CN106202964A - A kind of crude oil nonlinear optimization blending method based on partition initialization - Google Patents

Abstract

The invention discloses a kind of crude oil nonlinear optimization blending method based on partition initialization, during this method solving crude oil on-line blending, Properties of Optimization target is easily trapped into local extremum or causes the problem optimized without solving when using non-linear blending rule.Avoid because being absorbed in local extremum when optimizing or optimizing without solving, cause atmospheric and vacuum distillation unit seriously corroded, the situations such as the product of production is defective.Multiple proportion by weight is calculated particular by subregion, the feasible formula compositional optimization formula set of nonlinear optimization algorithm concoction optimization non-linear to crude oil equation solution is called successively based on these proportion by weight, and from feasible formula set, find the solution making target function value minimum, finally determine optimization of C/C composites.

Description

A kind of crude oil nonlinear optimization blending method based on partition initialization

Technical field

The present invention relates to crude oil on-line blending field, oil refining enterprise, the especially non-linear blending of crude oil be absorbed in local extremum or Optimize the situation without solving.

Background technology

At present, at petroleum and petrochemical industry, crude oil on-line blending optimisation technique has started to application of succeeding in recent years, this skill Art is presently mainly for the purpose of controlling the indexs such as the sulfur content of crude oil, acid number, naphtha yield, reduces the corruption of CDU device Degree of corrosion, makes device even running, ensures the part processing request of secondary process units simultaneously.

These parameters is respectively provided with and linearly adds and characteristic, uses general linear optimization technology can try to achieve blending formula.But It is in crude oil is concocted, there is also some and not there is the oil property linearly added with characteristic, such as crude oil Cetane number, directly affect To diesel cetane-number.When with this type of character for optimization aim, when using crude oil nonlinear optimization algorithm solving-optimizing solution, often Often it is absorbed in local extremum or optimizes the situation without solving, causing the on-line blending that cannot be normally carried out crude oil.When this occurs Can cause: 1) tempered oil character beyond CDU device allowed band, increased the weight of the corrosion of device；2) produce residual oil etc. and need secondary The oil property of processing exceeds standard, and is unsatisfactory for the requirement of secondary operations device；3) product produced is defective, it is impossible to dispatch from the factory pin Sell.As can be seen here, for meeting the normal production of refinery, this problem of solution is needed badly.

Summary of the invention

The present invention is directed to deficiency of the prior art, it is proposed that a kind of crude oil nonlinear optimization based on partition initialization is adjusted Conjunction method.During this method mainly solves crude oil on-line blending, relate to nonlinear optimization objective optimization without solve or The problem being absorbed in local extremum.

The technical scheme is that

A kind of crude oil nonlinear optimization blending method based on partition initialization, it comprises the following steps: concoct participation Each component formula scope of crude oil is divided into n part, obtains the little formula range in each component；In formula range little to each component Mass parts value carry out reconfiguring formation initial mass part proportioning, and then Optimization Solution obtains the optimization of C/C composites of component crude oil, And calculate the feasibility of each optimization of C/C composites, by feasible optimization of C/C composites compositional optimization formula set, the non-linear tune of based on crude Close optimization method and select one group of solution of optimum as final formula.

The present invention comprises the following steps:

S1, acquisition crude oil blending basic parameter: include that participation blending oil component number m, component oil property, each component are joined Side's scope XR_iAnd the constraints of Properties of Optimization scope, wherein, i represents the numbering of blending oil component, (i=1,2 ... m)；

S2, set up crude oil non-linear concoction optimization equation；

S3, each blending oil component formula range is divided into n part, i.e. obtains the little formula range of each blending component xR_{I, j,}J is the numbering of little formula range, (j=1,2 ... n) in representing each component；

S4, little formula range xR according to each blending component_i,j, calculate each blending component at the beginning of in little formula range Beginning mass parts value x_i,j(i=1,2 ... m；J=1,2 ... n), each component produces n initial mass part value；

S5, from n initial mass part value of every kind of blending component, all choose one the most successively, form each component new Initial mass part proportioning Xbe, raw k the combination of common property, k=n^m；

S6, from k of step S5 combination, select t group new initial mass part proportioning Xbe_t, the initial value of t takes 1；

S7, in the case of other blending basic parameter is constant, with Xbe_tFor initial formulation, call nonlinear optimization and calculate Method concoction optimization non-linear to crude oil equation solution optimization of C/C composites XS_t；

S8, judge optimization of C/C composites XS that solves in step S7_tWhether it is feasible solution, if feasible solution, then enters step S9, otherwise returns step S5；

S9, by optimization of C/C composites XS_tRecorded in optimization of C/C composites set XOS, enter step S10；

S10, make t=t+1, as t≤k, enter step S6, otherwise, terminate to solve feasible solution process；

S11, solve after the process of feasible solution terminates, compare several groups of feasible solutions in optimization of C/C composites set XOS, based on crude Non-linear concoction optimization equation, as optimization aim in the way of after blending, oil property content is minimum with target deviation, selects One group of optimum solution is as final formula.

In step S2 of the present invention, crude oil non-linear concoction optimization equation is as follows:

$\begin{array}{c}\begin{array}{cc}\min & \mathrm{\Φ}\left(X\right)=\underset{l=1}{\overset{h}{\Σ}}{\[{\mathrm{\λ}}_l(f{\left(X\right)}_l-{\mathrm{goal}}_l)\]}^2\end{array}\\ s.t.\left\{\begin{array}{c}\ΣX=1\\ X_{\min }\≤X\≤X_{\max }\\ {\mathrm{RangeL}}_l\≤f{\left(X\right)}_l\≤{\mathrm{RangeU}}_1\end{array}\right.\end{array}---\left(1\right)$

In formula (7), Φ (X) is the object function of crude oil non-linear concoction optimization equation；S.t. the non-linear tune of crude oil is represented Closing the constraints of optimization method, h represents that crude oil participates in the character number optimized, f (X)_lThe l kind optimized is participated in for crude oil The blending rule that matter is used, including linear processes blending rule；X is the formula of each blending component oil, X=[x₁,x₂… x_m], X >=0；X_min,iAnd X_max,iIt is respectively minima and maximum, the X of component oil i formula_min,i>=0, X_max,i≥0；λ_lFor former Oil participates in the weight of the l kind character optimized, λ_l≥0；goal_lThe optimization target values of the l kind character optimized is participated in for crude oil, goal_l≥0；rangeL_lThe target lower limit of the l kind character optimized, rangeU is participated in for crude oil_lThe l optimized is participated in for crude oil Plant the target upper limit of character.

The crude oil of the present invention during beending optimization, f (X)_lBy crude oil participate in optimize l kind character used non- Linear blend rule, CalcdCI employing quadratic nonlinearity regression model:

$f\left(x\right)=\underset{i=1}{\overset{m}{\Σ}}{\mathrm{CN}}_i{x}_i+\underset{i=1}{\overset{m}{\Σ}}\underset{t\≠j}{\overset{m}{\underset{t=1,}{\Σ}}}{\mathrm{\α}}_{i,t}{x}_i{x}_t---\left(2\right)$

In formula (1), m is the number of components participating in crude oil blending, and f (x) is crude oil Cetane number size after blending, CN_iIt it is group Divide the Cetane number size of crude oil i, x_iIt is that component crude oil i concocts accounting, x_tIt is that component crude oil t concocts accounting, α_i,t(i ≠ t) is The blending effect coefficient of blending component crude oil i Yu t, its size and the positive and negative nonlinear interaction reflected between different crude oils.

In step S3 of the present invention, each blending component formula range is divided into n part, i.e. obtains little formula range range_i,j(i=1,2 ... m；J=1,2 ... n), it is as follows that it calculates process:

Wherein i=1,2 ... m (3)

δ_i=(X_max,i-X_min,i)/n (4)

Wherein j=1,2 ... n (5)

x_min,i,j=X_min,i+δ_i× (j-1), x_max,i,j=X_min,i+δ_i×j (6)

In formula (2)-(5), XR_iFor the formula range of blending component i, X_max,iFor the formula maximum of blending component i, X_min,i For the formula minima of blending component i, m is the number of components participating in crude oil blending, and n is the number that formula range is bisected into, δ_iFor The interval that blending component i formula range is bisected into, xR_i,_jFor the little formula range of jth of blending component i, x_max,i,jFor blending group Divide the jth little formula range maximum of i, x_min,i,jJth little formula range minima for blending component i.

In step S4 of the present invention, according to xR_i,,_j, calculate initial mass part value x of each component_i,j(i=1,2 ... m；j =1,2 ... n), i.e. calculate xR_i,,_jIntermediate value as blending component i jth initial mass part be worth, its calculate process as follows:

x_i,j=(x_max,i,j+x_min,i,j)/2, wherein, i=1,2 ... m, j=1,2 ... n (7).

Beneficial effects of the present invention:

The crude oil nonlinear optimization blending method based on partition initialization of the present invention, solves crude oil on-line blending process In, Properties of Optimization target is easily trapped into local extremum or causes the problem optimized without solving when using non-linear blending rule.Avoid Because being absorbed in local extremum when optimizing or optimizing without solving, cause atmospheric and vacuum distillation unit seriously corroded, the feelings such as the product of production is defective Condition.

Accompanying drawing explanation

Fig. 1 is the flow chart of the present invention.

Detailed description of the invention

In the actual performance in certain oil refining enterprise and combine a concrete example with the present invention below, provide detailed calculating Journey and concrete operating process.This corporate boss to process in outstanding promise, Da Er mixing and Basra, Sha Chong, Sha Zhong, Oman etc. are former Oil, in order to reduce production cost, generally by sweet crude and sour crude hybrid process.For reduction Atmospheric vacuum dress of keeping the safety in production The extent of corrosion put, blending crude oil acid, sulfur content value etc. to control in the range of atmospheric and vacuum distillation unit design parameter.Simultaneously for improving The diesel cetane-number that secondary operations device produces, also using the Cetane number of crude oil as Properties of Optimization item.The present embodiment with Implement under premised on technical solution of the present invention, but protection scope of the present invention is not limited to following embodiment.

As it is shown in figure 1, a kind of crude oil nonlinear optimization blending method workflow based on partition initialization mainly includes Following steps:

I. crude oil blending basic parameter is obtained.

In conjunction with a concrete example, it is assumed that it is outstanding promise, Da Er mixed crude and bus that the blending of this batch participates in the crude oil of blending Draw Crude Oil, participate in Properties of Optimization and include that sulfur content, acid number, Cetane number and residual oil sulfur content, each oil are planted character and i.e. optimized about Bundle condition is as shown in table 1:

Table 1 participates in blending oil property and optimizes constraints

The blending that wherein Cetane number uses is returned then for quadratic nonlinearity regression model, is specifically shown in equation below:

$f\left(x\right)=\underset{i=1}{\overset{3}{\Σ}}{\mathrm{CN}}_i{x}_i+\underset{i=1}{\overset{3}{\Σ}}\underset{t\≠j}{\overset{3}{\underset{t=1,}{\Σ}}}{\mathrm{\α}}_{i,t}{x}_i{x}_t$

In formula, f (x) is crude oil Cetane number size after blending, CN_iIt is component crude oil outstanding person's promise, Da Er mixing and Basra In Cetane number size, x_i、x_tIt is the blending accounting participating in blending crude oil, α_i,t(i ≠ t) is the blending between blending component crude oil Effect coefficient, its size and the positive and negative nonlinear interaction reflected between different crude oils, concrete numerical value is as shown in table 2:

Table 2 participates in the blending effect coefficient between blending crude oil

	Jie Nuo	Reach mixed	In Basra
				Jie Nuo	0.00	-4.182	-17.6
Reach mixed	-4.182	0.00	10
				In Basra	-17.6	10	0.00

II. nonlinear optimization equation concocted by the crude oil set up such as formula (7) according to I..

III. each blending component formula range is divided into 3 parts, i.e. obtains the little formula range of each blending component, such as table Shown in 3:

Table 3 participates in the little formula range of blending crude oil

IV. according to table 3, the intermediate value of each little formula range is calculated as each component crude oil based on formula (5)～(6) Initial mass part is worth, and result of calculation is as shown in table 4:

Table 4 participates in initial mass part value of blending crude oil

Sequence number	Jie Nuo	Reach mixed	In Basra
				1	0.4	0.3	0.2
2	0.6	0.5	0.4
				3	0.8	0.7	0.6

V. from 3 initial mass part values of each component crude oil, all choose one, form new initial mass part proportioning, altogether Produce 27 groups, as shown in table 5:

New initial mass part proportioning that table 5 produces

VI. the 27 groups of initial mass part proportionings produced based on table 5, after normalization, directly use in MATLAB Fgoalattain algorithm is optimized algorithm, obtains feasible solution.

Table 6 initial formulation and corresponding optimum results

It is optimized by 27 groups of initial mass part proportionings and solves, in addition to 7 groups without feasible solution, remain 20 groups of initial mass parts Proportioning creates 3 groups of feasible solutions altogether, as shown in Table 6.

Feasible solution is as shown in table 7 with blending oil property.

The feasible solution of table 7 generation and blending oil property

Sequence number	Optimize feasible solution	Sulfur content	Acid number	Cetane number	Residual oil sulfur content
						1	0.45,0.45,0.1	1.62	0.30	26.93	3.22
2	0.5,0.4,0.1	1.74	0.29	26.64	3.50
						3	0.6,0.2,0.2	2.21	0.26	25.49	4.69

VII.

Object function Φ (X) value size in based on crude non-linear concoction optimization equation (7), selects final optimal solution.

3 groups of feasible solutions are substituted into crude oil non-linear concoction optimization equation (7) object function Φ (X), finds and make target letter The solution that numerical value is minimum, finally determines that optimization of C/C composites is [0.45,0.45,0.1].

Part that the present invention does not relate to is the most same as the prior art maybe can use prior art to be realized.

Claims (6)

1. a crude oil nonlinear optimization blending method based on partition initialization, is characterized in that it comprises the following steps: to ginseng It is divided into n part with each component formula scope of blending crude oil, obtains the little formula range in each component；Formula little to each component In the range of mass parts value carry out reconfiguring formation initial mass part proportioning, and then Optimization Solution obtains the optimization of component crude oil Formula, and calculate the feasibility of each optimization of C/C composites, by feasible optimization of C/C composites compositional optimization formula set, based on crude non-thread Property concoction optimization equation select optimum one group of solution as final formula.

Crude oil nonlinear optimization blending method based on partition initialization the most according to claim 1, is characterized in that it wraps Include following steps:

S1, acquisition crude oil blending basic parameter: include participating in blending oil component number m, component oil property, each component formula model Enclose XR_iAnd the constraints of Properties of Optimization scope, wherein, i represents the numbering of blending oil component, (i=1,2 ... m)；

S2, set up crude oil non-linear concoction optimization equation；

S3, each blending oil component formula range is divided into n part, i.e. obtains the little formula range xR of each blending component_i,j, j The numbering of little formula range, (j=1,2 ... n) in representing each component；

S4, little formula range xR according to each blending component_i,j, calculate each blending component first prothyl in little formula range Amount part value x_i,j(i=1,2 ... m；J=1,2 ... n), each component produces n initial mass part value；

S5, from n initial mass part value of every kind of blending component, all choose one the most successively, form each component the most initial Proportion by weight Xbe, raw k the combination of common property, k=n^m；

S6, from k of step S5 combination, select t group new initial mass part proportioning Xbe_t, the initial value of t takes 1；

S7, in the case of other blending basic parameter is constant, with Xbe_tFor initial formulation, call nonlinear optimization algorithm pair Crude oil non-linear concoction optimization equation solution optimization of C/C composites XS_t；

S8, judge optimization of C/C composites XS that solves in step S7_tWhether it is feasible solution, if feasible solution, then enters step S9, no Then return step S5；

S9, by optimization of C/C composites XS_tRecorded in optimization of C/C composites set XOS, enter step S10；

S10, make t=t+1, as t≤k, enter step S6, otherwise, terminate to solve feasible solution process；

S11, solve after the process of feasible solution terminates, compare several groups of feasible solutions in optimization of C/C composites set XOS, based on crude non-thread Property concoction optimization equation, as optimization aim in the way of oil property content and target deviation are minimum after blending, select optimum One group of solution as final formula.

Crude oil nonlinear optimization blending method based on partition initialization the most according to claim 2, is characterized in that step In S2, crude oil non-linear concoction optimization equation is as follows:

In formula (7), Φ (X) is the object function of crude oil non-linear concoction optimization equation；S.t. represent that the non-linear blending of crude oil is excellent Changing the constraints of equation, h represents that crude oil participates in the character number optimized, f (X)_lThe l kind character institute optimized is participated in for crude oil The blending rule used, including linear processes blending rule；X is the formula of each blending component oil, X=[x₁,x₂…x_m], X ≥0；X_min,iAnd X_max,iIt is respectively minima and maximum, the X of component oil i formula_min,i>=0, X_max,i≥0；λ_lParticipate in for crude oil The weight of the l kind character optimized, λ_l≥0；goal_lThe optimization target values of the l kind character optimized, goal is participated in for crude oil_l≥ 0；rangeL_lThe target lower limit of the l kind character optimized, rangeU is participated in for crude oil_lThe l kind character optimized is participated in for crude oil Target upper limit.

Crude oil nonlinear optimization blending method based on partition initialization the most according to claim 3, is characterized in that: crude oil During beending optimization, f (X)_lParticipate in, by crude oil, the non-linear blending rule that the l kind character of optimization is used, calculate ten Six alkane values employing quadratic nonlinearity regression models:

In formula (1), m is the number of components participating in crude oil blending, and f (x) is crude oil Cetane number size after blending, CN_iIt is component crude oil The Cetane number size of i, x_iIt is that component crude oil i concocts accounting, x_tIt is that component crude oil t concocts accounting, α_i,t(i ≠ t) is blending group Divide the blending effect coefficient of crude oil i Yu t, its size and the positive and negative nonlinear interaction reflected between different crude oils.

Crude oil nonlinear optimization blending method based on partition initialization the most according to claim 2, is characterized in that step In S3, each blending component formula range is divided into n part, i.e. obtains little formula range range_i,j(i=1,2 ... m；J=1, 2 ... n), it is as follows that it calculates process:

Wherein i=1,2 ... m (3)

δ_i=(X_max,i-X_min,i)/n (4)

Wherein j=1,2 ... n (5)

x_min,i,j=X_min,i+δ_i× (j-1), x_max,i,j=X_min,i+δ_i×j (6)

In formula (2)-(5), XR_iFor the formula range of blending component i, X_max,iFor the formula maximum of blending component i, X_min,iFor The formula minima of blending component i, m is the number of components participating in crude oil blending, and n is the number that formula range is bisected into, δ_iFor adjusting The interval that charge-coupled point of i formula range is bisected into, xR_i,_jFor the little formula range of jth of blending component i, x_max,i,jFor blending component The jth little formula range maximum of i, x_min,i,jJth little formula range minima for blending component i.

Crude oil nonlinear optimization blending method based on partition initialization the most according to claim 2, is characterized in that step In S4, according to xR_i,,j, calculate initial mass part value x of each component_i,j(i=1,2 ... m；J=1,2 ... n), i.e. calculate xR_i,,j Intermediate value as blending component i jth initial mass part be worth, its calculate process as follows:

x_i,j=(x_max,i,j+x_min,i,j)/2, wherein, i=1,2 ... m, j=1,2 ... n (7).