CN103870670A - Oil pipe corrosion degree forecasting method and device - Google Patents

Abstract

The invention discloses an oil pipe corrosion degree forecasting method and device, and belongs to the technical field of oilfield downholes. The method includes the steps: S1, selecting at least two groups of downhole working condition data and actual oil pipe corrosion rate data corresponding to the downhole working condition data, normalizing the selected data, and dividing the normalized data into a training data sample and an inspection data sample; S2, modeling the training data sample by combining a fuzzy linear regression method with a least square method to obtain an oil pipe corrosion forecasting model taking the downhole working condition data as dependent variables and taking forecasted oil pipe corrosion rate data as variables; S3, verifying correctness of the oil pipe corrosion forecasting model by the aid of the inspection data sample, executing S4 if the model passes inspection, otherwise, transposing S1, and selecting new data to repeatedly execute processes; S4, forecasting the corrosion degree of in-service oil pipes in a designated environment by the aid of the obtained oil pipe corrosion forecasting model. By the aid of the oil pipe corrosion degree forecasting method and device, forecasting accuracy of the corrosion degree of downhole oil pipes can be improved.

Description

A kind of tube corrosion degree Forecasting Methodology and device

Technical field

The present invention relates to oil field well technical field, particularly a kind of tube corrosion degree Forecasting Methodology and device.

Background technology

Along with the exploitation of oil gas field, the media such as sulfuretted hydrogen, carbon dioxide, chlorion, water and the microorganism that the High Temperature High Pressure polyphasic flow environment under Oil/gas Well contains can occur as attendants, make down-hole oil tube be subject to heavy corrosion, so not only affect the normal production of Oil/gas Well, and brought many difficulties to examination workover treatment.

The factor that affects Oil Tube Steel corrosion can be divided into environmental factor, material factor and mechanics factor etc.Wherein, environmental factor comprises temperature, partial pressure, corrosion products film, pH value, flow velocity, flow pattern, solution degree of supersaturation, bacterium etc.These factors affect the corrosion of oil pipe to some extent above, therefore predict that tube corrosion degree mainly sets about from this several respects factor.In Practical Project, the corrosion influence factors of down-hole oil tube is very complicated, for example: the variation of well depth, the difference of operating condition make concentration of medium and the working conditions etc. of down-hole diverse location, different times different from the process of construction cycle are complicated and changeable, tube corrosion degree varies.In addition, many gaps that exist such as tubing sub position and bimetallic corrosion problem, the problems such as interior oil film corrosion Inhibition Mechanism and down-hole string compound stress of managing are not yet studied at present thorough in this area.Due to still simulating down-hole actual behavior completely of existing corrosion test method and apparatus, therefore only pass through field working conditions Data Detection tube corrosion situation, or only all can not well carry out early warning to the corrosion failure of in-service oil pipe by existing corrosion prediction theoretical model and simulated condition corrosion test.

In prior art, by means of classical theory formula and laboratory simulation data, the corrosion condition of oil pipe is judged, conventional prediction extent of corrosion criterion has: (1) passes through CO ₂dividing potential drop judges: when

interval scale oil pipe heavy corrosion; And work as

time show the general corrosion of oil pipe; When

interval scale oil pipe does not corrode.(2) basis ratio in judgement: when time be CO ₂corrosion; When

for H ₂s corrosion.(3) according to Ca ²⁺/ HCO ₃ratio in judgement: when

and while having local water to exist, Ca ²⁺/ HCO ₃<0.5 shows that tube corrosion speed is lower, Ca ²⁺/ HCO ₃>1000 shows that tube corrosion speed is medium, Ca ²⁺/ HCO ₃=0.5～1000 represent oil pipe generation heavy corrosion.

Realizing in process of the present invention, inventor finds that prior art at least exists following problem:

In prior art, according to some classical theory formula or laboratory simulation data, tube corrosion is judged to early warning, often only judge the extent of corrosion of oil pipe for the single factors analysis that affects tube corrosion.And in fact, due to the complicacy of oil pipe subsurface environment, the method for prior art can not judge accurately to the corrosion condition of oil pipe.

Summary of the invention

In order to solve the problems of the prior art, the invention provides a kind of tube corrosion degree Forecasting Methodology and device, to improve the accuracy of down-hole oil tube extent of corrosion prediction.Technical scheme provided by the invention is as follows:

On the one hand, a kind of tube corrosion degree Forecasting Methodology is provided, pre-stored underground working data and the corresponding actual corrosion rate data of oil pipe in varying environments of organizing in database more, described underground working data comprise the data of multiple influence factors, described method comprises:

S1, chooses at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing is carried out to normalization, and the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with Fuzzy Linear Regression method and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, carries out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

S4, adopts the tube corrosion forecast model obtaining, and the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

Preferably, the group number of described training data sample is not less than the number of described dependent variable.

Preferably, described the data of choosing are carried out to normalization, are specially:

Determine the span z ∈ [z of the data of choosing _min, z _max], and the range scale of definite normalization

wherein z is the data of choosing,

for the data after normalization;

According to

with the described data of choosing, draw the data after normalization.

Preferably, described underground working data comprise CO ₂concentration, Cl ^-concentration, H ₂s concentration, temperature and pressure data.

Preferably, described combination Fuzzy Linear Regression method and least square method, carry out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe, is specially:

To typical linear regression model y _i=a ₁x _1i+ a ₂x _2i+ ... + a _mx _miuse least square method, and based on training data sample, determine a _j, wherein, x _jithe data of j influence factor in substitution i group training data sample, y _ithe actual corrosion rate data of oil pipe in substitution i group training data sample, j=1,2 ..., m;

According to with described training data sample and a that determines _j, determine c _j, wherein, H is default confidence value, and 0≤H≤1;

According to the subordinate function of Triangular Fuzzy Number

with a determining _j, c _j, draw Triangular Fuzzy Number A _j(a _j, c _j) value, as Estimates of Fuzzy Linear Regression Model y=A ₁x ₁+ A ₂x ₂+ ... + A _mx _min regression coefficient A _j, using this Estimates of Fuzzy Linear Regression Model as described tube corrosion forecast model.

Preferably, the described check data sample of described use is verified the correctness of described tube corrosion forecast model, is specially:

By tube corrosion forecast model described in the underground working data substitution in described check data sample, obtain oil pipe prediction corrosion rate data, ratio and predetermined threshold value comparison by the difference of the actual corrosion rate data of oil pipe corresponding these underground working data in described check data sample and described oil pipe prediction corrosion rate data with the actual corrosion rate data of this oil pipe, verify the correctness of described tube corrosion forecast model, if this ratio is no more than predetermined threshold value, be verified, otherwise, authentication failed.

On the other hand, a kind of tube corrosion degree Forecasting Methodology is provided, pre-stored underground working data and the corresponding actual corrosion rate data of oil pipe in varying environments of organizing in database more, described underground working data comprise the data of multiple influence factors, described method comprises:

S1, chooses at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing is carried out to normalization, and the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with the Fuzzy Nonlinear Return Law and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, carries out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

S4, adopts the tube corrosion forecast model obtaining, and the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

On the other hand, provide a kind of tube corrosion degree prediction unit, described device comprises:

Memory module, for pre-stored many group varying environment underground working data and the corresponding actual corrosion rate data of oil pipe, described underground working data comprise the data of multiple influence factors;

MBM, for carrying out: S1, choose at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing are carried out to normalization, the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with Fuzzy Linear Regression method and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, notifies prediction module to carry out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

Prediction module, for carrying out: S4, adopt the tube corrosion forecast model obtaining, the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

Preferably, the group number of described training data sample is not less than the number of described dependent variable.

Preferably, described MBM, specifically for:

Determine the span z ∈ [z of the data of choosing _min, z _max], and the range scale of definite normalization

wherein z is the data of choosing,

for the data after normalization;

According to

with the described data of choosing, draw the data after normalization.

Preferably, described underground working data comprise CO ₂concentration, Cl ^-concentration, H ₂s concentration, temperature and pressure data.

Preferably, described MBM, specifically for:

To typical linear regression model y _i=a ₁x _1i+ a ₂x _2i+ ... + a _mx _miuse least square method, and based on training data sample, determine a _j, wherein, x _jithe data of j influence factor in substitution i group training data sample, y _ithe actual corrosion rate data of oil pipe in substitution i group training data sample, j=1,2 ..., m;

According to

with described training data sample and a that determines _j, determine c _j, wherein, H is default confidence value, and 0≤H≤1;

According to the subordinate function of Triangular Fuzzy Number with a determining _j, c _j, draw Triangular Fuzzy Number A _j(a _j, c _j) value, as Estimates of Fuzzy Linear Regression Model y=A ₁x ₁+ A ₂x ₂+ ... + A _mx _min regression coefficient A _j, using this Estimates of Fuzzy Linear Regression Model as described tube corrosion forecast model.

Preferably, described MBM, specifically for:

By tube corrosion forecast model described in the underground working data substitution in described check data sample, obtain oil pipe prediction corrosion rate data, ratio and predetermined threshold value comparison by the difference of the actual corrosion rate data of oil pipe corresponding these underground working data in described check data sample and described oil pipe prediction corrosion rate data with the actual corrosion rate data of this oil pipe, verify the correctness of described tube corrosion forecast model, if this ratio is no more than predetermined threshold value, be verified, otherwise, authentication failed.

On the other hand, provide a kind of tube corrosion degree prediction unit, described device comprises:

Memory module, for pre-stored many group varying environment underground working data and the corresponding actual corrosion rate data of oil pipe, described underground working data comprise the data of multiple influence factors;

MBM, for carrying out: S1, choose at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing are carried out to normalization, the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with the Fuzzy Nonlinear Return Law and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, notifies prediction module to carry out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

Prediction module, for carrying out: S4, adopt the tube corrosion forecast model obtaining, the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

The beneficial effect that technical scheme provided by the invention is brought is:

The underground working data that detect based on reality and the corresponding actual corrosion rate data of oil pipe, by predicting corrosion rate data tube corrosion forecast model as variable by underground working data as dependent variable, oil pipe in conjunction with Fuzzy Linear Regression method and least square method foundation, and according to this model, oil pipe is carried out to extent of corrosion prediction, thereby can improve the accuracy of oil field well tube corrosion degree prediction.

Brief description of the drawings

In order to be illustrated more clearly in the technical scheme in the embodiment of the present invention, below the accompanying drawing of required use during embodiment is described is briefly described, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skill in the art, do not paying under the prerequisite of creative work, can also obtain according to these accompanying drawings other accompanying drawing.

Fig. 1 is the tube corrosion degree Forecasting Methodology process flow diagram that the embodiment of the present invention provides;

Fig. 2 is the tube corrosion degree Forecasting Methodology process flow diagram that the embodiment of the present invention provides;

Fig. 3 is the tube corrosion degree Forecasting Methodology process flow diagram that the embodiment of the present invention provides;

Fig. 4 is the tube corrosion degree prediction unit structural representation that the embodiment of the present invention provides;

Fig. 5 is the tube corrosion degree prediction unit structural representation that the embodiment of the present invention provides.

Embodiment

For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with accompanying drawing, embodiment of the present invention is described further in detail.

Embodiment mono-

The embodiment of the present invention provides a kind of tube corrosion degree Forecasting Methodology, the method is according to underground working data and the tube corrosion situation data collecting in engineering, arrange, first build the database of variable (actual corrosion rate) and dependent variable (corrosion influence factors), then use a kind of new algorithm that solves Fuzzy Linear Regression to set up tube corrosion forecast model, the final tube corrosion forecast model of setting up that utilizes is predicted the tube corrosion result under particular cases.

Figure 1 shows that a kind of tube corrosion degree Forecasting Methodology process flow diagram that the embodiment of the present invention provides, the method comprises the following steps:

S0: obtain underground working data and the corresponding actual corrosion rate data of oil pipe in many group varying environments, deposit in database.Wherein, underground working data comprise the data of multiple influence factors, for example, can comprise: temperature, pressure, corrosive concentration are (as CO ₂, Cl ^-, H ₂s concentration) etc.The actual corrosion rate define method of oil pipe is: the corrosion failure situation to field notes is carried out classification, and records corrosion pit depth, and the corrosion class of drafting has 4 grades: 1. without any corrosion; 2. slightly corrosion (box cupling, tube wall pit shape etc.); 3. heavy corrosion (box cupling, tube wall hole shape, sheet etc.); 4. corrosion failure (box cupling or pipe body corrosion perforation etc.), then, in conjunction with oil pipe inefficacy grade, according to corrosion pit depth and oil pipe tenure of use, the actual corrosion rate of definition oil pipe.Preferably, in this step, obtain the CO under many group different situations ₂, Cl ^-, H ₂s, temperature and pressure data and corresponding tube corrosion level data deposit in database.

S1: in database, choose at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing are carried out to normalization, the data after normalization are divided into training data sample and check data sample.Preferably, in this step, in database, choose CO ₂concentration, Cl ^-concentration, H ₂s concentration, temperature and pressure data (are that underground working data comprise CO ₂concentration, Cl ^-concentration, H ₂s concentration, temperature and pressure data) and the corresponding actual corrosion rate data of oil pipe, and these data are carried out to normalization processing.

Preferably, in this step, the group number of the data of choosing is greater than the number of the dependent variable described in step S2, and wherein, check data sample can be one group, and the group number of training data sample is not less than the number of described dependent variable.

S2: in conjunction with Fuzzy Linear Regression method and least square method, training data sample is carried out to modeling, obtain by underground working data the tube corrosion forecast model as variable as dependent variable, the actual corrosion rate data of oil pipe.

S3: service test data sample is verified the correctness of this tube corrosion forecast model, if upchecked, carries out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process.

S4: adopt the tube corrosion forecast model obtaining, the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

In above-mentioned steps S1, normalization is processed and is adopted linear-scale algorithm, can realize the yardstick data that raw data (measured data of storing in database) changed to desired range scale.Data after application dimensional variation are carried out modeling, can eliminate the model that uses the data of different scale scope to cause inaccurate.

Concrete, the process that the data of choosing are carried out can be as follows:

First, determine the span of the data of choosing, and the range scale of definite normalization.

If z represents the data of choosing, there is no the data sample of yardstick, its span is z ∈ [z _min, z _max], the data after the normalization obtaining after linear-scale algorithm process are used

represent, its span is $\tilde{z}\&Element;[{\tilde{z}}_{\min },{\tilde{z}}_{\max }].$ Then,, according to following formula and the data of choosing, draw the data after normalization.Z and

conversion formula between the two can be:

$\tilde{z}={\tilde{z}}_{\min }+\frac{z-z_{\min }}{z_{\max }-z_{\min }}({\tilde{z}}_{\max }-{\tilde{z}}_{\min })---\left(13\right)$

$z=z_{\min }+\frac{\tilde{z}-{\tilde{z}}_{\min }}{{\tilde{z}}_{\max }-{\tilde{z}}_{\min }}(z_{\max }-z_{\min })---\left(14\right)$

Above formula (13) has been realized the yardstick data that raw data changed to desired range scale, and formula (14) has realized yardstick data are changed to raw data again.(13) formula of utilization can selected underground working data and the corresponding actual corrosion rate data transformation of oil pipe thereof arrive new range scale (as, [1,1]) in.

Below in conjunction with Fuzzy Linear Regression method, illustrate shown in Fig. 1 the implementation method of S2 in flow process.

First introduce Fuzzy Linear Regression method, and classical linear regression analysis is similar, in Fuzzy Linear Regression, establishes variable y and its relevant dependent variable x ₁, x ₂..., x _mjust like lower linear relation:

y=A ₁x ₁+A ₂x ₂+…+A _mx _m （1）

Regretional analysis is to utilize known n group observation data y _i, x _1i, x _2i..., x _mi(wherein, i=1,2 ..., n), remove to estimate regression coefficient A _j.But in fuzzy linear regression analysis, think that model has ambiguity, i.e. regression coefficient A _jfuzzy number, so the match value of model

with observed reading y _ibetween deviation caused by this ambiguity.The general representation of typical linear regression model is:

y _i=a ₁x _1i+a ₂x _2i+…+a _mx _mi （8）

Conventionally get A _jfor Triangular Fuzzy Number A (a, c), its subordinate function is:

The subordinate function that is drawn y in formula (1) by above formula (2) is:

For making fitting function (1) to known n group observation data y _i, x _1i, x _2i..., x _mi(wherein, i=1,2 ..., n) matching is best, in linear regression analysis, must meet following 2 criterions simultaneously:

(a) must make the fuzzy amplitude sum minimum (being precision maximum) of each regression coefficient, that is:

$\min s=\underset{j}{\mathrm{\&Sigma;}}{c}_j\left|x_{\mathrm{ji}}\right|---\left(4\right)$

(b), according to certain confidence level H, all observation data y must can be covered _i, that is:

h _y(y _i)≥H,0≤H≤1 （5）

This criterion has ensured not have the y of degree of membership <H _i.

Can obtain according to formula (3) and formula (5):

$1-\frac{y_i-\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}}{\underset{j}{\mathrm{\&Sigma;}}{c}_j\left|x_{\mathrm{ji}}\right|}\&GreaterEqual;H,\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}-(1-H)\underset{j}{\mathrm{\&Sigma;}}{c}_j\left|x_{\mathrm{ji}}\right|\&le;y_i\&le;---\left(6\right)$

$\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}+(1-H)\underset{j}{\mathrm{\&Sigma;}}{c}_j\left|x_{\mathrm{ji}}\right|$

In Fuzzy Linear Regression method, convolution (4) and formula (6) can be converted into the problem that solves Fuzzy Linear Regression the linear programming shown in solving equation group (7):

$\left\{\begin{array}{c}\min s=\underset{j}{\mathrm{\&Sigma;}}{c}_j\left|x_{\mathrm{ji}}\right|\\ s.t\left\{\begin{array}{c}\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}-(1-H)\underset{j}{\mathrm{\&Sigma;}}{c}_j\left|x_{\mathrm{ji}}\right|\&le;y_i---\left(7\right)\\ \underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}+(1-H)\underset{j}{\mathrm{\&Sigma;}}{c}_j\left|x_{\mathrm{ji}}\right|\&GreaterEqual;y_i\end{array}\right.\end{array}\right.$

Solving equation group can obtain A after (7) _j(a _j, c _j) value, thereby obtain y _iwith dependent variable x ₁, x ₂..., x _mbetween function expression.

The embodiment of the present invention adopts the idea about modeling of above-mentioned Fuzzy Linear Regression, but do not adopt above-mentioned system of equations (7) to determine fuzzy coefficient, but calculate fuzzy coefficient in conjunction with Fuzzy Linear Regression method and least square method, the specific implementation method flow process that is illustrated in figure 2 step S2 in Fig. 1, can comprise the steps:

S21: the general type (as below (formula 15)) to typical linear regression model uses least square method, and based on training data sample, determines a _j, wherein, x _jithe data of j influence factor in substitution i group training data sample, y _ithe actual corrosion rate data of oil pipe in substitution i group training data sample, j=1,2 ..., m.

y _i＝a ₁x _1i+a ₂x _2i+…+a _mx _mi （15）

If the match value of above-mentioned model is utilize least square method to solve:

$\min Q=\min \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\left(y_i\stackrel{\&OverBar;}{y_i}\right)=\min \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\left[y_i\left(\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}\right)\right],i=\mathrm{1,2}..n;j=\mathrm{1,2}..m---\left(16\right)$

Known according to mathematical knowledge, when time, the Q in formula (16) obtains extreme value, that is, need to meet:

$\left\{\begin{array}{c}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}[y_i-\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}]=0\\ \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}[y_i-\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}]x_{\mathrm{<figure-callout\; id="1"\; label="i"\; filenames="HDA00002601066200021.png"\; state="\{\{state\}\}">i</figure-callout>}1}=0\\ \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}[y_i-\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}]x_{\mathrm{ij}}=0\\ ...\\ \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}[y_i-\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}]x_{\mathrm{iy}}=0\end{array}\right.---\left(17\right)$

Can be obtained the value a of Blur center of the Estimates of Fuzzy Linear Regression Model shown in formula (15) by system of equations (17) ₁, a ₂, a ₃..., a _m, can obtain the center die analog model y that typical linear returns _i=a ₁x _1i+ a ₂x _2i+ ... + a _mx _mi.

S22: according to following formula (10) and training data sample and a that determines _j, determine c _j.

Concrete, solve fuzzy amplitude c according to above-mentioned formula (6) _jprocess in, can obtain with following formula (10), according to formula (10) and training data sample and a that determines _jcan calculate c _j.Finally, can obtain the Estimates of Fuzzy Linear Regression Model shown in formula (11).

$\underset{j}{\mathrm{\&Sigma;}}{c}_j{x}_{\mathrm{ji}}=\frac{y_i-\underset{j}{\mathrm{\&Sigma;}}{a}_j{x}_{\mathrm{ji}}}{1-H}---\left(10\right)$

S23: according to the subordinate function of Triangular Fuzzy Number (shown in formula (2)) and a that determines _j, c _j, draw Triangular Fuzzy Number A _j(a _j, c _j) value, as Estimates of Fuzzy Linear Regression Model y=A ₁x ₁+ A ₂x ₂+ ... + A _mx _min regression coefficient A _j, using this Estimates of Fuzzy Linear Regression Model as described tube corrosion forecast model.This Estimates of Fuzzy Linear Regression Model can also be expressed as formula (11).

$y=\underset{j}{\mathrm{\&Sigma;}}(a_j,c_j)x_j---\left(11\right)$

Wherein, A _j(a _j, c _j) be Triangular Fuzzy Number, its value can change according to not coexisting within the scope of of z value.

CO ₂concentration, Cl ^-concentration, H ₂s concentration, temperature and pressure affect larger for the corrosion rate of oil pipe, comprise CO for underground working data ₂concentration, Cl ^-concentration, H ₂the situation of S concentration, temperature and pressure data, CO ₂concentration, Cl ^-concentration, H ₂the model of S concentration, temperature and pressure data and oil pipe prediction corrosion rate data can be as follows:

ΔV _corr=f(ΔCO ₂,ΔCl ^-,ΔH ₂S,ΔF,ΔT)（9）

Wherein, f () is Estimates of Fuzzy Linear Regression Model, by formula (11) substitution formula (9), can obtain tube corrosion forecast model:

V _corr=(a ₁,c ₁)*CO ₂+(a ₂,c ₂)*Cl ^-+(a ₃,c ₃)*ΔH ₂S+(a ₄,c ₄)*ΔF+(a ₅,c ₅)*ΔT （12）

Here, due to A _j(a _j, c _j) be Triangular Fuzzy Number, but not determine numerical value, so, in actual applications, z value can be set as the case may be, to determine each A _jnumerical value, for example, can be to all A _jthe value that z is set is a _j+ c _j.

In step S3, the correctness of service test data sample checking tube corrosion forecast model, can be specific as follows:

By this tube corrosion forecast model of underground working data substitution in check data sample, obtain oil pipe prediction corrosion rate data, ratio and predetermined threshold value comparison by the difference of actual oil pipe corresponding these underground working data in check data sample corrosion rate data and described oil pipe prediction corrosion rate data with the actual corrosion rate data of this oil pipe, verify the correctness of described tube corrosion forecast model, if this ratio is no more than predetermined threshold value, be verified, otherwise, authentication failed.Specifically formula as follows:

$\mathrm{\&Omega;}=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}|y_i-\underset{j}{\mathrm{\&Sigma;}}(a_j,c_j)x_{\mathrm{ji}}|}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{y}_i}---\left(19\right)$

In formula, the value of Ω is as long as in 30%(predetermined threshold value) in, think and be verified.If authentication failed, illustrates that the precision of model does not meet request for utilization, can re-execute step S1, S2, carry out data-optimized choosing, or collect after abundant data sample, then carry out modeling.

In the embodiment of the present invention, the underground working data that detect based on reality and the corresponding actual corrosion rate data of oil pipe, by predicting corrosion rate data tube corrosion forecast model as variable by underground working data as dependent variable, oil pipe in conjunction with Fuzzy Linear Regression method and least square method foundation, and according to this model, oil pipe is carried out to extent of corrosion prediction, thereby can improve the accuracy of oil field well tube corrosion degree prediction.

Embodiment bis-

The embodiment of the present invention provides a kind of tube corrosion degree Forecasting Methodology, pre-stored underground working data and the corresponding actual corrosion rate data of oil pipe in varying environments of organizing in database more, described underground working data comprise the data of multiple influence factors, as shown in Figure 3, the method can comprise the steps:

S1, chooses at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing is carried out to normalization, and the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with the Fuzzy Nonlinear Return Law and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, carries out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

S4, adopts the tube corrosion forecast model obtaining, and the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

In the embodiment of the present invention, the underground working data that detect based on reality and the corresponding actual corrosion rate data of oil pipe, by predicting corrosion rate data tube corrosion forecast model as variable by underground working data as dependent variable, oil pipe in conjunction with the Fuzzy Nonlinear Return Law and least square method foundation, and according to this model, oil pipe is carried out to extent of corrosion prediction, thereby can improve the accuracy of oil field well tube corrosion degree prediction.

Embodiment tri-

Based on identical technical conceive, the embodiment of the present invention provides a kind of tube corrosion degree prediction unit, and as shown in Figure 4, described device comprises:

Memory module 410, for pre-stored many group varying environment underground working data and the corresponding actual corrosion rate data of oil pipe, described underground working data comprise the data of multiple influence factors;

MBM 420, for carrying out: S1, choose at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing are carried out to normalization, the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with Fuzzy Linear Regression method and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, notifies prediction module to carry out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

Prediction module 430, for carrying out: S4, adopt the tube corrosion forecast model obtaining, the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

Preferably, the group number of described training data sample is not less than the number of described dependent variable.

Preferably, described MBM 420, specifically for:

Determine the span z ∈ [z of the data of choosing _min, z _max], and the range scale of definite normalization

wherein z is the data of choosing,

for the data after normalization;

According to

with the described data of choosing, draw the data after normalization.

Preferably, described underground working data comprise CO ₂concentration, Cl ^-concentration, H ₂s concentration, temperature and pressure data.

Preferably, described MBM 420, specifically for:

To typical linear regression model y _i=a ₁x _1i+ a ₂x _2i+ ... + a _mx _miuse least square method, and based on training data sample, determine a _j, wherein, x _jithe data of j influence factor in substitution i group training data sample, y _ithe actual corrosion rate data of oil pipe in substitution i group training data sample, j=1,2 ..., m;

According to

with described training data sample and a that determines _j, determine c _j, wherein, H is default confidence value, and 0≤H≤1;

According to the subordinate function of Triangular Fuzzy Number with a determining _j, c _j, draw Triangular Fuzzy Number A _j(a _j, c _j) value, as Estimates of Fuzzy Linear Regression Model y=A ₁x ₁+ A ₂x ₂+ ... + A _mx _min regression coefficient A _j, using this Estimates of Fuzzy Linear Regression Model as described tube corrosion forecast model.

Preferably, described MBM 420, specifically for:

By tube corrosion forecast model described in the underground working data substitution in described check data sample, obtain oil pipe prediction corrosion rate data, ratio and predetermined threshold value comparison by the difference of the actual corrosion rate data of oil pipe corresponding these underground working data in described check data sample and described oil pipe prediction corrosion rate data with the actual corrosion rate data of this oil pipe, verify the correctness of described tube corrosion forecast model, if this ratio is no more than predetermined threshold value, be verified, otherwise, authentication failed.

In the embodiment of the present invention, the underground working data that detect based on reality and the corresponding actual corrosion rate data of oil pipe, by predicting corrosion rate data tube corrosion forecast model as variable by underground working data as dependent variable, oil pipe in conjunction with Fuzzy Linear Regression method and least square method foundation, and according to this model, oil pipe is carried out to extent of corrosion prediction, thereby can improve the accuracy of oil field well tube corrosion degree prediction.

Embodiment tetra-

Based on identical technical conceive, the embodiment of the present invention provides a kind of tube corrosion degree prediction unit, and as shown in Figure 5, described device comprises:

Memory module 510, for pre-stored many group varying environment underground working data and the corresponding actual corrosion rate data of oil pipe, described underground working data comprise the data of multiple influence factors;

MBM 520, for carrying out: S1, choose at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing are carried out to normalization, the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with the Fuzzy Nonlinear Return Law and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, notifies prediction module to carry out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

Prediction module 530, for carrying out: S4, adopt the tube corrosion forecast model obtaining, the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

In the embodiment of the present invention, the underground working data that detect based on reality and the corresponding actual corrosion rate data of oil pipe, by predicting corrosion rate data tube corrosion forecast model as variable by underground working data as dependent variable, oil pipe in conjunction with the Fuzzy Nonlinear Return Law and least square method foundation, and according to this model, oil pipe is carried out to extent of corrosion prediction, thereby can improve the accuracy of oil field well tube corrosion degree prediction.

It should be noted that: the tube corrosion degree prediction unit that above-described embodiment provides is in the time carrying out the prediction of tube corrosion degree, only be illustrated with the division of above-mentioned each functional module, in practical application, can above-mentioned functions be distributed and completed by different functional modules as required, be divided into different functional modules by the inner structure of device, to complete all or part of function described above.In addition, the tube corrosion degree prediction unit that above-described embodiment provides and tube corrosion degree Forecasting Methodology embodiment belong to same design, and its specific implementation process refers to embodiment of the method, repeats no more here.

The invention described above embodiment sequence number, just to describing, does not represent the quality of embodiment.

One of ordinary skill in the art will appreciate that all or part of step that realizes above-described embodiment can complete by hardware, also can carry out the hardware that instruction is relevant by program completes, described program can be stored in a kind of computer-readable recording medium, the above-mentioned storage medium of mentioning can be ROM (read-only memory), disk or CD etc.

The foregoing is only preferred embodiment of the present invention, in order to limit the present invention, within the spirit and principles in the present invention not all, any amendment of doing, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.

Claims (14)

1. a tube corrosion degree Forecasting Methodology, it is characterized in that, pre-stored underground working data and the corresponding actual corrosion rate data of oil pipe in varying environments of organizing in database more, described underground working data comprise the data of multiple influence factors, described method comprises:

S1, chooses at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing is carried out to normalization, and the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with Fuzzy Linear Regression method and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, carries out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

S4, adopts the tube corrosion forecast model obtaining, and the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

2. method according to claim 1, is characterized in that, the group number of described training data sample is not less than the number of described dependent variable.

3. method according to claim 1, is characterized in that, described the data of choosing is carried out to normalization, is specially:

Determine the span z ∈ [z of the data of choosing _min, z _max], and the range scale of definite normalization

wherein z is the data of choosing,

for the data after normalization;

According to

with the described data of choosing, draw the data after normalization.

4. method according to claim 1, is characterized in that, described underground working data comprise CO ₂concentration, Cl ^-concentration, H ₂s concentration, temperature and pressure data.

5. method according to claim 1, it is characterized in that, described combination Fuzzy Linear Regression method and least square method, described training data sample is carried out to modeling, obtain predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe, be specially:

To typical linear regression model y _i=a ₁x _1i+ a ₂x _2i+ ... + a _mx _miuse least square method, and based on training data sample, determine a _j, wherein, x _jithe data of j influence factor in substitution i group training data sample, y _ithe actual corrosion rate data of oil pipe in substitution i group training data sample, j=1,2 ..., m;

According to

with described training data sample and a that determines _j, determine c _j, wherein, H is default confidence value, and 0≤H≤1;

According to the subordinate function of Triangular Fuzzy Number

with a determining _j, c _j, draw Triangular Fuzzy Number A _j(a _j, c _j) value, as Estimates of Fuzzy Linear Regression Model y=A ₁x ₁+ A ₂x ₂+ ... + A _mx _min regression coefficient A _j, using this Estimates of Fuzzy Linear Regression Model as described tube corrosion forecast model.

6. method according to claim 1, is characterized in that, the described check data sample of described use is verified the correctness of described tube corrosion forecast model, is specially:

By tube corrosion forecast model described in the underground working data substitution in described check data sample, obtain oil pipe prediction corrosion rate data, ratio and predetermined threshold value comparison by the difference of the actual corrosion rate data of oil pipe corresponding these underground working data in described check data sample and described oil pipe prediction corrosion rate data with the actual corrosion rate data of this oil pipe, verify the correctness of described tube corrosion forecast model, if this ratio is no more than predetermined threshold value, be verified, otherwise, authentication failed.

7. a tube corrosion degree Forecasting Methodology, it is characterized in that, pre-stored underground working data and the corresponding actual corrosion rate data of oil pipe in varying environments of organizing in database more, described underground working data comprise the data of multiple influence factors, described method comprises:

S1, chooses at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing is carried out to normalization, and the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with the Fuzzy Nonlinear Return Law and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, carries out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

S4, adopts the tube corrosion forecast model obtaining, and the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

8. a tube corrosion degree prediction unit, is characterized in that, described device comprises:

Memory module, for pre-stored many group varying environment underground working data and the corresponding actual corrosion rate data of oil pipe, described underground working data comprise the data of multiple influence factors;

MBM, for carrying out: S1, choose at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing are carried out to normalization, the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with Fuzzy Linear Regression method and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, notifies prediction module to carry out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

Prediction module, for carrying out: S4, adopt the tube corrosion forecast model obtaining, the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

9. device according to claim 8, is characterized in that, the group number of described training data sample is not less than the number of described dependent variable.

10. device according to claim 8, is characterized in that, described MBM, specifically for:

Determine the span z ∈ [z of the data of choosing _min, z _max], and the range scale of definite normalization

wherein z is the data of choosing, for the data after normalization;

According to with the described data of choosing, draw the data after normalization.

11. devices according to claim 8, is characterized in that, described underground working data comprise CO ₂concentration, Cl ^-concentration, H ₂s concentration, temperature and pressure data.

12. devices according to claim 8, is characterized in that, described MBM, specifically for:

To typical linear regression model y _i=a ₁x _1i+ a ₂x _2i+ ... + a _mx _miuse least square method, and based on training data sample, determine a _j, wherein, x _jithe data of j influence factor in substitution i group training data sample, y _ithe actual corrosion rate data of oil pipe in substitution i group training data sample, j=1,2 ..., m;

According to

with described training data sample and a that determines _j, determine c _j, wherein, H is default confidence value, and 0≤H≤1;

According to the subordinate function of Triangular Fuzzy Number with a determining _j, c _j, draw Triangular Fuzzy Number A _j(a _j, c _j) value, as Estimates of Fuzzy Linear Regression Model y=A ₁x ₁+ A ₂x ₂+ ... + A _mx _min regression coefficient A _j, using this Estimates of Fuzzy Linear Regression Model as described tube corrosion forecast model.

13. devices according to claim 8, is characterized in that, described MBM, specifically for:

By tube corrosion forecast model described in the underground working data substitution in described check data sample, obtain oil pipe prediction corrosion rate data, ratio and predetermined threshold value comparison by the difference of the actual corrosion rate data of oil pipe corresponding these underground working data in described check data sample and described oil pipe prediction corrosion rate data with the actual corrosion rate data of this oil pipe, verify the correctness of described tube corrosion forecast model, if this ratio is no more than predetermined threshold value, be verified, otherwise, authentication failed.

14. 1 kinds of tube corrosion degree prediction units, is characterized in that, described device comprises:

Memory module, for pre-stored many group varying environment underground working data and the corresponding actual corrosion rate data of oil pipe, described underground working data comprise the data of multiple influence factors;

MBM, for carrying out: S1, choose at least two group underground working data and the corresponding actual corrosion rate data of oil pipe thereof of storage, and the data of choosing are carried out to normalization, the data after normalization are divided into training data sample and check data sample;

S2, in conjunction with the Fuzzy Nonlinear Return Law and least square method, carries out modeling to described training data sample, obtains predicting the tube corrosion forecast model of corrosion rate data as variable by underground working data as dependent variable, oil pipe;

S3, uses described check data sample to verify the correctness of described tube corrosion forecast model, if upchecked, notifies prediction module to carry out S4, otherwise transposition S1, chooses new Data duplication and carry out flow process;

Prediction module, for carrying out: S4, adopt the tube corrosion forecast model obtaining, the in-service oil pipe under designated environment is carried out to extent of corrosion prediction.

Images