by temperature and meet Arrhenius
Equation；
Compared with Light ends for not had an effect compared with heavy component in S124, each virtual lump component,
And the accumulation after the reaction and component reaction compared with heavy component is to the generation quantity and product compared with Light ends
Raw speed makes a difference；
S125, all reactions are described using first order reaction kinetics model；
S126, course of reaction are only controlled by dynamic process, not by other process influences.
S13, structure hydrocracking reaction network；
The step S13 includes：
In N number of virtual lump component after division, the 1st product fraction of virtual lump component
The mean boiling point highest of oil, the mean boiling point of the product fraction oil of the virtual lump component of n-th is most
It is low；
I-th (1≤i≤N) virtual lump component includes i-1 in-degree and N-i out-degree；
Wherein, i represents the i-th node of virtual lump component；N represents the number of virtual lump component
Mesh, and each one node of virtual lump component correspondence, common N number of node.
For example, the reaction network for drawing course of reaction is as shown in Figure 4.Wherein, from a left side to
The mean boiling point of the product fraction oil of right virtual lump component is from low to high.In the reaction network
The total N of virtual lump component is 100.In partition process, the 1st virtual lump component
The mean boiling point highest of product fraction oil, it is understood that for virtual lump component is most heavy void
Intend lump component, and the mean boiling point of the product fraction of n-th virtual lump component oil is minimum,
It can be appreciated that virtual lump component is most light virtual lump component.Now by i-th lump
Component is extracted as shown in Figure 5 from reaction network schematic diagram.For i-th virtual lump group
For part, it can be considered comprising 100 i-th nodes of the digraph of node.For the node
For, amount to and include i-1 in-degree and N-i out-degree.For reaction system, i-1 enters
Degree represents the i-th node virtual lump component and receives the virtual lump group come from than i-th node
The product of the pyrolysis of oil fractions of the virtual lump component of i-1 heavier node of part, N-i out-degree
Represent the virtual lump component self-cleavage of the i-th node more virtual than i-th node to come from
The virtual lump component of N-i lighter node of lump component provides raw material.
It is S14, true according to the lumped reaction kinetics of the hydrocracking reaction network and hypothesis
Determine lumping kinetics equation；
According to involved by reaction network figure and lumped watershed hydrologic model equation assume that establishing description reacts
And ordinary differential system.In the present embodiment, equation group includes 100 ODEs.
Specifically, lumping kinetics equation is：
Formula (1)
Wherein, CiAnd CjRepresent the fraction oil quality point rate data of different virtual lump components；γi
The dynamics stoichiometric number of different virtual components is represented, different values represent different virtual lumps respectively
The reaction of formation and consumption reaction of component；N represents the number of virtual lump component；I and j distinguishes
Represent different virtual lump components；kaiRepresent matrix metadata.
S15, the fraction according to the product fraction oil that distillation test acquisition is simulated to feedstock oil
Oil quality point rate data, the feedstock oil is carried out digital simulation product fraction oil fraction oil
Mass fraction data and lumping kinetics equation, determine the institute of hydrocracking reaction rate matrix
There is M matrix metadata.
As shown in fig. 6, the matrix metadata is to include the lower triangular matrix of diagonal element, reaction
The mathematical form of rate matrix is through virtual lump component from gently including to can be converted into after permutatation
The upper triangular matrix of diagonal element.
It is assumed that all matrix metadata of hydrocracking reaction rate matrix, the matrix is
100x100 scales.But the matrix is lower triangular matrix according to preamble model hypothesis.Simultaneously because
The characteristics of reaction itself, each element characterizes the virtual lump group on reaction rate diagonal of a matrix
The heating rate coefficient for dividing, its absolute value should be equal to the generation of all other element in respective column
Number and.The hydrogenation reaction rate matrix structure diagram involved by the embodiment is now listed in Fig. 6 institutes
Show.Its matrix metadata of position a little is not zero in figure 6, and blank space its matrix element number
According to being zero.Include 5049 nonzero elements in whole matrix, these nonzero elements are this mould
Type parameter to be optimized is M matrix metadata to be optimized.
Above-mentioned steps S2 includes：
S21, the multiple matrix element numbers of exhaustive acquisition that preset times are carried out to M matrix metadata
According to group；
S22, M matrix metadata in each matrix element data group substituted into respectively it is described
Lumping kinetics equation calculates the fraction oil quality point rate data of product fraction oil；
S23, the fraction oil quality point rate data of the product fraction oil that will be calculated with corresponding technique
Under the conditions of according to experiment obtain product fraction oil fraction oil quality point rate data contrasted,
Obtain the calculating product fraction oil fraction oil quality point rate data with corresponding technique bar
The residual error of the fraction oil quality point rate data of the product fraction oil obtained according to experiment under part is minimum
When corresponding matrix metadata, redefine M matrix metadata.
In order to subsequent descriptions are convenient, the fraction oil quality point rate data of the product fraction oil that will be calculated
Calculated value is defined as, the fraction oil quality with the product fraction oil for calculating divides rate data in corresponding work
The fraction oil quality point rate data definition obtained according to experiment under the conditions of skill is experiment value.
Specifically, it is minimum with the residual error err of calculated value that experiment value is established in above-mentioned steps S21-S23
Initial reaction rate matrix matrix metadata.Specific optimization calculating process is as follows：
(1) assume that 5049 parameters to be optimized are the condition random number between 0-100, this group of bar
Part random number meets element absolute value on each column diagonal and is equal to other all elements in respective column
Algebraical sum this constraints.
(2) parameter to be optimized by assuming is simulated with feedstock oil using Runge-Kutta methods and steamed
Evaporate the fraction corresponding to curve and be distributed as primary condition, the inverse with 0 to reaction velocity is integration
Interval solve established comprising 100 lumping kinetics ODEs of ODE
Group.
(3) by solving result with corresponding process conditions product fraction oil simulation distillation curve institute
Corresponding fraction oil distribution curve carries out error calculation, and its computing formula is as follows.Count herein
During calculation, parameter p=1, q=1, N=100 and record error between the two are taken.
Calculated value is with the residual error err of experiment value：
Formula (2)
Wherein, CCal, iRepresent by calculating the i-th fraction oil quality of virtual lump component for obtaining
Point rate data, and CTest, iRepresent the fraction oil by testing the i-th virtual lump component for obtaining
Mass fraction data, N is the number of virtual lump component, and p and q is 0,1,2 or infinitely great.
(4) (1)-(3) process 10 is repeated6It is secondary, choose error most in all multiple errors
It is small to be worth as the matrix metadata of initial reaction rate matrix.As the initial of next step calculating
Value.
Above-mentioned steps S3 includes：
S31, with the mistake expressed by the absolute value of residual error err namely the difference of calculated value and experiment value
Difference function is used as object function to be optimized；
S32, in the step S2 optimize after M matrix metadata in each matrix
Metadata carries out the multiple populations of disturbance generation in the range of default value；
S33, according to disturbance after multiple populations obtain the residual error err respectively；
S34, using the reciprocal function of the residual error err as the fitness function of genetic algorithm, choosing
Take population at individual corresponding during fitness maximum；
S35, the population at individual is carried out into population duplication, it is individual as population male parent；
S36, the population male parent individuality is intersected and is made a variation and produced new population at individual,
Using the new population at individual as M matrix metadata after optimization.
Specifically, above-mentioned steps S31-S36 is based on genetic algorithm establishes experiment value with calculated value
The matrix metadata of the minimum reaction rate matrix of residual error.Specific optimization calculating process is as follows：
(1) each to M matrix metadata initial value matrix after optimization in the step S2
Element carries out small-scale perturbation, and range of disturbance is each of each matrix metadata initial value matrix
0.8-1.2 times of element numerical value.The initial matrix unit number that scale is 800 is generated by random perturbation
According to population.The purpose of this disturbance both ensure that " the outstanding property " of population oeverall quality, protect again
Demonstrate,prove " variation " of population, and the initial kind calculated as genetic algorithm optimization using the population
Group.
(2) reciprocal function of the error function shown in formula (2) fitting as genetic algorithm
Response function, calculates the fitness function value of each population at individual successively.Wherein error function meter
P=1 and q=1 is taken during calculation.Ensure that error is smaller, its related fitness is bigger.So that by mistake
The small matrix metadata population of difference is more obtained in that reservation.
(3) selecting individuality using the selective algorithm of genetic algorithm carries out population duplication, under
The population male parent of secondary calculating is individual；
(4) the population male parent individuality chosen is intersected and the generation new population that made a variation
Body, related crossing-over rate is 0.1-0.9, preferably 0.5, aberration rate is 0.0001-0.05, excellent
Elect 0.001 as；
(5) new population at individual is substituted into old population at individual, repeat step (2)-(4)
Until evolve reaching optimization aim, the computer algebra of whole genetic algorithm was 10000 generations.
Retain the population at individual of all different algebraically being related in calculating process 8x10 altogether6It is individual,
It is that the minimum individuality of error function value is joined as next step to choose fitness function maximum wherein
The initial value that number refine is calculated.
Specifically, above-mentioned steps rate matrix metadata is excellent based on genetic algorithm as primary data
Changing can also in concrete numerical value this detailed process of computation rate matrix correlation matrix metadata
It is interpreted as comprising the following steps：
A) using as experiment value with calculate the norm of value difference expressed by error function as to be optimized
The form of object function；
B the matrix element numerical value) for the initial reaction rate matrix of each optimization aim generation is adjacent
The population of random generation respective objects in domain, population at individual number is 20-400000；
C stochastic transformation) is carried out according to associated binary codes to each population, new population is formed；
D) by C) in the way of sequentially generate new individual, new individual number is 20-400000；
E) using object function to be optimized it is reciprocal or to the positive related function of the function as fitting
Response function, calculates the fitness function value of each population at individual successively；
F ideal adaptation) is selected according to probabilistic manner using " roulette " algorithm of genetic algorithm
The larger population of degree functional value carries out population duplication, individual as the population male parent for calculating next time；
G) the population male parent individuality chosen is intersected and the generation new population that made a variation
Body, related crossing-over rate is 0.1-0.9, and aberration rate is 0.0001-0.05；
H new population) is substituted into old population, repeat step C)-H) until evolving reaches optimization
Target, the computer algebra of whole genetic algorithm is 100-1000000 generations.
Specifically, in above-mentioned steps S4, based on Nonlinear Least-Square Algorithm optimize previous step by
The matrix metadata that genetic algorithm is established.Specific optimization calculating process is as follows：
(1) calculated corresponding to the maximum individuality of the fitness function for obtaining with by genetic algorithm
The iteration initial value that matrix metadata is calculated as nonlinear least square method.
(2) error function that whole calculating process is related to can be expressed as comprising 5049 related speed
The generalized function of rate matrix coefficient.It is expressed as follows：
Wherein, err and F (kai) total generalized error function is represented,
giRepresent the introduced error of each differential equation；
kaiRepresent reaction rate matrix metadata to be optimized；
N represents virtual lump group mesh.
Here, order
Again by fi(kaj) independent variable is replaced the following functional form of acquisition：
Next by comprising the n equation group f of unknown numberi(x(k))=0, i=1..n enters line translation and asks
Solution.
(3) assume the correlation matrix metadata with genetic algorithm acquisition as x(1), here it is assumed that initially
Parameter alpha1>0 and growth factor β>1, the allowable error of setting
ε>0, calculate F (x(k)) assume first that α=α1, k=1.
(4) α=α/β, k=1 are made.Successively calculating matrix unit number
According to
Form matrix
(5) because whole calculating process functional form is sufficiently complex, the solution procedure of its partial derivative
It is difficult to be expressed with analytical form, differential is approximately replaced using difference method herein.
(6) to avoid AkThere is singularity and introduces parameter alpha in matrix, solves equationCalculated direction d(k), and calculate the argument value of iteration next step
x(k+1)=x(k)+d(k)。
(7) new independent variable x is calculated again(k+1)Corresponding functional value F (x(k+1)).If
F(x(k+1))<F(x(k)) then prove that iterative calculation successfully passes the reduction of iterative target functional value.
(8th) step is then transferred to, (7th) step is otherwise transferred to.
(8) ifThen iterative calculation terminates, and makes last solutionOtherwise increase
Plus hunting zone, make α=α β, be transferred to 5) step recalculate.
(9) ifThen iterative calculation terminates, and makes last solutionOtherwise make
K=k+1, returns to (4th) step and recalculates.
In whole calculating process, because preamble genetic algorithm has caused genetic algorithm optimal solution
Compare " near " away from overall globally optimal solution.Therefore, choose initial in refine calculating process
Parameter and growth factor are smaller, and parameter selection is as follows in calculating process：α1=0.001,
β=1.05, Δ xj=0.05 and ε=0.001.
As shown in fig. 7, above-mentioned steps S7 specifically includes following steps：
S71,1000000 groups are produced to include 5049 using the core of multi-core computer 16 is simultaneously parallel
The array of the random number between individual 0~100；
S72,1000000 groups of random arrays to producing carry out the generation lump of 16 core parallel compositions
Kinetic reaction matrix；
S73, primary condition is distributed as with zero to corresponding anti-with feedstock oil mass fraction data
Answer air speed reciprocal for integrating range is calculated according to the matrix parallel of generation using Runge-Kutta methods
1000000 groups evaporate comprising the product corresponding to 100 lumping kinetics equation groups of the differential equation
The distribution curve of the fraction oil quality point rate data of part oil；
S74, the fraction oil that 1000000 set product fractions oil is calculated according to above-mentioned formula (2)
The distribution curve of amount point rate data and the fraction oil quality point rate data of actual product fraction oil
The difference of distribution curve, selection wherein corresponding to difference reckling comprising 5049 random numbers
Array as reaction rate matrix initial driving force parameter；
S75, it is carried out to each element in the initial driving force parameter array that is obtained
0.8-1.2 times of random perturbation produces 800 groups of initial populations of genetic algorithm parallel, and to each
Population at individual carries out binary coding；
S76, the binary coding grouping parallel that will be generated intersect, the binary system that variation generation is new
Code, and generation binary code is converted into corresponding genetic algorithm calculating new population, according to
The adaptive response function as genetic algorithm reciprocal of formula (2), parallel computation is initial successively
Population and the adaptive response functional value of new population；
S77, the adaptive response functional value for obtaining calculating are according to the mode of " roulette " by general
Rate randomly selects 800 groups of new populations, and follow-up calculating is participated in as population of future generation；
S78, the new population that will be generated are instead of initial population and record the two of all population at individual
Scale coding and all individual adaptive response functions and error function value, repeat S76-S77
The population that step 10000 time is obtained is used as final population；
S79, choose in calculating process that auto-adaptive function numerical value in all individualities is maximum namely error
The minimum individual initial value calculated as next step refine of function；
S80, each element for the optimum individual selected according to genetic algorithm are using three points value differences
The method divided is using 16 core parallel computation mode approximate calculation generalized error functions for each square
Array element data first derivative vector and second-order matrix, while set initial parameter, increase because
Son, allowable error and difference step size；
S81, carry out 16 core parallel compositions by matrix metadata and disturb and solve to forming matrix
Dependent equation group and then related search method is obtained for follow-up calculating is prepared；
S82, former variable is added and the direction of search forms new variables, and calculated corresponding to new variables
Enthalpy numerical value, it is right that the enthalpy numerical value of the enthalpy numerical value of new variables and former variable is carried out
Than judging, former variable is replaced with new variables if object function reduces, increase hunting zone, such as
The increase of fruit objective function value then reduces growth factor, and former variable keeps constant simultaneously；
S83, the solution vector used instead of S80 with new solution vector, repeat S80-S82 steps
Until error function value be less than 0.0001 untill, stop iterative calculation, and by the solution of gained to
Amount thereby determines that whole undetermined parameters of related lumping kinetics as the solution vector after refine,
Establish the mathematical form of lumping kinetics equation；
S84, according to establish lumping kinetics equation group with feedstock oil mass fraction data (i.e.
The fraction oil quality point rate number of the product fraction oil that distillation test is obtained is simulated to feedstock oil
According to) as the fraction oil quality point of the product fraction oil corresponding to the different air speeds of primary condition calculating
Rate data, and corresponding fraction oil quality point rate number can be calculated according further to different cutting schemes
According to distribution situation.
As shown in figure 8, X-coordinate represent fraction oil cutting temperature/degree Celsius, Y-coordinate represent evaporate
Part oil quality point rate data, are respectively adopted initial random seeds number, genetic algorithm and non-in figure
Linear least square calculates the product fraction oil quality point that lumping kinetics equation group is obtained
The distribution map of rate data divides the distribution map of rate data with the product fraction oil quality acquired in experiment
Comparison diagram.It can be found that because number of parameters to be optimized is numerous by initial random seeds number meter
The distribution map of the product fraction oil quality of calculation point rate data and the product fraction oil acquired in experiment
The distribution map of mass fraction data still has very big difference, and calculates the product for determining by genetic algorithm
The distribution map of product fraction oil quality point rate data and the product fraction oil quality point acquired in experiment
The distribution map of rate data is relatively close together but still has certain deviation, and eventually passes through eight times repeatedly
For the result of calculation for calculating refine acquisition steady conjunction then suitable with experimental result.Realize for producing
The high accuracy numerical fitting of product fraction distribution.The related residual calculated by genetic algorithm
Err=24.1186, and pass through the final residual of the step of non-linear least square method 7 iterative calculation acquisition
The corresponding reaction rate matrixes of difference err=3.9568. are numerous due to matrix metadata, only enumerate part
It is listed in table 1.Table 1 is shown to be calculated by various algorithms and obtains residual error and part matrix metadata
Numerical value.
Table 1
To choosing different collection sum and different cutting schemes and uses in the above method
The time of different computational methods consumption contrasts as follows respectively：
(1) in order to show algorithm involved in the present invention for the virtual lump involved by lumped model
The dependence of component number, is calculated with the lump of 20 lump 50 and 100 lumped models respectively herein
The distribution curve of related fraction oil quality point rate data, and result of calculation is entered with experimental result
Row contrast, comparing result is listed in shown in Fig. 9, X-coordinate representative products fraction oil cutting temperature
/ degree Celsius, Y-coordinate representative products fraction oil quality point rate data.
Result of calculation shows that lump component number is intended for the calculating of product fraction oil quality point rate
Close quite important.If it is desired to reaching fitting precision higher must assure that enough virtual group fractions
Amount.When virtual component is 20, result of calculation is very coarse to express product fraction oil
The general trend of amount point rate data, but product fraction oil quality point rate data cannot be given expression to completely
Distribution detailed information, no matter high boiling range fraction namely the virtual lump group of mean boiling point highest
The fraction oil quality point rate data or low boiling journey fraction fraction namely mean boiling point of part are minimum
The fraction oil quality point rate data of virtual lump component are not always the case；When virtual component is 50,
Result of calculation is preferable for the distribution and expression of the fraction oil quality point rate data of high boiling range fraction,
Because ripple of the distribution of the fraction oil quality of high boiling range fraction point rate data with cutting temperature
It is dynamic smaller, and what model coincide for the distribution of the fraction oil quality point rate data of low boiling journey fraction
It is rougher, because the distribution of the fraction oil quality of low boiling journey fraction point rate data is with cutting temperature
Fluctuation is larger；When virtual component is 100, fraction of the result of calculation for height boiling range fraction
It is preferable that the distribution of oil quality point rate data coincide, and had both reflected the fraction oil of high boiling range fraction
The gentle distribution of mass fraction data, is also demonstrated by the fraction oil quality point rate number of low boiling journey fraction
According to fluctuation be distributed.
(2) in order to show flexible practicality of the method for different faction cut schemes.It is existing
Two kinds of different product fraction oil cutting schemes and its according to the method described above calculating are obtained
Result is listed in shown in table 2 and table 3, and wherein table 2 and table 3 are the fraction oil of different cutting fractions
Amount point rate data distribution situation.
Table 2
Table 3
By the comparing of table 2 and table 3 it can be found that because the present invention can use more collection
Total component is divided and calculated to fraction oil, while its result of calculation fitting precision is very high.Cause
This result of calculation of the invention can have very broad sense for the flexible cutting scheme of fraction oil
With flexible applicability.
(3) in order to show acceleration effect of the parallel calculating method for calculating, herein using difference
Lump multi-component model calculated using different concurrent computational systems, during by related calculating
Between be listed in shown in table 4.
Table 4
By the data of table 4 it can be found that shortening as the increase for calculating core number calculates the time,
Computational efficiency can actually be increased.But for the shortening that unit concurrent computational system calculates the time
It is not in simple inversely prroportional relationship with the increase of calculating core number.When in calculating calculation less than etc.
This inversely prroportional relationship is obvious when 8, though the calculating time needed for when the center of calculating is for 16
So generally reduce but be simultaneously not up to 1/2 of calculating center when being 8, because parallel computation system
Caused by system internal hardware configuration.But from general, the introducing of parallel calculating method is certain
The calculating time can be greatly shortened.For example, for the whole calculating of 250 lumped reaction kinetics
It is about for process, when mutually being shut down using serial computing 150 days or so, and uses 16 cores parallel
Computing system can then ensure to complete identical evaluation work at 11 days or so.
As shown in Figure 10, the embodiment of the present invention additionally provides a kind of random seed number pretreatment minimum
Two structural representations for multiplying post processing parallel genetic lumping kinetics system, as shown in Figure 10, should
System includes：
Parameter primarily determines that module 101, for according to fraction oil quality point rate data and lump
Kinetics equation, determines all M matrix metadata of hydrocracking reaction rate matrix, institute
State under fraction oil quality point rate data are included in different technology conditions and steaming is simulated to feedstock oil
Evaporate the fraction oil quality point rate data of the product fraction oil that experiment is obtained and the feedstock oil is entered
The fraction oil quality point rate data of the product fraction oil of row digital simulation；
Parameter pretreatment module 102, for primarily determining that M matrix element in module to parameter
Data are optimized by random seed number preprocess method；
Parameter optimization module 103, for M matrix element after the optimization of parameter pretreatment module
Data proceed optimization by genetic algorithm；
Parameter determination module 104, for M matrix element number after the optimization of parameter optimization module
Optimized according to by nonlinear least square method, and determine M matrix element number after optimization
According to；
Model building module 105, for the M matrix element number determined according to parameter determination module
According to determining the model of lumping kinetics equation.
Mass fraction computing module 106, for the lump determined according to the model building module
The model of kinetics equation, to be simulated the product fraction oil that distillation test is obtained to feedstock oil
Fraction oil quality point rate data as primary condition, calculate the product corresponding to differential responses air speed
The fraction oil quality point rate data of product fraction oil；
Parallel computation module 107, for performing the lump by polycaryon processor parallel computation
The simulation work of kinetics equation.
Said system is one-to-one with the above method, and the implementation detail of the above method is also suitable
In the system, the present embodiment is not described in detail to the system.
In specification of the invention, numerous specific details are set forth.It is to be appreciated, however, that this
Inventive embodiment can be put into practice in the case of without these details.In some examples
In, known method, structure and technology is not been shown in detail, so as not to fuzzy to this specification
Understanding.
Although it will be appreciated by those of skill in the art that some embodiments described herein include it
Some included features are rather than further feature, but the spy of different embodiments in its embodiment
The combination levied means to be within the scope of the present invention and formed different embodiments.Example
Such as, in the following claims, the one of any of embodiment required for protection can
Mode is used in any combination.
Finally it should be noted that：Various embodiments above is merely illustrative of the technical solution of the present invention,
Rather than its limitations；Although being described in detail to the present invention with reference to foregoing embodiments,
It will be understood by those within the art that：It still can be to described in foregoing embodiments
Technical scheme modify, or which part or all technical characteristic are equally replaced
Change；And these modifications or replacement, the essence of appropriate technical solution is departed from of the invention each
The scope of embodiment technical scheme, it all should cover in claim of the invention and specification
In the middle of scope.
Claims (10)
1. a kind of random seed number pretreatment least square post-processes parallel genetic lumping kinetics
Method, it is characterised in that including：
S1, according to fraction oil quality point rate data and lumping kinetics equation, it is determined that hydrogenation splits
Change all M matrix metadata of reaction rate matrix, the fraction oil quality point rate packet
Include and be simulated the product fraction oil that distillation test is obtained to feedstock oil under different technology conditions
Fraction oil quality point rate data and the feedstock oil is carried out digital simulation product fraction oil
Fraction oil quality point rate data；
S2, M matrix metadata in step S1 is entered by random seed number preprocess method
Row optimization；
S3, in the step S2 optimize after M matrix metadata by genetic algorithm continuation
Optimize；
S4, to optimizing in step S3 after M matrix metadata by nonlinear least square method
Optimize, and determine M matrix metadata after optimization；
S5, the M matrix metadata determined according to step S4, determine lumping kinetics equation
Model；
The model of S6, the lumping kinetics equation determined according to the step S5, with to feedstock oil
The fraction oil quality point rate data of the product fraction oil that distillation test is obtained are simulated as first
Beginning condition, calculates the distillate mass fraction number of the product fraction oil corresponding to differential responses air speed
According to；
S7, step S2-S6 is performed into the lumping kinetics by polycaryon processor parallel computation
The simulation work of equation.
2. method according to claim 1, it is characterised in that the step S1 includes：
S11, the division virtual lump component of hydrocracking reaction；
S12, assume being hydrocracked lumped reaction kinetics；
S13, structure hydrocracking reaction network；
It is S14, true according to the lumped reaction kinetics of the hydrocracking reaction network and hypothesis
Determine lumping kinetics equation；
S15, the fraction according to the product fraction oil that distillation test acquisition is simulated to feedstock oil
Oil quality point rate data, the feedstock oil is carried out digital simulation product fraction oil fraction oil
Mass fraction data and lumping kinetics equation, determine the institute of hydrocracking reaction rate matrix
There is M matrix metadata.
3. method according to claim 2, it is characterised in that the step S11 includes：
S111, determine the feedstock oil be simulated under different technology conditions distillation test acquisition
Product fraction oil fraction oil quality point rate data and the mean boiling point of product fraction oil；
S112, the product for being simulated distillation test under different technology conditions according to the feedstock oil
The fraction oil quality point rate data of product fraction oil and the mean boiling point of product fraction oil, divide and add
The virtual lump component of hydrogen cracking reaction.
4. method according to claim 3, it is characterised in that the step S13 includes：
In N number of virtual lump component after division, the 1st product fraction of virtual lump component
The mean boiling point highest of oil, the mean boiling point of the product fraction oil of the virtual lump component of n-th is most
It is low；
The virtual lump components of i-th (1≤i≤N) are the i-th node, and i-th node includes i-1
In-degree and N-i out-degree；
Wherein, i represents the i-th node of virtual lump component；N represents the number of virtual lump component
Mesh, and each one node of virtual lump component correspondence, common N number of node.
5. method according to claim 2, it is characterised in that collect in the step S14
Total output equation is：
Wherein, CiAnd CjRepresent the fraction oil quality point rate data of different virtual lump components；γi
The dynamics stoichiometric number of different virtual lump components is represented, different values represent different virtual respectively
The reaction of formation and consumption reaction of lump component；N represents the number of virtual lump component；I and
J represents different virtual lump components respectively；kaiRepresent matrix metadata.
6. method according to claim 5, it is characterised in that the matrix metadata
It is to include the lower triangular matrix of diagonal element.
7. method according to claim 1, it is characterised in that the step S2 includes：
S21, the multiple matrix element numbers of exhaustive acquisition that preset times are carried out to M matrix metadata
According to group；
S22, M matrix metadata in each matrix element data group substituted into respectively it is described
Lumping kinetics equation calculates the fraction oil quality point rate data of product fraction oil；
S23, the fraction oil quality point rate data of the product fraction oil that will be calculated with corresponding technique
Under the conditions of according to experiment obtain product fraction oil fraction oil quality point rate data contrasted,
Obtain the calculating product fraction oil fraction oil quality point rate data with corresponding technique bar
The residual error of the fraction oil quality point rate data of the product fraction oil obtained according to experiment under part is minimum
When corresponding matrix metadata, redefine M matrix metadata.
8. method according to claim 7, it is characterised in that in the step S23,
Calculated value is with the residual error err of experiment value：
Wherein, CCal, iRepresent the fraction oil by calculating the i-th virtual lump component for obtaining
Amount point rate data, and CTest, iRepresent by testing evaporating for the i-th virtual lump component for obtaining
Part oil quality point rate data, N is the number of virtual lump component, p and q be 0,1,2 or
It is infinitely great.
9. method according to claim 8, it is characterised in that the step S3 includes：
S31, the error function expressed by the residual error err are used as object function to be optimized；
S32, in the step S2 optimize after M matrix metadata in each matrix
Metadata carries out the multiple populations of disturbance generation in the range of default value；
S33, according to disturbance after multiple populations obtain the residual error err respectively；
S34, using the reciprocal function of the residual error err as the fitness function of genetic algorithm, choosing
Take population at individual corresponding during fitness maximum；
S35, the population at individual is carried out into population duplication, it is individual as population male parent；
S36, the population male parent individuality is intersected and is made a variation and produced new population at individual,
Using the new population at individual as M matrix metadata after optimization.
10. a kind of random seed number pretreatment least square post-processes parallel genetic lumping kinetics
System, it is characterised in that including：
Parameter primarily determines that module, for according to fraction oil quality point rate data and collection total output
Equation is learned, all M matrix metadata of hydrocracking reaction rate matrix are determined, it is described to evaporate
Part oil quality point rate data are simulated distillation to feedstock oil in fact under being included in different technology conditions
Test the fraction oil quality point rate data of the product fraction oil of acquisition and the feedstock oil is counted
Calculate the fraction oil quality point rate data of the product fraction oil of fitting；
Parameter pretreatment module, for primarily determining that M matrix metadata in module to parameter
Optimized by random seed number preprocess method；
Parameter optimization module, for M matrix metadata after the optimization of parameter pretreatment module
Optimization is proceeded by genetic algorithm；
Parameter determination module, for leading to M matrix metadata after the optimization of parameter optimization module
Cross nonlinear least square method to optimize, and determine M matrix metadata after optimization；
Model building module, for the M matrix metadata determined according to parameter determination module,
Determine the model of lumping kinetics equation；
Mass fraction computing module, for the collection total output determined according to the model building module
The model of equation is learned, to be simulated evaporating for the product fraction oil that distillation test is obtained to feedstock oil
Part oil quality point rate data calculate the product corresponding to differential responses air speed and evaporate as primary condition
The fraction oil quality point rate data of part oil；
Parallel computation module, for performing the collection total output by polycaryon processor parallel computation
Learn the simulation work of equation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510771667.6A CN106709087A (en) | 2015-11-12 | 2015-11-12 | Random seed number preprocessing, least square postprocessing and parallel lumped kinetics method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510771667.6A CN106709087A (en) | 2015-11-12 | 2015-11-12 | Random seed number preprocessing, least square postprocessing and parallel lumped kinetics method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106709087A true CN106709087A (en) | 2017-05-24 |
Family
ID=58929717
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510771667.6A Pending CN106709087A (en) | 2015-11-12 | 2015-11-12 | Random seed number preprocessing, least square postprocessing and parallel lumped kinetics method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106709087A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113764046A (en) * | 2021-09-24 | 2021-12-07 | 中国石油化工股份有限公司 | Method for improving yield of high value-added product in catalytic diesel oil hydroconversion |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103914595A (en) * | 2014-04-01 | 2014-07-09 | 西北大学 | Modeling method of medium-temperature coal tar total-fraction hydrogen cracking lumping kinetic model |
-
2015
- 2015-11-12 CN CN201510771667.6A patent/CN106709087A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103914595A (en) * | 2014-04-01 | 2014-07-09 | 西北大学 | Modeling method of medium-temperature coal tar total-fraction hydrogen cracking lumping kinetic model |
Non-Patent Citations (3)
| Title |
|---|
| 李群勇: "加氢裂化反应器的建模和仿真", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》 * |
| 祝然: "结构导向集总新方法构建催化裂化动力学模型及其应用研究", 《中国博士学位论文全文数据库 工程科技Ⅰ辑》 * |
| 谷云格等: "油浆抽提油在Ni-Mo/SiO2-Al2O3催化剂上加氢脱除PAHs动力学", 《石油化工高等学校学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113764046A (en) * | 2021-09-24 | 2021-12-07 | 中国石油化工股份有限公司 | Method for improving yield of high value-added product in catalytic diesel oil hydroconversion |
| CN113764046B (en) * | 2021-09-24 | 2024-04-16 | 中国石油化工股份有限公司 | Method for improving yield of high value-added product of catalytic diesel oil hydro-conversion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103473598A (en) | Extreme learning machine based on length-changing particle swarm optimization algorithm | |
| CN110046408A (en) | Bounded-but-unknown uncertainty optimization method based on BP neural network differential and interval analysis | |
| Davis et al. | Efficient surrogate model development: optimum model form based on input function characteristics | |
| CN113130014B (en) | Rare earth extraction simulation method and system based on multi-branch neural network | |
| CN105608295A (en) | Multi-objective evolutionary algorithm (MOEA) and radial basis function (RBF) neural network optimization modeling method of coking furnace pressure | |
| CN108416439A (en) | Oil refining process Method For Predicting The Products and system based on variable weighting deep learning | |
| CN106709230A (en) | Method for preprocessing serial genetic lumping kinetics by using random function and post-processing serial genetic lumping kinetics by least squares | |
| CN110378035A (en) | It is a kind of that soft-measuring modeling method is hydrocracked based on deep learning | |
| CN106709087A (en) | Random seed number preprocessing, least square postprocessing and parallel lumped kinetics method | |
| CN112784331B (en) | Soil stress-strain relation determination method based on improved LSTM deep learning method | |
| Gampala et al. | Fast and intelligent antenna design optimization using machine learning | |
| CN106708785A (en) | Random seed number preprocessing, quasi-Newton postprocessing and parallel genetic lumped kinetics method | |
| CN106708784A (en) | Lumping kinetics method for random function preprocessing simplex post-processing parallel inheritance | |
| CN106709089A (en) | Lumping kinetics method for random seed number preprocessing least square post-processing distribution | |
| CN106709084A (en) | Random seed number preprocessing, least square postprocessing and serial lumped kinetics method | |
| CN106709096A (en) | Random function preprocessing, least square postprocessing and parallel genetic lumped kinetics method | |
| CN106709088A (en) | Lumping kinetics method for random function preprocessing simplex post-processing distribution inheritance | |
| CN106709092A (en) | Distribution heredity lumping kinetic method with random function preprocessing and simulated Newton postprocessing | |
| CN106709226A (en) | Distribution heredity lumping kinetic method with random function preprocessing and least square postprocessing | |
| CN106709571A (en) | Random seed number pretreatment and quasi-Newton post-treatment parallel genetic lumped kinetics method | |
| Caraka | Prediction of Euro 50 Using Back Propagation Neural Network (BPNN) and Genetic Algorithm (GA) | |
| CN106709086A (en) | Random seed number preprocessing, quasi-Newton postprocessing and serial genetic lumped kinetics method | |
| CN106709083A (en) | Random seed number preprocessing, simplex postprocessing and serial genetic lumped kinetics method | |
| CN106709085A (en) | Lumping kinetics method of random seed number pretreatment quasi-Newton after-treatment distribution inheritance | |
| CN106709095A (en) | Random function preprocessing, quasi-Newton postprocessing and serial genetic lumped kinetics method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170524 |
|
| RJ01 | Rejection of invention patent application after publication |